p705-20w

COMPOSITE
CONSTRUCTION
IN STEEL
AND CONCRETE
VIII
Proceedings of the Eighth
International Conference on Composite
Construction in Steel and Concrete
Edited by Gian Andrea Rassati,

Jerome F. Hajjar, and Roberto T. Leon
July 29 – August 2, 2017

SPRING CREEK RANCH
Jackson Hole, Wyoming, U.S.
Wright Runstad & Co.

COMPOSITE
CONSTRUCTION
IN STEEL
AND CONCRETE
VIII
Proceedings of the Eighth
International Conference on Composite
Construction in Steel and Concrete
Edited by Gian Andrea Rassati,

Jerome F. Hajjar, and Roberto T. Leon
July 29 – August 2, 2017

SPRING CREEK RANCH
Jackson Hole, Wyoming, U.S.
© AISC 2020
by
American Institute of Steel Construction
All rights reserved. This book or any part thereof must not be reproduced
in any form without the written permission of the publisher.
The AISC logo is a registered trademark of AISC.
ISBN 978-1-56424-062-0
The information presented in this publication has been prepared following recognized principles of design
and construction. While it is believed to be accurate, this information should not be used or relied upon
for any specific application without competent professional examination and verification of its accuracy,
suitability and applicability by a licensed engineer or architect. The publication of this information is not a
representation or warranty on the part of the American Institute of Steel Construction, its officers, agents,
employees or committee members, or of any other person named herein, that this information is suitable
for any general or particular use, or of freedom from infringement of any patent or patents. All represen-
tations or warranties, express or implied, other than as stated above, are specifically disclaimed. Anyone
making use of the information presented in this publication assumes all liability arising from such use.
Caution must be exercised when relying upon standards and guidelines developed by other bodies and
incorporated by reference herein since such material may be modified or amended from time to time sub-
sequent to the printing of this edition. The American Institute of Steel Construction bears no responsibility
for such material other than to refer to it and incorporate it by reference at the time of the initial publication
of this edition.
Printed in the United States of America
Cover image © Wright Runstad & Co., provided by Magnusson Klemencic Associates.
Abstract: These proceedings summarize the state-of-the-art in composite
construction worldwide, as presented at the eighth international conference
on Composite Construction in Steel and Concrete held at Spring Creek
Ranch (Jackson Hole, Wyoming, USA) from July 29 to August 2, 2017. The
papers contained in this volume cover a wide variety of topics, including
innovative composite structural systems, composite columns, composite
slabs, shear connectors, fire and seismic resistance of composite structural
systems, composite bridges, and practical applications.
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete iii
Preface
These proceedings summarize the state-of-the-art in composite construction worldwide,

as presented at the international conference on Composite Construction in Steel and
Concrete held at Spring Creek Ranch in Jackson Hole, Wyoming (USA) in July 2017. This
is the eighth in a series of conferences on this topic organized by the United Engineering
Foundation (then Engineering Conferences International and now by Continuing and
Professional Education @ Virginia Tech) aimed at assessing and synthesizing the most
recent advances in research and practice in the area of composite steel-concrete
construction. This conference was preceded by those held in Henniker, New Hampshire,
USA (1987), Potosi, Missouri, USA (1992), Irsee, Germany (1996), Banff, Canada (2000),
Kruger National Park, South Africa (2004), Tabernash, Colorado, USA (2008), and Palm
Cove, Australia (2013).
The papers contained in this volume cover a wide variety of topics, including innovative
composite structural systems, composite columns, composite slabs, shear connectors,
fire and seismic resistance of composite structural systems, composite bridges, and
practical applications. Seventy-six participants from seventeen countries participated in
four days of presentations, panel discussions, and dialogue addressing all aspects of
composite construction. The conference was organized and co-chaired by Dr. Roberto
Leon, Virginia Tech, Dr. Jerome F. Hajjar, Northeastern University, and Dr. Gian Andrea
Rassati, University of Cincinnati, U.S.A. The conference was generously supported by
the American Institute of Steel Construction (AISC) and the Virginia Tech Charles E. Via
Department of Civil and Environmental Engineering (VT/CEE).
The papers in the proceedings were peer reviewed as per the guidelines used for the
AISC Engineering Journal. The review process was administered by the editors of the
proceedings, who would like to thank all the authors and reviewers for their contributions.
The publication of the proceedings was supported by the AISC.
The editors will like to thank the staff of Continuing and Professional Education at Virginia
Polytechnic Institute and State University and particularly Ms. Kathy Laskowski, Director
of Facilities/Operations and Major Grants Coordinator at VT/CEE, for all their help in
organizing and coordinating the conference.
Gian Andrea Rassati, Jerome F. Hajjar, and Roberto T. Leon

April 2020
iv Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
Conference Participants
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete v
Organizing Committee
Chairs:
Roberto T. Leon, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA
Jerome F. Hajjar, Northeastern University, Boston, MA, USA
Gian Andrea Rassati, University of Cincinnati, Cincinnati, OH, USA
Scientific Committee
Ulrike Kuhlmann, University of Stuttgart, Germany

K. C. Tsai, NCREE, Taipei, Taiwan
Brian Uy, University of Sydney, Sydney, Australia
Jim Harries, Harris and Associates, Denver, Colorado, USA
Toko Hitaka, University of Kyoto, Kyoto, Japan
Larry Griffis, W.P. Moore and Associates, Austin, Texas, USA
Kent Harries, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
Yan Xiao, Hunan University, Hunan, China
Oreste Bursi, University of Trento, Trento, Italy
Mario Fontana, ETH, Zürich, Switzerland
Wolfgang Kurz, University of Kaiserslautern, Kaiserslautern, Germany
Dennis Lam, University of Leeds, Leeds, United Kingdom
Robert Tremblay, Ecole Polytechnique, Montreal, Canada
vi Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
TABLE OF CONTENTS
Composite Construction Design Codes and Structures
Rainier Square-concrete filled composite plate shear wall high-rise tower, Seattle, 1
WA
Brian Morgen, Hee Jae Yang, Ron Klemencic and John Hooper
G.T. Land Plaza: using composite structure for solving the challenges and 13
complexities of a tall, long and slender tower
Mark Sarkisian, Neville Mathias, John Gordon and Joanna Zhang
Multifunctional commercial buildings in steel and composite construction 25
Richard Stroetmann and Lukas Hüttig
Building scale composite systems 38
Aaron Mazeika and Ashpica Chhabra
Towards the use of high strength steel and concrete in high-rise construction 50
Richard Liew, Yan-Bo Wang and Du Yong
Development of a new push test for Eurocode 4 62
Stephen Hicks, Adrian Ciutina and Christoph Odenbreit
Development of European design guidance for steel-concrete (SC) structures in 74
nuclear power plant
Bassam Burgan and Eleftherios Aggelopoulos
The new Australia/New Zealand Standard on composite steel-concrete buildings, 86
ASNZS2327
Brian Uy, Stephen Hicks, Won Hee Kang, Huu-Tai Thai and Farhad Aslani
The new joint Australian and New Zealand Design Standard for steel and composite 98
bridges AS/NZS 5100.6 - Part 6: Steel And Composite Construction
Stephen Hicks, Brian Uy and Won-Hee Kang
Shear Connectors
Shear connections by headed studs close to the concrete edge 109
Ulrike Kuhlmann, Jochen Raichle and Ana M. Pascual
Considerations on composite beam connection due to plastic deformation capacity 121
and strength ratio
Yuko Shimada and Satoshi Yamada
Behaviour of composite dowels positioned close the surface of concrete slabs 132
Wolfgang Kurz, Yannick Broschart and Joanna Gajda
Pin shear connectors for steel-concrete composite constructions 144
Maik Kopp, Kevin Wolters, Markus Feldmann, Martin Claßen and Johannes Schäfer
Behavior of a new type of shear connectors for U-shaped steel-concrete hybrid 157
beams
Pisey Keo, Clémence Lepourry, Hugues Somja and Franck Palas
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete vii
Shear performance of nonparallel type twin perfobond steel plate shear 169
connectors: FEM analysis and experimental verification
Hai Chen, Yang Liu, Zixiong Guo, Yong Ye and Bahram Shahrooz
Composite Floor, Slab, and Diaphragm Systems

Experimental studies of composite slim floor beams 181
Dennis Lam, Therese Sheehan and Xianghe Dai
Slim-floor construction - deformation and load carrying behaviour 194
Johannes Schorr, Ulrike Kuhlmann and Gunter Hauf
Experimental and numerical study on mechanical behaviour of composite slabs 206
with closed profiled sheeting
Guochang Li, Xiao Zhang, Zhijian Yang and Fengwei Guo
Role of the floor system in the cyclic response of composite steel gravity framing 217
Sean Donahue, Michael Engelhardt, Patricia Clayton, Eric Williamson and Todd Helwig
A design approach for the serviceability limit state of composite steel-concrete 228
slabs
Gianluca Ranzi
Longitudinal shear resistance of ComFlor 210 239
Roland Abspoel and Jan Stark
Behavior of a sustainable composite floor system with deconstructable clamping 251
connectors
Lizhong Wang, Mark Webster and Jerome Hajjar
Preliminary assessment of a composite flooring system for reuse 263
Martin Nijgh, Mareike von Arnim, Marko Pavlović and Milan Veljković
Seismic collectors in composite steel deck diaphragms 274
Robert Fleischman, Anshul Agarwal, Haitham Ayyad, Richard Sause, Jim Ricles and
Chia-Ming Uang
Evaluation of strength and stiffness predictions for steel deck diaphragms with 286
structural concrete fill
Patrick O'Brien, W. Samuel Easterling and Matthew Eatherton
Composite Beams
Review of plastic moment resistance for composite beams 298
Markus Schäfer
Influence of the composite action on the load bearing and deformation behavior of 310
composite beams with and without profiled steel sheeting
Florian Eggert and Ulrike Kuhlmann
Effect of post-welding residual stresses on distortional buckling of steel-concrete 322
composite beams
Marian Gizejowski, Radoslaw Szczerba, Marcin Gajewski and Wioleta Barcewicz
viii Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
Strength of lateral-torsional buckling of a composite steel beam subjected to 334
reverse curvature bending
Satoshi Kitaoka, Ryoichi Kanno, Satoru Hiroshima, Koji Hanya, Keiichi Takada and
Fumihisa Yoshida
Deformation-based reliability concept for composite beams 346
Wolfgang Kurz, Karsten Geißler, Natali Kostadinova and Gregor Korpas
Effects of composite slab on shear strength of steel plate girders 358
Peter Wang and Maria Garlock
Composite Columns and Column Bases

Structural reliability of steel-concrete composite columns and frames 369
Mark Denavit
Calibration of the elastic flexural rigidity from ambient vibration measurements for 381
a building with encased composite columns
Tiziano Perea, Miguel A. Garcia, Manuel Ruiz-Sandoval, Roberto T. Leon, Mark D.
Denavit and Jerome F. Hajjar
Pushing the envelope of composite column design using high-strength materials 392
Amit Varma and Zhichao Lai
Strength of CFT Columns with Various Material Strengths and Sectional 404
Slenderness
Ho-Jun Lee and Hong-Gun Park
Buckling resistance of hybrid steel-concrete columns 416
Pisey Keo, Hugues Somja, Quang-Huy Nguyen and Mohammed Hjiaj
The seismic stability and ductility of steel columns interacting with concrete 428
footings
Hiroyuki Inamasu, Amit Kanvinde and Dimitrios Lignos
Performance based evaluation of a multiple CFT bridge pier with buckling 440
restrained bracings
Chi-Ho Jeon, Dong-Wook Kim and Chang-Su Shim
Seismic performance of an innovative composite column with replaceable steel slit 450
dampers: FEM analysis and design method
Yang Liu, Yifan Liao, Zixiong Guo, Yingting Lu, Xiaojuan Liu and Bahram Shahrooz
Life-cycle performance of concrete-filled steel tubular column subjected to 462
corrosion and impact
Lin-Hai Han and Chuan-Chuan Hou
Research on deformation limits of SRC columns based on experiment data 474
Jing Ji, Lei Zhang, Xiaolei Han, Tao You and Yuan Cao
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete ix
Composite Connections
Update of design provisions for moment-resisting joints between steel beams and 486
reinforced concrete columns
Luis Fargier-Gabaldon, Paul Cordova, Gustavo Parra-Montesinos and Gregory Deierlein
Innovative slip-critical blind bolts and connections for concrete-filled RHS columns 502
Wei Wang, Mingxiao Li, Yiyi Chen and Xiaogang Jian
Braced frame connections with square CFT columns penetrated by gusset plates 513
Peng Wang, Jianrong Pan, Jingmin Liang, Huashen Xie and Zhan Wang
Influence of composite slab on the nonlinear response of extended end-plate 526
beam-to-column steel joints
Roberto Tartaglia, Mario D'Aniello, Gian Andrea Rassati and James Swanson
Numerical investigation of moment-resisting slim-floor beam-to-column 538
connections
Cristian Vulcu, Rafaela Don, Adrian Ciutina and Dan Dubina
Improved beam-column connection for partially encased composite structure 550
Yiyi Chen, Guanghong Chuan and Jie Li
Girder-wall connections using concrete anchors in composite construction 561
Jian Zhao and Bo Yang
Steel-to-concrete joints with large anchor plates under shear loading 574
Jakob Ruopp and Ulrike Kuhlmann
Seismic column base connections: experiments, simulations, component and 586
building models, and design implications
Amit Kanvinde
Moment-rotation characteristics for deconstructable flush end plate composite 597
joints
Mark Bradford, Abdolreza Ataei and Xinpei Liu
Composite Lateral Force Resisting Systems

The influence of a composite slab on the seismic behavior of moment resisting 607
frames
Roberto Tartaglia, Mario D'Aniello, Gian Andrea Rassati and James Swanson
Concrete-filled steel tube dual lateral-force resisting system 619
Johnn Judd and Nipun Pakwan
An introduction to coupled composite core wall systems for high-rise construction 631
Amit Varma, Zhichao Lai and Jungil Seo
In-plane cyclic testing of concrete-filled sandwich steel panel walls with and 643
without boundary elements
Michel Bruneau and Yasser Alzeni
Cyclic shear behavior of steel-plate composite walls for high-rise buildings 655
Xiaodong Ji, Xiaowei Cheng and Xiangfu Jia
x Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
Experimental study on the seismic performance of new hybrid coupling wall 667
subassemblies with replaceable steel coupling beam
Yang Liu, Hao Lin, Zixiong Guo, Hongsong Hu and Bahram Shahrooz
Numerical investigation of composite detachable short links 678
Mariana Zimbru, Mario D'Aniello, Aurel Stratan, Raffaele Landolfo and Dan Dubină
Direct Analysis for structures made of high strength steel 690
Sw Liu, Sl Chan and Tj Li
Fire and Progressive Collapse

Performance of composite beams in fire: pre-test analysis of full-scale experiments 702
Joseph Main, Fahim Sadek, Jonathan Wigand, Jian Jiang, Lisa Choe, Selvarajah
Ramesh, Matthew Hoehler, Mina Seif and John Gross
Towards an engineering method for fire design of concrete-filled steel tube 714
columns with solid steel core
Martin Neuenschwander, Markus Knobloch and Mario Fontana
Fire behavior of blind-bolted connections to concrete filled tubular columns under 726
tension
Ana M. Pascual and Manuel L. Romero
Robustness design and possible collapse mechanisms of building frames in a fire 738
Kei Kimura, Fuminobu Ozaki and Ryoichi Kanno
Progressive collapse analysis of composite steel frames under elevated 750
temperature
Hussam Mahmoud and Chao Qin
Progressive collapse: the case of composite steel-concrete frames 762
Riccardo Zandonini, Nadia Baldassino, Fabio Freddi and Giacomo Roverso
Bridges
Concrete filled steel tubes for bridge applications 775
Charles Roeder and Dawn Lehman
Hot-dip galvanizing in steel and composite bridge constructions 787
Dieter Ungermann, Svenja Holtkamp and Dennis Rademacher
Operational testing of composite railway bridges with innovative composite dowels 800
Daniel Pak, Maik Kopp and Günter Seidl
Integral vs. seismic control in composite deck bridges 812
Uwe E. Dorka and Andrea Klarendic
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete xi
TABLE OF CONTENTS (continued)
Authors’ Index..........................................................................................................................825
Keyword Index.........................................................................................................................827
xii Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
RAINIER SQUARE: CONCRETE FILLED COMPOSITE PLATE SHEAR
WALL HIGH-RISE TOWER, SEATTLE, WA
Brian Morgen
Magnusson Klemencic Associates
Seattle, Washington USA
bmorgen@mka.com
Hee Jae Yang

hyang@mka.com
Ron Klemencic
rklemencic@mka.com
John Hooper
jhooper@mka.com
ABSTRACT
AISC 341-16 includes expansion of Section H Composite Braced-Frame and Shear-Wall

Systems to include Concrete Filled Composite Place Shear Walls (CF-CPSWs). This paper and
presentation focuses on the design of a high-rise building in Seattle that employs a CF-CPSW
coupled core wall lateral force-resisting system.
Recent high-rise office building construction has commonly utilized reinforced concrete core
walls for lateral resistance with composite steel floor framing. This mixture of concrete and steel
trades results in the concrete core construction being on the critical path of the overall project
schedule. The use of a CF-CPSW lateral system offers a potential alternative to concrete core
walls where the steel erector can install the floor framing concurrent with the erection of the dual
plate steel core and allow for the wall concrete to be infilled with the pouring of the composite
decks in a trailing sequence reducing the overall construction schedule. In addition, the steel
dual plate wall modules can be pre-fabricated in a shop and field spliced with the face plates
serving a permanent formwork for the concrete fill.
The Rainier Square project in Seattle is a 58-story, 850-feet tall, 1.4 million square foot mixed-
use office and residential tower with over 750,000 square feet of office space, approximately
200 residential units, 30,000 square feet of retail, and parking for over 1,000 vehicles. The
implementation of the CF-CPSW core was utilized to accelerate the construction schedule by
several months.
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete 1
INTRODUCTION
The 2016 edition of AISC 341 includes a new and expanded specification Section H7 for the
design of Concrete Filled Composite Plate Shear Wall (CF-CPSW) systems (American Institute
of Steel Construction, 2016). Composite plate shear walls are comprised of structural steel face
plates, tie-rods linking the face plates, and concrete infill. Composite action between the plates
and concrete fill can be achieved using either tie bars or a combination of tie bars and shear
studs.
The use of concrete filled composite plate shear walls in construction can provide a cost-
effective alternative to special reinforced concrete shear wall construction and reduce overall
project construction durations of the primary structural frame. Recent high-rise office building
construction has commonly utilized a reinforced concrete core wall for lateral resistance with
composite steel floor framing. This commonly results in the construction time of the concrete
core being on the critical path of the overall project schedule. The use of a CF-CPSW lateral
system offers a potential alternative to concrete core walls where the steel erector can install the
floor framing concurrent with the erection of the composite plate steel core and allow for the wall
concrete to be infilled with the pouring of the composite decks in a trailing sequence reducing
the overall construction schedule. In addition, the steel dual plate wall modules can be pre-
fabricated in a shop and field spliced with the face plates serving a permanent formwork for the
concrete fill.
Research over the past decade on the construction of dual-plate sandwich panel construction
has shown, with proper detailing of the face plate thickness and tie bar spacing, that the CF-
CPSW systems have a high strength and stiffness (similar to or greater than a similarly
proportioned reinforced concrete shear wall) and have a highly ductile seismic performance
(Alzeni & Bruneau, 2014) (Ramesh, Kreger, & Bowman, Behavior and Design of Earthquake-
Resistant Dual-Plate Composite Shear Wall Systems, 2013). Other positive system
characteristics include:
 The steel plate face plates and tie-rods act as permanent formwork for the concrete infill;
 The steel boundary elements and panels can be shop prefabricated and field spliced,
reducing construction time;
 The un-concreted panel assembly has the ability to carry the steel gravity floor framing
self-weight, allowing the shear wall and floor framing to advance several levels ahead of
the concrete infill operation;
 The steel face plates and tie-rods act as confinement for the unreinforced concrete,
increasing the confining pressure and permitting development of the full composite
capacity, with stable seismic energy dissipation;
 The concrete core and tie-rods provide plate stability and limit plate buckling enhancing
ductility of the composite wall system.
TALL BUILDING APPLICATION

The Rainier Square project in Seattle, Washington USA is a 58-story, 850-feet tall, 1.4 million
square foot mixed-use office and residential tower with over 750,000 square feet of office space,
approximately 200 residential units, 30,000 square feet of retail, and parking for over 1,000
vehicles. At projected completion in 2020, the project is slated to be Seattle’s second tallest
building.
The structural design was performed in accordance with the 2012 International Building Code
and City of Seattle Amendments, utilizing a non-prescriptive seismic design methodology
(Pacific Earthquake Engineering Research Center, Tall Building Initiative (PEER TBI), 2010)
2 Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
since selected code provisions are taken as exceptions, including allowable building height for
the proposed seismic lateral force-resisting system. A performance-based seismic design
procedure as allowed by the IBC and ASCE 7 is used to provide a level of safety and overall
building ductility equivalent to or greater than that of a prescriptively-designed building.
As illustrated in Figure 1, the building lateral resistance to wind and seismic lateral demands is
provided by a coupled shear wall core placed around the central elevator and stair core of the
tower supplemented by mid-height Buckling-Restrained Brace (BRB) outriggers and belt trusses
to enhance the overall stability and stiffness of the tower.
(a) (b) (c)

Fig 1. – Rainier Square Tower: (a) Architectural Rendering (courtesy Wright Runstad & Company), and
(b) & (c) ETABS computer model graphics
To demonstrate that the design is capable of providing code-equivalent seismic performance, a
three step analysis and design procedure was performed. The first phase of design is the
preliminary design phase where structural components are designed for code-level demands
(wind and seismic). For structural components that are anticipated to and detailed to yield based
on code-level seismic demands, they are proportioned to satisfy the minimum strength
requirements of the IBC. For structural components that are designated to remain elastic, an
initial design is performed based on estimated amplification factors to account for strength
demands coming from the maximum considered earthquake (MCE) analysis.
The second phase of the building design is a serviceability analysis for a service-level
earthquake (SLE) where evaluation of the primary system responses characteristics (such as
story drift) and structural components (such as coupling beams, shear walls, and buckling-
restrained braces) to demonstrate that the structure is serviceable when subjected to a
response spectrum corresponding to a 43-year return period. The final phase of the design
includes a nonlinear response history analysis verification of three-dimensional nonlinear
computer models considering the MCE event. As part of the MCE verification, acceptance
criteria are selected and established for Collapse Prevention performance under the MCE-level
demands.
Given the unique nature of the proposed lateral system and the limited use in building
construction, an independent peer review panel of experts in this building construction research
was employed by the Seattle Department of Construction and Inspections (SDCI) to assist in
the building permit review.
CF-CPSW CORE WALL DETAILING

Early schematic designs for the project included the central core being constructed of reinforced
concrete. Based on early pricing and schedule studies of the mid-height outrigger and belt truss
connections to the core, an alternate core construction utilizing composite plate shear walls was
introduced. Based on the construction schedule savings of several months, the project moved
forward with the CF-CPSW core construction system for the tower (see Figure 2).
Fig. 2 – CF-CPSW Multi-Cell Core Wall
The CF-CPSW wall system consists of a three-cell core at the building base which tapers to a
two-cell at the top of the office programming. Above the outrigger and belt-truss levels, a single
cell core extends through the residential programming to the roof. Composite wall thicknesses
range from 45” at the base to 21” thick at the top of the building. Face plates are 1/2-inch
minimum with more highly stressed regions as much as 3/4-inch thick. Tie-rods are specified as
1-inch diameter F1554 Grade 55 anchor rods and are spaced at 12-inch on center both
horizontal and vertical resulting in a tie-rod spacing-to-steel plate thickness ratio ranging from 16
to 24 and a tie reinforcing ratio of 0.55%. A ratio of 24 is less that the maximum recommended
tie-rod spacing-to-steel plate ratio limit of 30 as observed to match the high-ductility performing
test specimens from (Eom, Park, Lee, Kim, & Chang, 2009).
Also, the tie spacing of 12-inches each way is conservatively below the maximum limit for tie-
rod spacing-to-thickness ratio specified by AISC 341-16 equation (H7-1), with 𝐸 = 29000 𝑘𝑠𝑖,
𝐹𝑦 = 50 𝑘𝑠𝑖, and 𝑡 = 0.5 𝑖𝑛𝑐ℎ𝑒𝑠, which results in the following:
𝐸
𝑤1 = 1.8𝑡√ = 21.7 𝑖𝑛𝑐ℎ𝑒𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚
𝐹𝑦
The 1-inch diameter ASTM F1554 Grade 55 tie-rods meet the requirements of AISC 341-16
equations (H7-4), (H7-5), and (H7-6) and have welded connections to the face plates which
develop the full tension strength of the tie-rod.
As illustrated in Figure 3, the core wall corners, wall intersections, and wall ends are detailed as
tie-rod confined boundary elements to limit the onset of plate buckling and provide concrete
confinement. Research on concrete confined by rectangular steel tubes/plates with tie-rods
indicates increased ultimate strength and ductility of the section similar to that of added cross-tie
and confinement reinforcement in reinforced concrete detailing (Long & Cai, 2013).
(a) (b)
Fig. 3 – CF-CPSW Boundary Element Detailing: (a) Core Wall Corner Intersection, (b) Wall End
Much of the existing research on CF-CPSW systems to date has utilized planar wall systems
with and without boundary elements at the wall ends. To enable greater building lateral strength
and stiffness or to extend the system to greater building heights, planar walls are often needed
to be extended to flanged walls and/or core wall configurations where the introduction of
doorways into and out of the core for building services are required (stairs, elevator lobbies,
back of house, etc.). To maintain stiffness of the core, introduction of composite link beams over
the doorways are utilized to couple the wall panels and increase the overall core stiffness,
strength, and ductility.
Experimental research programs utilizing coupled wall panels with composite coupling beams
indicate ductile nonlinear performance (Eom, Park, Lee, Kim, & Chang, 2009). Plastic stress
distribution methods for calculating plastic moment capacity of the composite steel and concrete
cross section were found to be in good agreement with the load-carrying capacities of the tested
coupled wall specimens. The test results showed that CF-CPSW systems had excellent load-
carrying capacities, but in order to ensure ductile behavior of the system, detailing to prevent
early fracture of the welded connections at the coupling beam-to-wall interface is critical.
Subsequent experimental studies of concrete-filled composite plate coupling beams had similar
findings where the fracture of the web and flange plates at the interface of the coupling beam-to-
wall pier was identified as the critical limit state impacting ductility (Hu, Nie, & Eatherton, 2014),
(Nie, Hu, & Eatherton, 2014). To avoid welds at the critical demand interface at the coupling
beam-to-wall interface, the Rainier Square project utilizes detailing for the concrete-filled
composite plate coupling beams as illustrated in Figure 4.
The coupling beam web plates are lapped to the sides of the wall panel with the welded
connection away from the interface and location of maximum moment demand. The top and
bottom flange plates extend horizontally into the wall panel are welded to the inside face of side
plates to develop the capacity of top and bottom plates into the adjacent composite wall
elements.
(a) (b)
Fig. 4 – Composite Coupling Beam: (a) Elevation of Pre-Fabricated Coupling Beam Wall Panel, (b)
Coupling Beam Section
CODE-LEVEL ELASTIC STRENGTH DESIGN

The linear elastic analysis and design of the lateral force-resisting system subjected to the
design earthquake (DE) and wind loading is the first stage of the three-stage design and
verification process. The initial design is governed by the envelope of strength demands and
performance criteria for the DE and design wind loads. This step ensures that a minimum
strength for the building similar to that of traditional code-level demands is provided. The overall
objectives for the DE-level design are: (a) to proportion the members of the lateral system to
provide sufficient stiffness and strength to resist the DE demands, (b) to provide an initial design
for the core wall face plates, (c) to provide an initial design for the core wall coupling beams,
and (d) to provide an initial design for the buckling-restrained braces (BRBs) as part of the core
wall outrigger system. This design stage considers an elastic response spectrum analysis for
the DE-level event.
Given the height of the structure, a linear elastic design for ultimate wind demands ensures that
sufficient lateral system strength is provided to withstand the recommended wind demands as
provided by a site-specific wind tunnel testing study. The proportioning for the lateral force-
resisting elements is provided to satisfy the strength design requirements for the enveloped
strength demands from the DE and wind analyses. A comparison of the building shear and
overturning demands from the DE and wind demands is illustrated in Figure 5. As can be
observed, the wind demands generally control the proportioning of the lateral force-resisting
system with the overturning moment demand due to code wind strength demands being twice
that of the code seismic demands. Shear demands of the core generally follow this trend,
except for noted locations at lower floor levels where the DE demands exceed the accumulated
wind shear demand.
This observation, where wind demands are much larger than the DE demands and control the
proportioning of the core wall face plates, coupling beams, and BRBs, suggests that the building
lateral system and CF-CPSW core will likely exhibit limited yielding when subjected to MCE
demands. The nonlinear verification analyses confirmed this early design proportioning
observation as discussed later in this paper.
(a) (b)
Fig. 5 – Code-Level Building Demands: (a) Building Shear, (b) Building Overturning Moment
The linear elastic design for service-level earthquake (SLE) is the second stage of the three-
stage design and verification process of the lateral system. This step provides a second method
of providing a minimum strength for the lateral system. Acceptance criteria are selected such
that essentially elastic behavior of the structure is anticipated. The recommendations from the
PEER TBI (Pacific Earthquake Engineering Research Center, Tall Building Initiative (PEER
TBI), 2010) are utilized as a baseline for the design. The overall objectives for the SLE design
are: (a) to provide a second method for proportioning the members of the lateral system such
that sufficient strength and stiffness are provided and that the building is expected to behave in
an essentially elastic manner under a more frequent earthquake and (b) to verify that the initial
design for the lateral force-resisting system performed at the DE/wind analysis is satisfactory for
SLE demand levels.
NONLINEAR VERIFICATION
To achieve the objectives of the verification and design phases, a nonlinear computer model of
the structure is built where member sizes, material, and elastic and inelastic behaviors are
modeled as accurately as practical. The model is subjected to ground shaking, and the results
are analyzed. The nonlinear verification phase is utilized to illustrate that that lateral force-
resisting system meets at a minimum a Collapse Prevention performance objective. The
nonlinear analysis results are compared to a set of predefined acceptance criteria that are
selected such that nonlinear behaviors remain with limits established by testing or limits typically
deemed acceptable by the structural engineering profession.
The nonlinear verification model is constructed utilizing CSI Perform-3D (Computers and
Structures, Inc., 2016). The model includes inelastic member properties for ductile-detailed
elements that are anticipated to be loaded beyond their elastic limits under MCE demands.
These include the core wall axial/flexural response behavior, coupling beam rotational
demands, and outrigger BRB bracing element axial yielding. CF-CPSW shear resistance,
diaphragm slabs, and outrigger columns are desired to remain elastic under MCE demands,
and are designed and detailed as a force-controlled element action and are modeled with
effective stiffness elastic properties.
Inelastic vertical fiber elements are used for composite core walls to capture nonlinearity of
moment and axial load interaction. The fiber element is comprised of both concrete and steel
fibers. Coupling beams are modeled as nonlinear shear hinges with element stiffness, yield, and
degradation characteristics that are matched to composite coupling beam experimental testing
programs (Nie, Hu, & Eatherton, 2014). The following acceptance criteria as illustrated in Table
1 were selected for a Collapse Prevention performance objective under the MCE-level demands
for a Risk Category III structure, which includes a 0.8 adjustment factor to adjust from common
Risk Category II acceptance criteria.
Response Parameter Acceptance Criteria Value
Story Drift 3.0 x 0.8 = 2.4 percent taken as the average of 11 analyses;
4.5 x 0.8 = 3.6 percent maximum from any single analysis.
Residual Story Drift 1.0 x 0.8 = 0.8 percent taken as the average of 11 analyses;
1.5 x 0.8 = 1.2 percent maximum from any single analysis.
Coupling Beam Rotation 0.025 x 0.8 = 0.020 radian rotation limit, taken as the average of 11 analyses.
CF-CPSW Steel Plate Axial 0.016 x 0.8 = 0.013 in tension, and matching concrete compression limit, taken as the
Strain average of 11 analyses.
CF-CPSW concrete Based on confinement model by (Long & Cai, 2013), the strain limit is set as the strain
Compression Strain at 80% of the peak stress on the descending branch of the confinement curve. Confined
concrete strain limit of 0.0040 x 0.8 = 0.0032, taken as the average of 11 analyses.
BRB Deformation 2.5% x 0.8 = 2.0% compression and tension deformation, taken as the average of 11
analyses.
Table 1 – MCE Acceptance Criteria for Deformation-Controlled Actions
The nonlinear verification model was subjected to two suites of 11 pairs of site specific ground
motions determined using the conditional mean spectrum approach. One suite characterizes
relatively short period motion (corresponding to the primary 2nd/3rd mode of vibration), and the
other suite characterizes long period motion (corresponding to the primary 1st mode of
vibration).
As was observed in the detailed elastic design phase, the Rainier Square lateral system
proportioning was governed by wind design considerations, with the wind elastic strength
demands double or more of that of the design earthquake demands. The proportioning results
in an inherent seismic overstrength beyond designing solely for code-level earthquake strength
demands and suggested that limited system yielding might be observed in the nonlinear MCE
analyses. Figure 6 plots the average maximum tension and compression strains recorded in
the nonlinear responses at monitored locations at each wall end and wall intersection along the
height of the tower. The peak measured CF-CPSW face plate tension strain was approximately
0.003 in/in which is more than a factor of four below the target acceptance criteria (0.013 in/in)
and only about 1.5 times plate yield strain, which indicates an essentially elastic behavior of the
core face plates in flexure-induced tension demands.
The peak measured CF-CPSW concrete compression strains were approximately 0.001 in/in
which is observed to be well below the target acceptance limit of 0.0032 in/in. Peak concrete
compressive strain of 0.001 in/in is within the elastic range of stress-strain behavior for both a
confined concrete and an unconfined concrete condition, indicating an elastic response
behavior of the composite core wall in compression.
(a) (b)
Fig. 6 – CF-CPSW Maximum Strains: (a) Tension Strains, (b) Compression Strains
Other building response parameters also indicated limited nonlinear response for the lateral
force-resisting system under the MCE demands. As illustrated in Figure 7(a), the average
maximum inter-story drift of the building below the outrigger and belt truss level falls within
similar magnitude ranges. A similar trend for the drifts above the outrigger and belt truss level is
observed. The measured range of building floor plate corner drifts below the outrigger level are
approximately bounded by 0.25% to 0.75% whereas the measured range above the outrigger
level are approximately bound by 0.75% to 1.50% inter-story drift. For building response in the
short core direction, the outrigger level provides a stiff restraint limiting core and building
demands below the outrigger level. For building response in the long core direction, above the
outrigger level the core wall drops from a two-cell core to a single-cell core to the roof.
The single core above the outrigger level is less stiff than the lateral system of the building
below the outrigger level resulting in larger inter-story drifts in the building residential levels, but
still well below the targeted acceptance criteria of 2.4% under the MCE ground motion records.
(a) (b)
Fig. 7 – Representative Building Responses: (a) Maximum Building Inter-Story Drift, (b) Coupling Beam
Rotation
Figure 7(b) provides a representative coupling beam inelastic rotational response for a single
coupling beam bay up the height of the tower. It is observed that very limited rotational
demands are present in the core coupling beams below the outrigger level. This is supported
by the observation in Figure 7(a) with limited building drift demands in the lower portions of the
building. Above the outrigger level, coupling beam rotations are more pronounced, with
average maximum rotations approaching 1.5%, but falling well below the target acceptance
criteria of 2.0%.
Overall, as was suspected during the elastic design phase, the building response under MCE
demands undergoes very little inelastic deformation demands and responds in an ‘essentially
elastic’ response.
For building response parameters desired to remain elastic under MCE demands, they are
designed and detailed as a force-controlled element action as described in PEER TBI, with
appropriate response parameter criticality definition and strength reduction factors.
Specific to the CF-CPSW system, the design shear strength 𝜙𝑉𝑛𝑖 is determined utilizing AISC
341-16 equation (H7-7) as follows:
𝜙𝑉𝑛𝑖 = 𝜅𝐹𝑦 𝐴𝑠𝑤
where 𝜙 = 0.90, 𝐹𝑦 is the specified minimum yield stress of the face plates, and 𝐴𝑠𝑤 is the area
of the steel face plates. Given the configuration of concrete fill thickness and area of steel face
plates utilized on the Rainier Square project, the 𝜅-factor as defined by AISC 341-16 equations
(H7-8) and (H7-9) is nearly unity, with range of 0.97 to 1.0 depending on wall configuration.
Figure 8 plots the average maximum shear wall panel stresses for two representative shear wall
panels along the building height. Both average demands and average demands amplified by a
criticality factor of 1.5 are presented. The allowable shear stress of approximately 38 ksi is
indicated as being above the amplified shear wall demands with face plate thickness of 1/2-inch
nominal and more highly stressed regions near the building base and the outrigger levels at 3/4-
inch. The allowable shear stress is calculated by taking the allowable design shear strength
𝜙𝑉𝑛𝑖 and multiplying by a factor of 0.8 to account for the adjustment from Risk Category II to
Risk Category III acceptance criteria.
Overall, the MCE nonlinear analysis verified the preliminary elastic SLE, DE, and Wind design,
indicating the lateral force-resisting system meets at a minimum a Collapse Prevention
performance objective and provides a level of safety and overall building ductility equivalent to
or greater than that of a code prescriptive designed building.
Fig. 8 – Representative CF-CPSW Normalized Wall Shear Stresses
SUMMARY
The Rainier Square project in Seattle is a 58-story, 850-feet tall mixed-use office and residential
tower that utilizes a concrete filled composite plate shear wall (CF-CPSW) system for the central
core component of an outriggered core lateral force-resisting system. The use of prefabricated
steel plate panels allows for construction schedule savings when compared to a more traditional
reinforced concrete core wall system. A detailed design and nonlinear verification analyses of
the proposed Rainier Square lateral force-resisting system suggests that the lateral system
once designed to remain elastic under the governing ultimate wind demands, will likely have
limited inelastic demands under the MCE earthquake.
REFERENCES
Alzeni, Y., & Bruneau, M. (2014). Cyclic Inelastic Behavior of Concrete Filled Sandwich Panel Walls
Subjected to In-Plane Flexure. MCEER-14-0009.
American Institute of Steel Construction. (2016). Seismic Provisions for Structural Steel Buildings
(ANSI/AISC 341-16). Chicago, IL USA.
Eom, T.-S., Park, H.-G., Lee, C.-H., Kim, J.-H., & Chang, I.-H. (2009). Behavior of Double Skin
Composite Wall Subjected to In-Plane Cyclic Loading. Journal of Structural Engineering, 135,
1239-1249.
Hu, H.-S., Nie, J., & Eatherton, M. R. (2014). Internal Force and Deformation of Concrete-Filled Steel
Plate Composite Coupling Beams. Journal of Constructional Steel Research, 92, 150-163.
Long, Y.-L., & Cai, J. (2013). Stress-Strain Relationship of Concrete Confined by Rectangular Steel Tubes
with Binding Bars. Journal of Constructional Steel Research, 88, 1-14.
Nie, J.-G., Hu, H.-S., & Eatherton, M. R. (2014). Concrete Filled Steel Plate Composite Coupling Beams:
Experimental Study. Journal of Constructional Steel Research, 94, 49-63.
Pacific Earthquake Engineering Research Center, Tall Building Initiative (PEER TBI). (2010). Guidelines for
Performance-Based Seismic Design of Tall Buildings, Version 1.0.
Ramesh, S., Kreger, M. E., & Bowman, M. D. (2013). Behavior and Design of Earthquake-Resistant Dual-
Plate Composite Shear Wall Systems. West Layfayette, Indiana: School of Civil Engineering,
Purdue University.
Razvi, S., & Maatcioglu, M. (1999). Confinement Model for High-Strength Concrete. Journal of Structural
Engineering, 281-289.
G.T. LAND PLAZA: USING A COMPOSITE STRUCTURE TO
ADDRESS THE CHALLENGES OF A COMPLEX SLENDER TOWER
Mark Sarkisian, PE, SE, LEED
Partner, Skidmore, Owings & Merrill LLP
San Francisco, California
mark.sarkisian@som.com
Neville Mathias, PE, SE, LEED AP
Associate Director, Skidmore, Owings & Merrill LLP
neville.mathias@som.com
John Gordon, PE, SE, CEng MIStructE
Associate Director, Skidmore, Owings & Merrill LLP
john.gordon@som.com
Joanna Zhang, PE, SE, LEED AP
Associate, Skidmore, Owings & Merrill LLP
joanna.zhang@som.com
ABSTRACT
The G.T. Land Plaza project occupies a 37 hectare site within the new Central Business District
of Hangzhou in China. The mixed-use development comprises three independent building
structures that extend over much of the site, including a 43-story office and hotel tower (Tower A),
a 26-story residential tower (Tower B), and a retail podium building up to seven stories high
(Building C).
The architectural massing of Tower A defines a narrow rectangular bar building, 165 m by 33 m
in plan and 200 m tall to the top of the parapet. Various architectural features vary the regularity
of the building geometry, enhance the spatial experience for different building programs, and
present particular challenges in structural design. While the structure is predominantly of
reinforced concrete, structural steel elements and composite elements are integrated where
appropriate to deal with certain special features.
In addition to the typical static gravity and lateral analyses, response spectrum seismic analysis,
non-linear time-history seismic analysis, non-linear push-over seismic analysis, wind tunnel study,
vibration analysis, staged-construction analysis and thermal analysis were performed on the
building. To satisfy the Chinese regulations, the project underwent and passed a seismic Expert
Panel Review. Through a rigorous approach to the structural design, close collaboration between
design disciplines, and the strategic use of concrete, steel and composite structural components,
it was possible to provide a coordinated structural solution that satisfies both the target
performance goals for the overall structure and enhanced performance goals for critical elements.
Figure 1: Architectural rendering
INTRODUCTION
The G.T. Land Plaza occupies a 37 hectare site within the new Central Business District of
Hangzhou, located 5 km east of the city’s legendary West Lake and 700 m from the north bank
of the Qian Tang River. The site is flat, generally square in plan, and is occupied by a canal along
the north-east (project east) boundary. The mixed-use development comprises three independent
building structures that extend over much of the site, that are separated from each other by
seismic movement joints above grade. Tower A is a 43-story, 200 m tall building that will be the
focus of discussion in this paper. Tower A accommodates a hotel, office spaces and serviced
apartments, and provides retail spaces in the first five levels. Tower A is rectilinear in plan, with
its broad face aligned with the canal and the north-east (project east) boundary of the site (Figure
2). Adjacent to it, fronting the north-west (project north) boundary, Tower B is a 26-story, 110 m
tall, rectilinear building, accommodating retail in the first five levels above grade and serviced
apartments above. Occupying much of the remainder of the site, to the south of the two towers,
Building C is a retail podium building, up to seven stories tall. The roof over the center of Building
C supports a garden with an extensive water feature, symbolically evoking the West Lake as the
defining geographical feature of the city of Hangzhou.
Figure 2. Plan of the project at level 6, showing the three buildings

STRUCTURAL SYSTEM
The architectural massing of Tower A defines a long rectangular bar building, 165 m by 33 m in
plan and 200 m tall to the top of the parapet. Various architectural features vary the regularity of
this shape, and these will be described. The structure is predominantly of reinforced concrete,
with structural steel elements and composite elements integrated where necessary to deal with
certain special features.
Resistance to lateral wind and seismic loads is provided through a combination of reinforced
concrete core shear walls and reinforced concrete moment resisting frames. Reinforced concrete
shear walls are typically located within three distinct core areas surrounding the mechanical
rooms, stairwells and elevators. All major building columns participate in the moment resisting
frames, and these are typically located at 9.0 m on center along the broad face, and at 7.25 m /
9.50 m / 7.25 m on center in the narrow direction. The outermost lines of columns in this direction
are centered 24 m apart, and these define the width of the lateral system. Only a portion of the
core at the north end of the building extends beyond this zone. To limit column sizes as well as
to provide good ductility, composite steel reinforced concrete columns are provided in the lower
levels at certain locations.
The gravity system typically comprises reinforced concrete beams at 3 m on center, supporting
one-way spanning reinforced concrete floor slabs. Long span cantilever beams support the floor
slabs beyond the perimeter column lines. The reinforced concrete core shear walls and moment
resisting frames also resist gravity load.
Below grade, all three independent buildings share a common three-story deep basement that
extends over the entire site. The buildings are founded on a pile-supported reinforced concrete
mat foundation that varies in thickness. Delayed pour settlement joints are incorporated into the
basement and foundation structures, to accommodate the planned construction sequencing of
the individual buildings and the varying settlements that are expected to occur. The 3.0 m thick
mat under Tower A is supported by 840 no. 1000 mm diameter base-grouted bored piles. 690 of
these piles extend approximately 40 m below grade into a sand and gravel layer, while 150 more
heavily loaded piles under the north end of the tower extend approximately 65 m below grade into
bedrock.
SPECIAL FEATURES
Several major and complex architectural features exist that present particular challenges in the
structural design of Tower A:
Vertical slots. Two 13.5 m wide, multi-story tall vertical slots penetrate the building on its broad
face (Figure 3). These slots serve to subdivide the program elements of the mixed-use building
and allow more natural light to enter these spaces. In general, the program within the building is
subdivided vertically, with the serviced apartments, office and hotel occupying the northern,
central and southern portions respectively. The floors up to level 5 are largely occupied by retail.
Figure 3. Sections through Tower A
North slot with transfer trusses. The first of the two slots is located in the northern half of the
building, between level 1 and the underside of level 23, a slot height of 96.4 m. The location of
the slot provides a natural separation between the serviced apartments and the office space.
Provision of this slot involves eliminating a single line of columns below level 23, necessitating a
transfer system to support the columns above.
The single line of four columns is transferred by a series of two-story high, built-up steel transfer
trusses, located between levels 23 and 25, at the bottom of the column stacks (Figure 4). The
floors at these levels accommodate mostly mechanical space, and the higher floor-to-floor heights
provide a well-proportioned transfer truss system that is 12.3 m high and spans 18.0 m between
supports.
The primary transfer system is a simple A-frame, with two built-up steel box diagonals in
compression and a series of steel plates resisting the tension in the bottom chord. A secondary
transfer system is provided for redundancy. This takes the form of an inverted A-frame,
superimposed over the primary system, with built-up steel H-shape diagonals supporting the
tension forces induced by the column transfer. Compression forces in the top chord are resisted
by built-up steel H-shapes embedded within and acting compositely with the reinforced concrete
floor framing.
Figure 4. Construction photo of the transfer truss between levels 23 and 25

Four transfer trusses are provided to support the four columns. The innermost two trusses are
supported directly by steel columns embedded within the reinforced concrete walls of the north
core and central core. The steel columns extend one story above and below the height of the
transfer trusses. In addition, the bottom chord of each truss is extended one bay further on each
side through the use of steel shapes embedded within the core walls at level 23. The outermost
two trusses are supported by steel columns embedded within the reinforced concrete building
columns. As with the inner trusses, the steel columns extend one story above and below the
height of the trusses. Additionally, all members of the outer trusses extend one bay further each
way, with a single tension diagonal in the end bays.
At levels 23 and 24, within the 18.0 m wide zone defined by the clear span of the trusses, the floor
framing comprises steel beams supporting one-way spanning reinforced concrete floor slabs. At
level 25 and all floors above, the floor framing reverts to reinforced concrete beams and one-way
spanning slabs.
South slot with coupling trusses. The second of the two slots is located in the southern half of
the building, between level 25 and the underside of level 43, a height of 74.8 m, and separates
the office space from the hotel guest rooms. Two adjacent lines of columns are eliminated to
create this slot, which although only 13.5 m wide on the east face, widens to 23.4 m on the west
face.
A single 7.8 m tall story (level 43) occurs above this slot, above which is the main roof of the tower.
The floor space in this area is occupied by hotel program, mostly restaurants, kitchens, and other
back-of-house elements. Four single-story (7.8 m) high built-up steel trusses, spanning 27.0 m
between supports, are employed to support the floor framing at level 43 and roof level (Figure 5).
These act as a coupling structure, connecting the building on each side of the slot.
Figure 5. Coupling truss between level 43 and roof level

The trusses adopt a configuration similar to a Warren truss, typically with 4 bays of 6.75 m each.
The two inner trusses are supported by steel columns embedded within the reinforced concrete
columns and walls of the central core and south core. The two outer trusses are similar, but are
supported by steel columns embedded within the reinforced concrete building columns. These
trusses also extend one 9 m wide bay further each way. The steel columns supporting all trusses
extend down one story below the underside of the truss, to level 42. At level 43 and roof level,
within the 27.0 m wide zone defined by the clear span of the trusses, the floor framing comprises
steel beams supporting one-way spanning reinforced concrete floor slabs.
Long-span floors. To accommodate large open spaces within the hotel entrance lobby at level
1 and the two-story tall hotel reception area at level 6, a single line of columns is eliminated
between level 1 and the underside of level 8. All floors between these two levels are required to
span 18.0 m, with the deeper floor framing being accommodated within the higher floor-to-floor
heights at the lower levels of the tower. Above level 8, however, the lesser floor-to-floor heights
determine that these columns remain, requiring a further transfer system. Fortuitously, the
columns to be transferred are located under the building slot that separates the office space from
the hotel guest rooms in the upper levels of the tower, and as such these columns extend no
higher than level 25.
The primary transfer system takes the form of four single-story tall, built-up steel transfer trusses,
located between levels 8 and 9. This 4.5 m high story is occupied mainly by mechanical rooms.
The configuration of diagonals within the trusses closely resembles a Warren truss, and the
relatively shallow truss depths combined with the need to support seventeen floor levels
determine that the sizes of the diagonal and chord members are substantial. The two inner trusses
are supported by steel columns embedded within the reinforced concrete building columns. The
two outer trusses are similar, but extend one 9 m wide bay further each way. The steel columns
supporting all trusses extend one story above and below the height of the trusses.
Figure 6. Construction photos of the transfer truss between levels 8 and 9
A secondary transfer system is provided for redundancy. This system is a near identical
arrangement of transfer trusses located within the 4.5 m high mechanical floor near the top of the
transferred column stack, between levels 23 and 24. As such, the secondary system supports the
transferred columns as hangers, and steel column shapes encased within the reinforced concrete
columns are provided to accommodate the tensile forces of a hanging system. The steel column
shapes connect the primary and secondary transfer systems at levels 9 and 23 respectively.
At levels 8, 9, 23 and 24, within the 18.0 m wide zone defined by the clear span of the two truss
systems, the floor framing comprises steel beams supporting one-way spanning reinforced
concrete floor slabs. At floor levels in-between, the floor framing reverts to reinforced concrete
beams and one-way spanning slabs.
Cantilever trusses at hotel guest rooms. The architectural design extends the building beyond
the typical 33 m width, at two bays to the west of the south core, between levels 25 and 42. This
results in floors that extend approximately 10 m into column free space beyond the face of the
core. The floors are to accommodate hotel guest rooms that face west, with blank walls to the
north and south.
The original structural concept was to provide a series of vertical hanging steel truss frames,
aligned with the supporting walls and columns of the south core, to form the support structure on
the north and south sides of this space. To incorporate redundancy into the system, this
developed into X-braced hanging steel truss frames with the diagonal members resisting tension
or compression (Figure 7). The area within the space is column free. Floor framing spanning 18
m between trusses comprises steel beams supporting a composite metal deck floor slab. The
hanging trusses are supported by steel columns embedded within the wall and column elements
of the south core.
Figure 7. Cantilever trusses between levels 25 and 42
Long cantilever floor. Between levels 23 and 25, the building width extends to the west to create
a long, two-story high shelf. This extension occurs along much of the west elevation, between the
north and south cores. The depth of the extension matches that at the south core above level 25.
The extended space at level 24 is intended to accommodate hotel amenities and conference
rooms, whilst level 23 accommodates MEP equipment rooms.
Within the northern half of this extended space, the structural system typically incorporates a
series of 2.0 m deep reinforced concrete beams at level 24 that cantilever 10.5 m from each
column line. The 7.25 m backspan beam is similarly 2.0 m deep. These beams support
conventional reinforced concrete floor framing at level 24. Steel posts, located at the tips of the
cantilever beams, support steel framing with a composite metal deck slab at level 25. Similarly,
steel hangers supported at the underside of the cantilever beams support steel framing with a
composite metal deck slab at level 23.
At the north end, the extended space interfaces with the transfer trusses that span across the
north slot (described previously). At this location, the cantilever and backspan beams that would
have been supported by the transferred column line are replaced by a 4.5 m high, single-story
cantilever and backspan truss between levels 23 and 24. This is in turn supported by two of the
transfer trusses.
Within the southern half of the extended space, the reinforced concrete cantilever beams support
a swimming pool at level 24 (Figure 8). The structural system is similar to that in the northern half,
although the lower beams dictated by the pool’s depth eliminates level 23 below. In addition, the
cantilever and backspan beams taper at their common support to engage the main floor at level
23 and eliminate the short column.
Figure 8. Long cantilever beam supporting swimming pool at level 24

Similar to the structure at the north end, one of the cantilever and backspan beams supporting
the swimming pool are supported by the transfer trusses that support the long-span floors
(described previously).
Long cantilever trusses at hotel spa. At its southern end, the width of tower narrows from 33
m typically to only 13.5 m. This occurs over a 29 m length of the building beyond the south core.
Two lines of columns in the narrow direction are spaced only 6 m apart. A single east-west shear
wall replaces one of the columns in this area, to help mitigate the effects of lateral and torsional
loads.
Typically, this area of the tower is occupied by hotel guest rooms. The 7.8 m tall story at level 24,
however, accommodates hotel amenities, some of which have already been described. At the
southern end, between levels 24 and 25 only, the tower width extends dramatically from 13.5 m
to 33 m to match the typical tower width elsewhere. This extended area accommodates the hotel
spa treatment rooms.
The structural system in this area provides two single-story tall built-up steel trusses between
levels 24 and 25 that cantilever to the west of the main building structure (Figure 9). The first of
these trusses cantilevers 22.2 m from the nearer support column, and engages the second
support column 6 m behind it. The second truss cantilevers 19.5 m from the edge of the 6 m wide
shear wall which it fully engages with steel truss members embedded within. Both steel trusses
incorporate cross-braced panels to ensure redundancy, with vertical members at truss nodes.
Steel column shapes embedded within the supporting columns and walls extend one story above
and below the truss depths.
Figure 9. Long cantilever truss supporting hotel spa at level 24

In the area supported by the trusses at levels 24 and 25, floor framing comprises steel beams
supporting composite metal deck floor slabs. Steel cross-bracing on plan is provided to ensure
that the diaphragm forces are adequately transferred to the main structure and the lateral resisting
elements.
SPECIAL ANALYSES
The structural design was performed in accordance with the requirements of the relevant Chinese
codes and standards, including those for Zhejiang province. Building structures in Hangzhou are
designed for seismic design intensity 6 which represents a moderate seismic level. Considering
the height, length and slender form of Tower A, 100 year return basic wind pressure was used for
both the strength and drift code design checks. In addition to the code defined wind pressures,
the results of the wind tunnel study was also considered. Because of the various structural
irregularities resulting from the architectural design, certain special studies and analyses were
carried out. Some of these studies and analyses were mandated as a result of the seismic expert
panel review (EPR) process as a precondition to fulfilling the requirements for Design
Development approval. They included:
Wind tunnel test. The wind tunnel test and wind vibration analysis were conducted by College
of Civil Engineering and Architecture, Zhejiang University. The wind tunnel test was performed,
from 0o to 360o in 10o increments around the building. Both scenarios with and without future
buildings were modeled. (Lou, 2009) The wind loads from wind tunnel test were compared with
the wind load from the code and the most critical cases have been used for the structural design.
Figure 10: Wind environment simulation at the wind tunnel test (Lou, 2009)
Wind acceleration check. The accelerations at the highest occupied floor due to a 10-year return
wind were checked to ensure compliance with the 0.15 m/s2 code specified limit for residential
usage.
Vertical seismic force checks. The following elements were designed to resist a vertical seismic
force equivalent to 10% of seismic gravity load:
 The long cantilever trusses supporting the hotel spa at level 24.
 The cantilever trusses supporting the hotel guest rooms between levels 25 and 42.
 The cantilevered swimming pool structure at level 24.
 The coupling truss at level 42.
Seismic performance in the moderate earthquake. Certain key elements that are critical to the
performance of the building were designed to meet a higher seismic design standard than
conventional structural members. The following were designed to remain elastic in a moderate
earthquake (475-year return period):
 All transfer trusses.
Non-linear pushover analysis. A non-linear pushover analysis was performed for Tower A,
using the PERFORM-3D program. Various lateral load distribution scenarios were considered. It
was found that the inter-story drift ratio along the height of the building during a rare earthquake
event (2,475-year return period) was much less than 1% at all floors for the x- and y- directions.
Diaphragm slab stress analysis. To strengthen the tower structure above and below the two
large vertical slots in the building, the slabs over the entire floor at levels 1, 23, 24, 25, 43 and
roof level were increased to 180 mm thick. The slabs at levels 8 and 9 at the location of the
transfer truss over the hotel lobby were also increased to 180 mm. thick. Slab diaphragms in
critical areas were analyzed using semi-rigid diaphragms. Principal stresses were checked, and
where the allowable tensile stresses were exceeded, additional reinforcing bars, sufficient to resist
full tension load, were provided at slab mid-depths. In addition, reinforcing bars in all slabs in
critical areas were designed not to yield in the moderate earthquake.
Floor vibration check. The vibration performance of the following elements was checked for
compliance with the requirements of AISC Design Guide 11 “Floor Vibrations due to Human
Activity”:
In all cases, the estimated peak acceleration corresponding to the first vertical mode was
considerably less than the recommended limit of 0.5% g. This is because of the large participating
mass associated with vibration modes when compared to the walking excitation forces.
Core shear wall optimization. As the architectural design of Tower A developed, so did the
configuration of the lateral resisting system, in particular the shear walls. The extreme rectilinear
plan of the tower (165 m long by 33 m wide) determined that torsion control would be a major
feature of the structural design. Together with the high aspect ratio of the structure, it was
apparent that identifying a suitable configuration of shear walls would be paramount.
To minimize the impact on space within the tower, shear walls were arranged as much as possible
around and within the building cores (Figure 10). Three distinct core areas were established (north
core, central core and south core) to serve the three zones of the tower subdivided by the vertical
slots. To adequately resist torsion, the north and south cores were enhanced, through the addition
of an east-west wall extension to the west of the north core, as well as an isolated east-west shear
wall located south of the south core. Wall thicknesses were optimized as necessary, to bring the
centers of mass and rigidity as close as possible to each other. Similarly, where the provision of
walls was detrimental to the lateral performance of the structure, these walls were minimized as
much as possible or removed altogether. This is evident in the central core where walls along the
south and west sides of the core have been replaced by columns.
Figure 11. Core shear wall and floor framing configurations at various floor levels
Thermal analysis. An analysis was performed on Tower A to investigate the effects of thermal
variations on the behavior of the structure. The study also considered the effects of shrinkage
strains, in addition to thermal loads. Since the thermal demands on the structure depend on the
temperature at the time of construction, specifically at the time the delayed pour strips provided
to reduce shrinkage strains were poured, multiple construction scenarios with different
construction starting dates were considered. The thermal and shrinkage strain demands at
various points in time, during and after construction completion, with corresponding temperature
conditions which result from the governing scenarios, were then used to perform a structural
analysis of the full building model using the computer program ETABS. Various references were
used to compute the demands on structural elements considering thermal load and shrinkage,
and to estimate the implications of those demands on the structure including prediction of whether
cracking might occur and to propose localized strengthening of certain areas in order to control
cracking and/or improve strength resistance. Recommendations from those references were
evaluated in conjunction with results from the structural analysis of the building, which takes into
account the unique geometry and structural configuration of the building.
From the analysis results, it was found that the stresses caused by thermal changes were
relatively small, and were generally smaller in magnitude than those caused by shrinkage. It was
also found that the use of delayed pour strips was expected to reduce shrinkage strains by about
40%. The obtained stresses due to the combined effect of shrinkage and thermal loads were
generally very low throughout the building, and the highest tensile stresses generally occurred at
levels where the building slab was continuous from end to end which occurs at the floors near the
middle and top of the building. Tensile stresses were generally under 1.25 MPa, and on this basis,
no cracking would be expected in the slab under tension.
CONCLUSION
When considering only its rectilinear shape, the tall, long and slender Tower A is, by itself, a
challenging structure to engineer. In addition, the structure of the building exceeded several
prescriptive limits defined in the Chinese design codes causing it to require an expert panel review
(EPR) which was undertaken and passed.
The architectural design of Tower A called for radical formal moves that resulted in a complex
structure. Through a rigorous approach to the structural design and close collaboration between
all design disciplines, and a strategic use of concrete, steel and composite elements, it was
possible to provide coordinated structural solutions that met and exceeded the structural
performance and overall building structural design goals.
REFERENCE
Lou, W. J., Shen, G. H., & Wang, X. (2009). Wind Tunnel Test Report of G. T. Land Plaza.
Hangzhou, China: College of Civil Engineering and Architecture, Zhejiang University.
MULTIFUNCTIONAL COMMERCIAL BUILDINGS IN STEEL AND
COMPOSITE CONSTRUCTION
Prof. Dr.-Ing. Richard Stroetmann

Institute of Steel and Timber Construction, Technical University of Dresden
Dresden, Germany
Richard.Stroetmann@tu-dresden.de
Dipl.-Ing. Lukas Hüttig

Institute of Steel and Timber Construction, Technical University of Dresden
Dresden, Germany
Lukas.Huettig@tu-dresden.de
ABSTRACT
In the life cycle of commercial buildings the user requirements are changing usually manifold. In
addition to changing within a utilization form, e.g. office use, also different types of use, for in-
stance residential use, hotel and retail, may also be relevant. This requires a suitable building
design, which makes due to its supporting structures and façades, the floor plan layout and
building accesses, these kinds of usage possible. The following article presents results of the
ongoing research project P1118 "Multifunctional residential and commercial buildings in steel
and composite construction" (Stroetmann et al., 2015). This includes the ecological and eco-
nomic assessment of the buildings taking into account the convertibility, the use of steel and
composite constructions and the consideration of façades in the design and optimization ac-
cording to sustainability criteria.
1 INTRODUCTION
A relatively high vacancy rate of office buildings can be found in Germany and in many other
European countries (Figure 1). Especially in major cities, vacancy may become a visible prob-
lem. While, until the 70s of the last century, buildings were characterized by predominantly ro-
bust structures, which were planned for long-term use. In the subsequent time, increasingly
cost-effective investment properties were developed for special uses and early amortization of
investments. Negative consequences of this development are the lack of adaptability to chang-
ing requirements, decreasing marketability with possible structural vacancy and demolition of
the buildings.
The office and commercial building of the future should be planned in a way that the conse-
quences of social and sociological changes within its life cycle can be incorporated with low
monetary resources to avoid the vacancy of the property and to achieve a higher profitability.
For this, it is necessary to design the buildings for different types of uses. It requires a "robust"
primary structural system that allows the integration of a wide variety of uses without major
structural interventions.
The design of the structures gives the boundary conditions for the mechanical installations and
building finishing. Variable buildings are characterized by the fact that they require no or only
minor modifications of the structures when the type of use changes. Steel and composite con-
struction provides advantages related to conventional concrete structures because, besides
large spans and fewer columns, a reduction of building materials can be achieved. This leads to
less demand of building material and improves the ecological performance.
In the research project P881 the sustainability of steel and composite constructions for office
and administration buildings were investigated (Mensinger et al., 2016). Guidelines and plan-
ning tools, such as component catalogs and design software for office buildings were devel-
oped. By planning of buildings under defined design conditions, involving the relevant scenarios
of office configurations, the ecology and economy of the bearing structures was optimized.
Hence, multifunctional office and commercial buildings are the subject of the ongoing research
project P1118 (Stroetmann et al., 2015). Based on the project P881, various types of usage,
e.g. as office building, different types of residential, retail, hotel and gastronomy, will be consid-
ered in the design and optimization.
Fig. 1 – Changes in the vacancy rates of office space in selected European cities (Stroetmann &
Hüttig, 2016).
2 ASSESSMENT OF SUSTAINABILITY
The assessment of sustainability takes place by various certification systems. The British sys-
tem BREEAM and the US system LEED are the international leading systems. In Germany, the
equivalent systems DGNB (Deutsches Gütesiegel für Nachhaltiges Bauen) for public buildings
and BNB (Bewertungssystem Nachhaltiges Bauen für Bundesgebäude) for federal buildings
have been developed. Compared to BREEAM and LEED an equal rating of the three aspects
ecology, economy, socio-cultural and functional quality is taken by DGNB and BNB. Thus, the
German systems are also referred as systems of the second generation. In addition, the tech-
nical and the process quality are taken into account as in all areas relevant properties. Mean-
while, in addition to office and administration buildings other types of buildings, such as hotels,
commercial buildings, education centers, industrial buildings and parking garages can be as-
sessed.
2.1 ECOLOGICAL SUSTAINABILITY

The evaluation of the ecological quality follows by the Life Cycle Assessment (LCA) of buildings
and its components. This includes the phases of product manufacturing, construction, the use
stage and disposal. In addition credits and debits beyond the life cycle for reuse, recovery and
recycling will be considered. The environmental indicators (e. g. global warming potential and
primary energy demand) for building materials and products are collected in databases and en-
vironmental product declarations (EPD). An overview is given in (Stroetmann et al., 2016). In
several EPD only the phase of manufacturing and the credits and debits at the end of life are
considered. For the phases of erecting, use and disposal a lack of data can be observed. An
estimation can carried out from the construction processes and transportation distances.
The Assessment of buildings according to DGNB and BNB is carried out by defined reference,
target and limit values for the individual criteria. By comparison with the values of the assessed
building points will award. To determine the degree of the ecological quality, environmental indi-
cators will weighted by factors of importance. The reference, target and limited values are de-
fined for entire buildings but not for individual components or building systems. A similar proce-
dure, as applied for buildings, is possible for the evaluation of components and building systems
(Mensinger et al., 2016). Thereby, it is necessary to define the functional unit and to determine
target and limit values for the variations of relevant constructions. This allows classifying the
quality of the given solution. By using the factors of importance for the individual criteria, the
ecological performance can be determined similar to the systems of BNB and DGNB. The prin-
ciples of the method used in the research project P881 are shown in the Figure 2.
Fig. 2 – Life cycle assessment of structures in the context of the research project P881
(Mensinger et al., 2016).
2.2 Economical sustainability

The economic assessment takes into account the building-related life cycle costs (LCC) as well
as the efficiency, adaptability and the intrinsic value. Table 1 provides an overview of the criteria
given by BNB and DGNB for office and administration buildings and their shares in the overall
assessment. The construction costs of the buildings become less importance. The influence of
the follow-up costs as well as the efficiency, flexibility and intrinsic value increases. The efficient
use of buildings to the intended purpose is measured by the area efficiency factor Feff, which will
determined by the ratio of main usable area to gross floor area. At BNB the target value is
Feff = 0.75, the limit value is Feff < 0.48. The intrinsic value of a building increases by its adapta-
bility to different scenarios of use. Scopes of the evaluation are criteria like
− clear story height and building width,
− vertical access of the building,
− floorplan design (size of units, space for mechanical services, emergency exits etc.)
− construction of inner- and partition walls
− level of imposed loads and their reserves
− technical equipment for air condition, building automatization, building installation centers
− possibility of space use in small units, possibility to use renewable energies.
Depending on the type of building different requirements are placed on the mentioned criteria.
Space efficiency and adaptability can be achieved by appropriate floorplan design, building
width, clear story heights, building access, including relevant usage scenarios and appropriate
revisable technical installations. Flexible buildings are characterized by floors without or at least
optimal positioned inner columns involving various usage scenarios. Related to the construction,
the adaptability by BNB is rated positive, if at least 80 % of the inner walls are movable and at
least 50% of the floors are designed for an imposed load of  5 kN/m². Partition walls should be
connectable in each axis of the façade grid, not extend into the floor and ceiling structure, dust-
free mountable and reusable.
Under the conditions for a space-efficient and flexible building and the assumption that the costs
for operating and maintenance are largely independent on the bearing structures, the optimiza-
tion and comparison of building structures can be made on the manufacturing costs. Prices for
building materials, performance values for productivity and salaries depending on regional, tem-
poral and economic fluctuations. In the research project P881, a compilation of manufacturing
costs and work processes for bearing structures of buildings was made. This is based on re-
searches and experience of various companies from the German-speaking countries. The re-
sults of the parametric studies were rated as a percentage of the respective peak value (100 %).
This corresponds to a relative comparison, disposing the effect of these variations on the results
as far as possible. This applies as long as the relations of the expenses remain at the compared
systems.
Table 1. Criteria of the economic quality of office buildings (cf. BNB, DGNB).
content BNB DGNB
building-related life cycle costs 11,25 % 9,64 %
space efficiency flexibility and conversion feasibility
profitability, intrinsic value, 3,75 % 9,64 %
performance adaptability marketability
7,5 % 3,21 %
3 DESIGN OF BEARING STRUCTURES

From the building requirements and the specified grids of façades and columns, the boundary
conditions and design parameters for the bearing structures arises. This includes the spans of
slabs and beams, floor heights, imposed and additional dead loads, fire protection, the design of
components and installation spaces. A selection of appropriate construction systems, compo-
nents and design principles, which fulfill the requirement under consideration of sustainability
criteria in an optimum, has to be taken.
A systematic procedure for the design of the building structures is to deduce the essentials pa-
rameters of the relevant usage scenarios. The choice of scenarios should consider the building
location (e.g. city center or suburb) and the current and future needs. Furthermore, it is prefera-
ble to distinguish between floors within the building because certain uses preferable positioned
in certain floors. Typical positioning’s are underground parking and storage areas in the base-
ment, retail on the ground floor, office, residential and hotel uses on the upper floors. Specific
requirements resulting from the type of use, e. g., the width of the building, story height, im-
posed loads, access to the building and technical equipment. Vice versa, the suitability of the
building parameters to specific usage can be carried out. According to the previous descriptions,
in P1118 the essential characteristics of building parameters for the design of bearing structures
were examined. Some individual parameters are explained in the following.
The assumptions for dead and imposed loads are carried out in Europe by EN 1991-1-1 and the
country-specific national annexes. Figure 3 shows a comparison of the imposed loads, which
are applicable with a surcharge for partition walls for different types of usage taking into account
the German national annex. Floors for office, residential and hotel usage can be designed for a
relatively low imposed load (≤ 3.0 kN/m²). For retail, convention centers and gastronomy areas,
imposed loads up to 5.0 kN/m² have to be considered. Storage and archive space requiring
even higher loads. The amount of additional dead load depends on the ceiling construction. The
use of suspended ceilings and double floors creates a high degree of flexibility for the technical
installations and subsequent use changes. If the floors are rented to different users, the combi-
nation of both systems may be relevant.
Fig. 3 − Imposed loads and additional loads for partition walls for various types of use.
Fig. 4 − Convenient and possible building width and clear story height of various types of use.
Figure 4 shows convenient and possible building widths and story highs for various types of
usage. The left diagram indicates that a building width of 12 to 14 m allows several types of use.
The ceiling structures may design with or without internal columns. By positioning inner columns
the material consumption and the width of the floor constructions may be reduced, on the other
hand the flexible usage will be restricted. Looking at the clear story height, considerable differ-
ences as a result of the various usage requirements can be found. Parking levels can be built
with a clear height of  2 m. For residential and office uses clear heights  2.50 m are suitable.
For retail larger clear story heights are necessary depending on the sales areas.
4 DESIGN AND EVALUATION OF REFERENCE BUILDINGS

Based on the results of the research project P881, in the project P1118 buildings with different
degrees of variability will be optimized according to ecological and economic criteria. The inves-
tigations include the supporting constructions and foundations, the structural fire protection as
well as the façades. The assessment takes into account, in particular, the influence of variability
and the resulting improvement of marketability for a longer period of use. It will be proved, to
what extent a greater variability of the buildings leads to more effort in the supporting and fa-
çade constructions. Furthermore, the impact to the ecological and economic balance over the
life time of the building is in the focus of the examination. In the following, studies on a building
with high variability and the influence of the supporting and façade constructions on the ecologi-
cal performance are described.
Figure 5 shows the cross-section of the reference building with a high variability of use. A selec-
tion of floor systems without inner columns will be accessed. Due to the clear story heights of
3.2 m and imposed loads of 5.0 kN/m² in all upper floors, common usage types such as office,
residential, hotel use, medical practices and retail are made possible. The ground floor offers
due to its larger clear story height optimal conditions for the usage as a sales area. The two
lower floors serve as parking and storage areas. They will be included in the examinations at a
later stage. The following evaluations refer to the ground floor and the upper floors.
Multifunctional reference building with three

different floor systems (see Figure 6):
− downstand beams with composite slabs,
− additive floor with integrated beams (IFB)
− TOPfloor® Integral system.
Input variables:
− loads: dead load of construction,
additional dead load 1.5 kN/m²
imposed load 5.0 kN/m²
− construction steel grade: S460
− reinforcing steel grade: BSt500
− column spacing: 5.4 m
Fig. 5 – Cross section of a multifunctional reference building with clear story heights and total
building height with the floor systems of Figure 6.
For the reference building, three composite floor systems with a span of approximately 13.0 m
in transverse direction of the building were selected. This spans the building width without inner
columns. Figure 6 shows the positioning of the columns and beams in the floor plan as well as a
longitudinal and cross-sections of the floor systems. Figure 7 compares the demand of building
material for the supporting constructions. The quantities are shown separately for the different
types of materials with relationship to a square meter gross floor area (GFA). The mass of con-
crete has to multiply by a factor of 10. In the case of structural steel, the proportions of the col-
umns and beams are shown separately. The columns are hot rolled sections from the HEM-,
HEA- and HEB-series. Even if they have only a small portion of the building materials, its visibly
clear that the quantity of the column steel also increases as the floor weight increases.
The composite floor with downstand beams offers the advantage of high material efficiency and
low construction weight. This is particularly evident in the low demand for sectional steel. Be-
cause of the relatively large spans of 5.4 m, intermediate supports of the profile sheets are nec-
essary during concreting works.
Fig. 6 – Composite floor systems: Left – downstand beams with composite slabs; middle – addi-
tive floor with integrated floor beams (IFB); right – TOPfloor® Integral system.
Fig. 7 – Demand of building material for the different constructions (concrete in [10-1 kg/m²]).
The second variant is a slim floor system with integrated floor beams (IFB). High trapezoidal
metal sheets without composite effects and in-situ concrete (additive floor) was considered to
allow concreting without intermediate supports. The composite effect between the steel beams
and the in-situ concrete will reached by means of drilled web holes and through-mounted rein-
forcement. The additive floor system can be designed with a construction height of only 40 cm.
For the composite floors with downstand beams and slim floor systems typical additional com-
ponents for residential, office and commercial buildings, like floor screed and footfall sound insu-
lation, will be considered. Downwards, the structures are covered by suspended ceilings. In the
case of composite floors with downstand beams the technical installation can be integrated in
the level of the beams. To cross the beams the webs have openings in the regions of low shear
forces. In the case of the additive floors, the clear room height in the corridor area is reduced by
a suspended ceiling, above which the technical installation is located. The steel sections are
clad with 2.5 cm thick stone wool plates to achieve a fire resistance of 90 minutes.
The third variant is the TOPfloor® Integral system of the Swiss Wetter AG. The prefabricated
elements consist of two half honeycombed beams and a thin reinforced concrete slab. The
composite effect is reached by reinforcing bars welded on both sides to the web. The transport-
able elements have a width of 2.5 to 2.7 m, the distance of the beams is between 1.25 and 1.35
m. After laying, shear connectors between the elements will welded and the gaps between the
elements will grouted. The system can be carried out with the reinforced concrete slab in upper
or the lower position (negative position). Although the negative position leads, in particular for
long spans, to significant higher steel demand compared to the other floor systems, but this has
two essential advantages. On the one hand, the underlying concrete slab protects the overlying
steel beams against fire. Furthermore, the building technology can be accommodated in the
resulting cavity floor. A change of user does not require any intervention in the underlying story
of the building. The technical installation happens in cable ducts which are equipped with fire
protection. As a result, no further fire protection measures for the honeycomb beams are neces-
sary.
On the basis of the building materials demand, the environmental impacts caused by the sup-
porting constructions can be determined. Table 2 shows the primary energy demand and the
global warming potential of typical building materials. The ecological balances are made availa-
ble in environmental product declarations and databases, such as ÖKOBAUDAT. In these bal-
ances the phases considered in the lifecycle are reported according to EN 15643-2.
Figure 8 shows the primary energy demand and the global warming potential of the accessed
floor systems and columns related to a square meter gross floor area (GFA). The efficient ma-
terial utilization of the composite floor with downstand beams results in a low demand of building
materials, which has a positive effect on the ecological balance. In contrast, the TOPfloor® inte-
gral system results in significantly less favorable LCA values due to the small beam distances,
additional edge beams, the lower construction height and the steel chords in the compression
zone. In addition to the sectional steel and reinforcement, the galvanized composite and trape-
zoidal sheets have also a considerable impact in the ecological balance of the composite and
additive floors.
Table 2 – Primary energy demand and global warming potential of building materials.
PEges GWP
building material sources lifecycle phases
[MJ/t] [kg CO2/t]
IFBS,2013 A1-A3, C4, D
trapezoidal sheet 18808 1020
Ökobaudat,2015 A4, A5
steel sections and bauformunstahl e.V.,2013 A1-A3, D
12065 808.0
plates Ökobaudat, 2015 A4, A5, C1, C2
reinforcement 12753 760.8 Ökobaudat, 2015 A1-A5, D
InformationsZentrum Beton
concrete C20/25 465.8 73.7 A1-A5, C1-C4*, D
GmbH, 2013
*includes waste transport and separation of concrete, reinforcement and trapezoidal sheet
Fig. 8 – Primary energy demand and global warming potential of the supporting constructions.
5 SUSTAINABILITY AND INFLUENCE OF FACADE SYSTEM

Modern buildings impress with their exterior design with innovative façade solutions. The trend
is increasingly moving away from conventional perforated facades to strip and element facades
with a small proportion of opaque surfaces up to all-glass façades. Particularly for multifunction-
al buildings in steel and composite construction, element façades are preferred solutions be-
cause of the adaptability to changing user requirements.
The choice of the façade influences the ecological performance of a building, especially during
its use phase. Depending on the heat transfer coefficient of the façade system as well as the
size of the façade area, the heating and cooling energy demand increases or decreases. While
in the past all-glass and element façades have been classified as ecologically harmful, modern
systems with coated multi-pane insulation glazing and improved frame constructions can
achieve a heat transfer coefficient that is far below the allowed value and is comparable with
perforated façades.
In addition to the utilization phase, the construction and dismantling phases as well as the recy-
cling potential of the materials also influence the ecological balance of buildings. Up to now,
there are hardly any official data for façades available. In Table 3 preliminary data of four ele-
mentary facades for office and commercial buildings are listed. Figure 9 shows a comparison of
the non-renewable and total primary energy demand (PEne, PEges) as well as the global warming
potential (GWP). Three all-glass façade systems with frame constructions out of timber, steel
and aluminum as well as an element facade with aluminum cladding are considered. The eco-
logical data are basing on information from facade manufacturers and environmental product
declarations.
The comparison shows the sum of the expenses for the lifecycle phase’s construction, disman-
tling and recycling. As expected, the glass façade with timber frame shows the lowest ecological
impact due to the sustainable frame material. The all-glass facades with steel or aluminum
frames are significant less favorable. Furthermore, the higher aluminum content of the elemen-
tary façade leads to even higher values compared to the other systems. The total primary ener-
gy demand in the phases A1 to A5 and C1 to C4 is almost twice as high as the all-glass façade.
However, the recyclability of the aluminum leads to the fact that the additional impact is limited
to approx. 1/3 compared to the all-glass façade with steel or aluminum frame. For the all-glass
façade with timber frames the EPD assumes in the End of life phase that 90 % of the metals
can be recycled and 100 % of the timber will be used for thermal utilization.
Tab. 3 –Primary energy demand and global warming potential of different façade systems.
phase of production
recycling phase
and disposal
façade system Source
PEges GWP PEges GWP
[MJ/m²] [kg CO2/m²] [MJ/m²] [kg CO2/m²]
all-glass façade, (UniGlas, 2015
1358.3 64.35 -215,0 -12,60
timber frames BGT, 2012)
all-glass façade,
2155.3 121.03 -805.8 -47.90 (ift, 2012)
steel frames
all-glass façade,
2190.0 140.00 -847.4 -45.01
aluminum fra. (Eichhorn,
elementary façade, alu- 2010)
3830.0 225.00 -1998.5 -105.94
minum cladding and fra.
Fig. 9 – Primary energy demands and global warming potential of different façade systems in
the life cycle phases A1 – A5, C1 – C4 and D (recycling).
When evaluating the façade systems, the operating life up to a necessary replacement should
also be considered. According to the manufactures specifications the considered timber frame
facade has only an estimated operating life of 30 years, while the steel and aluminum facades
can achieve a life cycle of 50 to 60 years. In view of the high adaptability of the systems and the
physical properties, which are far above current legal requirements, longer life cycles are possi-
ble.
Figure 10 shows the primary energy demand of the various façade systems and supporting
structures for the reference building of Figure 6. The different construction heights of the floor
systems lead to different building heights and façade surfaces and their proportions in the eco-
logical balance (Table 4 and Figure 10). This is between 1/3 and 2/3 of the total primary energy
demand in the phases of construction, dismantling and recycling.
Figure 11 shows with the left graph a comparison of the ecological performance of the support-
ing structures for the different floor systems. The use of composite floors with downstand beams
leads to the best performance with a clear distance from the other systems. Considering the
various façade systems in the life cycle assessment in addition leads to a reduction of the dis-
tance. Consequently, the selection of floor systems with a low construction height results in sav-
ings of facade area, which leads to an improvement in the ecological balance.
Tab. 4 – Ratio of façade to gross floor area of the reference building for different floor systems.
floor system building height façade area GFAges FA/GFAges
composite slab, downstand beams 23.7 m 4731 m² 0.695
additive floor with IFB 21.9 m 4389 m² 6804 m² 0.645
TOPfloor® Integral system 22.2 m 4438 m² 0.652
Fig. 10 – Primary energy demand per year of life cycle for different combinations of supporting
constructions and façade system (total life cycle = 100 years).
6 CONCLUSION
Examples of changing social conditions are the demographic change and the growing aware-
ness of sustainability, which affect the user requirements of buildings. The development of the
vacancy rates shows that lots of commercial and office buildings are insufficient flexible for ad-
aptation to the changing user requirements. Resource conservation, recyclability, life cycle costs
and conservation of value, even under changing property conditions, are more in the focus of
planning.
Fig. 11 – Degree of ecological fulfillment of different supporting constructions and their combina-
tions with façade systems for a 100-year life cycle.
Steel and composite constructions offer significant advantages to increase the sustainability
related to conventional building structures. The comparatively low construction weights and de-
mand of building materials have positive effects on the structures up to the building foundations.
This leads to reductions of the expenses for construction, transportation, erection and waste at
the end of building life.
Subject of the ongoing AIF FOSTA research project P1118 is the sustainability of office and
commercial buildings by using steel and composite constructions. Aims and results are the
quantification of the environmental and economic sustainability of the supporting structures and
the development of recommendations, guidelines and tools for planning of these building types.
7 REFERENCES
bauforumstahl e. V. (2013). Environmental product declaration - Baustähle: Offene Walzprofile
und Grobbleche. Institut Bauen und Umwelt e. V. (Ed.). Declaration-number: EPD-BFS-
20130094-IBG1-DE. Valid until: 24th October 2018.
BGT Bischoff Glastechnik AG (2012) Environmental product declaration -
Mehrscheibenisolierglas. ift Rosenheim GmbH (Ed.). Declaration-number: M-EPD-MIG-001021.
Valid until: 1st May 2017.
Eichhorn, D. (2010) Kurzfassung - Ökobilanzierung der Lindner ECO® Fassade. Lindner Fas-
saden GmbH (Ed.). Arnstorf.
EN 1993-1-2 (2010). Eurocode 3: Design of steel structures – Part 1-2: General rules – Struc-
tural fire design; German version EN 1993-1-2:2005 + AC:2009.
IFBS (2013): Environmental product declaration - Profiltafeln aus Stahl für Dach-, Wand- und
Deckenkonstruktionen. Institut Bauen und Umwelt e.V. (Ed.). Declaration-number: EPD-IFBS-
2013211-D. Valid until: 13th January 2018.
ift Rosenheim GmbH (2012) Environmental product declaration – Muster-EPD Stahl-
/Edelstahlfassaden. ift Rosenheim GmbH (Ed.). Declaration-number: M-EPD-SFA-XXX. Valid
until: 1st November 2017.
InformationsZentrum Beton GmbH (2013): Environmental product declaration - Beton der
Druckfestigkeitsklasse C 20/25. Institut Bauen und Umwelt e.V. (Ed.). Declaration-number:
EPD-IZB-2013411-D. Valid until: 25th Juley 2018.
Stroetmann R., Faßl Th., Hüttig L. (2016). Nachhaltige Geschossbauten in Stahl- und Verbund-
bauweise. In Stahlbaukalender 2016, Abschnitt 10, Seite 571 – 665. Berlin: Wilhelm Ernst &
Sohn.
Mensinger, M., Stroetmann, R., Feldmann, M., Eisele, J., Zink, J., Lingnau, V., Kokot, K. (2016).
Nachhaltige Büro- und Verwaltungsgebäude in Stahl- und Stahlverbundbauweise (Final Report
Research Project FOSTA P881). Düsseldorf: Verlag und Vertriebsgesellschaft mbH.
Ökobaudat (n.d.). Retrieved December 16, 2015, Information portal for sustainable building of
the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear
Safety, http://www.oekobaudat.de/en/
Stroetmann, R., Eisele, J., Schach, R., Hüttig, L., Trautmann, B.; Harzdorf, A. (2015) Einflüsse
der Stahl- und Verbundbauweise auf die Lebenszykluskosten und Vermarktungsfähigkeit multi-
funktionaler Büro- und Geschäftshäuser. IGF-Project-No. 18659 BG; FOSTA P1118. Manuscript
in preparation.
Stroetmann, R, & Hüttig, L. (2016) Multifunctional commercial buildings in steel and composite
construction. Report of IABSE Congress Stockholm – Challenges in Design and Construction of
an Innovative and Sustainable Built Environment, 482-489.
UniGlas GmbH & Co. KG (2015) Environmental product declaration - Konstruktion der
UNIGLAS Facade. ift Rosenheim GmbH (Ed.). Declaration-number: EPD-UGF-0.01. Valid until:
1st May 2017.
BUILDING SCALE COMPOSITE SYSTEMS
Aaron Mazeika, PE SE AIA

Skidmore, Owings & Merrill LLP
Chicago, USA
aaron.mazeika@som.com
Ashpica Chhabra, PE SE
Skidmore, Owings & Merrill LLP
Chicago, USA
ashpica.chhabra@som.com
ABSTRACT
Composite construction is often defined at the component level, with individual structural
members being composite elements, or with the selection of different materials for each element
based on specific structural demands. This paper will instead focus on composite structural
systems at the building- and multi-building scale. A selection of building design solutions will be
discussed where conventional structural components have been employed in innovative
configurations resulting in large scale composite structural systems that offer distinct structural
and economic advantages over more conventional design solutions.
CONVENTIONAL COMPOSITE CONSTRUCTION

Composite materials have been in use in building construction for thousands of years, from wattle
and daub walls that were used as far back as the Neolithic period, through to modern carbon fiber
reinforced polymers that have been introduced to building structures in just the past few years.
For the most part during this period, composite construction has developed with a focus on the
micro scale. The interaction of materials of different properties resulting in a material with an
amalgam of the properties of its constituent parts (wattle and daub, steel reinforced concrete,
glass fiber reinforced polymer, etc.). Composite construction can also be considered as a
structural system at the building component level. Composite steel floor framing beams and
concrete filled steel tube columns being examples of this application. This paper is intended to
highlight the use of composite construction at the macro or building system scale.
COMPOSITE BUILDING CORE

In high rise building construction it is common for the building core to reduce in size up through
the height of the tower, as certain core components drop off. A three zone direct elevatored office
building requires three elevator banks from the ground to the top of zone 1, two from the top of
zone 1 to the top of zone 2, and one from the top of zone 2 to the highest floor. This erosion of
the building core with height typically allows floor area used for the core below to be returned to
rentable office space at the upper floors, however there are limitations to this approach. Office
space that is too far from the building façade will receive insufficient natural daylight and will not
be useful as rentable office space. While in office buildings this space can be useful for other
functions, such as file storage etc., in other program types this space can be without value. A
specific example of this would be the common building typology of a mixed used tower with a
hotel constructed atop an office tower. This program mix is increasingly commonplace with many
examples built around the world. What makes this program stacking difficult is the combination of
different core requirements and different lease span (distance from core to façade) requirements
of the two different programs. The office levels typically have high core requirements due to the
large occupancy populations that need to be transported at peak arrival, departure and lunch time.
The office levels also typically favor a lease depth of 12-13m. A hotel floor plate by contrast has
a much smaller core requirement due to lower and less concentrated elevator traffic, and a
preferred lease depth of approximately 10m. The trivial solution to these conflicting requirements
would be a step change in the building width, with a narrower hotel tower sitting atop a wider
office tower. A more elegant form may be a tapered tower, but this introduces a new problem of
variable depths to the hotel rooms, an issue that complicates the work of the hotel operators and
interior designers. This design problem was elegantly solved in the design of the Jin Mao Tower
in Shanghai, China. This 420m tall composite tower consists of 52 stories of office floors beneath
a 36 story tall 5-star hotel. To address the different core requirements, rather than taper or step
the perimeter of the tower, the design team instead elected to hollow out the core and turn the
unused space into a dramatic 30 story tall atrium (see figure 01).
Figure 01-Jin Mao Tower Section (Left); Typical Hotel Level Plan (Right)
For the Wujiang Greenland Tower, in Suzhou, China, this concept of an upper level atrium hotel
was developed one stage further. The lower level barrel shaped office floor plan sat beneath a
hotel floor plan consisting of two curved single-loaded hotel bars facing north and south (see
figure 03). This arrangement allowed for the middle third of the core to be eroded forming an
atrium which passed completely from the west side to the east side of the building, giving the
building its signature ‘eye of the needle’ form (See figure 02). The resulting atrium has dramatic
views out to the west and east. While adding an important architectural element, the atrium in this
configuration resulted in the core being split in two, greatly reducing its structural stiffness in the
upper half of the tower. To correct this the two isolated building cores in the upper half of the
building are laced together using structural steel bracing to form a composite ‘coupled core’.
Wujiang Greenland Structural System

The lateral force resisting system for the tower consists of a ‘frame shear-wall’ structure
supplemented by structural steel outriggers and belt trusses. It consists of reinforced concrete
walls around vertical circulation and mechanical service zones at the center of the floor plate, and
a moment resisting frame, with columns located at the perimeter of the floor plate.
Figure 02- Architectural Rendering
The reinforced concrete core is separated into two portions in the upper levels of the building by
the hotel lobby (see figure 03 for building section). These two portions of the core are laced
together using a structural steel diagonalized truss, which allows the two portions of the core to
act compositely as one overall rectangular section. The moment resisting frame consists of steel
spandrel beams and encased composite columns at hotel levels and reinforced concrete beams
and encased composite columns at the office levels.
Floor framing consists of structural steel beams supporting composite metal deck slabs at hotel
levels and conventional reinforced concrete beams and slabs at office levels. Two of the columns
on each of the longer faces of the building are connected to the reinforced core with structural
steel outrigger trusses at mechanical levels and stiffen the tower lateral force resisting system in
the weak axis of the building. Belts Trusses are provided at these levels to engage axial stiffness
contributions of all perimeter columns. Core walls will vary from 1800mm thick at the base to
350mm thick at the top, with column sizes varying from 2100mm diameter to 800mm diameter.
Figure 03 - Building Elevations (Left); Hotel Level framing plan (Top Right); Typical lower level
framing plan (Bottom Right)
Elastic Analysis
ETABS was used as the elastic analysis software, with modeling assumptions and loadings
defined per the requirements of the Chinese codes. During the initial design process the columns
and walls are sized to meet the axial stress ratio limits as defined in the Chinese code. The steel
bracing members at the coupled core, outriggers and belt truss members were sized using in-
house optimization program that proportions structures such that internal work is done with a
minimum volume structure with the target of achieving maximum material efficiency without
exceeding a drift limit of H/500 under wind tunnel loads.
Mode 1: 7.3s (Translation) Mode 2: 4.8s (Translation) Mode 3: 2.9 (Torsion)
Figure 04 - Dynamic properties of building
Seismic Engineering
PERFORM 3D was used as the nonlinear time history analysis software. Seven sets of site
specific time histories very provided by the local seismic consultant. The analysis is performed to
meet performance targets set by the Chinese code as well as by the expert panel. The connecting
braces were required to remain non yielding under rare earthquake. Hence, the inelastic energy
dissipation is achieved by yielding of core link beams and perimeter moment frame beams (see
figure 05). The drifts under rare earthquake time histories were within the code limit.
Figure 05: Elements ratios as multiples of yielding limit state under rare earthquake time
histories; R/C link beams; Composite Moment Frame columns; R/C Moment Frame beams;
Steel Moment Frame Beams; Steel Diagonals at coupled core and outrigger/ belt truss levels.
(from left to right)
Connection Analysis and Design
The coupled-core wall lateral system consists of cross-brace members that join into an embedded
truss inside the core wall. In addition to the cross-braces, the members embedded in the core
wall are divided into four types: primary embedded verticals, secondary embedded verticals,
primary embedded braces and secondary embedded braces (see figure 06). The preliminary
sizing of embedded members was based on maintaining steel-to-steel load path at the
connections. The primary embedded braces continue the same section as the diagonal braces.
These braces transfer the vertical component of the forces along the diagonals to the primary
embedded verticals. For embedded verticals under net compressive forces, gradually increasing
contribution from composite action is assumed as studs are introduced along the member length.
The secondary embedded members are sized for the resultant forces due to the slight kink
between the cross-brace and the embedded primary brace.
Primary
embedded braces
Primary embedded
vertical
Secondary
embedded vertical
Secondary
embedded braces
Figure 06 - Embedded truss at the split core levels
Additionally, the connection was analyzed using general purpose finite element software Strand
7, to verify that the stresses in the steel do not exceed the design limits of Q345GJ steel and that
the stress level in concrete would not exceed the design capacity of C60 concrete in tension or
compression. The target performance for the connection is to remain elastic under moderate
earthquake forces. The forces were based on the elastic ETABS model. Based on the results of
the analysis, it was found that concrete cracks under moderate earthquake forces but the
embedded steel members and the longitudinal reinforcement have enough reserve capacity to
transfer the load (see figure 07). Since the embedded steel is designed to resist all the tension
forces by itself, the connection has enough capacity to resist earthquake forces.
Figure 07: Finite element modeling of core bracing connection; stresses in embedded steel,
concrete compression/ tension and reinforcement compression/tension (from left to right)
Resistance to Racking Motion

Resistance to racking between individual building halves is provided by having full diaphragm
connectivity at levels 77-79 and level 51 as well as from link bridges at all upper levels. A
horizontal axial spring connecting diagonally opposite corners of the split core walls is used to
measure the racking motion in the analysis model. If there is no connection between the split core,
there is a net 108mm spring deformation under moderate earthquake time histories. Once the
split cores are tied together with vertical cross-bracing this measured displacement reduces to
89mm. If the diaphragm connectivity at levels 77-79 and Level 51 is considered the maximum
displacement reduces considerably to 6.4mm. Alternatively if connectivity at all the link bridges is
considered, the maximum displacement reduces to 3mm. In the building as designed with cross
braces, discrete diaphragm connectivity and link bridges the maximum displacement due to
racking motion is 2mm
COUPLED BUILDINGS
After connecting two distinct building cores within a single building for enhanced performance, a
logical development of the thought would be to study the connection of building cores in two
adjacent buildings. As high rise towers become increasingly slender, the structural efficiency is
greatly reduced and the construction cost increases exponentially. This raises the possibility of it
being more cost effective to construct two slender towers that are structurally connected together,
rather than being conventional independent towers. The logic being that when combined, the
effective slenderness of the coupled towers would be lower as it would be based on the plan
dimension of the combined structural system. This idea was developed further in SOM’s
competition entry for the Zhengzhou Greenland Towers project (see figure 08).
The competition brief called for a twin tower configuration on the site of new district planned on
the outskirts of Zhengzhou, China. After developing multiple twin tower forms, the architectural
design team challenged the structural design team to explore how the decision to build twin towers
could enhance the structural performance and thus yield a more efficient design solution. These
explorations yielded two concepts that shaped the final building form.
Figure 08: Architectural rendering
Firstly, by connecting the cores of two slender buildings together with a structural steel truss, the
towers could mutually stiffen each other, with the core of building B acting as an outrigger column
to the core of building A, and vice versa. This allowed each of the towers to be more slender than
cost effective for an individual tower. It also informed the plan configuration of each tower. As the
outrigger behavior between the two towers would only work when one tower was moving in the
direction towards or away from the other tower, this mutual stiffening was only advantageous in
one direction of movement for each of the towers. Consequently, the towers would each need to
have sufficient stiffness acting alone when deflecting perpendicular to its neighboring tower, but
would be supplemented by the outrigger effect when displacing towards or away from its neighbor.
This resulted in a need for a rectangular plan and core, with the long axis of each building running
parallel to each other.
Secondly it was recognized that the connected towers approach allowed the use of an offset core
plan configuration, that would normally not be satisfactory for towers of this height. For twin tower
configurations, there is generally an architectural concern about view blockage, where one tower
blocks the views from one face of the other tower and vice versa. Consequently, for twin tower
configurations there is typically a low value side of the floor plate. It is therefore logical to design
the plan configuration of the tower such that the majority of rentable floor area is biased towards
the preferred views. This can be achieved through the use of an offset core configuration. The
building core components are relocated from their more typical location at the center of the floor
plate and are instead offset to the side of the building with the least beneficial views. As well as
improving the quality of views, this configuration also yields a clean rectangular office floor plan,
rather than the more typical ring configuration resulting from the use of a central core. This plan
configuration was most famously employed at the Inland Steel Building, completed in Chicago, IL
in 1958 (see fig. 09), and has remained a desirable yet uncommon configuration since that time.
Figure 09 - Inland steel building – Typical floor plan with offset core configuration
While the offset core configuration offers many architectural benefits, this layout complicates the
structural engineering design as the offset core introduces a significant torsional irregularity to the
structure, which results in a torsional building response when the building is subjected to lateral
wind or seismic loads. A significant eccentricity is introduced between the midpoint (for wind) and
center of mass (for seismic) of the floor plate which are closer to the center of the office floor plate,
and the center of stiffness of the lateral force resisting systems which is closer to the center of the
building core. The torsional response that this creates can increase motion perception in the tower,
complicate the detailing of the exterior wall, and increase the cost of the structural system to
control these issues. For these reasons this plan configuration has typically seen limited use in
office buildings up to approximately 20 stories tall. Here too the connected twin towers concept
offers significant benefit. As well making a flexural connection between the two offset cores,
allowing each building core to act as an outrigger to the other, there is the potential to make a
diaphragm connection between the two towers also. When connected together by a stiff
diaphragm, each tower can no longer twist independently about its own center of stiffness and
the torsional response is restricted to the combined building twisting around the center point
between the two towers, significantly increasing the torsional stiffness. In addition to the stiffness
increase, as the towers can now be thought of as one combined building structure, the torsional
eccentricity is eliminated as the center of mass, center of geometry and center of stiffness are all
now located at the center point between the towers.
Zhengzhou Greenland Structural System
Zhengzhou is located in a region of high seismicity, and moderate wind. In the stronger east-west
direction, the lateral system of the towers consists of a reinforced concrete core and a perimeter
moment frame with 9m column spacing. In the transverse direction, the lateral system consists of
the reinforced concrete shear wall core, and a perimeter moment frame with closely spaced
columns. At the top of the office zone where the hotel floor is set back, the transfer trusses also
work as outrigger trusses connecting the core and perimeter columns. The hotel amenity bridge
floors at the top of the building connect the cores of the two towers, and the amenity floor slabs
provide diaphragm connectivity between the two towers (see figure 10).
Connecting Diaphragm
Bridge Truss
Floor hung from bridge

truss
Outrigger Transfer truss
Figure 10 - Connected level plan (top left); Typical plan with two offset core floor plates
(bottom left); weak axis structural system diagram (right)
Unconnected, each tower is relatively slender and has a low torsional stiffness. Due to the offset
core configuration, the buildings have a significant torsional response when subjected to lateral
loads. By tying together, the twin towers at the top with two floor diaphragms, the displacement
of each tower is constrained to the other, and the overall torsional stiffness is significantly
improved. This results in the torsional period being reduced by 18% (see figure 11). Additionally,
by connecting the two towers together, overall building drift under lateral loading is reduced by
10%.
Torsional mode without bridge truss T=4.0 s Torsional mode with bridge truss T=3.3s
Figure 11: Enhanced torsional performance with connected bridge trusses
BUILDING CLUSTERS
This connected towers concept was further developed in the design of a competition entry for a
multi tower mixed use development in Sudirman City, Indonesia. An architectural design idea was
developed to interconnect the residential and hotel towers with pedestrian bridges, allowing all
the residential occupancies to share a common amenity space. Given the programmatic desire
to interconnect the buildings, a series of studies were undertaken to determine the most beneficial
configurations to allow the four connected towers to mutually support each other and maximize
the structural efficiency of the cluster of buildings as a whole. For residential towers, where floor
area in proximity with the façade is at a premium, the optimal architectural design solution often
results in elongated towers with a pronounced ‘narrow’ dimension and ‘wide’ dimension. While
enhancing views this configuration makes the structural engineering design of the towers difficult,
as additional material is required to provide sufficient structural strength and stiffness in the
narrow, weak direction. The structural concept behind the connected bridges was to tie the weak
direction movements of one building to the strong axis of another building. Through this approach,
the strong axis of a building where excess strength and stiffness is available, is used to strengthen
the weak axis of the supported building.
Pinwheel Configuration
Below is a study comparing weak axis core stresses and dynamic behavior of rectangular cores
with a plan aspect ratio of 1:4 tied together in various configurations. Connecting two rectangular
bars oriented in the same direction results in no improvement in core stresses and makes the
torsional behavior worse than that of an individual building. Connecting two rectangular buildings
orthogonal to each allows one building to assist the other. However, connecting the buildings in
a pinwheel configuration allows each building to provide support and be supported.
.
Mode 1 – 7.0s Translation Mode 1 – 7.1s Translation Mode 1 – 4.0s Translation Mode 1 – 3.1s Translation
Mode 2 – 2.3 s Translation Mode 2 – 2.7s Translation Mode 2 – 3.1 s Translation Mode 2 – 3.1s Translation
Mode 3 – 0.95s Torison Mode 3 – 2.2s Torison Mode 3 – 1.9 s Torison Mode 3 – 2.4s Torison
Figure 12: Connecting multiple buildings study
Sudirman City Structural System
Similar idea as discussed above was used for the lateral system for the Sudirman city competition.
Instead of four buildings as shown in figure 12, there buildings were interconnected with
skybridges to control the building drift. For the tallest residential building, building drift exceeds
the drift limit when designed for gravity loads alone. However, it was found that the building drift
can be controlled by interconnecting to other towers with sky bridges as shown in figure 13.
55
50
45
40
35 Connected
Towers
30
Story No
25
Connection
20
Limit
15
10
5
0
0 0.01 0.02 0.03
Drift
Figure 13 –3D analysis model of the connected towers (left); Drift along weak axis of the
tallest tower (right)
CONCLUSIONS
Composite construction offers many possibilities for new efficient building configurations when
considered in the context of building scale and multi-building scale composite systems. The use
of different structural materials within a building structure based on the suitability of a material for
the specific structural function is well established. Reinforced concrete is commonly used for
axially loaded elements such as building cores and perimeter columns, while structural steel is
commonly used for flexural elements such as floor framing, and certain special connecting
elements such as outrigger and belt trusses. These conventional approaches at the building scale
can be also applied to the multi-building scale for further efficiency gains, when the building
geometries and configurations can be established to leverage these potential benefits.
TOWARDS THE USE OF HIGH STRENGTH STEEL AND CONCRETE
IN HIGH-RISE CONSTRUCTION
J Y Richard Liew
Department of Civil & Environmental Engineering, National University of Singapore
ceeljy@nus.edu.sg
Yan-Bo Wang
Department of Civil & Environmental Engineering, National University of Singapore
ceewyb@nus.edu.sg
Yong Du
College of Civil Engineering, Nanjing University of Technology, Nanjing, China
yongdu_mail@njtech.edu.cn
ABSTRACT
Concrete filled steel tubular column comprising a hollow steel tube infilled with concrete has
been used widely in high rise buildings. Although modern design codes provide guides on
concrete filled steel tubular members, they do not cover their applications involving high
strength concrete and high tensile steel. To fill this research gap, new tests have been
conducted on concrete filled steel tubular members with ultra-high strength concrete (fck up to
190N/mm2) and high tensile steel (fy up to 780N/mm2). In this paper, a design guide has been
proposed for concrete filled steel tubular members based on an extension of Eurocode 4
method for concrete compressive strength up to 190N/mm2 and high tensile steel with yield
strength up to 550N/mm2. More than 2030 test data on concrete filled steel tubes with normal
and high strength materials have been analysed to formulate this design guide.
1 INTRODUCTION
High strength construction materials are now attractive owing to their economic and
architectural advantages. For columns at the lower floors and basements of a high-rise
building, the use of high strength materials has an advantage of reducing the members’ cross-
sectional size. This helps to increase the usable floor space and reduce the overall cost of
investment in high-rise building. However, material brittleness could be one of the problems
for high strength concrete and local buckling may be a problem for structural members with
high tensile steel. To overcome these problems, one solution is to use composite structural
members, especially concrete filled steel tubes as columns [1-3], where the ductility and
strength of the concrete core can be enhanced by the confinement effect from the steel tubes
while the local buckling of the steel tube can be delayed or even prevented by the concrete
core. However, modern design codes do not cover their applications involving high strength
concrete (HSC) and high tensile steel (HTS).
The material strength limitation of various design codes for composite columns are compared
in Table 1. EC 4 [4] is applicable to composite columns with normal weight concrete of strength
classes C20/25 to C50/60 and steel grades S235 to S460. AISC 360-10 [5] only applies to
composite columns with normal weight concrete cylinder strength from 21 N/mm2 to 70 N/mm2
and steel yield strength up to 525 N/mm2. The Chinese Code [6] only applies to composite
columns with concrete cylinder strength from 25 N/mm2 to 67 N/mm2 and steel yield strength
from 235 N/mm2 to 420 N/mm2. The Japanese Code [7] allows the use of high strength
concrete with fck up to 90 N/mm2. Composite structural members generally exhibit better
ductility and higher buckling resistance compared with steel or reinforced concrete members.
However, EC 4 [4] gives a narrower range of material strength for steel and concrete
compared with EC 2 [8] and EC 3 [9, 10] for concrete and steel structural design, respectively.
Table 1. Limitation on characteristic strength (N/mm2) in modern design codes

Steel yield strength Concrete cylinder strength
Codes
(N/mm2) (N/mm2)
ANSI/AISC 360-10:2010 ≤ 525 21 ~ 70
GB 50936:2014 235 ~ 420 25 ~ 67
AIJ, 1997 235 ~ 440 18 ~ 90
EN 1992-1-1:2004 N.A. 12 ~ 90
EN 1993-1-1:2005
235 ~ 700 N.A.
EN 1993-1-12:2005
EN 1994-1-1:2004 235 ~ 460 20 ~ 50
This highlights the dearth of a new design guide to fill the gap by allowing the design of CFST
columns with concrete cylinder strength up to 90 N/mm2 and steel of yield strength up to 550
N/mm2. In this paper, a design guide has been proposed for concrete filled steel tubular
members based on an extension of Eurocode 4 method EC 4 [4, 12] with special
considerations for the high strength concrete and the high tensile steel. The design method
was calibrated against the test data collected worldwide.
2 NEW TESTS AND COLLECTED TEST DATA

A total of 45 new tests, including short column (Figure 1) and beam-column (Figure 2) have
been conducted to provide additional data of concrete filled steel tubular members with ultra-
high strength concrete (UHSC, fck up to 190N/mm2) and high tensile steel (fy up to 780N/mm2).
For short column, tests have been carried out on three series of specimens under concentric
compression. For Series 1 tests, 11 small scale specimens were prepared with trial tests on
UHSC infilled in S275 and S355 steel tubes under different loading conditions. For Series 2
tests, 8 larger specimens were tested for further verification with both of UHSC and normal
strength concrete (NSC). For Series 3 tests, 17 specimens were tested to evaluate the
performance of CFST columns in which 12 specimens used the UHSC and S700 steel tubes
and 5 specimens employed the UHSC and mild steel. The CFST column specimens in Series
1, 2 and 3 tests are shown in Table 2 where λ is the relative slenderness, δ is the steel
contribution ratio as in EC 4 and Ntest,u is the testing values of the ultimate resistance.
Table 2. Configuration details and test results for short CFST specimens
L fy fck Ntest,u
Series No. Steel sections λ δ
(mm) (MPa) (MPa) (kN)
S1-1-1(a) 250 CHS114.3×3.6 403 - 0.088 - 486
S1-1-2(a) 250 CHS114.3×6.3 428 - 0.094 - 1039
S1-1-2(b) 250 CHS114.3×6.3 428 - 0.094 - 990
S1-2-1(a) 210* CHS114.3×6.3 428 173.5 0.110 0.493 2866
S1-2-1(b) 210* CHS114.3×6.3 428 173.5 0.110 0.493 2595
Series 1 S1-3-1(a) 250 CHS114.3×3.6 403 173.5 0.142 0.326 2422
S1-3-1(b) 250 CHS114.3×3.6 403 173.5 0.142 0.326 2340
S1-3-2(a) 250 CHS114.3×3.6 403 184.2 0.145 0.313 2497
S1-3-2(b) 250 CHS114.3×3.6 403 184.2 0.145 0.313 2314
S1-3-3(a) 250 CHS114.3×6.3 428 173.5 0.131 0.493 2610
S1-3-3(b) 250 CHS114.3×6.3 428 173.5 0.131 0.493 2633
S2-1-1 600 CHS219.1×5 380 - 0.109 - 1190
Series 2
S2-1-2 600 CHS219.1×5 380 51.6 0.142 0.520 3118
S2-1-3 600 CHS219.1×5 380 185.1 0.191 0.232 7837
S2-1-4 600 CHS219.1×5 380 193.3 0.193 0.224 8664
S2-2-1 600 CHS219.1×10 381 - 0.109 - 3050
S2-2-2 600 CHS219.1×10 381 51.6 0.130 0.700 5241
S2-2-3 600 CHS219.1×10 381 185.1 0.168 0.394 9085
S2-2-4 600 CHS219.1×10 381 193.3 0.170 0.384 9187
S3-1-1 450 SHS150×8 779 152.3 0.173 0.688 6536
S3-1-2 450 SHS150×8 779 157.2 0.175 0.682 6715
S3-1-3 450 SHS150×8 779 147.0 0.174 0.696 6616
S3-1-4 450 SHS150×8 779 164.1 0.177 0.672 7276
S3-1-5 450 SHS150×8 779 148.0 0.174 0.695 6974
S3-1-6 450 SHS150×8 779 - 0.150 - 3695
S3-2-1 450 SHS150×12 756 152.3 0.171 0.777 8585
S3-2-2 450 SHS150×12 756 157.2 0.173 0.771 8452
Series 3 S3-2-3 450 SHS150×12 756 147.0 0.172 0.783 8687
S3-2-4 450 SHS150×12 756 164.1 0.174 0.764 8730
S3-2-5 450 SHS150×12 756 148.0 0.172 0.782 8912
S3-2-6 450 SHS150×12 756 - 0.154 - 6456
S3-3-1 450 SHS150×12.5 446 152.3 0.149 0.655 5953
S3-3-2 450 SHS150×12.5 446 157.2 0.151 0.648 5911
S3-3-3 450 SHS150×12.5 446 147.0 0.150 0.663 6039
S3-3-4 450 SHS150×12.5 446 164.1 0.153 0.638 6409
S3-3-5 450 SHS150×12.5 446 148.0 0.150 0.662 6285
*
Loaded only on the concrete core which was 210mm in height.
(a) Circular specimen (b) Square specimen

Fig.1 Test set-up of short column
On the other hand, there were 9 slender CFST beam-column specimens of which 6 were
circular sections (CS) and 3 were square sections (SS). The steel sections include 6 hot
finished circular hollow mild steel sections, one hot finished square hollow mild steel section,
and two welded square hollow HTS sections. All the sections are Class 1 sections according
to EC 3. The column details and test results are shown in Table 3, where e0 is the load
eccentricity to apply an end moment, Neo, to the column, and L is the effective length between
the centres of the end roller supports, as shown in Figure 2.
Table 3. Configuration details and test results for CFST beam-column specimens
Steel fy fu Ea fck Ecm e0 L Ntest,u
No.
section(s) (MPa) (MPa) (GPa) (MPa) (GPa) (mm) (mm) (kN)
CS-1 CHS219.1×16 374 551 202 186 68 0 4195 6324
CS-2 CHS219.1×16 374 551 202 181 68 20 3640 4389
CS-3 CHS219.1×16 374 551 202 176 68 50 3640 3246
CS-4 CHS273×10 412 521 204 180 68 0 4195 8592
CS-5 CHS273×10 412 521 204 184 68 50 4450 5083
CS-6 CHS273×16 401 522 203 180 68 50 4450 5284
SS-1 SHS200×12.5 465 559 206 183 68 20 3640 5187
SS-2 SHS200×12* 756 825 199 176 68 20 3640 7136
SS-3 SHS200×12* 756 825 199 177 68 50 3640 4997
*
For 6mm-thick mild steel backing strip used in welded sections, fy = 325MPa, fu =
467MPa, and Ea =201GPa
e0
LVDT1 LVDT2
L
LVDT3
L/2
L/4
(a) Instrumentaion (b) Circular specimen (c) Square specimen

Fig.2 Test set-up of beam-column
In addition, the test database expands the work of Goode [13] to include 2033 test results on
CFST columns with additional test data on CFST columns with ultra-high strength concrete
reported above and elsewhere. The new test data includes steel tubes infilled with ultra-high
strength concrete with cylinder compressive strength greater than 90N/mm2. Test specimens
involving short and long CFST members subjected to compression, uniaxial bending, and bi-
axial bending, are categorized for comparison with EC 4 predictions. Tests on encased
columns, columns with stainless steel and aluminium steel sections are excluded. Tests
involving preload effect, sustained loading for creep and shrinkage studies and dynamic
loadings are not included. CFST columns with Class 4 slender sections, in which the d/t ratio
exceeds the class 3 limit stipulated in EC 4, are also excluded although they were included
originally in Goode’s database [13]. In this database, the concrete compressive cylinder
strength is in the range between 8.5N/mm2 to 243N/mm2, and the steel yield strength ranges
from 178N/mm2 to 853N/mm2. The ratio of column height over section smaller dimension is
between 0.67 and 60, and the relative slenderness 𝜆𝜆̅ ranges from 0.02 to 1.30.
3 INFLUENCE OF CONCRETE STRENGTH

The average test/prediction ratios against concrete cylinder strength are shown in Table 5.
The ratios are categorized into three groups based on concrete strength. By studying the
values which are not bracketed, the average test/EC 4 value for each type of CFST specimens
is greater than unity indicating a conservative average prediction by EC 4, both for normal and
high strength concretes.
Table 5 also provides the percentage of conservative prediction (i.e., test/EC 4 prediction ≥
1.0) which reflects the reliability of the prediction by EC 4. For high strength concrete
(50N/mm2 < fck ≤ 90N/mm2), the percentages of all circular columns and the axially loaded
rectangular cross sections are lower than those of their counterparts with normal strength
concrete (fck ≤ 50N/mm2). For ultra-high strength concrete (fck > 90N/mm2), the percentages of
all columns are lower than those of their counterparts with normal strength concrete. This
reflects the increasing complexity and severity as the concrete strength increases. To allow
for this in design, the effective compressive strength of concrete is adopted in accordance with
EC 2 [8] for high strength concrete and ultra-high strength concrete. The effective strength is
determined by multiplying a reduction factor η with the concrete strength given as:
1.0 − ( f ck − 50 ) 200 50 N/mm 2 < f ck ≤ 90 N/mm 2
η = (1)
0.8 f ck > 90 N/mm 2
For ultra-high strength concrete with fck > 90N/mm2, it is recommended that η = 0.8 should be
adopted and the increase of concrete strength due to confinement effect from steel tube
should be ignored. With the introduction of reduction factor η, the effective compressive
strengths are given in Table 4 for various high strength concrete classes. Accordingly, the
secant modulus for high and ultra-high strength concrete should be modified based on the
effective strength as
𝐸𝐸cm = 22[(𝜂𝜂 ⋅ 𝑓𝑓ck + 8)⁄10]0.3 (2)
The percentage of reduction in the secant modulus for high strength concrete is less as
compared to the compressive strength as shown in Table 4.
Table 4. Effective compressive strength and modified secant modulus of HSC

Strength classes C55/67 C60/75 C70/85 C80/95 C90/105
Effective compressive strength
54 57 63 68 72
(N/mm2)
Reduction ratio 2.5% 5.0% 10.0% 15.0% 20.0%
Modified secant modulus
38.0 38.6 39.6 40.4 41.1
(N/mm2)
Reduction ratio 0.7% 1.3% 2.8% 4.3% 5.9%
Table 5. Influence of concrete strength on test/EC 4 prediction ratios for CFST columns
Compressive cylinder strength of concrete
Type of column ≤50
≤90 N/mm2 >90 N/mm2
N/mm2
Nos. 295 130 44
Test/EC4 ≥ 66.8% 59.2% 47.7%
1 (99.3%) [66.9%]{97.7%} <100%>|100%|
Axially loaded
1.068 1.023 1.016
circular cross Av.
(1.355) [1.062]{1.383} <1.298>|1.646|
section
0.136 0.111 0.104
St. Dev.
(0.169) [0.132]{0.190} <0.139>|0.153|
Nos. 383 60 22
Test/EC4 ≥ 85.9% 68.3% 81.8%
1 (97.4%) [83.3%]{98.3} <100%>|100%|
Axially loaded
1.186 1.039 1.085
circular column Av.
(1.388) [1.075]{1.339} <1.195>|1.512|
0.246 0.110 0.095
St. Dev.
(0.267) [0.121]{0.162} <0.093>|0.157|
Nos. 240 66 46
Test/EC4 ≥ 82.1% 71.2% 69.6%
1 (98.8%) [81.8%]{98.5%} <78.3%>|91.3%|
Circular beam-
1.192 1.086 1.008
column Av.
(1.352) [1.136]{1.356} <1.121>|1.378|
0.217 0.182 0.172
St. Dev.
(0.237) [0.189]{0.216} <0.205>|0.266|
Nos. 282 63 39
Test/EC4 ≥ 80.1% 68.3% 56.4%
Axially loaded 1 (99.6%) [90.5%]{96.8%} <89.7%]|100%|
rectangular cross 1.122 1.068 1.032
Av.
section (1.287) [1.118]{1.330} <1.136>|1.321|
0.150 0.123 0.093
St. Dev.
(0.196) [0.117]{0.168} <0.099>|0.132|
Nos. 101 40 12
Test/EC4 ≥ 62.4% 70.0% 58.3%
Axially loaded 1 (94.1%) [77.5%]{95.0%} <91.7%>|100%|
rectangular 1.059 1.057 1.095
Av.
column (1.220) [1.099]{1.321} <1.212>|1.458|
0.140 0.134 0.206
St. Dev.
(0.172) [0.140]{0.177} <0.193>|0.233|
Nos. 160 23 27
Rectangular beam- Test/EC4 ≥ 73.1% 87.0% 70.4%
column 1 (98.1%) [87.0%]{100%} <85.2%>|100%|
1.107 1.099 1.044
Av.
(1.338) [1.128]{1.461} <1.089>|1.314|
0.279 0.112 0.115
St. Dev.
(0.341) [0.102]{0.148} <0.117>|0.124|
Nos. 1461 382 190
Test/EC4 ≥ 77.3% 67.0% 62.6%
1 (98.3%) [78.3%]{97.6%} <90.0%>|97.9%|
All test data 1.133 1.052 1.034
Av.
(1.339) [1.094]{1.361} <1.175>|1.440|
0.210 0.132 0.132
St. Dev.
(0.240) [0.141]{0.186} <0.165>|0.224|
Notes:
1) For the value1, (value2), [value3], {value4}, <value5> and |value6| in the table, value1
is based on the characteristic strengths of steel and concrete; (value2) is based on
design strengths; [value3] is based on characteristic strengths with reduction factor η for
concrete; {value4} is based on design strengths with reduction factor η for concrete.
<value5> is based on characteristic strengths with reduction factor η and neglect of
confinement for concrete; |value6| is based on design strengths with reduction factor η
and neglect of confinement for concrete.
2) The design partial factor is 1.5 and 1.0 for concrete and steel, respectively.
3) This table does NOT include test specimens with class 4 section as in EC4.
4) Av.=Average value; St.Dev.=Standard Deviation.
The predictions of CFST members using EC4 based on the effective strength and modified
secant modulus are given in
Table 5 where values are in the brackets [ ] and < >. It is found that the percentages of all
columns with high strength and ultra-high strength concretes are comparable with those of
their counterparts with normal strength concrete. The comparable reliability with normal
strength concrete has been achieved for high strength concrete and ultra-high strength
concrete with the introduction of concrete strength reduction factor η in Eq. (1) and by
neglecting the confinement effect for ultra-high strength concrete. Hence, the modified EC4
method can be extended to CFST columns with higher concrete strength.
5.0 ≤ 50MPa ≤ 90MPa > 90MPa

71.9% 18.8% 9.3% Characteristic Value
Design Value
4.0
Ratio Test/EC4
3.0
2.0
1.0
0.0
0 50 100 150 200 250
Concrete Cylinder Strength (N/mm2)
Fig.3 Comparison of test/EC4 prediction ratio against concrete strength
Table 5 gives the percentages for all the test data in terms of concrete strength. Overall, the
percentage decreases with increasing concrete strength (refer to values not in the brackets).
Considering the reduction factor η and ignoring the concrete confinement effect, the
percentages of Test/EC4 prediction ≥ 1.0 of high strength concrete and ultra-high strength
concrete are higher than those of their counterparts with normal strength concrete. The design
values (refer to values in ( ), { } and | |) are also provided in
Table 5 with the introduction of partial factors of 1.5 and 1.0 for concrete and steel,
respectively. When the codes specified design values are compared with the test results, the
percentages of under-prediction (i.e., test/EC4 prediction < 1.0) are less than 3%.
The Test/EC4 ratios are plotted in Figure 3, with the effective strength and modified modulus
of elasticity applied for concrete with compressive strength higher than 50N/mm2. Most data
lies close to unity, except one data showing a rather high test/EC4 value (=3.39, characteristic
value). This data was taken from tests by Assis et al.[14]. The test/prediction ratio was given
as 1.98 in Ref.[14], implying there might be problem in this test. There is also difference
between the authors’ prediction and the prediction in Ref.[14], but it remains unknown. Figure
3 shows that, in order to further extend the EC4 scope to include the ultra-high strength
concrete, more test data could be done for CFST columns with concrete compressive strength
between 120 N/mm2 to 150 N/mm2. Alternatively, the use of ultra-high strength concrete for
CFST column is made possible if a condition on the strain compatibility between steel and
concrete material is observed as described in Section 5.
4 INFLUENCE OF STEEL STRENGTH

The average test/prediction ratios against steel yield strength are shown in Table 6. All the
values are based on the reduction factor η in Eq. (1) and they neglect the confinement for
ultra-high strength concrete. The ratios are categorized into three groups based on steel
strength. It is observed that the average ratio of test/EC4 prediction for each type of column
with mild steel (fy ≤ 460N/mm2) is greater than unity. However, this is not true for columns with
high tensile steel (fy > 460N/mm2). This might be due to the lack of consideration of the strain
compatibility between steel and concrete. Thus, for CFST columns with steel strength fy >
460N/mm2, the reliability of the proposed method needs to be further investigated.
Table 6 also gives the percent of test/EC4 prediction > 1.0 and the average prediction of all
test data in terms of steel yield strength. The ratios between test and design prediction are
also provided. The average test/EC4 prediction ratio for CFST with high tensile steel is higher
than those with mild steel, but the standard deviation is higher indicating that wide scattering
of results is observed for specimens with high tensile steel. When the codes’ design values
are compared with the test results, the percentage of under-prediction (i.e., test/EC4 prediction
< 1.0) is less than 5%, although the average ratio is higher than those with mild steel.
Table 6. Influence of steel strength on test/EC4 prediction ratios for CFST columns
Yield strength of steel
Types of column
≤460N/mm2 ≤550N/mm2 >550N/mm2
Nos. 450 5 14
Test/EC4 ≥ 71.6% 40.0% 28.6%
Axially loaded circular
1 (99.6%) (40.0%) (100%)
cross section
Av. 1.093 (1.399) 0.922 (1.133) 0.975 (1.180)
St. Dev. 0.150 (0.189) 0.200 (0.215) 0.068 (0.081)
Nos. 414 38 13
Test/EC4 ≥ 85.7% 89.5% 92.3%
Axially loaded circular
1 (97.6%) (100%) (92.3%)
column
Av. 1.152 (1.378) 1.399 (1.532) 1.160 (1.270)
St. Dev. 0.167 (0.210) 0.544 (0.526) 0.112 (0.122)
Nos. 346 6 -
Test/EC4 ≥ 82.4% 33.3%
-
Circular beam-column 1 (98.0%) (83.3%)
Av. 1.175 (1.356) 1.032 (1.367) -
St. Dev. 0.211 (0.236) 0.213 (0.268) -
Nos. 308 21 55
Test/EC4 ≥ 84.7% 100% 65.5%
Axially loaded rectangular
1 (99.0%) (100%) (100%)
cross section
Av. 1.135 (1.324) 1.132 (1.310) 1.048 (1.146)
St. Dev. 0.147 (0.189) 0.071 (0.096) 0.089 (0.117)
Nos. 145 8 -
Test/EC4 ≥ 67.6% 87.5%
Axially loaded rectangular -
1 (95.2%) (87.5%)
column
Av. 1.078 (1.262) 1.152 (1.328) -
St. Dev. 0.141 (0.181) 0.267 (0.331) -
Rectangular beam-column Nos. 187 8 15
Test/EC4 ≥ 73.8% 87.5% 100%
1 (98.4%) (100%) (100%)
Av. 1.099 (1.338) 1.061 (1.385) 1.221 (1.459)
St. Dev. 0.255 (0.319) 0.071 (0.119) 0.219 (0.168)
Nos. 1850 86 97
Test/EC4 ≥ 78.9% 84.9% 69.1%
All test data 1 (98.3%) (94.2%) (99.0%)
Av. 1.128 (1.357) 1.226 (1.410) 1.079 (1.216)
St. Dev. 0.181 (0.222) 0.409 (0.395) 0.141 (0.165)
The Test/EC4 ratios are also plotted in Figure 4. It is observed that more than 90% test data
are from CFST columns with mild steels. The test data is insufficient to establish the validity
of using the high tensile steels according to EC 4. The following section provides additional
guideline to limit the use of high tensile steel by selecting matching grades of steel and
concrete materials for composite construction.
5.0
≤ 460MPa ≤ 550MPa Characteristic Value
91.0% 4.2% Design Value
4.0
> 550MPa
Ratio Test/EC4
3.0 4.8%
2.0
1.0
0.0
150 250 350 450 550 650 750 850 950
Steel Yield Strength (N/mm2)
Fig.4 Comparison of test/EC4 prediction ratio against steel strength
5 STRAIN COMPATIBILITY BETWEEN STEEL AND CONCRETE

For high strength concrete filled steel tubular columns subjected to compression, it is
necessary to ensure that yielding of the steel section occurs before the concrete core reaches
its maximum stress. Otherwise, the full plastic resistance of the composite section cannot be
achieved due to brittle failure of high strength concrete after reaching the maximum stress.
Hence, the selections of steel grade and concrete class have to ensure that the yield strain of
steel is smaller than the compressive strain of concrete at the peak stress. The yield strain of
steel and the strain of concrete at peak stress may be calculated as [8&9]:
Steel yield strain: ε y = f y Ea (3)

Concrete strain at peak stress (‰): ε c1 = 0.7 f cm0.31 (4)
where 𝑓𝑓cm = 𝑓𝑓ck + 8 is the mean compressive strength of concrete at 28 days, in N/mm2. Ea
= 210000 N/mm2 is the elastic modulus of steel tube. It is noted that the calculation for the
strain of concrete at peak stress ignores the confinement effect from the steel tubes. Steel
with yield strength greater than 550N/mm2 may be used provided that a more accurate
assessment on the concrete strain at peak stress, considering the tri-axial confinement effect
from the steel tube, is carried out.
Table 7. Compatibility between steel and concrete materials for CFST columns
S235 S275 S355 S420 S460 S500 S550

C25/30 √ √ √ √ × × ×
C30/37 √ √ √ √ × × ×
C35/45 √ √ √ √ √ × ×
C40/50 √ √ √ √ √ × ×
C45/55 √ √ √ √ √ √ ×
C50/60 √ √ √ √ √ √ ×
C55/67 √ √ √ √ √ √ ×
C60/75 √ √ √ √ √ √ ×
C70/85 √ √ √ √ √ √ √
C80/95 √ √ √ √ √ √ √
C90/105 √ √ √ √ √ √ √
C110/- √ √ √ √ √ √ √
C130/- √ √ √ √ √ √ √
C150/- √ √ √ √ √ √ √
C170/- √ √ √ √ √ √ √
C190/- √ √ √ √ √ √ √
Notes: “√” indicates compatible materials and “×” is not recommended.
Table 7 gives the recommendation on the matching grades of steel and concrete suitable for
use in CFST columns. This is based on the condition εy < εcl and the experimental observations
on the new test data presented in [1-2]. It is recommended that the steel tubular sections up
to Grade S550 may be used with concrete class up to C190, although test evidence by Liew
et al. [3] shows that the strain of the confined concrete at peak stress of CFST is much higher
than the concrete without any lateral confinement. Alternatively, the maximum steel strength
can be determined according to the concrete characteristic strength with strength class up to
C190 using the following expression:
= (
f y min 0.7 Ea ( f ck + 8 )
0.31
, 550 ) (5)
6 CASE STUDY
The M-N interaction curves for a CFST column with section shown in Figure 5 are determined
according to the proposed design method given in above.
N
Npl,Rd A
500
Npm,Rd C
25
500
Npm,Rd D
2
B M
Mpl,Rd Mmax,Rd
Fig.5 CFST column for case study and illustration for M-N interaction curve
For the CFST columns used in high-rise buildings, they are usually subject to combined effects
of high compression force and bending moment. These effects could be effectively resisted
by increasing the concrete and steel strengths as shown in Figure 6. The combined use of
C190 concrete and S550 steel shows superior performance on improving the overall
resistance and thus it is more economical in terms of reducing steel and concrete materials,
using less weld consumable, reducing the cost of fire protection and fabrication, reducing
construction time and increasing usable floor area.
50000
45000
40000
35000
Axial Force (kN)
30000
25000
20000
15000
10000 C50/60+S355
5000 C90/105+S460
C190/-+S550
0
0 1000 2000 3000 4000 5000 6000 7000
Bending Moment (kN.m)
Fig.6 Comparison between CFSTs with varying both concrete and steel strengths
7 CONCLUSIONS
Based on calibration with 2033 test data, the current EC 4 method can be safely extended to
the design of concrete filled steel tubular (CFST) members with steel strength up to 550N/mm2
and concrete compressive cylinder strength up to 190N/mm2, with the following modifications
and restrictions:
• This design guide is meant for CFTS members with at least Class 3 steel section. Class
4 steel section should not be used.
• Matching grades of steel and concrete materials should be used. Table 7 provides a
guide to select the grade of steel and class of concrete for the design of CFST
members to avoid the crushing of the core concrete before yielding of steel section.
• A strength reduction factor should be applied for high strength concrete with cylinder
strength greater than 50N/mm2 but less than 90N/mm2. Accordingly, the secant
modulus of concrete should also be modified.
• For ultra-high strength concrete with compressive cylinder strength higher than
90N/mm2 but less than 190N/mm2, a conservative approach is to adopt the concrete
strength reduction factor of 0.8 and further ignore the concrete confinement effect.
Steel with yield strength greater than 550N/mm2 may be used with ultra-high strength
concrete provided that a more accurate assessment on the concrete strain at peak
stress, considering the tri-axial confinement effect from the steel tube, is carried out.
• The key parameters for numerical analysis of ultra-high strength concrete under lateral
confinements were obtained from tests on UHSC under various confinement pressure.
These parameters were recommended so that proper analysis and design can be
carried on steel tubular members infilled with UHSC for them to be used economically
in high-rise buildings.
The design recommendations proposed in this paper will endow structural engineers with the
confidence to use high strength materials in a safe and economic manner to design high-rise
buildings. The authors aware that there is a size effect needs to be investigated in view of the
smaller size specimens tested in the laboratory relative to those use in practice. Although the
size effect has been partially accounted for in the partial factors in design codes, more tests
on larger size columns are needed to gain confident in designing composite columns with high
strength materials.
REFERENCES
[1] Liew J.Y.R., Xiong D.X. (2012). Ultra-high strength concrete filled composite columns for
multi-storey building construction. Advances in Structural Engineering, 15(9), 1487-1503.
[2] Liew J.Y.R., Xiong M.X., Xiong D.X. (2014). Design of high strength concrete filled tubular
columns for tall buildings. Journal of high-rise buildings, 3(3), 1-7.
[3] Liew, J.Y.R. (2004). Buildable design of multi-storey and large span steel structures,
Journal of Steel Structures, 4(2). 53-70.
[4] EN 1994-1-1 (2004). Eurocode 4: Design of composite steel and concrete structures –
Part 1-1: General rules and rules for buildings.
[5] AISC 360-10 (2010). Specification for structural steel buildings. American Institute of
Steel Construction (AISC), Chicago, USA.
[6] GB 50936 (2014). Technical code for concrete-filled steel tubular structures. Ministry of
Housing and Urban-Rural Development of the People’s Republic of China, in Chinese. 6
[7] AIJ (1997). Recommendations for design and construction of concrete filled steel tubular
structures, Architectural Institute of Japan, Japan.
[8] EN 1992-1-1 (2004). Eurocode 2: Design of concrete structures – Part 1-1: General rules
and rules for buildings.
[9] EN 1993-1-1 (2005). Eurocode 3 - Design of steel structures – Part 1-1: General rules
and rules for buildings.
[10] EN 1993-1-12 (2007). Eurocode 3 – Design of steel structures – Part 1-12: Additional
rules for the extension of EN 1993 up to steel grades S700.
[11] Liew J.Y.R., Xiong M.X. (2015). Design guide for concrete filled tubular members with
high strength materials – An extension of Eurocode 4 method to C90/105 concrete and
S550 steel. Research Publishing Singapore.
[12] EN 1994-1-2 (2005). Eurocode 4: Design of composite steel and concrete structures –
Part 1-2: General rules – Structural fire design.
[13] Goode C.D. (2008). Composite columns – 1819 tests concrete filled steel tube columns
compared with Eurocode 4. The Structural Engineer, 86(16), 19 August.
[14] Assi I.M., Qudeimat E.M., Hunaiti Y.M. (2003) Ultimate moment capacity of formed and
lightweight aggregate concrete filled steel tubes. Steel and Composite Structures,
3(3):199-212.
DEVELOPMENT OF A NEW PUSH TEST FOR EUROCODE 4
Stephen J. Hicks
General Manager Structural Systems
Heavy Engineering Research Association
Auckland, New Zealand
stephen.hicks@hera.org.nz
Adrian Ciutina
Professor, Politehnica University Timişoara
Department of Overland Communication Ways, Foundation and Cadastral Survey
Timişoara, România
adrian.ciutina@upt.ro
Christoph Odenbreit
ArcelorMittal Chair of Steel and Façade Engineering
University of Luxembourg, Laboratory of Steel and Composite Structure
Luxembourg-Kirchberg, Luxembourg
christoph.odenbreit@uni.lu
ABSTRACT
The standard push test specimen in the current version of Eurocode 4 does not provide any
information on what modifications should be made when profiled steel sheeting is introduced.
Whilst the lack of information was intended to encourage innovation, this has sometimes lead to
test results implying that stud connectors possess a lower resistance and ductility than
assumed in current Standards. From full-scale beam and companion push tests in Europe, it
has been shown that modifying the push test through the introduction of a normal force to the
face of the test slabs provides comparable load-slip performance to that encountered within a
beam. Within the work programme for developing the second generation of Eurocodes, this
paper presents part of the background to Sub-task 1 of SC4.T3, whose aim is to develop an
improved push test for Eurocode 4 when stud connectors are welded within profiled steel
sheeting.
INTRODUCTION
STANDARD PUSH TEST

The forces that occur in the concrete flange of a composite beam are shown schematically in
Fig. 1(a). Should a distributed load q be introduced, the vertical shear forces are affected such
that ∆V = Vl – Vr = q × ∆l. If the distributed load q acts on the concrete flange, as well as the
longitudinal shear force Fℓ, a compression force q × ∆l exists at the interface between the
concrete and the top flange of the steel beam.
(a)
(c)
(b)
Fig. 1 - (a) Internal forces within: a composite beam; (b) a push test (Roik and Hanswille, 1987);
and (c) determination of characteristic resistance and slip capacity from push test load-slip
curve according to Eurocode 4.
The load-slip performance of shear connectors has been historically established from small-
scale push specimens of the type shown in Fig. 1(b). By applying a load to the top end of the
steel-section, the load-slip behaviour of the connectors can be determined. This type of
specimen is known as a ‘push test specimen’ in Eurocode 4 (EN 1994-1-1, 2004) and, apart
from slight variations in its geometry, has hardly changed since its inception in the early 1930’s
(Roš, 1934). The internal forces in the push specimen are shown in Fig. 1(b) to enable direct
comparisons to be made with those in a composite beam. The forces Fℓ are transferred through
the concrete in a similar way as a composite beam (N.B. the recess at the bottom of the slab is
optional in the standard test in Eurocode 4). The moment P × e, resulting from the eccentric
load introduction, causes tension in the studs and compression at the interface between the
concrete and the flange of the steel section. In the Eurocode 4 standard test, the magnitude of
the tension forces in the studs Ften is therefore affected by frictional forces developing at the
interface between the test slabs and the strong floor µ × P (where µ is the friction coefficient); if
these frictional forces are eliminated, Ften increases, which has been shown to reduce the shear
resistance of the studs by approximately 30% (Hicks and McConnel, 1997).
EVALUATION OF THE RESULTS FROM PUSH TESTS
Characteristic resistance PRk

According to current requirements of Eurocode 4, if 3 nominally identical push tests are carried
out, and the deviation of any individual result from the mean value does not exceed 10%, the
characteristic resistance of a shear connector PRk is defined as 0.9 times the minimum failure
load per stud (see Fig. 1(c)). Letting Pe,m be the mean measured resistance per connector from
three nominally identical push tests, and Pe,min,n=3 be the lowest of the three measured
resistances per connector, this requirement can be written as follows:
PRk = 0.9Pe,min,n = 3 provided that Pe,e − Pe,m ≤ 0.1Pe,m (1)
where Pe,e is each extreme (maximum or minimum) measured resistance.

If the scatter of results exceeds the 10% limit (i.e. Pe,e – Pe,m> 0.1Pe,m), the test evaluation
should be carried out according to EN 1990, Annex D (2005).
Equation (1) is effectively based on the EN 1990 provisions for the evaluation of a characteristic
value from a small number of test results, when the coefficient of variation Vr is known from a
significant number of previous tests (i.e. ‘VX known’). Although there may be some debate on
the exact value for the coefficient of variation from historical push tests, it can be deduced from
EN 1990 that, for a set of three results with extreme measured resistances Pe,e less than 10%
of the mean value Pe,m, the method in Eurocode 4 implies that Vr = 11% (Johnson, 2012).
It would be unwise to assume that Vr = 11% is always satisfied since, from a recent
investigation undertaken in the UK (Smith and Couchman, 2010), 3 of the 9 groups of nominally
identical push tests did not satisfy the 10% limit for the scatter of results. Furthermore, if three
nominally identical tests are not undertaken, it would be impossible to verify the Eurocode 4
requirement for the scatter of the results. Therefore, hypothetically, if only one push test was
undertaken, the characteristic resistance could be at least 20% lower than expected.
Characteristic slip capacity δuk

The ductility of a shear connector is measured by the slip capacity δu, which is defined in
Eurocode 4, B.2.5 as the slip corresponding to the point where the characteristic resistance of
the connector intersects the falling branch of the load-slip curve (see Fig. 1(c)). The
characteristic slip capacity δuk is taken as 0.9 times the minimum test value of δu,min, such that
δ uk = 0.9δ u ,min,n = 3 . Alternatively, the characteristic properties of a shear connector can be
determined by a statistical evaluation of all of the results according to EN 1990. It is therefore
implied by Eurocode 4, that the coefficient of variation for the slip capacity is Vr = 11% when
δ uk = 0.9δ u min,n =3 , but no information is given to the designer on when it is appropriate to use EN
1990.
PUSH TESTS ON HEADED STUDS WELDED WITHIN TRAPEZOIDAL PROFILED STEEL
SHEETING TRANSVERSE TO A BEAM
Questions have arisen on the appropriateness of using the reduction factors contained within
Eurocode 4, owing to the fact that the failure mechanisms of studs in trapezoidal profiled steel
sheeting are quite different to those experienced in solid slabs. For example, when push tests
are conducted on studs welded favourably or centrally within the ribs of a sheet (see Fig. 2), a
typical failure mode known as concrete cone pull-out occurs (Hawkins and Mitchell,1984; Lloyd
and Wright, 1990). In this case, the whole cone together with stud rotates and is pulled out of
the slab, carrying with it a wedged-shaped pyramidal portion of concrete (see Fig. 2(d)); in
these cases the axial tension in the stud can be significant, which has been measured in some
special test specimens to be in the order of 30% of the longitudinal shear resistance (van der
Sanden, 1996). Due to the tension and rotation of the stud, the concrete slab can separate from
the profiled steel sheeting relatively early in push tests, which brings into question whether it is
entirely appropriate to neglect the compression at the interface between the concrete and the
steel section that would occur in a composite beam subjected to a uniformly distributed load
together with its self-weight (see Fig. 1(a)).
Fig. 2 - Dimensions of profiled steel sheeting and studs in the (a) Central (b) Favourable (c)
Unfavourable position
The Eurocode 4 rules for partial shear connection are based on two independent studies
(Johnson and Molenstra, 1991; Aribert, 1990). These studies assumed that, in solid concrete
slabs and composite slabs using profiled steel sheets prevalent in the 1980’s, the characteristic
slip capacity of 19 mm diameter studs was approximately δuk = 6 mm (see Fig. 1(c)). The rules
for partial shear connection in Eurocode 4 were limited to situations where the required slip did
not exceed 6 mm. Studs were deemed to be ‘ductile’ in those situations.
Push tests in Australia (Patrick, 2004) suggested that studs welded within the ribs of modern
trapezoidal profiled steel sheeting possess lower resistance and ductility than assumed in
Eurocode 4. To investigate whether the claims were affected by the thinner (0.75 mm), high
strength steel decking (550 MPa) that is common in Australasia, 24 push tests with trapezoidal
decking were undertaken at Imperial College London in 2003 (Hicks, 2007a). Each test slab
had three concrete ribs with studs through-deck welded in the two ribs to the top of the
specimens. However, it was found that the specimens with two studs per rib in the favourable
position (nr = 2F) gave consistently lower resistances than those with one stud per rib (nr = 1F),
irrespective of the stud layout. From an inspection of the specimens it was clear that the results
had been affected by an artificial ‘back-breaking’ failure mode, which was characterised by the
last studded rib at the top of the specimen rotating, causing a horizontal crack to appear across
the full width of the test slab (see Fig. 3(a)). It was believed that this failure was caused by the
couple of internal forces in the last studded rib at the top of the specimen (see Fig. 3(b)). To
remedy this situation, it appeared that the provision of an unstudded rib at the top of specimen
eliminated this artificial back-breaking failure mode.
(b)
(a)
Fig. 3 - (a) Back-breaking failure experienced in Imperial College London tests (b) Internal forces
causing rotation of the last studded rib at the top of the specimen
An obvious question regarding stud connectors in push specimens is whether their behaviour is
comparable to that which would occur in a full-scale beam. In an attempt to address this
question, tests on a 10.0, 5.0 and 11.4 m span composite beam, together with 12 companion
push tests were undertaken in the UK between 2004 and 2008 (Hicks, 2007b; Hicks and Smith,
2014). To enable the internal forces to be evaluated, the steel beams were instrumented with
strain gauges on the top and bottom flange at cross-sections corresponding to the shear
connector positions. As well as determining the build-up of axial force, the internal bending
moments were evaluated to check the reliability of the strain gauge readings. The slip
distribution at the shear connection was established from horizontally mounted transducers,
which monitored the relative displacement between bars cast in the concrete behind the shear
connector positions and the top flange of the beam. The internal load–slip curves for the shear
connectors were therefore evaluated by plotting the change in axial force at each cross-section
against the corresponding slip.
The load-slip curves for studs with the lowest failure load in the beam tests are compared to
their companion push tests in Fig. 4. As can be seen from these plots, there is no similarity in
performance of studs for these two types of specimen. In Fig. 4, the slips measured in the push
tests are well below the levels achieved in the beam and, if considered in isolation, would
suggest that the studs should not be taken to be ‘ductile’ (which has adverse implications for
partial shear connection design). Furthermore, the tests showed that the resistance per stud for
nr = 3F was no better than nr = 2F, thereby indicating that the design equations in BS 5950-3.1
(1990) and ANSI/AISC 360-16 (2016) were unconservative by up to 45%. As a direct result of
this work, an amendment was made to BS 5950-3.1 (2010).
(a) (b)
(c)
Fig. 4 – Load-slip curve for through-deck welded studs in beam test compared to standard push
test for: (a) one stud per rib nr = 1F; (b) two studs per rib nr = 2F; and (c) nr = 3F three studs
per rib in the favourable position, respectively.
DEVELOPMENT OF AN IMPROVED PUSH TEST FOR HEADED STUDS WELDED WITHIN

TRAPEZOIDAL PROFILED SHEETING
NORTH AMERICAN INVESTIGATIONS

It was considered that the reason for the poor performance of studs welded within trapezoidal
sheeting in push tests was due to the absence of the compression force at the interface
between the concrete and the flange of the steel section, which exists in real composite beams
from the floor loading (see Fig. 1(a)). Easterling et al. (1993) can be credited for attempting to
remedy the problem of poor load-slip performance by modifying the standard push test through
the introduction of a normal force to the face of the test slabs. In these North American
investigations normal forces equivalent to between 0 and 20% of the vertical load, were applied
directly over the centre-line of the steel section. In total, 234 push tests and 4 full-scale
composite beam tests were undertaken. It was concluded that “only a 5% normal load in the
push test causes a lot of variability in stud strength” (Rambo-Roddenberry, 1994). It was also
deemed that the results from push tests with a normal force equivalent to 10% of the vertical
load compared favourably with the performance of the four 9.0 m span companion beam tests.
AUSTRALIAN INVESTIGATIONS
Bradford et al. (2006) found that the standard push test shown in Fig. 1(b) was inappropriate for
composite slabs using trapezoidal decking. It was found that the specimen tended to twist
(similar to the ‘back-breaking failure’ shown in Fig. 3), and fail in a brittle and premature way.
The standard push test was modified to a single-sided arrangement and tested in the horizontal
position. A total of 10 push specimens were tested, where normal forces equivalent to between
0 and 10% of the longitudinal load was applied along the edges of the specimen at an
eccentricity of 600 mm, thereby applying a hogging moment over the centre-line of the steel
section. The eccentric loading was chosen to introduce the beneficial effect on the stud
connector resistance from the presence of this hogging moment (from a compression force
being applied to the base of the stud), and to reflect the loading conditions on two 8.05 m span
companion composite beam tests (Ranzi et al., 2009).
EUROPEAN INVESTIGATIONS
Taking inspiration from the earlier North American and Australian modifications a new test was
devised in the UK, which was thought to better reflect the conditions that exist in a real beam.
As good quality in situ load-slip data existed from three full-scale composite beam tests, the
new test could be calibrated against this performance (Hicks, 2007b; Hicks and Smith, 2014).
Also, rather than developing a completely new specimen, it was proposed to modify the
standard specimen given in Eurocode 4, in the interests of developing a relationship with
historical push test resistances. Due to the possibility of different friction coefficients at the base
of the test slabs affecting the repeatability of the tests, it was decided to develop a self-
contained rig that could be disassembled and erected in different locations without the need of
a strong floor.
The improved push rig is shown in Fig. 5(a). The loading system consists of vertical jacks
applying the longitudinal shear force, accompanied with horizontal jacks applying a uniform
normal force to the face of the test slabs. From testing a total of 14 nominally identical
specimens that had been constructed from a single concrete mix using the same details that
had been provided in the companion beam tests (with nr = 1F and nr = 2F), the following levels
of normal force were applied (taken as a proportion of the longitudinal force): 0; 4%; 8%; 12%;
and 16%. The load-slip curves for these tests are presented in Fig. 5(b) and Fig. 5(c), which are
compared with those measured in the beam tests. It was considered that the results with a 12%
lateral load provided the closest match with the load-slip behaviour from the beam tests.
A further two tests with a lateral load of 12% were conducted to evaluate the characteristic
values for nr = 1F and nr = 2F. By employing the test evaluation procedure given in Eurocode 4,
the improved push rig delivered characteristic resistance and slip capacity values that were
comparable to those achieved in the companion the beam tests (Hicks and Smith, 2014). The
improved push rig was subsequently used to investigate the effect of a number of key variables
on the load-slip arrangement of headed stud connectors by Smith and Couchman (2010) whom
undertook 27 push tests.
The development work described above was extended through the major RFCS DISCCO
research project (Aggelopoulos et al., 2016), which included both beam and push tests. A total
of 64 push specimens were tested, where normal forces equivalent to between 0 and 16% of
the longitudinal load were applied directly over the steel beam (concentrically) or through the
slab at different eccentricities to the steel beam. The eccentric loading was chosen to introduce
the beneficial effect on the headed stud connector resistance from the presence of this hogging
moment and to reflect the loading conditions in composite floors. Also, as well as considering
that the normal force increased in proportion to the applied longitudinal shear force (as would
occur in a beam), the maximum required normal force was established and fixed before the
longitudinal shear force was applied (i.e. a constant normal force).
(b)
(a)
(c)
Fig. 5 - (a) Improved push test rig developed in the UK; Comparison of load-slip behaviour for the
new improved test with that measured in beam tests for: (b) nr = 1F; and (c) nr = 2F)
In order to define reasonable values for the normal force, a typical floor bay within a building
was considered (Nellinger et al., 2017), which consisted of an internal simply-supported
composite beam which, in turn, supported a double-span composite slab. To reflect current
practice, steel grades of S235 and S355 were considered, together with a concrete grade of
C30/37. Furthermore, two different composite slabs with a deck height of 58 and 80 mm were
considered in the study. An imposed load of qk= 3,5 kN/m² was considered in all cases. Using
the current Eurocode 4 design provisions (i.e. rigid-plastic material was considered), the spans
of the beams and slab were varied to determine the range of normal forces v versus
longitudinal shear forces T that would be expected in practice. The results from this study are
presented graphically in Fig. 6(a), which showed that the degree of transverse loading v/T
ranged from 5 to 10%.
(a)
(b)
Fig. 6 - (a) Degree of transverse loading v/T (b) Improved push test developed within the RFCS
DISCCO project
In some of the subsequent push tests where no normal force was applied to the face of the test
slabs, a ‘torsional failure mode’ was reported which was characterised by the last studded rib at
the top of the specimen rotating, causing a horizontal crack to appear across the full width of
the test slab (see Fig. 3). In subsequent push tests that were carried out in the DISCCO project,
it was considered important to include a normal force in the push tests, in order to suppress this
artificial failure mode.
Through the comprehensive test programme within the DISCCO project, similar findings to that
reported in the earlier investigations were made, in that the resistance and ductility of stud
connectors welded within the ribs of trapezoidal profiled steel sheeting were strongly affected
by the magnitude of the normal force applied to the face of the test slabs (Nellinger et al.,
2017). The improved push test arrangement that was developed by the University of
Luxembourg within this RFCS project is shown in Fig. 6(b).
DEVELOPMENT OF A NEW STANDARD PUSH TEST FOR THE SECOND GENERATION OF

EUROCODE 4
Due to the favourable comparisons with full-scale composite beam tests, as well as similar
levels of lateral force being found to be appropriate in the earlier North American and Australian
research programmes, it is recommended that besides the standard push test procedure for
shear connectors, a distinct push test should be used for stud connectors welded within the ribs
of trapezoidal profiled steel sheeting to ensure that this popular form of construction is
supported by Eurocode 4. The proposed test arrangement that is currently being considered is
presented in Fig. 7.
1 cover 15 mm 4 reinforcement:
ribbed bars with 450 ≤ fsk ≤ 550 MPa
2 bedded in mortar, gypsum or similar
steel section: HE 260 B or 254 x 254 x 89 kg. UC
3 recess optional
Fig. 7 – Proposed test specimen for headed stud connectors welded within trapezoidal profiled
sheeting
To avoid the premature back-breaking failure mode observed previously (see Fig. 3), and
provide comparable load-slip performance of shear connectors in full-scale beams, it is
proposed to provide a lateral force normal to the face of the test slabs. From the above review
of previous international push tests, the value of the lateral force has ranged between 5 to 12%
of the longitudinal shear force. As a compromise, it is currently proposed that a normal force not
greater than 10% of the longitudinal shear force should be applied in the improved push test.
This upper value is consistent with that used in the North American test programme whose
results were used to form the basis of the design rules for stud connectors in the AISC
Specification (ANSI/AISC 360-16, 2016).
CONCLUSIONS
From the considerable variations that have been observed in the behaviour of stud connectors
in terms of resistance and ductility, it would appear that a clear case exists for the need to
standardize the push specimen test regime when trapezoidal profiled steel sheeting is
employed. From a review of international experiments (comprising 12 full-scale composite
beams and 361 push tests), it was found that the application of a normal force to the face of the
slabs in push specimens eliminated an unrealistic premature back-breaking failure mode.
Moreover, the application of a normal force also provided comparable characteristic resistances
and slip capacities to that achieved in companion full-scale beams. From a consideration of this
wealth of international data, a push test specimen for headed stud connectors welded within
trapezoidal profiled sheeting is proposed for the second generation of Eurocode 4. Given that
the value of the maximum normal force proposed is similar to that used to form the basis for the
provisions in the 2016 AISC Specification, it is hoped that the new Eurocode 4 test will
encourage international harmonization for push tests, which will lead to more consistent
experimental data and improved design equations in the future.
ACKNOWLEDGMENTS
Whilst the present authors are members of the Eurocode 4 Project Team SC4.T3 ‘Revised
rules for shear connection in the presence of modern forms of profiled sheeting’, any views
expressed in this paper do not necessarily reflect those of the entire Project Team. The draft
revisions to Eurocode 4 are still at a draft stage and may be subject to change. Any views
expressed in this paper may not necessarily reflect those of sub-committee responsible for
Eurocode 4 (CEN/TC250/SC4).
REFERENCES
Aggelopoulos E.S., Lawson, R.M, Kuhlmann, U., Eggert, F., Odenbreit, C., Nellinger, S., Lam,
D., Dai, X, Sheehan, T. and Obiala, R. (2016). “Development of Improved Shear Connection
Rules in Composite Beams (DISCCo)”, Draft Final Report, RFSR CT 2012-00030.
ANSI/AISC 360-16. (2016) Specification for Structural Steel Buildings. American Institute of
Steel Construction, Chicago, IL.
Aribert J-M. (1990). “Dimensionnement de poutres mixtes en connection partielle”. Mixed
Structures Including New Materials. IABSE Symposium, Brussels, International Association for
Bridge and Structural Engineering. 60. pp. 215–220
Bradford, M.A., Filonov, A., Hogan, T.J., Ranzi, G. and Uy, B. (2006). “Strength and ductility of
shear connection in composite T-beams”. 8th International Conference on Steel, Space &
Composite Structures, Kuala Lumpar, Malaysia, 15-17 May, pp. 15-26
BS 5950-3.1 (1990). Structural use of steelwork in buildings: Part 3: Section 3.1: Code of
practice for design of simple and continuous composite beams, British Standards Institution,
London.
BS 5950-3.1+A1 (2010) Structural use of steelwork in buildings: Part 3: Section 3.1: Code of
practice for design of simple and continuous composite beams, British Standards Institution,
London.
Smith, A.L. and Couchman, G.H. (2010). “Strength and ductility of headed stud connectors in
profiled steel sheeting”. Journal of Constructional Steel Research, 66. pp. 748-754
Easterling, W.S., Gibbings, D.R. and Murray, T.M. (1993). “Strength of Shear Studs in Steel
Deck on Composite Beams and Joists” Engineering Journal, American Institute of Steel
Construction, Second Quarter. pp. 44-55
EN 1990 (2005). Eurocode: Basis of structural design. European Committee for
Standardization, Brussels
EN 1994-1-1 (2004). Eurocode 4: Design of composite steel and concrete structures - Part 1-1:
General rules and rules for buildings, European Committee for Standardization, Brussels.
Hawkins, N. M. and Mitchell, D. (1984). “Seismic response of composite shear connections”. J.
Struct. Eng. 110, 9. pp. 2120–2136.
Hicks, S.J. and McConnel, R.E. (1997) “The shear resistance of headed studs used with
profiled steel sheeting”. In Composite Construction in Steel and Concrete III, Buckner, C.D.,
Shahrooz, B.M, ASCE: New York. pp. 325-338
Hicks S. (2007a) “Resistance and ductility of shear connection: full-scale beam and push tests”.
Sixth International Conference on Steel and Aluminium Structures (ICSAS ‘07), Beale, RG (ed).
Oxford Brookes University, UK, pp. 613-620
Hicks, S. (2007b) “Strength and ductility of headed stud connectors welded in modern profiled
steel sheeting”. Structural Engineer. 85(10). pp. 32-38.
Hicks, S.J. and Smith, A.L. (2014). “Stud Shear Connectors in Composite Beams that Support
Slabs with Profiled Steel Sheeting”. Structural Engineering International. 24(2). pp. 246-253
Johnson, R.P. (2012). Designers’ Guide to Eurocode 4: Design of Composite Steel and
Concrete Structures EN 1994-1-1 (Second Edition), London, Thomas Telford.
Johnson, R.P. and Molenstra N. (1991) “Partial shear connection in composite beams for
buildings”. Proc. Inst. Civ. Eng. 91(Part 2). pp. 679–704.
Lloyd, R. M. and Wright, H. D. (1990). “Shear connection between composite slabs and steel
beams”. Journal of Constructional Steel Research. 15. pp. 255–285.
Nellinger, S., Odenbreit, C., Renata Obiala, R. and Lawson, R.M. (2017). “Influence of
transverse loading onto push-out tests with deep steel decking”, Journal of Constructional Steel
Research, 128. pp. 335-353
Patrick, M. (2004). Composite beam shear connection design and detailing practices for
Australian steel decks, University of Western Sydney, Report CCTR-CBSC-001-04, Sydney.
Rambo-Roddenberry, M. (1994). “Behavior and Strength of Welded Stud Shear Connectors”,
PhD Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, VA.
Ranzi, G, Bradford, M.A., Ansouriana, P., Filonov, A., Rasmussen, K.J.R., Hogan, T.J., Uy, B.
(2009). “Full-scale tests on composite steel–concrete beams with steel trapezoidal decking”.
Journal of Constructional Steel Research, 65(7). pp. 1490-1506
Roik, K. and Hanswille, G. (1987) “Zur Dauerfestigkeit von Kopfbolzendübeln bei
Verbundträgern”. Der Bauingenieur. 62. pp. 273–285.
Roš, M. (1934) “Les constructions acier-béton système «Alpha»”. L’Ossature Métallique. 3(4).
pp. 195-208
Smith, A.L. and Couchman, G.H. (2010). “Strength and ductility of headed stud connectors in
profiled steel sheeting”. Journal of Constructional Steel Research. 66. pp. 748-754.
van der Sanden, P.G.F.J. (1996). The behaviour of a headed stud connection in a ‘new’ push
test including a ribbed slab, Tests: Main report. BKO Report 95-15, Eindhoven.
DEVELOPMENT OF EUROPEAN DESIGN GUIDANCE FOR STEEL-
CONCRETE (SC) STRUCTURES IN NUCLEAR POWER PLANT
Bassam A. Burgan
Director
The Steel Construction Institute (SCI)
Ascot, UK
b.burgan@steel-sci.com
Eleftherios S. Aggelopoulos
Senior Engineer
The Steel Construction Institute (SCI)
Ascot, UK
e.aggelopoulos@steel-sci.com
ABSTRACT
New European design guidance for SC structures has been developed following completion
of a major research project. The project conducted full and large scale tests on structural
members, connections, members subjected to thermal conditions representative of loss of
coolant accident (LOCA) and members in fire. It also addressed design issues relating to the
non-composite construction stage.
The results of the tests have been used in the validation of numerical models. Extensive finite
element parametric studies were also performed to study configurations beyond those tested
and a large database of test results from international sources has been assimilated.
The project led to the development of European design guidance for SC structures following
the principles of the Eurocodes. This paper discusses the guidance developed for the
structural design of SC members in ambient conditions.
NOTATION
total area of the two steel plates ( );
area of the concrete;
area of the compression steel plate ;
area of the tension steel plate ;
total cross-sectional area of the shear reinforcement
width of the cross-section;
effective depth of SC section ( /2 )
Young’s modulus of steel;
elastic modulus of concrete;
characteristic compressive cylinder strength of concrete at 28 days;
yield strength of the steel plates;
characteristic yield strength of the shear reinforcement
shear modulus of elasticity of steel
total depth of SC section
second moment of area of the SC member;
second moment of area of the steel plates;
second moment of area of the uncracked concrete section;
shear resistance of a single stud calculated from EN 1994-1-1, clause 6.6.3.1(1);
. longitudinal spacing of the shear reinforcement;
concrete infill thickness;
thickness of the tension steel plate;
thickness of the compression steel plate;
thickness of the steel plate where both plates are of equal thickness (i.e. ;
out-of-plane shear resistance of the SC member;
, in-plane shear resistance of the SC member;
yield strain of the compression steel plate
compression plate slenderness / ;
reduction factor given in EN 1993-1-1, clause 6.3.1.2 as a function of the relative
slenderness, ̅ ;
partial factor for concrete, given in EN 1992-1-1;
∗
partial factor for member resistance;
partial factor for cross-section resistance of steel, given in EN 1993-1-1.
INTRODUCTION
SC is the name for a generic steel-concrete composite construction system comprising two
steel plates connected by a grid of tie bars with structural concrete between the plates. The
plates act as load bearing formwork during the placement of the concrete and hence no
additional formwork is required. In the completed condition the plates act as reinforcement to
the concrete. The tie bars, apart from holding the two plates together during construction,
provide transverse shear reinforcement. Composite action between the steel plates and the
concrete core is achieved through the use of headed shear studs welded to the steel plates.
The construction form is shown in Figure 1.
Fig. 1 – Typical SC member prior to concreting

A European funded four year project investigating the behaviour of SC structures was recently
completed (SCIENCE, 2017). Structural member tests were conducted to study bending and
out-of-plane shear failure. Connection tests studied wall-to-wall, floor-to-wall and
foundation-to-wall behaviour. LOCA tests were performed at small scale with highly controlled
boundary conditions under thermal loading and at large scale (including comparative
reinforced concrete specimen tests) under combinations of thermal and mechanical loading.
Fire tests were conducted on floor, floor-to-wall and wall large scale specimens. This was
supported by FE parametric studies. Design guidance following the format of the Eurocodes
was developed using the results from this project and other studies available in the literature.
A reliability analysis to Annex D of EN 1990 (2002) was performed to derive partial safety
factors. The guidance has been trialled by applying it to the design of the “Diesel Ultime
Secours (DUS)” building (Diesel Generator Building), which is being used as a post Fukushima
safety measure on all EDF nuclear reactor sites in France. This paper focuses on the ultimate
limit state design of SC members.
MEMBERS IN TENSION
The resistance of an SC member in tension is conservatively taken as the resistance of the
steel plates without any contribution from the concrete. The latter is determined in accordance
with EN 1993-1-1 (2005). This assumption is reasonable since the tensile resistance of the
concrete is negligible.
MEMBERS IN COMPRESSION
Cross-section resistance to compression
The compression resistance of SC members is given (in accordance with EN 1994-1-1:2004)
by:
0.85
(1) ,
The factor of 0.85 applied to the concrete contribution is used in EN 1994-1-1 for concrete
encased or partially concrete encased steel sections. The factor may be taken as 1.0 for
concrete filled steel sections. The case of a SC member would justify a higher factor than 0.85
due to the better containment of the concrete compared with an encased section. However,
in the absence of specific data, 0.85 is conservatively retained.
Local plate buckling
Equation 1 does not account for local (plate) buckling and global (member) buckling. Plate
buckling may be ignored provided that the plate slenderness given by / , is within the
limits shown in Table 1.
Table 1: Limits to stud spacing
Steel Grade S235 S275 S355 S460
37 34 30 26
The limits in Table 1 are based on a large number of test results (Akiyama et al., 1991; Usami
et al., 1995; Miyauchi et al., 1996; Choi and Han, 2009) as shown in Figure 2. The critical
buckling strain (normalised by the yield strain) is plotted as a function of the plate slenderness
parameter ( / ). The Euler buckling curve is also shown for a partially fixed ended strut
(with effective length factor 0.7) and provides a good fit to the experimental results as
previously observed (Zhang et al., 2014).
The limits in Table 1 correspond to a plate slenderness parameter limit of 1.25. This ensures
ductile behaviour following plate buckling with no sudden reduction in the member resistance
and a gradual redistribution of the force from the compression plate to the concrete. Such
behaviour has been demonstrated. For example, in one test (Choi and Han, 2009), although
both plates buckled at around 60% of the yield strain, the specimen was able to resist at least
20% more load than that which caused plate buckling. The plate slenderness limits in Table 1
are also consistent with the limits in JEAC 4618 (2009).
Fig. 2 - Plate buckling strain as a function of plate slenderness parameter
Member resistance to compression

A member in compression should also satisfy the following relationship to prevent global
buckling:
(2) ∗ 1.0
, /
where
is the design value of the compression force;
, is the characteristic value of the plastic resistance to compression given by Equation 1
with the partial factors set to unity;
is the buckling reduction factor which depends on member slenderness
∗
is the partial factor, taken as 1.15 for global buckling.
The relative slenderness is given by:
,
(3) ̅
is the elastic critical normal force. To calculate , EN 1994-1-1 stipulates the use of an
effective flexural stiffness, given by:
(4)
Where is a correction factor accounting for the reduced stiffness of cracked concrete and
should be taken as 0.6. A method for calculating the flexural stiffness allowing for the effect of
partial shear connection (slip between the plates and the concrete) was also considered, as
was the effect of setting 1.0 in Equation 4. The sensitivity of the flexural reduction factor,
, to the three methods of calculating the flexural stiffness is shown in Figure 3 to be low, with
Equation 4 leading to conservative results. In this Figure, s1 and s2 are spacing of shear studs
in the direction of the compression force and orthogonally to it and are used to achieve different
levels of shear connection. Based on this comparison, Equation 4 obtained from EN 1994-1-1
is recommended.
Fig. 3 – Buckling reduction factor, , calculated using different methods of estimating the
flexural stiffness
OUT-OF-PLANE BENDING
The resistance to out-of-plane bending is calculated from stress block analysis neglecting the
tensile strength of concrete (see Figure 4).
Fig. 4 – Assumed plastic stress distribution in SC members in bending
The design plastic moment resistance is given by:
(5) , , , 1 , / ∗
2 2 2
where
, is the axial resistance of the compression plate = min ( , );
, is the axial resistance of the tension plate = min ( , 0.8 );
, is the resistance of the concrete in compression = ;

is a factor defining the effective height of the compression zone 0.8 for
50 MPa;
is a factor to account for the idealisation of the concrete stress distribution = 1.0 for
50 MPa;
is the concrete strength = , where can be taken as 0.85 for bending;
is the depth to the neutral axis from the top of the concrete;
is the number of studs on the compression plate between the point of maximum
moment and the nearest point of contraflexure;
is the number of studs on the tension plate between the point of maximum moment
and the nearest point of contraflexure;
∗
is the partial factor for bending to be taken as 1.24.
For SC members with nominally identical steel plates, is zero and the design plastic moment
resistance is:
∗
(6) , , /
For SC members in bending with slender compression plates that fall outside the limits of
Table 1, the bending resistance can be calculated using Equation 5, ignoring any contribution
from the compression plate.
Figure 5 shows a comparison of bending resistance from experiments with values obtained
using Equation 5. The Figure shows test results from published sources (Takeuchi et al., 1999
and Sener and Varma, 2014, also presented in Sener et al., 2015). It also shows two results
from bending tests carried out in SCIENCE project, one of which was a ‘hybrid’ specimen
(austenitic stainless steel compression plate and carbon steel tension plate). The bending
resistance was under-predicted by 9.1% for the carbon steel specimen and by 9.4% for the
hybrid specimen.
Fig. 5 - Comparison of experimental and predicted nominal values of plastic moment

resistance
OUT-OF-PLANE SHEAR
The design models of EN 1992-1-1 are adopted for out-of-plane shear. Although out-of-plane
shear reinforcement is always present in SC members (in the form of tie bars), the resistance
is calculated for members with and without out-of-plane shear reinforcement in accordance
with EN 1992-1-1. This is because, in certain cases (e.g. widely spaced tie bars), the equation
for unreinforced members may govern. The spacing of tie bars in SC members should not
exceed , the effective depth of the member.
Members without shear reinforcement
The shear strength of members without shear reinforcement is governed by the concrete
strength, the tensile reinforcement ratio provided by the bottom steel plate (which is taken as
the smaller of the actual ratio or 2%) and the section depth. For members with no shear
reinforcement, the design shear resistance is given by:
(7) , , 100 / ∗
The minimum value of , is given by:
(8) ,
where
1 200/ 2.0
is the tensile reinforcement ratio ( / 0.02)
/ 0.2
is the axial force in the cross-section due to external loading ( 0 for compression)
/
, 0.18
0.035
0.15
In Equations 7 and 8, all dimensions are in [mm], areas in [mm2], forces in [N] and strengths
∗
and stresses in [MPa]. The partial factor, , for out-of-plane shear for members with no
shear reinforcement is taken as 1.28.
Figure 6 shows a comparison of values calculated using Equation 7 (with ∗ set to unity) and
experimental results (from Sener et al., 2016, Takeuchi et al., 1999 and from SCIENCE, 2017).
The ratio of experimental to predicted values, / , are ploted as a function of the ratio
of the bending moment to shear force divided by the depth of the section ( / ) at the critical
section. The design model underestimates the shear resistance particularly for smaller values
of ( / ). This is partly because Equation 7 ignores arching action.
Fig. 6 - Predicted nominal shear resistance against experimental results for specimens
without shear reinforcement
Members with shear reinforcement
For members with shear reinforcement, the design shear resistance is given by the smaller
of:
∗
(9) , /
.
and
(10) , / cot tan
where
is the lever arm. If there is no axial force, a value of 0.9 may be used [mm]
is the angle between the concrete compression strut and the member axis
perpendicular to the shear force, limited to 1 2.5 (i.e. 21.8 to 45°)
0.6 1 /250 , where is in MPa as provided in EN 1992-1-1, clause
6.2.3(3). EN 1992-1-1 allows the use of a higher value of if the design stress
in the shear reinforcement is less than 0.8
. is the longitudinal spacing of shear reinforcement. . should not exceed 0.75 .
In equations 9 and 10, all dimensions are in [mm], areas in [mm2] and strength in [MPa]. ∗ is
the partial factor for out-of-plane shear (members with shear reinforcement), taken as 1.44.
Comparison of tests (from Hong et al., 2009, Takeuchi et al., 1998 and from SCIENCE, 2017)
against predictions using Equation 9 and 10 is presented in Figure 7. In all cases, the
calculations were performed for a strut inclination angle of 21.8°. The comparisons are for
specimens with shear reinforcement in the form of tie bars. Predictions using the simplified
modified compression field theory (MCFT) by Bentz et al. (2006) are also plotted. Equation 9
is generally conservative. In the case of the test performed by Hong et al. (2009) with
/ = 3.65, failure occurred due to shear and bending interaction (rather than pure shear
failure).
Fig. 7 - Predicted nominal shear resistance against experimental results for specimens with
shear reinforcement
IN-PLANE SHEAR
The resistance to in-plane shear is calculated in accordance with the relationship originally
developed by Ozaki et al. (2004):
1
(11) , ∗
2
3
1
4 21
0.7 2
∗
In Equation 11, all dimensions are in [mm] and strengths and moduli in [MPa]. is the partial
factor for in-plane shear, taken as 1.32.
The results from using Equation 11 (with ∗ set to unity) are compared against experimental
results in the literature in Figure 8. In almost all cases, Equation 11 is conservative and
reasonably accurate. The experimental results in Figure 8 are taken at the point where the
steel plate yields in the test. This is consistent with the basis of Equation 11. Therefore, in
most cases, the conservatism is in part due to the fact that members can sustain higher loads
in the post-yield stage (as demonstrated in the tests) until concrete failure occurs. However,
further research is required to allow exploitation of this additional strength.
Fig. 8 - Comparison between experimental in-plane shear results (Seo et al., 2016) and the
nominal in-plane shear resistance predicted by Equation 11
SHEAR CONNECTION REQUIREMENTS

A limit on the minimum degree of shear connection is necessary so as not to adversely affect
the strength and stiffness of SC members. This is achieved by using a maximum stud spacing
given by:
(12a) for the tension plate
,
0.8
(12b) for the compression plate
,
where
, is the longitudinal shear force per unit length of member, given by:
(13) ,
where
is the distance of the centroid of the plate from the neutral axis;
The above requirement is based on the methodology used in EN 1994-2 (2005) for distributing
studs along the length of a bridge beam, as well as similar requirements in the Bi-Steel Manual
(2003), KEPIC SNG (2009) and AISC N690 (2015). It offers a practical way of choosing the
spacing with a sensible level of conservatism.
CONCLUSIONS
Guidance for the design of SC structures that follows the Eurocode format has been developed
using the findings from a recently completed research project and other published data. Partial
factors which can be used with the nominal design values of the resistance have been
calculated for the new design equations and comparisons made with test results.
ACKNOWLEDGMENT
This work was funded by the European Union's Research Fund for Coal and Steel (RFCS)
under grant agreement no. RFSR-CT-2013-00017 and Electricité De France (EDF). The
technical work was performed by The Steel Construction Institute (UK), University of Surrey
(UK), Teknologian tutkimuskeskus VTT (Finland), VTT Expert Services Ltd (Finland),
Karlsruhe Institute of Technology (Germany), SMP Ingenieure im Bauwesen GmbH
(Germany), EGIS Industries SA (France), Electricité De France (France), Ecole Normale
Superieure de Cachan (France), Ecole Speciale des Travaux Publics (France), Centre
Technique Industriel de la Construction Métallique (France), Efectis (France).
REFERENCES
AISC N690-12 + N690s1-15 (2015). Specification for Safety-Related Steel Structures for
Nuclear Facilities (including supplement No. 1), American Institute of Steel Construction
(AISC).
Akiyama, H., Sekimoto, H., Fukihara, M., Nakanishi, K. and Hara, K. (1991). A compression
and shear loading test of a concrete filled steel bearing wall. 11th International Conference on
Structural Mechanics in Reactor Technology (SMiRT 11), Vol. H, Tokyo.
Bentz, E.C., Vecchio, F.J. and Collins, M.P. (2006). Simplified modified compression field
theory for calculating shear strength of reinforced concrete elements. ACI Structural Journal,
July-August 2006, 614-624.
Choi, B.J. and Han, H.S. (2009). An experiment on compressive profile of the unstiffened steel
plate-concrete structures under compression loading. Steel and Composite Structures, 9(6),
439-454.
Corus (2003). Bi-Steel Design and Construction Guide (2nd ed.). Corus UK Ltd.
EN 1990:2002. Eurocode: Basis of structural design.
EN 1992-1-1:2004. Eurocode 2: Design of concrete structures - Part 1-1: General rules and
rules for Buildings.
EN 1993-1-1:2005. Eurocode 3: Design of steel structures — Part 1-1: General rules and rules
for buildings.
EN 1994-1-1:2004. Eurocode 4: Design of composite steel and concrete structures — Part 1-
1: General rules and rules for buildings.
EN 1994-1-2:2005. Eurocode 4. Design of composite steel and concrete structures. General
rules. Structural fire design.
EN 1994-2:2005. Eurocode 4. Design of composite steel and concrete structures. General
rules and rules for bridges.
Hong, S.-G., Lee, K.-J., Park, D.-S., Ham, K.-W. and Lee, H.-W. (2009). Out-of-plane shear
strength of steel plate concrete walls dependent on bond behaviour. 20th International
Conference on Structural Mechanics in Reactor Technology (SMiRT 20), Espoo, Finland,
August 9-14, 2009.
JEAC 4618 (2009). Japan Electric Association: Technical Code for Aseismic Design of Steel
Plate Reinforced Concrete Structures. Translated into English in January 2012 by the
Obayashi Corporation.
KEPIC-SNG (2009). Specification for Safety-Related Steel Plate Concrete Structures for
Nuclear Facilities.
Miyauchi, Y. et al. (1996). Experimental study on a concrete filled steel structure. Summaries
of Technical Papers of Annual Meeting, Vol. 3, Architectural Institute of Japan.
Ozaki, M., Akita, S., Osuga, H., Nakayama, T. and Adachi, N. (2004). Study on steel plate
reinforced concrete panels subjected to cyclic in-plane shear. Nuclear Engineering and
Design, 228, 225-244.
SCIENCE (2017). SC for Industrial, Energy and Nuclear Construction Efficiency. Research
Fund for Coal and Steel (RFCS), RFSR-CT-2013-00017.
Sener, K.C. and Varma, A.H. (2014). Steel-plate composite walls: Experimental database and
design for out-of-plane shear. Journal of Constructional Steel Research, 100, 197-210.
Sener, K.C., Varma, A.H. and Ayhan, D. (2015). Steel plate composite (SC) walls: Out-of-
plane flexural behaviour, database and design. Journal of Constructional Steel Research, 108,
46-59.
Sener, K.C., Varma, A.H. and Seo, J. (2016). Experimental and numerical investigation of the
shear behavior of steel-plate composite (SC) beams without shear reinforcement. Engineering
Structures, 127, 495-509.
Seo, J., Varma, A., Sener, K. and Ayhan, D. (2016). Steel-plate composite (SC) walls: In-plane
shear behaviour, database and design. Journal of Constructional Steel Research, 119, 202-
215.
Takeuchi, M., Fujita, F., Funakoshi, A., Shohara, R., Akira, S., and Matsumoto, R. (1999).
Experimental Study on Steel Plate Reinforced Concrete Structure, Part 28 Response of SC
Members Subjected to Out-of-Plane Load (Outline of the Experimental Program and the
Results). Proceedings of the Annual Conference of Architectural Institute of Japan, (in
Japanese), 1037- 1038.
Takeuchi, M., Narikawa, M., Matsuo, I., Hara, K. and Usami, S. (1998). Study on a concrete
filled structure for nuclear power plants. Nuclear Engineering and Design, 179(2), 209-223.
Usami, S., Akiyama, H., Narikawa, M., Hara, K., Takeuchi, M. and Sasaki, N. (1995). Study
on a concrete filled steel structure for nuclear power plants (part 2). Compressive loading tests
on wall members. 13th International Conference on Structural Mechanics in Reactor
Technology (SMiRT 13), Porto Alegre, Brazil, August 13-18, 1995.
Varma, A., Zhang, K., Chi, H., Booth, P. and Baker, T. (2011). In-plane shear behaviour of SC
composite walls: Theory vs. experiment. 21st International Conference on Structural
Mechanics in Reactor Technology (SMiRT 21), New Delhi, India, November 2011.
Zhang, K., Varma, A.H., Malushte, S.R. and Gallocher, S. (2014). Effect of shear connectors
on local buckling and composite action in steel concrete composite walls. Nuclear Engineering
and Design, 269, 231-239.
THE NEW AUSTRALIA/NEW ZEALAND STANDARD ON COMPOSITE
STEEL-CONCRETE BUILDINGS, ASNZS2327
Brian Uy
School of Civil Engineering, The University of Sydney
New South Wales, Australia
brian.uy@sydney.edu.au
Stephen J. Hicks
Won-Hee Kang
Centre for Infrastructure Engineering, Western Sydney University
w.kang@westernsydney.edu.au
Huu-Tai Thai
School of Engineering and Mathematical Sciences, La Trobe University
Victoria, Australia
tai.thai@latrobe.edu.au
Farhad Aslani
School of Civil, Environmental and Mining Engineering, The University of Western Australia
Western Australia, Australia
farhad.aslani@uwa.edu.au
ABSTRACT
This paper will provide a summary of the new Australia-New Zealand Standard on the design of
composite steel-concrete buildings ASNZS2327 which is due for release in 2017. This standard
which has included over five years of development has harmonized design provisions across
the Tasman Sea, in steel, concrete and composite structures. This project was carried out in
parallel with the standard for bridge structures (ASNZS 5100.6) and thus there was significant
commonality and cross-over. ASNZS2327 covers the design of structural elements such as
slabs, beams, columns and joints, as well as addressing the system behaviour for serviceability
and fire which are highly pertinent for the design of steel and composite structures. Methods
for the design for earthquake have also been included. The standard allows for the increase in
concrete compressive strength up to 100 MPa and steel strengths up to 690 MPa. In addition,
the standard has involved significant amounts of structural reliability studies to be carried out.
This paper will provide a summary of some of the salient issues as well as highlighting future
areas of interest which may be considered for future revisions.
INTRODUCTION
The concept of steel-concrete composite construction generally produces structural behaviour
where the overall response is greater than the sum of the parts. This concept holds true for
composite beam behaviour where the introduction of longitudinal shear connection can provide
flexural stiffness and strength of a member which is greater than the constituent parts, namely
the structural steel section and the reinforced concrete slab. Throughout the latter half of the
twentieth century this concept has been further applied to composite steel-concrete columns to
produce column stiffnesses and strengths which are greater than the sum of the parts of the
steel section and the reinforced concrete elements. These benefits have been further exploited
by taking advantage of the confinement effects that steel tubes can provide to interior concrete
infill and the subsequent benefits provided by the concrete infill on delaying local buckling of the
steel shell. This concept is now so widespread that in the last calendar year, more than 50 % of
all tall buildings constructed worldwide, utilized composite frames, typically incorporating
concrete filled steel columns (Council of Tall Buildings and Urban Habitat, 2016).
In Australia, builders of the recently completed Perth Tower (the tallest tower in Perth), chose to
adopt composite construction throughout the entire structure and a concrete filled steel column
solution. In order to secure this type of solution the builder pre-ordered and stored all spirally
welded steel tubes in the columns (see Figure 1) to ensure steel cost fluctuations were
minimized and construction costs were able to be controlled (Australian Steel Institute, 2010).
The emphasis in Australia has been mainly focussed on construction economy when it relates
to steel and steel-concrete composite structures.
(a) Completed building (b) Building during construction
Fig. 1 - BHP Billiton Tower, Perth, Australia
In New Zealand the emphasis in steel and steel-concrete composite construction whilst also
concerned with construction economy, is heavily influenced by seismic behaviour and
performance. The new 14-storey 20 Customhouse Quay in Wellington (see Figure 2) is due to
be completed in 2017 to replace the earthquake damaged BP House on Wellington’s
waterfront. This building was designed post Christchurch earthquake and utilizes base isolators
(Schouten, 2015), resulting in an estimated seismic resilience of up to 180% of code. The
building incorporates many steel-concrete composite components, including a steel diagrid
external frame and long span composite cellular beams.
Fig. 2 – 20 Customhouse Quay, Wellington, New Zealand
DRAFT BUILDING STANDARD FOR COMPOSITE STEEL-CONCRETE, ASNZS 2327

The Australian Standard for composite steel-concrete structures in buildings, AS2327.1 only
ever covered the design of simply supported composite beams (Standards Australia, 2003). A
major initiative some 5 years ago involved ensuring that all forms of composite systems,
including beams, slabs, columns and joints would be covered for design and has resulted in the
Australia/New Zealand harmonisation of the standard, ASNZS2327 (Standards
Australia/Standards New Zealand 2017a). The standard table of contents is shown below and
salient features of the standard will described herein
SECTION 1 GENERAL
SECTION 2 DESIGN OF COMPOSITE SLABS
SECTION 3 DESIGN OF COMPOSITE BEAMS
SECTION 4 DESIGN OF COMPOSITE COLUMNS
SECTION 5 DESIGN OF COMPOSITE JOINTS
SECTION 6 DESIGN OF COMPOSITE FLOOR SYSTEMS
SECTION 7 SYSTEM DESIGN FOR FIRE RESISTANCE
SECTION 8 DESIGN FOR EARTHQUAKE
APPENDICES
DESIGN OF COMPOSITE SLABS

Section 2 of ASNZS2327 covers the comprehensive design of composite slabs, (see Figure 3).
The intent of this section is to cover the strength and serviceability design of composite slabs
utilising metal decking. Issues including flexural strength, longitudinal shear and vertical shear
provisions are covered in this section. Concepts of partial interaction are also considered and
this section also links quite closely to that being proposed for testing in the Appendices of the
standard. Furthermore, post-tensioned concrete construction is also extremely prominent in
Australian buildings and recent innovations into post-tensioning concrete slabs with metal
decking have been carried out (see Figure 4). One of the major issues is the changes that need
to be introduced to deal with the presence of the metal decking for serviceability and strength
provisions and these will be considered as part of this section, namely the concepts of non-
uniform shrinkage (Al Deen et al., 2015).
Fig. 3 – Types of interlock for composite slabs

(Standards Australia/Standards New Zealand 2017a)
Fig. 4 – Post-tensioned composite slabs
DESIGN OF COMPOSITE BEAMS

Section 3 of ASNZS2327 covers the comprehensive design of composite steel-concrete beams.
This section covers the design of composite beams, considering flexural strength, shear
strength and combined actions as well as serviceability provisions. Partial shear connection
approaches are also highlighted for the design of simply supported and continuous beams.
This section also considers the design of composite beams using hollow core slabs as
illustrated in Figure 5, (Uy and Bradford, 2007).
Fig. 5 – Precast hollowcore units
Fig. 6 – Alternative steel beam types (Standards Australia/Standards New Zealand 2017a)
DESIGN OF COMPOSITE COLUMNS

This section covers the design of composite columns and closely follows the approach in the
AS/NZS 5100 Part 6 (Standards Australia/Standards New Zealand, 2017b). The design of
composite columns for strength, stability incorporating axial force, uniaxial and biaxial bending
is considered. In particular, the important effects of confinement are covered by this section.
Furthermore, the capacity factor for concrete in compression is proposed to be 0.65 based on
reliability analyses using the design assisted by testing method provided in EN 1990 Annex D.8
(European Committee for Standardization, 2002). The reliability analyses were carried out for
1583 CFST columns included in Tao et al.’s database (Tao et al., 2008). Figure 7 shows the
analysis results along the squash load ratio of steel, where the average capacity factors for
concrete are above 0.65. If a practical range of squash load ratio such as 0.1~0.6 is
considered, the average capacity factor will increase even further.
1
0.9
0.8
0.7
Capacity factors
0.6
0.5
0.4
0.3
Capacity factor for steel (φ)
0.2
Capacity factor for concrete (φc)
0.1
Average of φc
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Squash load ratio of steel
Fig. 7 – Calibrated capacity factors for concrete in compression when the capacity factor steel
is fixed to be 0.9
One major area of innovation which this section will cover includes the use of high strength
concrete (cylinder strengths up to 100 MPa) and high strength steel (nominal yield strengths up
to 690 MPa) (Aslani et al., 2015a, b). In addition the standard has involved calibration and has
been extended to permit spiral welded tubes to be used in design (Aslani et al., 2015c and
Aslani et al., 2017).
Fig. 8 – Spiral welded composite columns, Aslani et al. (2017)
DESIGN OF COMPOSITE JOINTS
This standard has also involved in the development of a section for strength and serviceability
design of semi-rigid joints including beam-to-beam and beam-to-column joints as shown in
Figure 9. For beam-to-column joints, the column could be either open sections with/without
stiffeners or hollow sections with/without infilled concrete. The design of joints to hollow section
columns is based on the stiffness model of Thai and Uy (2016) which was calibrated with
experimental results of 44 available tests on bolted endplate beam-to-CFST column joints.
Beam-to-beam joint Beam-to-column joint
(b) Beam-to-beam joint

(a) Plan view
Through reinforcement
Stiffener
(c) Beam-to-column joint

(d) Composite joints (Thai et al., 2017)
Fig. 9 - Composite steel-concrete beam and slab systems, Australia
DESIGN OF COMPOSITE FLOOR SYSTEMS

The intent of this section is to address system behaviour particularly for deflections and
vibrations for panels. This will then give designers the ability to take into account the beneficial
effects of system behaviour in addressing these important serviceability provisions which
sometimes penalize steel frame structures from a design perspective (Steel Construction
Institute, 2012).
Fig. 10 – Deflection requirements for composite floor systems
SYSTEM DESIGN FOR FIRE RESISTANCE

This section is also intended to give guidance on design for fire using a system based
approach, which acknowledges that for indeterminate systems there is a significant degree of
redundancy that provides additional structural capacity within a fire that is unable to be
addressed considering single elements within a building. State of the art approaches for dealing
with this will be provided herein (Steel Construction Institution, 2006 and Abu et al., 2011).
Fig. 11 – Collapse mechanism for an isolated panel

DESIGN FOR EARTHQUAKE
This section covers the design for earthquake of composite frame systems. One of the major
elements of the section is the specific guidance provided to Australian and New Zealand
designers in accordance with AS1170.4 and NZS1170.5 (Standards Australia 2004 and
Standards New Zealand 2007).
APPENDICES
This section of the proposed standard is meant to provide standard test methods for a number
of specific issues which are covered. Push test methods for establishing the strength, stiffness
and ductility of shear connectors will be outlined in this section as will test methods for
establishing the strength characteristics of composite slabs incorporating profiled steel
sheeting. In addition, provisions for evaluating design resistance from tests will also be
presented. Finally, ASNZS 2327 will be one of the first international composite design
standards to present provisions for beams with both regular and isolated web-openings, thereby
supporting the use of long-span cellular beams (Steel Construction Institution, 2011).
Fig. 12 – Method for determining the longitudinal shear capacity of headed shear studs in
voided concrete slabs made composite with steel
CONCLUSIONS AND FURTHER RESEARCH
Whilst there is ongoing research into structural steel and some of the technical issues
associated with materials and systems, it is felt that further research will be punctuated by
approaches that provide paradigm shifts in the design of steel and steel-concrete composite
building structures. Some of the more prominent issues that will promote these paradigm shifts
include precast and prefabricated construction, deconstructability and new and higher
performance materials.
In particular steel and steel-concrete composite framed buildings have the ability to be designed
and constructed with deconstruction in mind. Advanced economies around the world are now
looking toward reuse as a potential for addressing the issue of shortages in natural resources in
future. Composite action has the ability to reduce steel usage and deconstruction has the
ability to provide reuse options. Reduce and reuse strategies are far superior to recycling
options and have the ability to provide much greater benefits to society from the perspectives of
sustainability.
ACKNOWLEDGEMENTS
The authors would like to thank the members of the BD-32 Committee for all their hard work to
date; in particular, the Standards Australia Project Manager, Declan Robinson. As members of
the BD-32 Committee, the present authors have prepared this paper according to what they
consider to be the significant changes to the existing AS 2327.1. However, the standard is still
being prepared for public comment and, as a consequence of this, may be subject to change.
Any views expressed in this paper may not necessarily reflect those of the other members of
the Committee, nor those of Standards Australia and Standards New Zealand
REFERENCES
Abu A.K., Ramanitrarivo V., Burgess I.W., (2011) Collapse Mechanisms of Composite
Slab Panels in Fire, Journal of Structural Fire Engineering, 2 (3), 205-216.
Al-Deen, S., Ranzi, G., Uy, B. (2015) Non-uniform shrinkage in simply-supported composite
steel-concrete slabs. Steel and Composite Structures, 18(2), 375-394.
Aslani, F., Uy, B., Tao, Z., Mashiri, F., (2015a) Behaviour and design of composite columns
incorporating compact high-strength steel plates, Journal of Constructional Steel
Research, 107, 94-110.
Aslani, F., Uy, B., Tao, Z., Mashiri, F., (2015b) Predicting the axial load capacity of high-
strength concrete filled steel tubular columns, Steel and Composite Structures, 19, 967-
993.
Aslani, F., Uy, B., Hicks, S., Kang, W.H., (2015c) Spiral welded tubes – imperfections, residual
stresses, and buckling characteristics, The Eighth International Conference on Advances
in Steel Structures, July 21–24, Lisbon, Portugal.
Aslani F., Uy B., Hur J. and Carino P., (2017) Behaviour and design of hollow and concrete-
filled spiral welded steel tube columns subjected to axial compression, Journal of
Constructional Steel Research, 128, 261-288.
Australian Steel Institute (2010) Perth tower agape to grand views, Steel Australia, 34 (6), 18-
20.
Council on Tall Buildings and Urban Habitat (2016) The Skyscraper Surge Continues in 2015,
“The Year of 100 Supertalls”. CTBUH Year in Review: Tall Trends of 2015, and Forecasts
for 2016, http://www.skyscrapercenter.com/research/CTBUH_ResearchReport_2015YearInReview.pdf
accessed 15/05/2017.
European Committee for Standardization (2002) EN 1990:2002 Eurocode: Basis of structural

design, Brussels, 2002.
Schouten, H. (2015) $80m hi-tech tower to replace Wellington’s Perth House,
http://www.stuff.co.nz/business/67536882/80m-hi-tech-tower-to-replace-Wellingtons-BP-House, accessed
15/05/2017.
Standards Australia (2003) Composite structures, Part 1: Simple supported beams, AS
2327.1:2003, New South Wales, Australia.
Standards Australia (2007) Australian Standard, AS 1170.4—2007 Structural design actions
Part 4: Earthquake actions in Australia.
Standards Australia/Standards New Zealand (2017a) Composite Structures, AS/NZS 2327-
2017, Sydney, Australia.
Standards Australia/Standards New Zealand (2017b), AS/NZS 5100.6: 2017, Bridge Design,
Part 6: Steel and composite construction, Sydney/Wellington, Australia/New Zealand.
Standards New Zealand (2004) New Zealand Standard, NZS 1170.5—2004 Structural design
actions Part 5: Earthquake actions in New Zealand.
Steel Construction Institute (2006) Fire Safe Design: A New Approach to Multi-Storey Steel-
Framed Buildings, SCI P288, United Kingdom.
Steel Construction Institute (2011) Design of composite beams with large web openings, SCI
P355, United Kingdom.
Steel Construction Institute (2012) Design of floors for vibration: a new approach, SCI P354,
United Kingdom.
Tao Z., Uy B., Han L.H., He S.H., (2008) Design of concrete-filled steel tubular members
according to the Australian Standard AS 5100 model and calibration, Australian Journal of
Structural Engineering, 8 (3), 197-214.
Thai H.T., Uy B., (2016) Rotational stiffness and moment resistance of bolted endplate joints
with hollow or CFST columns, Journal of Constructional Steel Research, 126, 139-152.
Thai H.T., Uy B., Yamesri, Aslani F., (2017) Behaviour of bolted endplate composite joints to
square and circular CFST columns, Journal of Constructional Steel Research, 131, 68-82.
Uy B. and Bradford M.A., (2007) Composite action of structural steel beams and precast
concrete slabs for the flexural strength limit state, Australian Journal of Structural
Engineering, 7, 123-133.
THE NEW JOINT AUSTRALIAN AND NEW ZEALAND DESIGN STANDARD FOR STEEL
AND COMPOSITE BRIDGES AS/NZS 5100.6 - PART 6: STEEL AND COMPOSITE
CONSTRUCTION
Stephen J. Hicks
General Manager Structural Systems
Brian Uy
Professor and Head of School
School of Civil Engineering, The University of Sydney
brian.uy@sydney.edu.au
Won-Hee Kang
Senior Lecturer in Structural Reliability
Centre for Infrastructure Engineering, Western Sydney University
W.Kang@westernsydney.edu.au
ABSTRACT
This paper presents some of the innovations that are included within the new Bridge Design
Standard for Steel and Composite Construction AS/NZS 5100.6, which will be the first
harmonized standard between Australia and New Zealand for the design of bridges. As Chairs
of the Committees responsible for AS/NZS 5100.6 and AS/NZS 2327, the authors of this paper
present the challenges faced from the introduction concrete compressive strengths up to
100 MPa and quenched and tempered steels with a yield strength up to 690 MPa. Perhaps one
of the most innovative aspects of this standard is the introduction of an appendix that provides
design rules for steel products that are not manufactured to Australia and New Zealand
standards. This appendix is underpinned by rigorous structural reliability analyses undertaken
by Australian and New Zealand researchers, which included the present authors of this paper.
INTRODUCTION
This paper provides an overview of the new Australasian Bridge Design Standard for Steel and
Composite Construction AS/NZS 5100.6 (2017). Building on earlier steel design standard
harmonisation initiatives, such as the cold-formed steel structures standard AS/NZS 4600
(2005), AS/NZS 5100.6 is the first joint Australian and New Zealand design standard for
bridges. It has been a catalyst for further harmonization activities in design standards for steel
construction, such as the development of the new standard for steel and concrete composite
buildings DR AS/NZS 2327 (2016). In the future, it is hoped that this work may lead to a
harmonization of the existing AS 4100 (1998) and NZS 3404 (1997) into a joint Australian and
New Zealand steel structures standard. The structure of the current Australasian structural steel
standards follows that of many other international standards and is presented in Fig. 1.
Fig. 1 - Current structure of Australasian structural steel standards.
Steel bridges have historically been used in both Australia and New Zealand. Two of the most
iconic bridges in these countries are the Sydney Harbour Bridge (see Fig. 2(a)) and the
Auckland Harbour Bridge: both of which used imported steel from Dorman Long, UK. Due to its
limited domestic steel supply, New Zealand continues to use imported steel in many of its
bridges such as the Te Rewa Rewa Bridge in New Plymouth shown in Fig. 2(b) (Mulqueen,
2011). The success of using imported structural steel in New Zealand is reflected in a market
share of 50% for steel in construction, rising to 80% in Christchurch.
(b)
(a)
Fig. 2 - (a) Sydney Harbour Bridge, Australia (b) Te Rewa Rewa Bridge, New Zealand.
A number of technical and political challenges were presented in the development of

AS/NZS 5100.6. On the technical side, the introduction of higher strength steel and concrete
meant that many of the existing composite design provisions given in AS 5100.6 (2004) had to
be reconsidered. In both technical and political terms, the requirement by New Zealand bridge
designers to include steels manufactured to overseas standards resulted in reliability analyses
being undertaken to demonstrate that the same margins of safety for designs using steels
manufactured to AS and AS/NZS standards were maintained. Many of the other changes that
will be presented in this paper drew on work that underpin the design provisions given in the
structural Eurocodes (EN 1994-2 (2005) and EN 1993-1-9 (2005)) as well as AISC 360-16
(2016).
AS/NZS 5100.6 consists of 15 Sections and several Appendices. The structure is presented
below in the following subheadings. Where significant changes have been made compared to
the existing AS 5100.6, these are highlighted and an overview of the background work given
SECTION 1, 2 & 3 - SCOPE AND GENERAL, MATERIALS & GENERAL DESIGN

REQUIREMENTS
AS/NZS 5100.6 is concerned with the design of structural steelwork in bridges together with
steel-concrete composite members, including composite beams and composite columns. In
addition, for consistency with the concrete bridges design standard AS 5100.5 (2017), concrete
compressive cylinder strengths f′c of up to 100 MPa are permitted. The standard applies to the
design of other steel components of bridges including steel piers, steel railings and sign
structures. The scope does not, however, include bridges with orthotropic plate decks, cold-
formed steel members other than structural hollow sections, steel elements less than 3 mm
thick nor steel members where the value of yield stress fy > 690MPa. In a similar way as its
predecessor, the structural steel product standards recognized are AS/NZS 1163 (2016),
AS/NZS 1594 (2002), AS/NZS 3678 (2016), AS/NZS 3679.1 (2016) and AS/NZS 3679.2
(2016). However, the new addition is quenched and tempered steel plate produced according to
AS 3597 (2008).
A design life of 100-years is assumed in the AS(/NZS) 5100: 2017 suite of standards. In a
similar manner to North American practice, a global factor approach is adopted, where the
design resistance is calculated by multiplying the nominal (characteristic) resistance by a
capacity reduction factor φ (cf. partial factor approach, where φ = 1/γM). Whilst this approach is
simple to apply in structural steel and reinforced concrete design, it can prove problematical to
apply in composite design as the equations for nominal capacity can consist of up to four
different materials. As a consequence of this, the existing AS 5100.6 adopts a hybrid approach,
where the capacity factors in composite columns are applied to the individual material
components with φ = 0.9 for structural steel together with reinforcing steel and φc = 0.6 for
concrete; this hybrid approach is maintained in AS/NZS 5100.6.
ALTERNATIVE STEEL PRODUCTS

According to AS/NZS 1170.0 (2002), it is based on the philosophy and principles set out in ISO
2394 (1998) (AS 5104 (2005) identical). In probability-based design, the probability of failure Pf
is the basic reliability measure that is used. An alternative measure is the reliability index β and
is related to the probability of failure Pf by:
Pf = Φ (− β ) (1)
where Φ is the cumulative distribution function of the standardised normal distribution
The target reliability index is related to the expected social and economic consequences from a
design failure. According to ISO 2394 and AS 5104, the suggested reliability index for ultimate
limit state design is β = 3.8, which corresponds to the case when the consequence of failure is
great (the highest level) and the relative costs of safety measures are moderate. Design values
of resistances Rd are defined such that the probability of having a more unfavourable value is
as follows:
P (R ≤ Rd ) = Φ (− α R β ) (2)
where αR is the First Order Reliability Method (FORM) sensitivity factor for resistance.
For a dominating resistance parameter, ISO 2394 and AS 5104 recommend αR = 0.8.
Therefore, the design value for resistance corresponds to the product αRβ = 0.8 × 3.8 = 3.04
(equivalent to a probability of the actual resistance falling below the design resistance of 1 in
845 = 0.0012). The remaining safety is achieved in the specification of actions.
In the interests of closer economic relations, the New Zealand Steel Bridge Group together with
other key-stakeholders confirmed the wish to revise the Steel and composite bridge design
standard AS 5100.6 as a joint AS/NZS document in 2012. Due to the long history of
successfully using overseas structural steels with the design standard NZS 3404, it was
required by New Zealand designers that steel products conforming to EN 10025 (2004), JIS
G 3106 (2004) and JIS G 3136 (2005) should be supported by the resulting AS/NZS 5100.6. In
response to concerns that the use of overseas steels might cause an erosion to the safety
margins required by AS/NZS 1170.0, a structural reliability investigation was undertaken by
Kang et al. (2015a), which considered both the material variability and geometric manufacturing
tolerances.
Some typical results from the reliability analyses are presented in Fig. 3, which shows the
relationship between the capacity factor φ and the reliability index β . As can be seen, for the
target reliability index β = 3.04, the calibration results are almost identical within rounding
errors, viz. the capacity reduction factor φ = 0.94, irrespective of the different manufacturing
tolerances; this finding was confirmed in further structural reliability work on columns, composite
columns and composite beams, and was later shown by Uy et al. (2015) that it could also be
safely extended to a wider range of steel products manufactured to Korean (KS), Chinese (GB)
and American (ASTM) standards, which are currently recognized for design in Singapore by the
Building and Construction Authority (BC1, 2012). Therefore, the existing capacity factor of φ =
0.90 given in AS 5100.6 for beams in bending is on the conservative side for steel sections
complying with BS EN 10025, JIS G 3106 and JIS G 3136. This finding led to a proposed
normative appendix to AS/NZS 5100.6 together with design rules for these alternative steel
materials.
Unfortunately, on comparing the differences between the traceability and conformity
requirements in the 2016 edition of the product standards (AS/NZS 1163, AS/NZS 3678,
AS/NZS 3679.1 and AS/NZS 3679.2), the committee agreed that further work would need to be
undertaken in the future to resolve these differences. Therefore, the rules for alternative steel
materials are currently provided within a New Zealand-only appendix.
(a) ϕ=0.94 at β = 3.04 (b) ϕ=0.95 at β = 3.04
Fig. 3 - Capacity factor versus reliability index for compact sections using products complying with
manufacturing tolerances given in (a) EN 10034 (1993) and (b) JIS G 3192 (2005)
SECTION 4 & 5 - METHODS OF STRUCTURAL ANALYSIS & STEEL BEAMS

Owing to the fact that finite element methods of analysis are now available in most engineering
software packages, bridges are often being designed using these tools. However, there is a
general paucity of information at a standards level on how different structures should be
modelled. To remedy this situation, provisions are given in AS/NZS 5100.6 on modelling
material behaviour in different types of analyses, together with required geometric
imperfections.
One potential buckling mode that may occur when intermediate restraints are flexible is now
recognized in AS/NZS 5100.6. The mode of buckling features one or two half wavelengths over
the span, with the restraint positions in each half wavelength being displaced by the buckling.
This mode occurs in multi-girder and ladder deck bridges during construction when there is no
plan bracing; the only bracing is in planes (triangulated bracing or stiff cross-girders) between
beam pairs. These planes offer torsional restraint to the main beams, by virtue of the vertical
stiffness of the main beams themselves. The mode of buckling with torsional restraints is shown
in Fig. 4, for a single half wave in a simply supported span (whilst the mode is illustrated with
stiff cross-girders, it can equally apply to beams with triangulated bracing). Once the deck slab
has been cast, the cross-girders effectively form inverted U-frames and these may be used to
provide some restraint to the bottom flanges adjacent to intermediate supports
Fig. 4 - Buckling mode with intermediate torsional restraints
SECTION 6 & 7 - COMPOSITE BEAMS & COMPOSITE BOX GIRDERS

The most common form of shear connector in composite construction is the headed stud. A
structural reliability study was undertaken specifically for AS/NZS 5100.6 (Hicks and Jones,
2013), which considered the results from 113 push tests. This work demonstrated that the
following equations for the design shear capacity fds can be used for stud connectors embedded
in solid concrete slabs and encasements:
fds = φ 0.70d bs
2
fuc (3)
or
fds = φ 0.29d bs2 fcy′ E c (4)
whichever is smaller.
where φ is the capacity reduction factor, which may be taken as φ = 0.8 (previously given as φ =
0.85 in AS 5100.6), dbs is the nominal diameter of the shank of a stud connector, but 16 mm ≤
dbs ≤ 25 mm; fuc is the ultimate tensile strength of the stud material, but not greater than 500
MPa; f'cy is the characteristic strength of the concrete at the age considered, but 16 MPa ≤ f'cy ≤
100 MPa; Ec is the modulus of elasticity of concrete at the age being considered, which may be
( ) (
taken as: E c = ρ 1.2 0.043 fcmi for fcmi ≤ 40 MPa; or E c = ρ 1.2 0.024 fcmi + 0.12 for fcmi > 40 )
MPa, ρ is the density of concrete (kg/m³) and fcmi is the mean value of the in situ compressive
strength.
In AS/NZS 5100.6, Equation (3) has been extended to include high strength structural bolts as
shear connectors. In these circumstances, the constant of 0.70 is replaced with 0.50.
From reliability analyses of 84 push tests using channel shear connectors, it was found that
many of the international rules were on the unconservative side (Hicks et al., 2016; CSA S16-
09, 2009; AISC 360-16, 2016). To remedy this situation, the following equation for the design
shear capacity fds was developed
fds = φ 33.1(t f + 0.5t w )Lsc fcy′ (5)
where tf, tw and Lsc is the flange thickness, web thickness and length of the channel shear
connector in millimetres, respectively.
SECTION 8 & 9 – TRANSVERSE MEMBERS AND RESTRAINTS & MEMBERS SUBJECT TO

AXIAL TENSION
Section 8 brings together the requirements for transverse members and restraints, and their
design requirements, which are based on AS 4100 and NZS 3404. The Section also covers the
requirements of AS 4100 and NZS 3404 for restraints to compression members.
Section 9 covers the design of members subject to axial tension forces that are statically
loaded. Members subject to fatigue loading should also be assessed in accordance with
Section 13 of AS/NZS 5100.6 (see below).
SECTION 10 & 11 - MEMBERS SUBJECT TO AXIAL COMPRESSION & MEMBERS

SUBJECT TO COMBINED ACTIONS
In Eurocode 4, composite columns are limited to steel with a nominal yield strength of
235 MPa ≤ fy ≤ 460 MPa and a concrete characteristic compressive strength of 20 MPa ≤ f'cy ≤
50 MPa. Through structural reliability work that considered an extensive database of 1583 test
results by Kang et al. (2015b), it was found that the existing capacity factors of φ = 0.9 for the
steel and φc = 0.65 for the concrete were justified for the design equations given in the existing
AS 5100.6. However, it was also shown that the provisions for composite columns can be
extended in AS/NZS 5100.6 to permit fy values of up to 690 MPa and f'cy values of up to
100 MPa.
SECTION 12 & 13 – CONNECTIONS & FATIGUE

According to Taplin et al. (2013), the design truck loading originally used in Australia was
derived from the AASHTO H20-S16 combination, where the loading was increased by
approximately 35% and the drive and trailer axles were replaced by tandem axle sets.
Subsequent weigh-in-motion (WIM) data suggested that the average extreme daily events
exceeded the load effects given by this design truck loading (Heywood, 1995). To remedy this
situation, the SM1600 loading model was developed to ensure new bridges would possess
sufficient resilience to future productivity enhancements in road transport. The SM1600 loading
model represents the W80, A160, M1600 and S1600 traffic design loads in AS 5100.2 (2017).
The A160 axle load and M1600 moving traffic load are presented in Fig. 5 (the S1600 static
traffic load is similar to the latter, except that the UDL is increased to 24 kN/m and the 4 ×
360 kN tri-axle set is reduced to 4 × 240 kN).
The fatigue loading provisions in AS 5100.2 (2017) are equivalent to the damage-tolerant
method given in EN 1993-1-9 and the International Institute of Welding (Hobbacher, 2016) in
that, although a 100-year design life is considered in the design of the bridge, the fatigue life is
based on 75-years because of the following assumptions (AS 5100.2 Supp 1, 2007): the belief
that bridges will be inspected regularly and that intervention will occur when fatigue damage is
detected; and the uncertainty associated with the prediction of the fatigue life.
For road bridges, the current fatigue design traffic load effects are determined from 70% of the
effects of a single A160 axle load (Fig. 5(a)), or 70% of the M1600 moving traffic load excluding
the UDL (Fig. 5(b)), whichever is more severe. In both cases, a load factor of 1.0 is used and
the load effects are increased by the dynamic load allowance α. The loads shown in Fig. 5
should be placed within any design traffic lane to maximize the fatigue effects for the
component under consideration. When multiple traffic lanes are used, the current design
approach is to undertake a Palmgren‐Miner summation.
Fig. 5 - Road traffic loads according to AS 5100.2 (a) A160 axle load (b) M1600 moving traffic
load (all dimensions in metres)
To provide greater alignment with international practice, AS/NZS 5100.6 has adopted the IIW
provisions for fatigue design. In addition, to simplify bridge design, the damage equivalent
approach from EN 1993-1-9 has been specifically developed for the loading given in AS 5100.2
(Hobbacher et al., 2016), which has led to the following simple equation for fatigue verification
of road bridges:
γ Ff ∆σ E ,2 = γ Ff [(1 + α )λ∆σ max ] ≤ φMf ∆σ c (6)
where γFf is the load factor for equivalent constant amplitude stress ranges (taken as γFf = 1.0),
∆σE,2 is the characteristic value of equivalent nominal stress range for 2 million cycles, α is a
dynamic load allowance from AS 5100.2, λ is the damage equivalent factor (see Equation (6)
below), ∆σmax is the maximum stress range caused by the fatigue loads specified in AS 5100.2,
φMf is the capacity reduction factor (for high consequence of failure taken as: 0.85 for the
‘damage tolerant’ method, where there will be regular inspection; or 0.75 for the ‘safe life’
method, where there will be little/no inspection over the design life of the structure) and ∆σc is
the fatigue resistance of the detail at 2 million cycles (the detail category).
The damage equivalent factor for road bridges is given by:
λ = λc λL λR λY λM (7)
where λc is for vehicles per day, λL is for consideration of span, λR is the route factor, λY is the
service life and λM is the effect of multiple lanes (if not yet covered in load assumptions).
A similar expression as Equation (7) is given for rail bridges and, for ease of use, the various
damage equivalent factors are given in tabular form. For cases when the designer wishes to
squeeze out the very last remaining capacity from a design or assess existing bridges, the
Palmgren‐Miner summation is supported within an appendix to AS/NZS 5100.6.
SECTION 12 & 13 – BRITTLE FRACTURE & TESTING OF STRUCTURES OR ELEMENTS

Given the introduction of quench and tempered steel plate, the steel types for brittle fracture
have been extended to include AS 3597 products. Also, to remedy the unsatisfactory situation
that the permissible service temperatures according to steel type and thickness are only
appropriate for steels currently made in Australia or New Zealand, rules for evaluating the notch
toughness of non-domestic steel products are given. It is hoped that these rules will be
incorporated within AS/NZS 1163, AS/NZS 1594, AS/NZS 3678 and AS/NZS 3679.1 in due
course.
CONCLUSIONS
Bridge design – Part 6: Steel and composite construction, is the only part of the AS 5100 suite
of standards that has been revised as a joint AS/NZS standard. Overseas steels that have
historically been used in New Zealand bridge design for the last 24 years are to be included in
an appendix. Following the international trend of using less natural resources, design rules for
higher strength steel and concrete are given. The new design rules within the proposed
AS/NZS 5100.6 provide greater alignment with international best practice and, in some cases,
significant improvements are given.
ACKNOWLEDGEMENTS
The authors would like to thank the members of the BD-090-06 Committee for all their hard
work to date; in particular, the Standards Australia Project Manager, June Chen, responsible for
Part 6 and the complete AS 5100 suite of standards. As members of the BD-090-06
Committee, the present authors have prepared this paper according to what they consider to be
the significant changes to the existing AS 5100.6. However, the standard is still being prepared
for public comment and, as a consequence of this, may be subject to change. Any views
expressed in this paper may not necessarily reflect those of the other members of the
Committee, nor those of Standards Australia and Standards New Zealand.
REFERENCES
ANSI/AISC 360-16. (2016). Specification for Structural Steel Buildings. American Institute of
Steel Construction, Chicago, IL.
AS 3597 (2008). Structural and pressure vessel steel - Quenched and tempered plate,
Standards Australia, Sydney, Australia.
AS 4100 (1998). Steel structures, Standards Australia, Sydney, Australia.
AS 5100.2 (2017). Bridge design, Part 2: Design loads, Standards Australia, Sydney, Australia.
AS 5100.2 Supp 1 (2007). Bridge design – Design loads – Commentary (Supplement to AS
5100.2 - 2004), Standards Australia, Sydney, Australia.
AS 5100.5 (2017). Bridge Design, Part 5: Concrete, Standards Australia, Sydney, Australia.
AS 5100.6 (2004). Bridge Design, Part 6: Steel and composite construction, Standards
Australia, Sydney, Australia.
AS 5104 (2005). General principles on reliability for structures, Standards Australia, Sydney,
Australia.
AS/NZS 1163 (2016). Cold-formed structural steel hollow sections, Standards
Australia/Standards New Zealand, Sydney/Wellington, Australia/New Zealand.
AS/NZS 1170.0 (2002). Structural Design Actions, Standards Australia/Standards New
Zealand, Sydney/Wellington, Australia/New Zealand.
AS/NZS 1594 (2002). Hot-rolled steel flat products, Standards Australia/Standards New
AS/NZS 3678: 2016, Structural steel - Hot-rolled plates, floorplates and slabs, Standards
AS/NZS 3679.1 (2016) Structural steel – Part 1: Hot-rolled bars and sections, Standards
AS/NZS 3679.2 (2016). Structural steel – Part 2: Welded I sections, Standards
AS/NZS 4600 (2005). Cold-formed steel structures, Standards Australia/Standards New
AS/NZS 5100.6 (2017). Bridge Design, Part 6: Steel and composite construction, Standards
BC1 (2012). Design guide on use of alternative structural steel to BS 5950 and Eurocode 3,
Building and Construction Authority, Singapore, See http://www.bca.gov.sg (accessed
30/01/14)
CSA S16-09 (2009). Design of steel structures. Canadian Standards Association, Ontario.
DR AS/NZS 2327 (2016). Composite steel-concrete construction for buildings, Standards
EN 1993-1-9 (2005). Eurocode 3: Design of steel structures – Part 1-9: Fatigue, European
Committee for Standardization, Brussels, Belgium.
EN 1994-2 (2005). Eurocode 4: Design of composite steel and concrete structures – Part 2:
General rules and rules for bridges, European Committee for Standardization, Brussels,
Belgium.
EN 10025-1 et seq. (2004). Hot rolled products of structural steels, General technical delivery
conditions, CEN, Brussels, Belgium.
EN 10034 (1993). Structural steel I- and H-sections – Tolerances on shape and dimensions,
CEN, Brussels, Belgium.
Heywood, R (1995) Live loads on Australian bridges – statistical models from weigh-in-motion
data. Australian Civil Engineering Transactions CE37(2), pp. 107–116
Hicks, S., Cao, J., McKenzie, C., Chowdhury, M. and Kaufusi, R. (2016). “Evaluation of shear
connectors in composite bridges”, Research Report 602, NZ Transport Agency, Wellington,
https://www.nzta.govt.nz/resources/research/reports/602/
Hicks, S. and Jones, A. (2013). “Statistical Evaluation of the Design Resistance of Headed Stud
Connectors Embedded in Solid Concrete Slabs”. Structural Engineering International. 23(3), pp.
269-277.
Hobbacher, A. (2016). Recommendations for Fatigue Design of Welded Joints and
Components, IIW Collection, Springer International, Switzerland.
Hobbacher, A.F., Hicks, S.J., Karpenko, M., Thole, F. and Uy, B. (2016). “Transfer of
Australasian bridge design to fatigue verification system of Eurocode 3”. Journal of
Constructional Steel Research, 122. pp. 532–542.
ISO 2394 (1998). General principles on reliability for structures, International Organization for
Standardization, Geneva, Switzerland.
JIS G 3106 (2004). Rolled steels for welded structure, Japanese Standards Association, Tokyo,
Japan.
JIS G 3136 (2005). Rolled steels for building structure, Japanese Standards Association,
Tokyo, Japan.
JIS G 3192 (2005). Dimensions, mass and permissible variations of hot rolled steel sections,
Japanese Standards Association, Tokyo, Japan.
Kang, W-H, Hicks, S, and Uy, B. (2015a). “Safety factors for the resistance of steel sections”.
Australian Journal of Structural Engineering. 16(2). pp. 116-128.
Kang, W-H., Uy, B., Tao, Z. and Hicks, S. (2015b). “Design strength of concrete-filled steel
columns”. Advanced Steel Construction. 11(2), pp. 165-184.
Mulqueen, P.C. (2011). “Creating the Te Rewa Rewa Bridge, New Zealand”, Structural
Engineering International, 21(4). pp. 486-491.
NZS 3404.1 (1997). Steel structures standard, Standards New Zealand, Wellington, New
Zealand.
Taplin, G., Deery, M., van Geldermalsen, T., Gilbert, J., Grace, R. (2013) “A new vehicle
loading standard for road bridges in New Zealand”, Research Report 539, NZ Transport
Agency, Wellington.
Uy, B., Hicks S. and Kang, W-H. (2015). “Australasian advances in steel & composite
structures to enhance cross-border practice”. 17th ASEP International Convention (17AIC),
Pasig City, Philippines.
SHEAR CONNECTIONS BY HEADED STUDS
CLOSE TO THE CONCRETE EDGE
Prof. Dr.-Ing. Ulrike Kuhlmann

University Stuttgart, Germany
sekretariat@ke.uni-stuttgart.de
Dr.-Ing. Jochen Raichle
Consultancy Prof. Dr. -Ing. Kuhlmann, Ostfildern-Nellingen, Germany
Raichle@ing-nellingen.de
Dr.-Ing. Ana M. Pascual

ana.pascual@ke.uni-stuttgart.de
ABSTRACT
Headed studs arranged close to the concrete edge present a limited carrying capacity because
of the splitting forces induced in the concrete. In the past years several investigations have been
carried out at the Institute of Structural Design of the University of Stuttgart on the shear behavior
of the headed studs close to the concrete edge (originally called horizontally lying shear studs)
under static and cyclic loading, considering shear in longitudinal and vertical direction. As a result,
the equations for the design resistance of the headed studs and the fatigue strength have been
provided and included in EN 1994-2 in the Annex C.
Meanwhile, further studies of the behavior have led to possible improvements and enhancements
of the existing rules and these investigations and results are reported in this paper. New re-
searches have especially dealt with the behavior of the connection under cyclic loading in order
to assess the fatigue strength under vertical shear and the influence of the edge distance. As a
consequence, the design equations for the headed stud close to the concrete edge or close to
the concrete surface have been redefined and a proposal has been developed to be included in
the future version of EN 1994.
1. INTRODUCTION
Independently, if the stud is in vertical or horizontal position, there may be a reduction of the
connection capacity due to the splitting forces (Fig. 1) that appear when the center of the stud is
close to the edge of the concrete. The failure mode depends directly on the edge distance ar (see
Fig. 2). On the other hand, this type of connection allows interesting forms of construction, such
as building edge beams in car park construction or slim-floor beams, see (Lam, Dai, Kuhlmann,
Raichle, & Braun, 2015).
Fig. 1 Concrete splitting under longitudinal Fig. 2 Headed studs close to the
shear concrete edge
In bridge construction, the horizontal studs may serve to longitudinally connect the steel girder to
the concrete slab, as for instance in an arch bridge. Similarly, an arrangement of horizontal studs
represents the basis of an innovative composite bridge girder, where the upper flange of the girder
has been replaced by two layers of horizontally lying studs (Fig. 3 and Fig. 4), the lower is em-
bedded into a prefabricated concrete slab and the upper is in concrete in-situ, therefore a reduc-
tion in the steel consumption is achieved together with the constructional advantages of prefabri-
cation. Additionally, this section can be enhanced by using a corrugated steel web for the girder,
which improves the transversal bending capacity and the longitudinal connection.
Fig. 3 Prefabricated composite bridge girder Fig. 4 New section with two layers of horizon-
with horizontally lying studs tally lying studs
The study of the particular behavior of the headed studs close to the concrete edge has been the
objective of several researches, see (Breuninger & Kuhlmann, 2001) (Breuninger, 2000)
(Kuhlmann & Breuninger, 2000) as well as (Kürschner, 2003) (Kürschner & Kuhlmann, 2004)
(Kürschner & Kuhlmann, 2005) under static longitudinal and vertical shear and the combination
of both. Furthermore, also the fatigue strength has been under analyses. The results of these
works are included in Annex C of (EN 1994-2), so they can be used in practice.
Nonetheless, there are still some aspects that require a further development, this is the case of
the fatigue strength of the connection under vertical shear, as it occurs under the fatigue loading
derived from wheel loads, see Fig. 5. In this framework, this paper presents the proposal of en-
hancement of the Annex C based on the works from (Raichle & Kuhlmann, 2016) (Raichle, 2015).
Regarding the fatigue strength under vertical shear, the experimental program developed and the
evaluation of the results is described.
Fig. 5 Fatigue loading under transverse shear
2. STATE OF THE ART AND ENHANCEMENT

There are different geometrical parameters that affect the behavior of the stud connection, but
the most important one is the effective edge distance ar', which is the distance between the bolt
axis and the axis of the reinforcement (Fig. 6). Additionally, it is necessary to distinguish between
middle and edge position, since the edge position is more unfavorable due to the tensional forces
on the studs, not balanced as in the middle position.
The research on the existing rules for headed studs that cause splitting forces in the direction of
the slab thickness in (EN 1994-2) Annex C leads to some technical corrections in order to provide
more appropriate values of strength for common applications and to extend the application scope.
According to the investigations from (Breuninger, 2000) to prevent the pull-out of the headed studs
for the case of an edge position, a length v of the headed studs is required (Fig. 6). These geo-
metrical conditions for the edge position are also included in the Annex C, (EN 1994-2) in clause
C1 (2). Nonetheless, they are currently under revision because they lead to extremely long studs,
and a further research is being carried to introduce alternatively the application of the fastening
technology, (EN 1992-4) in order to verify the pull-out forces.
middle position section A-A edge position
h
ch+ ds /2 n
h
ch+ ds /2 a d
cv+ ds /2 cv+ ds /2
a r,o
a'r,o
a'r,o= a'r a'r,o= a'r
b
cv+ ds /2 cv+ ds /2
ds s dl
Fig. 6 Geometrical parameters in headed studs close to the concrete edge
The design shear resistance of the headed studs in longitudinal and vertical direction can be
determined according to the equation (C.1) and equation (C.4) of (EN 1994-2) Annex C respec-
tively. In both cases the coefficient kv considers whether the arrangement of studs is at edge or
middle position. For simplification, this factor is the same for headed studs in case of longitudinal
and vertical shear. However, (Kürschner & Kuhlmann, 2004) recommends 1.25 for kv to calculate
the shear resistance under vertical shear in middle position. This position and loading in the con-
nection are quite common in practice and the simplified normative regulation (kv=1.14) is too con-
servative and does not lead to economic results, so 1.25 may be used instead.
(Kürschner & Kuhlmann, 2004) studied also the influence of a multi-row array of shear studs under
vertical shear load. The results proved that the vertical shear strength of each row of studs may
be calculated individually. However, the numerical investigations in (Kürschner & Kuhlmann,
2004) verified that for a distance av between the rows of studs smaller than 200 mm, the total
capacity is reduced, and the factor η has to be considered, as it is given in eq. (1).
0.7 ≤ η = av / 200 mm ≤ 1.0 with av ≥ 100 mm (1)
Moreover, the existing rules in Annex C EN 1994-2 section C.1 (3) allow the dimensioning of the
stirrup reinforcement for the splitting forces induced longitudinal shear. Nonetheless, the stirrups
are also required for vertical shear, as it was determined by (Kürschner & Kuhlmann, 2004). Cur-
rently, the tensile force in the stirrups Td according to eq. (C.2) in section C.1 (3) should be im-
proved according to eq. (2) with Fd,L and Fd,V being the stud shear forces in longitudinal and vertical
direction.
Td =0.3 Fd,L+ Fd,V (2)
Related to the fatigue strength of the headed studs under longitudinal loading according to
(Raichle, 2015), the equation (C.5) is derived from (Breuninger & Kuhlmann, 2001) (Kürschner,
2003) (Kürschner & Kuhlmann, 2004) for normal weight concrete:
(ΔPR)m N = (ΔPL, c)m Nc (3)
With
ΔPR Fatigue strength based on difference of longitudinal shear force per stud
ΔPL,c Reference fatigue strength at Nc = 2 × 106 according to Table C.1 in (EN 1994-2)
m Slope of the fatigue strength curve with m = 8
N Number of force range cycles
In Table C.1 the maximum fatigue strength for longitudinal shear without edge influence ΔPL,c is
35.6 kN per stud. This value for a stud of diameter 22 mm leads to a fatigue strength ΔτR of 95
MPa. However, the current standard EN 1994-1-1, section 6.8.3 (3) provides a maximum fatigue
strength for longitudinal shear Δτc not higher than 90 MPa, that means 34.2 kN / stud. The maxi-
mum fatigue strength should be therefore adjusted in Table C.1 to 34.2 kN to be consistent.
On the other hand, the equation (C.5) only makes reference to the longitudinal shear, but it is
assumed that the splitting forces due to fatigue appear similarly under vertical loading. In this
paper the study of the fatigue behavior of the headed studs close to the concrete edge under
vertical shear is presented and also an interaction equation of fatigue strength under the longitu-
dinal and vertical shear, see (Raichle, 2015). The research results also indicate that a fatigue
analysis under shear stress is required for the stirrups.
3. EXPERIMENTAL INVESTIGATIONS
The test specimens correspond to the specimens from the static tests with headed studs at the
edge position under vertical shear from (Kürschner & Kuhlmann, 2004). The most important pa-
rameter on the connection behavior according to previous investigations is the effective edge
distance ar,o´ (definition, see Fig. 6). This parameter also in this case was analyzed and served to
define the two test series: (QE1) with an edge distance of 100 mm and (QE2) with an edge dis-
tance of 50 mm, see Fig. 7. An edge position was adopted on the safe side, because it is more
unfavorable than the middle position. For each series a specimen was tested under static loads.
For the series (QE1), with the distance of 100 mm, 10 specimens were tested, and for the edge
distance of 50 mm, 6 test specimens. The headed studs were of a length of 150 mm and had a
diameter of 22 mm. The concrete compressive strength was individually measured and varied
between 30 and 35 MPa (cylinder strength). Sinusoidal load cycles at similar maximum load level
and different load ranges were applied.
800
Top view
section A-A (series QE1)
150 150
300
a a
a = 165
1300
section A-A (series QE2)

a
200 100
300
A A
Fig. 7 Geometrical parameters in headed studs close to the edge
The cyclic loading was introduced by pressure applied with a girder on the slab in order that the
load was distributed through the slab and all the studs were equally loaded (Fig. 8).
The typical failure of headed studs without edge influence appears due to the fatigue fracture
at the collar of the headed studs. The fatigue crack increases as long as the residual cross
section is no longer able to bear the load. Later on, the failure of the rest of the headed studs
occurs and there is a sudden shearing-off of the studs. This type of failure was also observed
in the specimens tested here, with the studs close to the concrete edge and vertically loaded,
see Fig. 9. In all the specimens, cracks appeared in the concrete between the headed studs.
Additionally, cracks also occurred at the surface of the concrete.
Fig. 8 Test rig of fatigue tests under transverse shear
Fig. 9 Headed studs shearing-off as failure mode. a) View on test specimen, b) View on fac-
ture surface
The second type of failure observed was the concrete breakout on the top (Fig. 10). Outgoing
from the steel part two cracks grew in the concrete that finally joined in a common crack. From
the reinforcement, specifically from the stirrups, in spite of separation cracks compression struts
appeared, which allowed to continue the cyclic loading for a while. A strut-and-tie model illustrates
the tension stresses transferred by the reinforcement that are in balance with the compression
struts transmitted by the concrete. The load taken by the reinforcement prevents the crack for-
mation between the dowels and a sudden failure.
So, together with the shearing-off of the headed studs and the concrete breakout, there was a
third failure mode that took place in two of the tested specimens, the fatigue fracture of the rein-
forcement, see Fig. 11. In both cases, the fatigue fracture was in the upper bending. The failure
of the reinforcement illustrates the necessity to consider in the connection design the verification
of its strength together with the appropriate definition of the mandrel diameter.
In summary, under longitudinal and vertical shear load when the headed studs are close to the
concrete edge a concrete crushing takes place that results on the reduction of the stiffness. Di-
rectly above the stud collar local compressive stresses appear in the concrete, which lead to the
concrete failure. The increasing flexibility in the area of the stud base generates not only the shear
stress but also an unfavorable bending stress at this point, so that, a fatigue crack in the base of
the stud appears. The progressive cracking in the concrete and in the headed studs eventually
leads to the failure either by shearing-off of the stud or concrete break-out.
Fig. 11 Fatigue failure of a stirrup
Zug
a'r
Druck
Pv Pv Pv Riss
Pv Pv Pv Pv
Betondruckversagen
Fig. 10 Concrete breakout failure mode, a) view on test specimen b) belonging strut-and-tie
model
4. FATIGUE STRENGTH UNDER VERTICAL SHEAR

Based on these test results a statistical evaluation has defined an S-N curve, see (Kuhlmann &
Raichle, 2007) (Raichle, 2015) (Raichle & Kuhlmann, 2016). The equation follows in principle eq.
(3) for longitudinal cyclic loading except for the reference value.
(ΔPR)m N = (ΔPV, c)m Nc (4)
With
ΔPR Fatigue strength based on difference of vertical shear force per stud
ΔPV,c Reference fatigue strength at Nc = 2 × 106 according to Table 1
m Slope of the fatigue strength curve with m = 8
N Number of force range cycles
The fatigue strength curves of horizontally lying headed studs under longitudinal and vertical
shear for the two edge distances (ar,´ = 50 mm and ar,´ = 100 mm) and without edge influence are
shown in Fig. 12. For horizontally lying headed studs under longitudinal and vertical shear and an
effective edge distance ar,o´ of 100 mm the fatigue strength is lower than for studs without an edge
influence. Therefore, there has to be a reduction also for an effective edge distance greater than
100 mm. For ar,o´ of 125 mm the fatigue strength of headed studs without edge influence is
reached which therefore give the upper limit.
Load difference DPc,V [kN]

100
80
60
40 a v x
w y
20
10
f
8
6
Small edge distance Small edge distance Headed studs without
Nc = 2∙106
4 vertical shear longitudinal shear edge influence
a: ar' ≥ 100 mm v: ar' ≥ 100 mm x: d = 25 mm
2 f: ar' = 50 mm w: ar' = 50 mm y: d = 22 mm
1
10.000 100.000 1.000.000 10.000.000 100.000.000
Cycles N [-]
Fig. 12 Fatigue strength S-N curves
Based on the test results (Kuhlmann & Raichle, 2007) (Raichle, 2015) (Raichle & Kuhlmann,
2016) and the experiences on the fatigue behavior of horizontally lying headed studs under lon-
gitudinal loading as well as headed studs without an edge influence the fatigue strength for
headed studs with edge influence under transverse loading has been derive. As a result, the
Table C1 of the Annex C (EN 1994-2) should include the values of ΔPV,c for the fatigue strength
under vertical shear force, Table 1 is proposed to replace it.
Table 1. – Fatigue strength ΔPR for headed studs close to the concrete
edge.
Longitudinal shear Vertical shear
ar′ (mm) ΔPL,c (kN) ΔPV,c (kN)
50 24.9 8.9*
100 34.2(35.6) 27.7*
≥ 125 34.2(35.6) 34.2

NOTE Intermediate values should be determined by linear interpolation
⃰ In case of cracked concrete slab due to tension a reduction with k = 0.5
is necessary.
The parametric studies in (Kuhlmann & Raichle, 2008), (Raichle & Kuhlmann, 2016) show that
for concrete strength fck<30 MPa and stud diameter of 19 mm some changes of the data in Table
C.1 (EN 1994-2) are required. These changes could be included in the table adding the following
text:
In case of concrete strength fck<30 MPa the fatigue strength ΔPc should be reduced with fck/30.
The values in Table C.1 are valid for stud diameter ≥22 mm. In case of stud diameter of 19 mm
the fatigue strength should be reduced to 75% of the values on Table C.1.
In case of cracked concrete slab due to tension a reduction with k=0.5 is necessary (Kuhlmann &
et al., 2013) (Kuhlmann & Maier, 2012)In the experiments, see Fig. 11 also fatigue cracking of
the stirrups was observed. These cracks occurred in the mandrel region. Therefore, besides the
verification of the strength capacity of the stirrups by using the eq. (2), it is necessary to check
their fatigue capacity with (EN 1992-1-1) Section 6.8.4 taking into account the mandrel diameter.
4. INTERACTION OF LONGITUDINAL AND VERTICAL SHEAR

For some constructions, for example composite beams without upper steel flange, longitudinal
and transverse loading can occur simultaneously. The longitudinal shear loading may be intro-
duce by global bending, whereas the local load introduction by the wheel load may induce vertical
shear loading. Static experiments of (Kürschner & Kuhlmann, 2004) under the combined action
of both showed that in the edge position the interaction should be taken into account.
Since no additional fatigue tests have been carried out for this case of combined loading, a design
recommendation, which is on the safe side, is given at this point. A typical equation for the inter-
action proof is the following expression eq. (5):
0,75 0,75
𝛾𝐹𝑓 𝛥𝑃𝐿,𝐸2 𝛾𝐹𝑓 𝛥𝑃𝑉,𝐸2
( ) + ( ) ≤1 (5)
𝛥𝑃𝐿,𝑐 ⁄𝛾𝑀𝑓,𝑆 𝛥𝑃𝑉,𝑐 ⁄𝛾𝑀𝑓,𝑆
ΔPL,E2 and ΔPV,E2 can be determined according to clause 6.8.6.2 (EN 1994-2).
The value 0.75 for the exponent in the equation (5) is chosen conservatively in absence of an
extensive test program. Similarly, as for the static resistance an interaction should only be ob-
served for the edge position and not for the middle position (Kürschner & Kuhlmann, 2004) this is
also assumed to be the case for fatigue.
For the stirrups a fatigue design as given in (EN 1992-1-1) clause 6.8.4 is necessary. The mandrel
diameter should be considered.
5. CONCLUSIONS
The ultimate resistance as well as the fatigue strength of horizontally lying studs or better headed
studs close to the concrete edge are strongly influenced by the effective edge distance ar´ (see
Fig. 2) because the shear forces may cause splitting forces, which lead to an earlier failure as for
headed studs positioned without any edge influence. Therefore, not the horizontal or vertical ar-
rangement of the shear studs is of importance but the distance to the concrete edges. So the title
of this chapter, which is at the moment named “Headed studs that cause splitting forces in the
direction of the slab thickness” should be changed accordingly when taken over in a new version
of Eurocode 4. For the moment it is also planned to transfer this content of the existing Annex C
of EN 1994-2 to the future EN 1994-1-1, as the rules are of course not only valid for bridges but
also for buildings and it is more due to the history of development that it is placed in the bridge
part Eurocode 4.
This paper summarizes some of the improvements which have been suggested for the conversion
of this Annex to the new version of Eurocode 4 which is under development at the moment. Some
of the improvements are simply clarifications, some are also real enhancements due to research
which has been realized in the meantime. This especially concerns the fatigue behavior under
transverse shear loading for which design rules are introduced similar to what has been included
for longitudinal shear.
Some open questions are still left. This concerns the interaction of longitudinal and vertical shear
but also the geometric restrictions given so far in order to prevent pull-out failure for horizontally
lying studs in edge position. For the latter research has started to consider also the experience
from the fastening technique, meanwhile given in EN 1992-4.
These new developments now allow a design for both static and fatigue loading in longitudinal
and vertical direction and the development and application of interesting new composite sections.
REFERENCES
Breuninger, U. (2000). Zum Tragverhalten liegender Kopfbolzen unter Längsschub-
beanspruchung. Dissertation, Institute of Structural Design, University of Stuttgart, Issue
No. 2000-1, ISSN 1439-3751.
Breuninger, U., & Kuhlmann, U. (2001). Tragverhalten und Tragfähigkeit liegender
Kopfbolzendübel unter Längsschubbeanspruchung. In: Stahlbau, Volume 70 , Issue 11,
Berlin: Ernst & Sohn., S. 835–845.
EN 1992-1-1. (2004). Eurocode 2. Design of concrete structures – Part 1-1: General rules and
rules for buildings. Brussels, Belgium: Comité Européen de Normalisation.
EN 1992-4. (2015). Eurocode 2, draft. Design of concrete structures – Part 4: Design of fastenings
for use in concrete, prEN1992-4. Brussels, Belgium: Comité Européen de Normalisation.
EN 1994-2. (2005). Eurocode 4. Design of composite steel and concrete structures. Part 2:
General rules and rules for bridges. Brussels, Belgium: Comité Européen de
Normalisation.
Kuhlmann, U., & Breuninger, U. (2000). Behaviour of Horizontally Lying Studs with Longitudinal
Shear Force. Hajjer, J. et al. (Ed.): Proceedings of Composite Constructions in Steel and
Concrete IV, May 28 to June 2, Banff, Alberta, Canada. pp. 438-449.
Kuhlmann, U., & et al. (2013). Sustainable Steel-Composite Bridges in Built Environment (SBRI),
Final Report, RFSR-CT-2009-00020, RFCS-Publication, EUR 26322 EN EUROPEAN
COMMISSION. Directorate-General for Research and Innovation Directorate G —
Industrial Technologies Unit G.5 — Research Fund for Coal and Steel, RTD-
PUBLICATIONS@ec.europa.eu.
Kuhlmann, U., & Kürschner, K. (2004). Structural Behaviour of Horizontally Lying Shear Studs.
Leon, R. and Lange, J. (Ed.): Proceedings of Composite Constructions in Steel and
Concrete V, July 18t to 23, Kruger National Park, South Africa. pp. 534-543.
Kuhlmann, U., & Maier, P. e. (2012). Optimizing bridge design by improved deterioration models
through fatigue tests. Frangopol, D. and Biondini, F. (Ed.): Proceedings of 6th International
Conference on Bridge Maintenance, Safety and Management (IABMAS), Stresa, Lake
Maggiore, Italy. pp. 1816-1821.
Kuhlmann, U., & Raichle, J. (2007). Fatigue Behaviour of Horizontally Lying Shear Studs. In:
Eligehausen, R. (Ed.), Connections between Steel and Concrete. Proceedings of the 2nd
International Symposium 4th to 7th September, Stuttgart, Germany, Vol. 2, pp. 1359-1368.
Kuhlmann, U., & Raichle, J. (2008). Ermüdungsverhalten liegender Kopfbolzen infolge
Querschub. Institute of Structural Design, University of Stuttgart. Research report,
research project FE 15.0407/2004/CRB for Bundesanstalt für Straßenwesen (bast), Issue
99, Publisher: Bremerhaven: Wirtschaftsverlag NW, Verlag für Neue Wissenschaft. ISBN
978-3-8650.
Kuhlmann, U., & Raichle, J. (2008). Headed Studs Close to the Concrete Surface - Fatigue
Behaviour and Application. Leon, R. (Ed.): Proceedings of Composite Constructions in
Steel and Concrete VI, July 20.-24., Colorado, USA. pp. 26-38.
Kürschner, K. (2003). Trag- und Ermüdungsverhalten liegender Kopfbolzen im Verbundbau.
Dissertation, Institute of Structural Design, University of Stuttgart, Issue No. 2003-4, ISSN
1439-3751.
Kürschner, K., & Kuhlmann, U. (2004). Trag- und Ermüdungsverhalten liegender
Kopfbolzendübel unter Quer- und Längsschub. In: Stahlbau, Volume 73, Issue 7, pp. 505–
516, Berlin: Ernst & Sohn.
Kürschner, K., & Kuhlmann, U. (2005). Mechanische Verbundmittel für Verbundträger aus Stahl
und Beton. In: Stahlbaukalender 2005, Editor: U. Kuhlmann, pp. 455–534, Berlin: Ernst &
Sohn.
Lam, D., Dai, X., Kuhlmann, U., Raichle, J., & Braun, M. (2015). Slim-floor construction - design
for ultimate limit state. In: Steel Construction, Volume 8, Issue 2, p. 79-84 DOI:
10.1002/stco.201510019, Berlin: Ernst & Sohn. – ISSN 1867-0539.
Raichle, J. (2015). Randnahe Kopfbolzen im Brückenbau. Dissertation, Institute of Structural
Design, University of Stuttgart, Issue No. 2015-2, ISSN 1439-3751.
Raichle, J., & Kuhlmann, U. (2016). Randnahe Kopfbolzen – Ermüdungsverhalten unter
Querschub. In: Stahlbau, Volume 85, Issue 2, pp. 153–160, Berlin: Ernst & Sohn.
CONSIDERATION ON COMPOSITE BEAM CONNECTION DUE TO
PLASTIC DEFORMATION CAPACITY AND STRENGTH RATIO
Yuko Shimada
Faculty of Engineering, Chiba University
Chiba, Japan
yshimada@faculty.chiba-u.jp
Satoshi Yamada
FIRST, Tokyo Tech
Yokohama, Japan
yamada.s.aa@m.titech.ac.jp
ABSTRACT
This paper is a fundamental study of quantitatively relationship between plastic deformation
capacity and strength in composite beam. First, analytical method of composite beam based on
the previous experiment results is shown. Specifically, there are three steps in the analysis;
invalid area of beam web near scallop is set according to previous experiment results, stress
balance between tension and compression in the section is kept under monotonic loading, and
deformation of a section is calculated assuming Bernoulli-Euler theory. Secondly, the
relationship between plastic deformation of composite beam and geometric condition of beam-
to-column connection or material characteristics by parametrical study. In this paper, the effect
of steel strength and width-to-thickness ratio of column is focused. All composite beam in this
paper is presupposed fracture of lower beam flange. Thus, the strain near scallop at lower
flange is used as an index that shows progression of plastic deformation.
1. INTRODUCTION
In seismic design, the detail of beam-to-column connections should prevent early fracture to
retain seismic performance of buildings. In order to guarantee sufficient strength of connection
for the strength that it makes plastic hinge at beam edge, the maximum strength of connection
is designed by connection coefficient  such as Eq.(1). The formula is stated in Technical
criteria references on structure of buildings in Japan [1], and it should be generally used in steel
structure for designing safety building against earthquake.
j Mu   b M p (1)
here, jMu is the maximum strength of connection,  is connection coefficient, and bMp is full-
plastic moment of beam.
Generally, beams of building are postulated as bare steel on the calculation of strength in the
present seismic design, and composite effect by concrete slab on steel beam is neglected. In
previous studies, it has been already known that the strength of composite beam is greater than
bare steel beam by moving the neutral axis from center of web to near concrete slab [2]-[6].
Moreover, the plastic deformation of composite beam which presupposed fracture of lower
beam flange is smaller than bare steel beam, however, it is not clear that composite beam-to-
column connection can keep enough energy absorption capacity as bare steel beams. In order
to judge whether composite beam-to-column connection have enough energy absorption
capacity or not, it should evaluate not only the growth of strength by composite effect but also
the deference of the plastic deformation capacity between composite beam and bare steel
beam.
The authors have been considered the relationships between strength and deformation
capacity of composite beam recently. In [7], the plastic strength was defined as a strength when
the plastic hinge is made at the composite beam end, and in [8], connection coefficient of
composite beam was defined as a ratio of the maximum strength of composite beam
connection to the plastic strength of composite beam.
The connection coefficient of composite beam is defined according to the definition of the
connection coefficient of bare steel beam. The recommended value of connection coefficient of
bare steel beam in seismic design is decided by considering the relationships between plastic
deformation capacity and steel grade, strain hardening, connection detail. However, the
connection coefficient of composite beam is not used in seismic design. This reason is that it
has not been examined the relationships between connection coefficient and plastic
deformation capacity of composite beam yet. It should be clarify the effects of above elements
and other elements such as concrete strength and slab thickness for the connection coefficient
of composite beam to understand the recommended value for seismic design.
This paper presents an algorithm of composite beam, which has RHS column, H-shaped beam,
and slab with deck plate to achieve preliminary trend of plastic deformation capacity. To
validate the performance of our algorithm, experimental results in [8] is used. The composite
effect of slab to the plastic deformation capacity of composite beam is examined based on the
algorithm.
2. Algorithm of composite beam under one-way positive bending

The algorithm in this paper is based on the analysis method shown in [9]. Assuming Bernoulli-
Euler theory, a moment and a curvature in a section are obtained according to calculation of
stress balance in the section. The relationship between bending moment and rotation angle at
beam end are calculated by integrating these moments and curvatures in each section from
beam-to-column connection to beam end. There are two differences between this algorithm and
previous analysis method; existence of concrete slab and representation of moment
transmission efficiency of beam web.
Under random loading as similar as earthquake, the plastic deformation capacity of steel beam
is affected by loading protocol. In this paper, the aims are grasping basic trends between plastic
deformation capacity of composite beam and sizes of steel members, material characteristics of
steel beam. Therefore, load is one-way positive moment.
In the positive-moment region of a composite beam, the effective width of concrete slab is
changing according to the location of beam section, however, the effective width in this paper is
determined as same as column width as shown Figure 1, regardless the location of section.
Beam width
Concrete slab thickness

adobe the steel deck
Slab elements
Fig. 1 – Concrete slab setting
Previous studies pointed out that moment transmission efficiency of beam web decreased at
the beam end when the partial loss of beam section to the weld access hole and local out-of-
plane deformation of RHS column flange are occurred. These are represented in our algorithm
as shown in Figure 2. If the beam end is welded to the enough thick column flange, the
maximum bending moment of composite beam connection is obtained as cjMu [6]. While there is
local out-of-plane deformation, the area of the beam web near beam end is not transferred
stress. The height of this area (hr) is calculated by the balance of each stress block as shown in
Figure 2. In the calculation of moment in the connection, it is assumed that the lower beam
flange reach to ultimate stress, and the upper beam flange beam web reach to yield stress. If
the moment in case that the all of the compression side of beam web is not transferred stress is
larger than cjMu, the stress balance is re-set according that stress of upper beam flange
decreases than yield stress. Here, cjMu is originally not considered weld access hole, therefore,
cjMu in this paper is modified for existence of weld access hole.
Steel hysteresis model is based on the results of coupon test. That is, the tension side of steel
is used the relationship between nominal stress and nominal strain obtained by each coupon
test of beam flange and beam web. While, the compression side of steel is used the following
equipment; this is obtained by replacing from the tension side value. When the nominal strain is
larger than the uniform elongation, the calculation is continued as the strain progress during the
stiffness of the tension side is zero.
Concrete hysteresis model is shown in Figure 3. In the compression side of concrete, the
maximum strain is assumed 0.3%, the maximum stress is assumed 1.3 c. When the
compression strain is smaller than the maximum strain, the hysteresis model is used the
formula modified from the relationship between stress and strain shown in [10]. While, the
compression strain is larger than the maximum strain, the strain progress during the stiffness of
the tension side is zero. In the tension side of concrete, the calculation is continued as the
strain also progress during the stiffness of the tension side is zero.
The length from
column surface to
the top of weld
access hole
The area of non-

transmission
bending stress
(a) Case bM<cjMu (b) Case bM=cjMu (c) Case bM>cjMu
C  N fy  N fu bM: Beam bending moment

hw ( b ) 
N wy fu: Tension strength of beam flange
    
M ( b )  N fu d j  t bf  C t bf / 2  t dp  t s / 2  N wy hw ( b ) t bf / 2  d j  S r  hw ( b ) / 2  fy: Yield strength of beam flange
bw: Yield strength of beam web
In case of bM<cjMu or bM=cjMu c: Compression strength of concrete slab
  Hb: Beam height
1  4 cj M u  M ( b ) 
hwU 
2

 d j  2 S r  hw ( b )   d j 2
 2 S r  hw ( b ) 
N wy
 Bb: Beam width

 tbf: Beam flange thickness
hwL  hwU  hw (b ) tbw: Beam web thickness
Sr: Weld access hole height
hr  d j  2 S r  hwU  hwL ts: Concrete slab thickness adobe the deck
tdb: Steel deck height
In case of bM>cjMu
Bc: Column width

hwL  t bf / 2  d j  S r 
 M ( b )  cj M u  C=c ts Bc: Concrete strength
 t bf / 2  d j  Sr 
2
 
 2 hw ( b ) t bf / 2  d j  S r  hw ( b ) / 2 
N wy
 Nfu=fu tsf Bb: Beam flange ultimate strength
 
Nfy=fy tsf Bb: Beam flange yield strength
hr  d j  2 S r  hwL Nwy=wy tbw : Beam web yield strength
dj= Hb-2tbf: Beam inner height between both
flanges
Fig. 2 – Setting & condition of moment transmission efficiency
[0    0.3%]
   
  1.3 f c  2  
0.003  0.003 
[ 0.3%   ]
  1.3 f c
Fig. 3 – Concrete hysteresis model
3. The comparison between test results and analysis results
In order to examine the validation of the algorism, the analysis results are compared to test
results. The test results including specimen number are based on the database for completely
composite beam under cyclic loading tests [8]. According that the skeleton curve under cyclic
loading test results is equivalent to the relationship between moment and rotation angle under
monotonic loading, the comparisons are conducted. The skeleton curve is obtained by
relationship between moment and rotation angle in test results [11], [12].
Figure 5 shows the comparisons of relationship between moment and rotation angle. The
specimen name is based on the database of composite beam [8]. The analysis results are
generally smaller than test results. According to the results, the beam height has little effect on
the elastic stiffness and the moment. From Ex01 to Ex03 results, the effect of moment
transmission efficiency is little. Also, the ratio of section area of web to flange has also little
effect on the moment. In Figure 5, the comparisons of strain progress of the lower beam flange
3000
M[kNm] Test result 2000 M[kNm] 2500 M[kNm]
2500 2000
1500
2000
1500
1500 Analysis 1000
1000
1000 result 500
500 500
θ[rad] θ[rad] θ[rad]
0 0 0
0 0.01 0.02 0 0.01 0.02 0 0.01 0.02
ε ε ε
0.10 0.10 0.10
0.08 Analysis 0.08 0.08
0.06
result 0.06 0.06
0.04 0.04 0.04
0.02 Test result 0.02 0.02

0.00 0.00 0.00
0 0.01 0.02 0 0.01 0.02 0 0.01 0.02
D01 D02 D03
1200 M[kNm] 1000 M[kNm] 1200 M[kNm]

1000 800 1000
800 800
600
600 600
400
400 400
200 200 200
0 0 0
0 0.01 0.02 0 0.01 0.02 0 0.01 0.02
ε ε ε
0.10 0.10 0.10
0.08 0.08 0.08
0.06 0.06 0.06
0.04 0.04 0.04
0.02 0.02 0.02
0.00 0.00 0.00
0 0.01 0.02 0 0.01 0.02 0 0.01 0.02
Ex01 Ex02 Ex03

Fig. 5 – The comparison between the analysis results and test results
at the smallest section of web are also shown. Generally the analysis results are corresponded
to test results. According to these progresses, the effects of beam height and the moment
transmission efficiency are also little.
4. Parametric studies
4.1 Procedure of the studies
In order to clarify the effect of steel strength and the size of columns and beams to plastic
deformation capacity of composite beam, parametric studies was conducted. The common
parameters are steel kinds (beams are SN400B and columns are BCR295), the beam length
(2,350mm), steel yield stress, and detail of concrete slab. The steel yield stresses are used the
average values based on the survey of the steel stresses [13], that is, the yield stress of column
is 392N/mm2, the yield stress of beam is 305N/mm2. The concrete slabs are designed with
similar deck to previous test [7], that is, deck height is 75mm and slab thickness on the deck is
100mm. The tension strength of concrete is designed 21N/mm2.
There are three parameters in this examination; the combination of the column and beam sizes,
the existence of concrete slab, and the yield ratio of beam material. The beams in this study are
set RH-400x200x8x13 and RH-600x200x11x17 according to the general beam size of middle-
and-low rise buildings. The columns are set basically □350x350x12 or □350x350x19 in both
beam. In case that the beam is RH-400x200x8x13, □300x300x16 and □300x300x19 are also
added. This is because that these beam and column combinations have almost 1.3 on the
connection coefficient of bare steel beam, however, the value of connection coefficient of
composite beam are valid. The purpose of changing the combinations subcomponents is
considered to the moment transmission efficiency. Also, the deference of slab effect is
considered according to two size of beam and concrete slab with constant thickness. The yield
ratio of beam material are set two case; 71.6% and 67.4%, which are corresponded to the sum
or difference to the average value and the standard deviation of yield ratio [13].
Table 1 – Parameter List & energy dissipation capacity

Wp
Beam Column Yield Ratio
sα cα Strain 2% Strain 5% Strain 10%
[mm] [mm] (σy/σu) Bare steel Composite Bare steel Composite Bare steel Composite
67.4% 1.31 1.29 3.0 2.9 21.2 14.9 61.8 43.4
□-300×300×16 1.24 1.24 3.0 3.0 18.2 13.2 53.1 36.2
71.6%
67.4% 1.31 1.33 3.0 3.0 21.2 14.7 61.8 44.0
□-300×300×19 1.24 1.28 3.0 3.0 18.2 13.0 53.0 36.5
71.6%
H-400×200×8×13
67.4% 1.30 1.20 2.7 2.5 16.1 6.6 50.0 26.8
□-350×350×12 1.23 1.15 2.7 2.5 12.9 6.3 41.7 21.6
71.6%
67.4% 1.31 1.32 3.0 3.0 21.2 15.3 61.8 42.1
□-350×350×19
71.6% 1.24 1.27 3.0 3.1 18.2 13.2 53.1 36.8
67.4% 1.15 1.06 3.0 2.7 11.2 8.2 41.5 24.8
□-350×350×12 1.09 1.01 3.0 2.7 9.9 7.9 31.1 20.6
71.6%
H-600×200×11×17
67.4% 1.22 1.17 3.7 4.3 17.1 16.4 62.3 62.5
□-350×350×19 1.16 1.14 3.6 4.4 12.9 13.6 49.8 50.4
71.6%
*’strain -%’ shows the strain value of the lower beam flange
4.2 The relationships between moment and rotation angle

Figure 6 shows the relationships between moment and rotation angle in bare steel beam and
composite beam obtained by the algorism as showed at chapter 2. Here, every rotation angle is
divided by bp (the idealized rotation angle of bare steel beam to elastic-theory deformations
when the moment is the full-plastic moment). Generally, the moment of composite beam is 1.5
times larger than the moment of bare steel beam, although rotation angle of composite beam is
smaller than the value of bare steel beam. When the beam height is 600mm and skin-plate is
relatively thin, both moments of composite beam and bare steel beam are smaller than the
moment with thick skin-plate. Moreover, deformation capacity of composite beam is affected by
beam height.
2.0 2.0
M/bMp M/bMp
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
1.0 1.0
□-350×19 Comp. □-350×19 Comp.
0.8 0.8
□-350×12 Comp. □-350×12 Comp.
0.6 0.6
□-350×19 Non-slab □-350×19 Non-slab
0.4 □-350×12 Non-slab 0.4 □-350×12 Non-slab
0.2 0.2
θ/bθp θ/bθp
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Yield Ratio 67.4%, H-600x200x11x17 Yield Ratio 67.4%, H-400x200x8x13

2.0 2.0
M/bMp M/bMp
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
1.0 1.0
□-300×19 Comp. □-300×19 Comp.
0.8 0.8
□-300×16 Comp. □-300×16 Comp.
0.6 0.6
0.4 □-300×16 Non-slab 0.4 □-300×16 Non-slab
0.2 0.2
θ/bθp θ/bθp
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

2.0 2.0
M/bMp M/bMp
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
1.0 1.0
□-350×19 Comp. □-350×19 Comp.
0.8 0.8
□-350×12 Comp. □-350×12 Comp.
0.6 0.6
0.4 □-350×12 Non-slab 0.4 □-350×12 Non-slab
0.2 0.2
θ/bθp θ/bθp
0.0 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Fig. 6 – The relationships between moment and rotation angle
Figure 7 shows the comparisons of strain progress of the lower beam flange at the smallest
section of web. The horizontal axis is dimensionless rotation angle bp as same value in
Figure 6. When bp is under 2.0, the strain of each model are similar regardless the existence
of concrete slab, beam height, and moment transmission efficiency. After that, the strain of
composite beam are increased immediately than bare steel beam through bp is small. This
trend is remarkable in case the beam height is 400mm, or skin-plate is relatively thin. On the
other hand, yield ratio is little affected to the relation of strain and rotation angle. One of the
effective reasons that plastic deformation capacity of composite beam is smaller than the value
of bare steel beam is rise of the plastic neutral axis of beam [7]. In composite beam, the plastic
neutral axis moves to near the upper beam flange. Therefore, plastic deformation capacity of
composite beam decreases according to plastic area of composite beam is smaller than bare
steel beam accompany with strain hardening.
10 The lower 10 The lower

9 flange strain 9 flange strain
8 [%] 8 [%]
7 7
6 6
5 □-350×19 Comp. 5
4 □-350×12 Comp. 4 □-350×19 Comp.
3 □-350×19 Non-slab 3 □-350×12 Comp.
2 2
□-350×12 Non-slab
1 1
θ/bθp θ/bθp
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

8 [%] 8 [%]
7 7
6 6
5 5
4 □-300×19 Comp. 4 □-300×19 Comp.
3 □-300×16 Comp. 3 □-300×16 Comp.
2 2
1 1
θ/bθp θ/bθp
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

8 [%] 8 [%]
7 7
6 6
5 □-350×19 Comp. 5
4 □-350×12 Comp. 4 □-350×19 Comp.
3 □-350×19 Non-slab 3 □-350×12 Comp.
2 2
□-350×12 Non-slab
1 1
θ/bθp θ/bθp
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Fig. 7 – The relationships between strain and rotation angle
4.3 Plastic energy dissipation capacity and damage of beam
The plastic deformation capacity of beams is deeply corresponded to the strain of the lower
beam flange at the smallest section of web as shown above. The strength of composite beam is
generally smaller than the value of bare steel beam, therefore, plastic energy dissipation
capacity of composite beam and bare steel beam are focused in this section. According to
integrate the moment at the beam end and rotation angle, plastic energy dissipation capacity Wp
is obtained. Wp in composite beam or bare steel beam have almost linear relationship to the
strain of the lower beam flange after these beams are yield as shown in Figure 8. Wp in
composite beam is smaller than the value of bare steel beam after the strain of the lower beam
flange is 5%, this trend is remarkable when the beam height is relatively small and moment
transmission efficiency of web is small according to column flange is thin. Plastic energy
dissipation capacity in composite beam cWp and plastic energy dissipation capacity in bare steel
beam bWp of each parametric model are also shown in Table 1.
70 Wp 70 Wp
[kNm･rad] [kNm･rad]
60 60
□-350×19, Comp. □-350×19, Comp.
50 □-350×12, Comp. 50 □-350×12, Comp.
40 □-350×19, Non-slab 40 □-350×19, Non-slab
□-350×12, Non-slab □-350×12, Non-slab
30 30
20 20
10 10
The lower flange strain [%] The lower flange strain [%]
0 0
0 2 4 6 8 10 0 2 4 6 8 10

Fig. 8 – Plastic energy dissipation capacity and strain at the lower beam flange relationships
In order to show the effect of slab and member size, plastic energy dissipation ratio Wp/bnWp is
decided. The numerator is plastic energy dissipation capacity in each parametric model Wp. The
denominator is the plastic energy dissipation capacity in the bare steel beam (bnWp) which have
RH-400x200x8x13 as beam, □350x350x12 as column, and 67.4% as member yield ratio. The
reason that the model is selected as the denominator of the ratio is its connection coefficient as
bare steel beam is 1.3. As shown chapter 1, in seismic design in Japan, the connection
coefficient of bare steel beam is very important and it is recommended to over 1.3 in steel beam.
Therefore, using the plastic energy dissipation capacity in bare steel beam which connection
coefficient is 1.3 as the denominator, the plastic energy dissipation ratio means the structural
margin of plastic deformation capacity in composite beam or bare steel beam against demand
value of bare steel beam in seismic design.
In this section, three timings are focused as the example of plastic deformation; the strain of the
lower beam flange is 2%, 5%, and 10%. According to the stress and strain relation of the lower
beam flange, 2% strain is showed when it is just after yield, 5% strain is showed when strain
hardening is progressed, and 10% strain is showed when it is just before fracture occurred.
Figure 9 shows plastic energy dissipation ratio Wp/bnWp in these three timings. When the strain
of the lower beam flange is 2% as shown Figure 9 (a) & (d), Wp/bnWp in composite beam have
little deference to Wp/bnWp in bare steel beam. However, the strain of the lower beam flange is
5% and 10%, Wp/bnWp in composite beam decreased 30-50% than Wp/bnWp in bare steel beam.
The decreasing of Wp/bnWp in composite beam is remarkable when column flange thickness is
12mm, that is, moment transmission efficiency is small. In case of beam height is 600mm which
the connection coefficient in bare steel beam is originally small, the differences of Wp/bnWp
between composite beam and bare steel beam is little. As shown in Figure 9 (d)-(f), high yield
ratio is affected to the connection coefficient in bare steel beam is small, however, the
decreasing of Wp/bnWp in composite beam is small. Moreover, if beam and column combinations
have sufficient plastic deformation capacity as bare steel beam as shown gray area in Figure 9
(a)-(c), concrete slab make plastic deformation capacity of the beam small.
Wp/bnWp Wp/bnWp Wp/bnWp

1.4 1.4 1.4
□-300×16 □-300×19 □-300×16 □-300×19 □-300×16 □-300×19
□-350×12 □-350×19 H-400 □-350×12 □-350×19 H-400 □-350×12 □-350×19 H-400
1.2 □-350×12 □-350×19 ←H-600 1.2 □-350×12 □-350×19 ←H-600 1.2 □-350×12 □-350×19 ←H-600
Non-slab Non-slab Non-slab
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0

1.00 1.10 1.20 1.30 sα 1.00 1.10 1.20 1.30 sα 1.00 1.10 1.20 1.30 sα
(a) Y.R. 67.4%, strain=2% (b) Y.R. 67.4%, strain=5% (c) Y.R. 67.4%, strain=10%
Wp/bnWp Wp/bnWp Wp/bnWp
1.4 1.4 1.4
□-300×16 □-300×19 □-300×16 □-300×19 H-400 □-300×16 □-300×19
□-350×12 □-350×19 H-400 □-350×12 □-350×19 □-350×12 □-350×19 H-400
1.2 □-350×12 □-350×19 ←H-600 1.2 □-350×12 □-350×19 ←H-600 1.2 □-350×12 □-350×19 ←H-600
Non-slab Non-slab Non-slab
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0

1.00 1.10 1.20 1.30 sα 1.00 1.10 1.20 1.30 sα 1.00 1.10 1.20 1.30 sα
(d) Y.R. 71.6%, strain=2% (e) Y.R. 71.6%, strain=5% (f) Y.R. 71.6%, strain=10%
*’strain -%’ shows the strain value of the lower beam flange
Fig. 9 – Plastic energy dissipation ratio & connection coefficient in bare steel beam relationships
5. Conclusion
Based on previous cyclic loading test of wholly composite beam, the algorithm of composite
beam with RHS column, deck-plate typed concrete slab is shown. In order to achieve
preliminary trend of plastic deformation capacity of composite, parametric studies was
conducted. Focused on the energy dissipation ratio as comparing to the plastic deformation
capacity between composite beam and bare steel beam, obtained results shows that energy
dissipation capacity in composite beam are almost similar to bare steel beam just after beams
have been yield. However, energy dissipation capacity in composite beam are decreased after
plastic deformation of beams have been increased. The trend is remarkable in case that both
beam height and moment transmission efficiency are small. Moreover, if beam and column
combinations have sufficient plastic deformation capacity as bare steel beam, concrete slab
make plastic deformation capacity of the beam small.
REFERENCES
[1] Editorial committee of technical criteria references on structure of buildings (2007).
Technical criteria references on structure of buildings Ver. 2007.
[2] Igarashi S, Inoue K, Kim SE, & Tada M (1983). Experimental studies on the Elastic-
plastic hysteretic behavior of fully composite beams subjected to earthquake type of loading.
Journal of Structural and Construction Engineering, AIJ, 333, 63-72, JAPAN
[3] Tateyama E, Inoue K, Sugimoto S, & Matsumura H (1988). Study on ultimate bending
strength and deformation capacity of H-shaped beam connected to RHS column with through
diaphragms. Journal of Structural and Construction Engineering, AIJ, 389, 109-121, JAPAN.
[4] Okada K, Oh SH, & Yamada S (2003). Effect of joint efficiency at beam-to-column
connection on ductility capacity of composite beams. Journal of Structural and Construction
Engineering, AIJ, 573, 185-192, JAPAN.
[5] Ishii T, & Morita K (2001). Experimental Study on Fracture Behavior of Composite
Beam-to-SHS Column Connection. JSSC Journal of Constructional Steel, JSSC, 8(31), 65-79,
JAPAN.
[6] Tanaka T, Suita K, Asakura N, Tsukada T, Uozumi N, & Takatsuka K (2015). Influence
of floor slab on deformation capacity - Deformation capacity of welded beam-to-column
connection subjected to repeated plastic strain Part 5 -. Journal of Structural and Construction
Engineering, AIJ, 707, 127-136, JAPAN.
[7] Yamada S, Kishiki S, & Shimada Y (2010): Strength Evaluation of Fully Composite
beam under positive bending. Proceedings of JSSC (18), JSSC, 93-98, JAPAN.
[8] Shimada Y, Yamada S & Yasuda B (2016). Calculation Method of Connection
Coefficient of Composite Beam. Journal of Structural and Construction Engineering, AIJ, 724,
1005-1014, JAPAN.
[9] Yamada S, Jiao Y & Kishiki S (2015). Evaluation Method of Plastic Deformation
Capacity of Steel Beam governed by Ductile Fracture at the Toe of the Weld Access Hole.
Journal of Structural and Construction Engineering, AIJ, 711, 767-777, JAPAN.
[10] Japan Concrete Institute (1996). Concrete Handbook the Second Edition, 240, JAPAN.
[11] Kato B & Akiyama H (1968). The Ultimate Strength of the Steel Beam-Column (Part 4),
Journal of Structural and Construction Engineering, AIJ, 151, 15-20, JAPAN.
[12] Akiyama H & Takahashi M (1990). Influence of Bauchinger Effect on Seismic
Resistance of Steel Structures, Journal of Structural and Construction Engineering, AIJ, 418,
49-57, JAPAN.
[13] Fujisawa K, Ichinose Y, Sugimoto M, & Sonoda M (2013). Statistical Study on
Mechanical Properties and Chemical Compositions of SN Steels. Summaries of Technical
Papers of Annual Meeting of AIJ, AIJ, stru.-3, 699-700, JAPAN.
BEHAVIOR OF COMPOSITE DOWELS POSITIONED CLOSE TO
THE SURFACE OF CONCRETE SLABS
Prof. Dr.-Ing. Wolfgang Kurz

Yannick Broschart, M.Sc.
Institute of Steel Structures, University of Kaiserslautern,
Paul-Ehrlich-Straße 14, 67663 Kaiserslautern, Germany
wolfgang.kurz@bauing.uni-kl.de
yannick.broschart@bauing.uni-kl.de
Dipl.-Ing. Joanna Gajda

BORAPA Ingenieurgesellschaft mbH,
Luxemburger Straße 1, 67657 Kaiserslautern, Germany
gajda@borapa.de
ABSTRACT
Composite dowels are efficient shear connectors that become more and more common in use for
composite beams. They can be used as an economic alternative to headed studs with various
advantages.
Modern concrete slabs are dimensioned increasingly thinner and often include several additional
functions, for example air conditioning or thermal activation of the concrete. For those applications
composite dowels can be positioned horizontally in the concrete slab (Fig. 1). Existing
investigations on composite dowels describe different failure modes such as splitting of the
concrete, local compression of the concrete, prying-out of a concrete cone or plasticizing of the
steel.
At University of Kaiserslautern composite dowels with positions close to the free surface of the
concrete slab were investigated experimentally by push-out-tests and on the basis of numerical
simulations. In addition, the positive influence of a spiral reinforcement in the dowels was
analyzed. Further primary and secondary failure mechanisms were observed for this specific
position of the composite dowels. These investigations and a design approach will be presented
within the paper.
INTRODUCTION
In composite slabs longitudinal shear forces have to be transferred between the concrete section
and the steel profiles. This shear connection is conventionally realized by headed studs. As an
alternative, composite dowels can be installed. Existing investigations and regulations on
composite dowels include various failure modes of the structure. These are summarized and
illustrated schematically within Table 1 in compliance with [Berthellemy 2011] and
[Heinemeyer 2011].
Table 1 – Compilation of observed failure modes of composite dowels
Failure pattern
Symbol Failure mode
longitudinal section / cross section
Psp splitting
Plc local compression of the concrete
Psh shearing of the composite dowel
Ppo pry-out of the concrete
Ppl plasticizing of the steel
Pedge edge failure
Pspo semi-pry-out of the concrete
Specifically, if the position of the composite dowels is close to the free concrete surface, as to be
seen in Fig. 1, an additional failure mode can occur in which the concrete cover at the edge of the
concrete slab spalls. This failure mode was called ‘edge failure’. At University of Kaiserslautern
push-out-tests with puzzle-shaped composite dowels positioned close to the free surface were
performed in the context of a research project aiming to respect these effects within the design
approaches. Furthermore, the experimental test results were compared to the outputs of
numerical simulations.
Edge
Edge position
Middle
position position
Fig. 1 – Possible positions of puzzle-shaped composite dowels close to the free surface (left),
affected concrete cuboid (center), fracture area after the spalling (top right) and semi-pry-out of
half concrete cones (bottom right)
EXPERIMENTAL INVESTIGATIONS
The behavior of composite dowels close to the free surface of concrete slabs was investigated
experimentally with the help of push-out-tests. In total, 46 push-out-tests were performed with
varying constructive and geometric parameters as well as with different material properties. Table
2 summarizes the investigated parameters. Two types of push-out specimens were realized: PO,
representing the edge position of the composite dowels, and PO A, representing the middle
position, as to be seen in Fig. 3 and Fig. 4. The PO A specimens were produced by bending the
steel teeth alternately to the left and to the right.
Table 2 – Overview of the varied parameters and material properties

Parameter Range of variation
P – Puzzle shape (r)*
Composite dowel geometry
K – Clothoid shape
4 mm (r)
Composite dowel thickness tw 6 mm
8 mm
S 235 JR (r)
Steel grade
S 355 JR
C 25/30 (r)
Concrete strength class C 30/37
C 35/45
100 mm (r)
Concrete slab thickness dc
150 mm
200 mm
Concrete slab width bc
300 mm (r)
25 mm (r)
Embedment depth of the composite dowel he
50 mm
non-existent
1Ø8
Reinforcement per composite dowel AB 2Ø6
3Ø6
2Ø8 (r)
non-existent (r)
Spiral reinforcement per composite dowel Aw ø 50 mm, ds=5 mm, sw=25 mm
ø 60 mm, ds=5 mm, sw=25 mm
* (r) reference test
Fig. 2 – Designation of the different parts of the composite dowel
The first series consisted of first trial tests to define the dimensions of the subsequent specimens.
As a consequence, the thickness of the concrete slab was set to 100 mm, comparable to slim
sandwich constructions. Further dimensions of the specimens can be seen within Fig. 4. The
width of the supports as well as the relation between the height and the width of the specimen
were chosen with regard to receiving concrete compression struts with an angle of approximately
45°.
a b c
Fig. 3 – Puzzle geometry after bending (a, b), spiral and stirrup reinforcement (c)
The reference type of reinforcement consisted of two stirrups per concrete dowel placed 12,5 cm
apart from each other with diameters of 8 mm. Furthermore, a spiral reinforcement was installed
in some specimens to improve the hydrostatic compression stress area in front of the steel teeth
by a confinement of the concrete. Underneath the lowest half steel teeth in test specimen,
polystyrene cuboids were installed (Fig. 4) to avoid the contact between steel and concrete in this
area, because this would have fuged the measurement of the resistance of the full composite
dowels. At the bottom of the concrete slabs, a tension rod was applied to avoid drifting apart.
The load on the specimen was increased up to 100 kN with a speed of 4 mm/min and afterwards
decreased to 20 kN. This cycle was repeated for 25 times, in order to set aside the adhesive
friction, before the load was raised to the failure of the specimen with a speed of 0,25 to 0,5
mm/min. The load increase was paused at load level steps of 50 kN to determine the quasi-static
load-displacement diagram.
Experimental setup of the push-out-tests
PO push-out-tests PO A push-out-tests
Edge position of the composite dowel Middle position of the composite dowel
Fig. 4 – Experimental setup of the two types of push-out-tests
During the experimental procedure, the load was measured as well as the slip δ, which describes
the relative displacement between the steel profile and the concrete section in direction of the x-
axis. Additionally, the relative displacement Δ of the two concrete slabs to one another in direction
of the z-axis was registered with the help of inductive displacement transducers as well as the
thickness variation of the concrete slabs in direction of the y-axis (Fig. 5).
DVOL
DVOR V Front
H Back
DVUL L Left
DVUR
R Right
O Top
U Bottom
S Slip δ
A Uplift Δ
D Thickness
variation
Fig. 5 – Geometric assumptions for the design approach in xz-view (left) and yz-view (right)
The following table shows the varied properties of seven push-out-specimens (PO) without a
spiral reinforcement: the thickness of the concrete slab bc, the depth of the embedding of the steel
web into the concrete slab he, the tensile strength of the concrete fctm and the yield strength of the
steel fy.
Table 2 – Properties of the tested specimens without spiral reinforcement

Without spiral reinforcement PO 34 PO 36 PO 38 PO 40 PO 42 PO 44 PO 46
Thickness of the concrete slab bc [mm] 100 100 150 100 100 150 150
Depth of the embedding he [mm] 25 25 25 25 50 25 50
Tensile strength of the concrete fctm [MPa] 2,2 2,2 2,2 2,9 2,9 2,9 2,9
Yield strength of the steel fy [MPa] 362 296 296 296 296 296 296
The typical crack pattern of the PO specimens is shown within Fig. 6. The splitting of the concrete
slabs within the xz-plane could not be seen at the surface. The first surface cracks appeared as
splitting cracks underneath the steel teeth in direction of the z-axis and successively grew towards
the centre of the concrete slabs with an increasing incline towards the supports. The final surface
cracks initiated the spalling of the concrete cover and the edge failure, starting at the upper end
of the concrete slab and growing downwards in direction of the x-axis immediately.
Fig. 6 – Typical crack pattern after the failure of the specimen using the example of PO 36
The load-slip-diagram in Fig. 7 shows the testing machine load as a function of the slip of the
composite dowels in direction of the x-axis. The final failure mode of the specimens is edge failure.
In spite of the slim structure elements, a distinctly ductile behavior could be observed.
350
300
250
200
Load P [kN]
PO 34
150
PO 36
PO 38
100 PO 40
PO 42
50 PO 44
PO 46
0
0 2 4 6 8 10 12 14 16 18 20
Slip δ [mm]
Fig. 7 – Load-slip-diagram of the PO specimens
NUMERICAL SIMULATIONS
The experimental tests were numerically simulated with the FE software ABAQUS 6.14-2 from
Simulia. The software does not only consider geometric non-linearity but also material non-
linearity. As a matter of fact the modelling of the concrete’s tension and compression stress
behavior decisively influences the simulation results and should be represented as realistically as
possible. The so-called ‘Concrete Damage Plasticity Model’ (CDP) was most appropriate to
simulate the experimental results and to show a realistic failure pattern. The approximation of the
concrete’s material properties is schematically presented in Fig. 8.
Tension behavior Compression behavior
t: tension
c: compression
d: damage variable
E0: initial elastic modulus
Ɛ̃pl: equivalent plastic strain
Ɛel: elastic strain
Fig. 8 – Uniaxial tension and compression stress behavior of the concrete [Simulia 2014]
The following figures (Fig. 9 and Fig 10) show the load-slip-curves of the simulations in
comparison to those of the experiments. Up to a certain point, there is a high conformity. But the
decrease of the resistance of the composite dowels cannot be sufficiently simulated due to the
underlying model of tension behavior. In reality, if concrete reaches its tensile fracture strain, a
crack occurs and the transferable stress declines immediately, while in the simulation’s material
model some load bearing capacity has to remain for numerical reasons.
Additionally, the tensile-damage-rate of the concrete is illustrated (Fig. 9, right). The visible
pattern is similar to surface cracks in direction of the x-axis which initiated the edge failure (Fig. 6).
350
300
250
Load P [kN]
200
PO 35 test
Versuch PO 35
150
Simulation
PO PO
35 simulation
100 35
Versuch
PO PO 36
36 test
50 Simulation PO
PO 36 simulation
36
0
0 2 4 6 8 10
Slip δ [mm]
Fig. 9 – Comparison of the tested and the numerically simulated load-slip-curves of PO

specimens (left) and tensile-damage-rate of the concrete (right)
900 900
800 800
700 700
600 600
Load P [kN]
Load P [kN]
500 POA
PO A 55 500 POA
PO A 66
Versuch Versuch
test
test
400 POA5 400 POA
PO A 66
PO A5
Simulation
simulation Simulation
simulation
300 300
POA
PO A 99 POA
PO A 10
Versuch
test Versuch
test
200 200
POA
PO A 99 POA
PO A 10
100 Simulation
simulation 100 Simulation
simulation
0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Slip δ [mm] Slip δ [mm]
Fig. 10 – Comparison of the experimentally tested and the numerically simulated load-slip-
curves of PO A specimens
CONSIDERATIONS ON THE LOAD TRANSFER MECHANISMS

On the basis of the measurement results and the numerical simulations, load transfer
mechanisms were modelled and are illustrated three-dimensionally at the bottom of Fig. 11 in a
simplified way. The load-slip-diagram of PO 36 can be partitioned into five sections. The
successive decline of the specimen’s stiffness is an indicator for the transition of the load transfer
mechanism. The first decline of the stiffness (Fig. 11-A) is considered to occur because of the
detaching of the adhesive friction. Independently of this effect, in sections I and II (Fig. 11) the
concrete struts were considered to start at the hydrostatic compression stress area in front of the
steel tooth. At their other end the cone-shaped struts are supported by the concrete slab. To
obtain the state of equilibrium, the compression struts have to be connected by circularly arranged
tension struts. This tension ring is interrupted by the steel plate of the composite dowel. As a
matter of fact and corresponding to the measurement of the stirrup’s strain and the thickness
variation of the concrete slab in direction of the y-axis, the first crack area seems to appear inside
of the concrete slab parallel to the xz-plane (Fig. 11-B). This crack area cannot be seen as a
surface crack. In sections III and VI (Fig. 11) further tension struts occur behind the hydrostatic
stress area because of the connection of the concrete in front of the steel tooth to the concrete
area behind the steel tooth. Point C (Fig. 11) can be interpreted as the plasticizing of the steel
teeth. The second crack area appears in the yz-plane, as a consequence of the previously
mentioned tension struts reaching the tensile strength of the concrete, and is visible at the
specimen’s surface (Fig. 11-D). Immediately after the occurrence of the first crack area, the
compression struts transform with regard to the newly shaped concrete cuboids and form two
elliptical compression cones. To design the schematic illustration more clearly, these two elliptical
compression cones were not shown right after their occurrence but only afterwards in Fig. 11-V.
Finally, the concrete tension in direction of the z-axis reaches the concrete tensile strength and
the third crack area appears in xy-plane, initiating the edge failure (Fig. 11-E).
A Detaching of adhesive friction
B Splitting crack in xz-plane
C Plasticizing of the steel teeth
D Crack area in yz-plane
E Crack area in xy-plane

(edge failure)
I+II III+IV D V E
Fig. 11 – Load transfer mechanisms and resulting crack areas
DEVELOPMENT OF A CALCULATION MODEL

On the basis of the previously presented considerations on the load transfer mechanisms, a
design approach was developed for the failure mode edge failure. Therefore, geometric
assumptions were made, which are illustrated within Fig. 12 for one half of the concrete slab at
one side of the composite dowel. Into this concrete area half the steel tooth force ½·P spreads
out. Unavoidably, a part of this force spreads out in z-direction into the concrete cover because
of its stiffness. The resulting force flow can be idealized by a strut-and-tie model. As a first
assumption the present force ½·P was split into one half entering the upper concrete slab and
one half entering the concrete cover. This force component consequently has an amount of ¼·P.
A more reasonable apportionment of this force will subsequently be taken into account by the
calibration factor η. Corresponding to this model, the resulting vertical component Fz depends
on the height of the concrete cover cz and on the distance between the steel tooth and the
reinforcement stirrup sx. Definitively, the vertical component Fz has to be transferred into the
stirrup reinforcement by concrete tension. The model assumes a load diffusion angle α. As a
result, an operational crack area Acr,op can be imagined on the level of the stirrup reinforcement.
This area is not equivalent to the true crack area. It only serves as an operational value to
calculate the resistance against edge failure, which is assumed to occur as soon as the present
stress within Acr,op reaches the concrete’s tensile strength fct. Furthermore, there is a horizontal
force component Fy because of the offset of the steel tooth force relative to the concrete cuboid
in direction of the y-axis, which causes a bending of the concrete cuboid around the x-axis and
leads to additional bending stresses within Acr,op. The superposition of these stress distributions
is schematically shown within Fig. 12, too. The exceedance of fct also initiates the longitudinal
crack formation in direction of the x-axis which is specific for this failure mode.
y . .x σct ≤ fct
concrete slab x y
z z
Lcr,x,op
steel tooth
sx
Acr,op Acr,op Acr,op Acr,op
Fz
¼·P Fz α ez ez
Fy
cz Fz
Fy
ey
½ bc
Fig. 12 – Geometric assumptions for the design approach in xz-view (left) and yz-view (right)
Corresponding to these assumptions, a design approach was derived. The value of the crack
area Acr,op can be quantified as follows. All quantities are shown within Fig. 12.
𝑏𝑏𝑐𝑐 𝑏𝑏𝑐𝑐
𝐴𝐴𝑐𝑐𝑐𝑐,𝑜𝑜𝑜𝑜 = 𝐿𝐿𝑐𝑐𝑐𝑐,𝑥𝑥,𝑜𝑜𝑜𝑜 ∙
= 𝑐𝑐𝑧𝑧 ∙ tan(𝛼𝛼) ∙ (1)
2 2
Consequently, the section modulus around the x-axis W x,op of the concrete cuboid underneath
Acr,op can be characterized by:
𝑏𝑏 2
𝐿𝐿𝑐𝑐𝑐𝑐,𝑥𝑥,𝑜𝑜𝑜𝑜 ∙ � 𝑐𝑐 � 𝑐𝑐𝑧𝑧 ∙ tan(𝛼𝛼) ∙ 𝑏𝑏𝑐𝑐 2 (2)
2
𝑊𝑊𝑥𝑥,𝑜𝑜𝑜𝑜 = =
6 24
The vertical force component affecting Acr,op has the value:
𝑃𝑃 𝑒𝑒𝑧𝑧 𝑃𝑃 𝑐𝑐𝑧𝑧
𝐹𝐹𝑧𝑧 = ∙ = ∙ (3)
4 𝑠𝑠𝑥𝑥 8 𝑠𝑠𝑥𝑥
The bending moment to the concrete cuboid is:
𝑃𝑃 𝑒𝑒𝑦𝑦 𝑃𝑃 𝑏𝑏𝑐𝑐
𝑀𝑀𝑥𝑥 = 𝐹𝐹𝑦𝑦 ∙ 𝑒𝑒𝑧𝑧 = ∙ ∙ 𝑒𝑒 = ∙ ∙ 𝑐𝑐 (4)
4 𝑠𝑠𝑥𝑥 𝑧𝑧 32 𝑠𝑠𝑥𝑥 𝑧𝑧
On the basis of these values, the resulting stress within Acr,op can be calculated. The failure mode
‘edge failure’ was assumed to occur as soon as this stress reaches the tensile strength of the
concrete fctm:
𝐹𝐹𝑧𝑧 𝑀𝑀𝑥𝑥 1 𝑃𝑃
𝜎𝜎𝑐𝑐𝑐𝑐 = 𝜂𝜂 ∙ � + � = 𝜂𝜂 ∙ ∙ ≤ 𝑓𝑓𝑐𝑐𝑐𝑐𝑐𝑐 (5)
𝐴𝐴𝑐𝑐𝑐𝑐,𝑜𝑜𝑜𝑜 𝑊𝑊𝑥𝑥,𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡𝑡(𝛼𝛼) 𝑏𝑏𝑐𝑐 ∙ 𝑠𝑠𝑥𝑥
With the help of this design approach, the experimentally obtained resistances were recalculated
using the following formula. With an assumed load diffusion angle of 40°, which seemed
appropriate due to numerical simulations, the calibration factor was set to η = 1,67 to achieve
plausible values for the theoretically calculated resistance per steel tooth.
𝑡𝑡𝑡𝑡𝑡𝑡(𝛼𝛼)
𝑃𝑃𝑡𝑡ℎ𝑒𝑒𝑒𝑒 = ∙ 𝑏𝑏𝑐𝑐 ∙ 𝑠𝑠𝑥𝑥 ∙ 𝑓𝑓𝑐𝑐𝑐𝑐𝑐𝑐 = 1,4 ∙ 𝑏𝑏𝑐𝑐 ∙ 𝑠𝑠𝑥𝑥 ∙ 𝑓𝑓𝑐𝑐𝑐𝑐𝑐𝑐 (6)
𝜂𝜂
With the calibration factor η the apportionment of the force component ½·P into the upper
concrete slab and into the concrete cover can be respected as well as the linear deviation between
the theoretical assumptions of the model and the true experimental conditions. For instance, the
load transfer mechanisms are based on strut-and-tie models while in reality, the load spreads out
three-dimensionally into the concrete cuboid. Additionally, the integration of non-linear factors is
possible. Therefore, further experimental investigations would be necessary to define these
parameters in a statistically reliable way.
A comparison between the experimental results and the theoretically calculated resistances is
shown in Table 3. The deviation lies between -16% and +15%.
Table 3 – Recalculation of the experimentally tested resistances per composite dowel

Without spiral reinforcement PO34 PO36 PO38 PO40 PO42 PO44 PO46
Experimental
Pexp [kN] 64,3 63,3 85,8 55,0 50,0 72,5 60,3
resistance
Theoretical
Ptheo [kN] 60,9 60,9 91,3 46,2 46,2 69,3 69,3
resistance
Deviation from Pexp to Ptheo -5% -4% +7% -16% -8% -4% +15%
With respect to the specimens with an installed spiral reinforcement (Fig. 3c), a linear
magnification factor λspiral = 1,4 was inserted additionally as a first proposal. This leads to a
deviation between the experimentally tested and the theoretically calculated resistances of
specimens with a spiral reinforcement of -15% to +24%.
After a statistical evaluation and an evaluation on safety factor γc, the complete proposed
calculation of the design value of the resistance against edge failure of composite dowels
positioned close to the concrete surface in consideration of the integration into the safety concept
of the Eurocodes results in:
1 1
𝑃𝑃𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒,𝑅𝑅𝑅𝑅 = ∙ 𝑃𝑃𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒,𝑅𝑅𝑅𝑅 = ∙ 0,80 ∙ 1,4 ∙ 𝑏𝑏𝑐𝑐 ∙ 𝑠𝑠𝑥𝑥 ∙ 𝑓𝑓𝑐𝑐𝑐𝑐𝑐𝑐 = 0,90 ∙ 𝑏𝑏𝑐𝑐 ∙ 𝑠𝑠𝑥𝑥 ∙ 𝑓𝑓𝑐𝑐𝑐𝑐𝑐𝑐 (7)
𝛾𝛾𝑉𝑉 1,25
OUTLOOK
In general, the decisive failure mode is influenced by various parameters. Not all of them have
been investigated sufficiently and taken into account within the design approach yet. Furthermore,
a greater amount of data is necessary to allow statistically reliable propositions. Therefore, the
investigations will be continued in the context of a subsequent research project, which has started
in April 2017. The project is performed by University of Kaiserslautern in cooperation with RWTH
Aachen University and is aimed at developing a consistent design approach for composite
dowels. It includes further experimental investigations on composite dowels positioned close to
the free surface of the concrete slabs as well as corresponding numerical simulations. The long-
term goal is to integrate a consistent design approach for composite dowels into the Eurocode as
an alternative to headed studs.
ACKNOWLEDGEMENTS
The presented test results were achieved by the financial support of the state Rheinland-Pfalz
during the pilot project InnoProm with funds of MBWWK and EFRE. The project’s promotion
number was 964-52708-1/81029978.
REFERENCES
Berthellemy, J.; et al. (2011). Zum Tragverhalten von Verbunddübeln – Teil 1: Tragverhalten unter
statischer Belastung. Stahlbau 80, Heft 3, Ernst & Sohn Verlag, Berlin, Germany.
Broschart, Y. (2016). Development of a design approach for the failure mode ’edge failure’ of
composite dowels positioned close to the surface of concrete slabs. MSc thesis, University
of Kaiserslautern, Germany.
Broschart, Y.; Gajda, J.; Kurz, W. (2018). Untersuchungen zum Tragverhalten randnaher
Verbunddübelleisten. Stahlbau 87, Heft 5, Ernst & Sohn Verlag, Berlin, Germany.
DIBt Deutsches Institut für Bautechnik (2013). Allgemeine bauaufsichtliche Zulassung:
Verbunddübelleisten. Approval number Z-26.1-23, Berlin, Germany.
European Committee for Standardization (2002). Eurocode: Basis of structural design.
EN 1990:2002 + A1:2005 + A1:2005/AC:2010. Brussels, Belgium.
European Committee for Standardization (2004). Eurocode 4: Design of composite steel and
concrete structures - Part 1-1: General rules and rules for buildings. EN 1994-1-1:2004 +
AC:2009. Brussels, Belgium.
Gajda, J. (2019): Loadbearing effect of concrete dowels close to the concrete surface. PhD thesis
(not published yet), Technische Universität Kaiserslautern, Germany.
Gajda, J.; Kurz, W. (2013). Tragverhalten randnaher Betondübel unter Längsschub-
beanspruchung. Stahlbau 82, Heft 9, Ernst & Sohn Verlag, Berlin, Germany.
Gajda, J.; Kurz, W. (2013). Analysis of near to free surface located concrete dowels in composite
structures. Composite Construction in Steel and Concrete VII, ASCE American Society of
Civil Engineering, North Queensland, Australia.
Heinemeyer, S. (2011). Zum Trag- und Verformungsverhalten von Verbundträgern aus
ultrahochfestem Beton mit Verbundleisten. PhD thesis, Aachen, Germany.
Kopp, M.; Wolters, K.; et al. (2018) Composite dowels as shear connectors for composite beams
– Background to the design concept for static loading. Journal of Constructional Steel
Research, Amsterdam, Netherlands.
Simulia (2014). Abaqus 6.14 Analysis User’s Guide – 23.6.3 Concrete damaged plasticity.
Dassault Sytèmes Simulia Corp., Providence, USA.
PIN SHEAR CONNECTORS FOR STEEL-CONCRETE COMPOSITE
CONSTRUCTIONS
Maik Kopp
RWTH Aachen University, Division 10.2 - Construction Management
Aachen, Germany
maik.kopp@zhv.rwth-aachen.de
Kevin Wolters
RWTH Aachen University, Institute of Steel Construction
Aachen, Germany
k.wolters@stb.rwth-aachen.de
Markus Feldmann
RWTH Aachen University, Institute of Steel Construction
Aachen, Germany
feldmann@stb.rwth-aachen.de
Martin Claßen
RWTH Aachen University, Institute of Structural Concrete
Aachen, Germany
mclassen@imb.rwth-aachen.de
Johannes Schäfer
RWTH Aachen University, Laserline GmbH
Mülheim-Kärlich, Germany
johannes.schaefer@laserline.com
ABSTRACT
The use of high-strength steel as well as ultra-high-performance concrete (UHPC) in composite
constructions enable slender, wide spanning and highly sustainable composite elements in
architecturally sophisticated structures. Compared to normal-strength materials, high-strength
concrete and steel allow a significant reduction of the dimensions of load-bearing elements.
Thus, for certain fields of application the theoretical thicknesses of concrete chords can be
reduced to a few millimeters. However, for such slender composite structures and concrete
chords the applicability of conventional shear connectors like headed studs is no longer given.
The systematic connection of the composite cross sections requires novel, small-scaled shear
connectors transmitting anchorage and shear forces through the composite joint.
This paper deals with the load-bearing and slip capacity of the novel small-scaled pin shear
connectors due to static forces. The experimental and numerical results as well as the resulting
design equations are presented.
INTRODUCTION
Steel-concrete composites combine the favorable material properties of structural steel and
concrete in an optimal way. Concrete is arranged in the compression area and steel in the
tensile zone. To ensure the composite between both materials, effective connectors are
necessary, which have a high load as well as a sufficient deformation capacity (Hanswille,
Schäfer & Bergmann, 2010).
The usage of high-strength materials (e.g. high-strength steel and UHPC) increases the load
capacity of the composite structure significantly and allows smaller dimensions of the load-
bearing elements respectively (Feldmann, Hegger, Hechler & Rauscher, 2007) (Heinemeyer,
Gallwoszus & Hegger, 2012) (Feldmann et al., 2014). In certain applications the theoretical
thicknesses of concrete chords can be reduced to less than 30 mm, so that shear connectors
like headed studs (EN 1994-1-1, 2010) or composite dowels (Feldmann, Kopp & Pak, 2016) are
non-executable. Therefore novel small-scale shear connectors are developed.
In a national research project the pin-structures (Figure 1) were investigated with regard to the
static load-bearing capacity and slip capacity (Claßen, Gallwoszus, Kopp & Schäfer, 2015)
(Schäfer et al., 2015) (Claßen et al., 2015) (Kopp et al., 2016). By means of push-out tests the
local resistance was determined. The global load-bearing behavior was investigated in
composite beam and plate tests. Numerical simulations supported the experimental test series.
Based on the experimental and numerical results, engineering models were derived to describe
the load-bearing mechanisms (concrete and steel failure). Finally, design equations for
composite construction with pin structures as shear connectors were determined.
Fig. 1 - (a) Terms of the pin structures and (b) example of group arrangement
PRODUCTION AND MATERIALS

Pin-welding process
The pin-connectors are produced directly from the welding wire using a novel welding process,
based on the so called Cold Metal Transfer (CMT) procedure. In this process the wire is moved
towards the base material until it is welded on due to a short circuit (Figure 2 a, b).
Subsequently the wire is pulled backwards without any electricity supply (Figure 2 c). At a
coordinated point a renewed electricity supply cuts and melts the wire. The material is formed to
a sphere due to surface tension (Figure 2 d-f).
Fig. 2 - (a) Terms of the pin structures and (b) example of group arrangement
Current pins can be reliably produced up to a height of about 5 mm depending on the wire
material and diameter. Several pin-geometries, including varying height and diameters of the
pins, are considered in the investigations.
Materials
In order to take advantage of the multiple benefits of the innovative welding process for
applications in steel-concrete composites, the combination of small-scale pins with high-strength
and ultra-high-performance concrete is effective. Subsequently, the material properties of the
used pins, steel and the concrete are assembled.
During the experimental investigations two different welding wires with 0.8 mm and 1.2 mm are
tested. First, austenitic welding wires ER308LSi are combined with an austenitic steel sheets
(AISI 304) with a thickness t = 3.0 mm. The nominal tensile strength of ER308LSi is about 590
N/mm². Furthermore, welding wires out of low-alloy steel (ER70S-6) are used for tests with steel
sheets ASTM A633 (t = 3.0 mm).
In order to achieve a completely form-locked and force-locked embedding of small-scale pin
connectors in the concrete microstructure, a small grain size and high fluidity of the concrete are
desirable from the perspective of concrete technology. These requirements are fulfilled in a fine-
grained concrete mixture, which was developed for applications with textile-reinforced concrete
(Brockmann, 2006). Therefore, in the experimental tests, this fine-grained concrete is combined
with both welding wires. The used fine-grained concrete has a maximum grain diameter of 0.6
mm and a compressive strength of about 70 N/mm². Due to its high binder content and the
small fraction of coarse aggregates, the fine concrete has a smaller elastic modulus and greater
ductility compared to normal concretes of the same strength.
Furthermore, the combination of pin shear connectors with ultra-high-performance concrete
(UHPC) is analyzed in the context of the current research project (Claßen et al., 2015). Due to
the combination of optimized mixture designs with steel fibers UHPC is a dense structured
concrete. This concrete has an extremely high characteristic compressive strength (>180
N/mm²), bending tensile strength (>40N/mm²) and elastic modulus (>50,000 N/mm²).The
addition of steel fibers influences the ductility and strength affirmatively. A sufficient ductility and
a crack bridging effect can only be guaranteed with high tensile strength and with high
elongation at rupture of the fibers (Leutbecher, 2008).
EXPERIMENTAL INVESTIGATIONS
Local load-bearing behavior of pin-connectors due to shear loading (push-out tests)
Several test series were performed to determine the load-bearing and slip capacity of pin-
connectors (Kopp et al., 2016). The shear capacity of a single pin was investigated by push-out
tests (cp. Figure 3) based on the standard push-out test for headed studs of EN 1994-1-1,
Annex B (EN 1994-1-1, 2010). Overall 24 test series, each with ten push-out specimen, were
performed with regard to shear capacity, initial stiffness and deformability.
Fig. 3 - Push-out test (a) without and (b) with horizontal bearing
The first test of each series was performed path-controlled until failure. Based on the maximum
force all other test samples were preloaded with 25 force-controlled load cycles (between 5 and
40% maximum load) to loosen the adhesion bond and subsequently loaded path-controlled until
failure. During the experiments, the slip in the composite joint was measured with inductive
displacement transducers.
The influence of the horizontal bearing of the concrete slabs was also examined. The
experiments show a clear dependence of the failure mode to the horizontal bearing. If a
horizontal gaping of the composite joint is prevented by a bracing of the concrete slabs mainly
steel failure (pin shearing) occurs. If the horizontal bracing of concrete slabs is omitted, the
usual failure is prying-out of a concrete cone around the pin. Thus, the ultimate loads of
experiments without horizontal bracing are on average 25 % below the fracture loads of
experiments with horizontal bearing. This effect can be explained by the simultaneous
interaction of shearing and tensile forces in the composite joint.
Fig. 4 - Force-slip curve of (a) austenitic steel and (b) construction steel
Exemplarily the test curves of austenitic and structural steel combined with high-strength
concrete (HSC) are plotted in Fig. 5. Load-bearing capacity increases significantly when using
thicker and higher pins (Figure 4 a, B.1-06 compared to B.1-10, austenitic steel). The results of
the tests using structural steel are characterized by a higher scatter and lower ultimate loads
(Figure 4 b). This might be caused by a better thermal conductivity of structural steel compared
to austenitic steel and therefore a worse connection between pin and base material. In
consequence, pin shearing often occurs in the heat-affected zone underneath the pin-foot.
Table 1 shows selected results of the currently greatest possible pin-heights. The usage of
UHPC increases the load-bearing and decreases the deformation capacity δuk compared to
HSC.
Table 1 - Test results series B.1 (push-out tests, maximum pin-heights)

Øwire hpin Cini,m Pmax PRk δuk
Series Steel plate / wire Concrete
[mm] [mm] [N/mm] [N/Pin] [N/Pin] [mm]
B.1-06A AISI 304 / ER308LSi HSC 0.8 4.35 5980 670.9 454.8 0.63
B.1-10A AISI 304 / ER308LSi HSC 1.2 5.10 6852 1268.0 918.7 1.07
B.1-11AU AISI 304 / ER308LSi UHPC 0.8 4.35 8912 971.1 459.0 0.44
B.1.12AU AISI 304 / ER308LSi UHPC 1.2 5.10 12 611 1601.5 571.8 0.62
B.1-16S ASTM A633 / ER70S-6 HSC 0.8 4.77 5825 874.8 616.1 0.81
B.1.18S ASTM A633 / ER70S-6 HSC 1.2 4.82 5850 970.1 558.0 1.28
B.1-19SU ASTM A633 / ER70S-6 UHPC 0.8 4.77 7051 1027.5 647.2 0.36
B.1-20SU ASTM A633 / ER70S-6 UHPC 1.2 4.82 11 469 1471.5 853.5 0.37
Fig. 5 - (a) Computer tomography (CT) system with specimen and (b) Load-slip-curve of a push-
out test with CT images
During the experimental tests, push-out tests with the aid of computer tomography (CT) were
performed. Hereby, it was possible to look into the composite specimens during some load
levels. Figure 5 (a) shows the CT system with the developed load introduction. The load
introduction consists of an acrylic glass pipe and two round steel plates. The upper steel plate
has an opening to position the specimen within the acrylic glass. The glass enabled 3D CT
images without interferences due to density difference. Figure 5 (b) exemplifies a load-slip-curve
of a push-out tests with two CT images at different times. The upper image (1, green frame)
shows the first steel crack in the lower left pin, which resulted in a load drop (green dotted circle
in the diagram). In the lower CT image (2, red frame) both the steel cracks (lower left and upper
right pin) as well as a concrete pry-out cone (lower right pin) are visible at the end of the test.
These combined failures resulted in the total collapse of the specimen.
Global load-bearing behavior (beam tests)

To research the global load-bearing characteristics of pin-structures in composite construction,
bending tests of composite beam (Series C.1) and composite slabs (Series C.2) are performed
to detect the required slip capacity of the pin shear connectors. Furthermore, it is supposed to
verify the transferability of the local mechanisms on the global load-bearing and deformation
behavior. The beam test program included six tests with positive (sagging) bending stress and
two tests with negative (hogging) bending stress. In Series C.2 four composite slab tests are
performed to investigate the behavior of the pin-structures in cracked concrete.
Table 2 - Test results series C.1 (beams)

Øwire hpin η L Mmax δmax
Series Steel wire Concrete
[mm] [mm] [%] [m] [kNm] [mm]
C.1-01A ER308LSi HSC 1.2 5.10 100 2.60 17.6 0.11

C.1-02A ER308LSi HSC 1.2 5.10 50 2.60 14.8 1.11
C.1-03A ER308LSi HSC 1.2 5.10 100 3.75 15.1 0.11
C.1-04A ER308LSi HSC 1.2 5.10 50 3.75 10.9 0.63
C.1-05SU ER70S-6 UHPC 1.2 4.82 100 3.15 43.7 0.05
C.1-06SU ER70S-6 UHPC 1.2 4.82 50 3.15 25.3 0.04
C.1-07SN ER70S-6 HSC 1.2 4.82 100 3.00 -15.4 0.05
C.1-08SN ER70S-6 HSC 1.2 4.82 50 3.00 -7.2 0.01
The design of the composite beams under sagging moment ensures the plastic neutral axis to
be located in the composite joint in order to reach an optimum use (concrete in compression
area, steel in tension zone). The individual beams differ regarding span length, materials and
degree of shear connection η. Parameters and results of series C.1 are shown in Table 2 and
Figure 6.
For all beam tests 4-point bending test setup is chosen. It is obvious that a higher degree of
shear connection leads to a higher load-bearing capacity and lower vertical deflection
respectively less slip in the composite joint. By the use of UHPC the load-bearing capacity
increases at the expense of a lower deformability. In case of a full shear connection, the load
can be increased until a failure of the compressive zone occurs, while there is shear failure of
the pins in case of 50 percent degree of shear connection.
Fig. 6 - Force-deflection curves of composite beam tests with (a) austenitic steel combined with
HSC and (b) structural steel with UHPC
NUMERICAL INVESTIGATION
Structure of FE-model and material characteristics
To examine the influences of pin-geometry (dimensions of pin-head, pin-shank and pin-foot),
pin-distance, bonding depth (pin-height) and strength of steel and concrete on the shear
behavior of the pin-structures, parametric studies were performed using the finite-element
software ABAQUS. The push-out test is modelled with C3D8R volume elements with non-linear
material characteristics (Figure 7). Double symmetry is exploited to reduce computing time. The
model was validated by the experimental results (Claßen et al., 2016).
Material models
HSC Welding wire ER308LSi
Compression Tension Compression and tension
Stress-strain Stress-crack opening Stress-strain
Elastic-plastic
Concrete Damage Plasticity (CDP)
(strain-hardening)
Material parameters CDP-Parameters Material parameters
Comp. strength fcm = 73.0 N/mm²
Tensile strength fct = 4.0 N/mm²
Dilation Angle Ψ = 35° Yield strength fy = 350 N/mm²
Young’s modulus Ecm = 31 200 N/mm²
Eccentricity ε = 0.3 Tensile strength fu = 570 N/mm²
Poisson’s ratio ν = 0.2
Strength ratio fb0/fc0 = 1.16 Young’s modulus E = 170 000 N/mm²
Sargin parameter D=0
Damage parameter K = 0.667 Poisson’s ratio ν = 0.3
Plastic strain εc1 = 0.0046
Crack energy Gf = 42.0 N/mm²
Fig. 7 - Exemplary material characteristics and parameters in the numeric model
The plastic concrete behavior is modelled using the Concrete Damage Plasticity (CDP) theory
(Kueres, Stark, Herbrand & Claßen, 2015). An elastic-plastic material behavior including strain
hardening after the yielding point is allocated to the steel components. Contact definitions are
introduced to prevent penetration between steel and concrete. While the tangential behavior is
characterized by friction (µ=0.2), there is a “hard-contact” in normal direction.
Influence of pin-distance and number of pins in a group
To examine the influence of pin-distance to the load-bearing capacity, a group of four pins was
modelled and the distance was varied systematically. Thereby, a ratio of a/hpin < 2.25 decreases
the shear capacity. Higher ratios cannot increase the capacity because a concrete compression
failure occurs (Figure 8 a). A pin-distance a > 2.3 hpin is recommended. In case the distance lies
beneath this limit (e.g. a = 7.5 mm) the shear capacity decreases with a greater number of pins
in a group (Figure 8 b).
Fig. 8 - (a) Related shear capacity P/Pmax subject to the pin-distance ratio a/hpin;
(b) related shear capacity P/Pmax as a function of the number of pins in a group
Influence of pin-geometry (head, shank, foot)

Besides pin-geometries A to C several pin-diameters are executable. Figure 9 (a) shows the
influence of the shank-diameter on the shear capacity, which increases nearly quadratic. This
correlation is similar to headed studs. Also the ratio between dhead to dshank respectively dfoot to
dshank influences the shear capacity as far as the ratio lies beneath 1.4 (Figure 9 b). A greater
head-diameter ensures the necessary anchorage against a pull-out failure of the connection.
Fig. 9 - (a) Influence of pin-shaft diameter on shear capacity related to the standard pin type B;
(b) influence of different pin-head and pin-foot ratios
Influence of bonding depth

The simulations with varying bonding depths show a bad influence on shear capacity for ratios
hpin/dshank beneath 4, the effect is almost linear (Figure 10 a). There is no increasing capacity for
higher ratios because of a limited height of the zone which is activated for a support of the
concrete compression strut hz = 4 dshank – dhead (Figure 10 b). For lower ratios the compression
strut cannot be built up completely.
Fig. 10 - (a) Influence of the bonding depth on shear capacity related to the standard pin type B;
(b) development of concrete pressure at the pin-foot for different bonding depths
Influence of concrete and steel strength

Furthermore the effect of varying concrete compressive strengths on shear capacity is
investigated. The shear capacity increases affine to the square root of the product of
compressive strength and modulus of elasticity until steel failure occurs (Figure 11 a). Finally a
parametrical study, where steel failure is enforced by high-performance concrete, examines the
influences of steel strength and shank-diameter. The shear capacity increases proportional to
ultimate strength and to the square of the shaft-diameter (Figure 11 b).
Fig. 11 - (a) Influence of the concrete compressive strength on shear capacity;

(b) influence of shank diameter and steel strength
DESIGN OF PIN SHEAR CONNECTORS

Based on the findings in the previous chapters an approach was developed for the design of
composite joints using small-scaled pin-connectors. As some phenomenological parallels exist
between pin-structures and headed studs, the approach of the latter connecters according (EN
1994-1-1, 2010) can be transferred.
The shear capacity limited by concrete failure modes is influenced by the shank-diameter, the
ratio hpin/dshank as well as the concrete compressive strength. Considering an empirical factor
0.4, the maximum shear capacity can be expressed as
2.045
𝑃𝑅𝑚,𝐶𝑜𝑛𝑐𝑟𝑒𝑡𝑒 = 0.4 ∙ 𝛼 ∙ 𝑑𝑆ℎ𝑎𝑛𝑘 ∙ √𝑓𝑐𝑚 ∙ 𝐸𝑐𝑚
ℎ𝑃𝑖𝑛 ℎ𝑃𝑖𝑛
where: 𝛼 = 0.16 ∙ ( ) + 2.18 for 2.5 ≤ ≤ 4.0
𝑑𝑆ℎ𝑎𝑛𝑘 𝑑𝑆ℎ𝑎𝑛𝑘
ℎ𝑃𝑖𝑛
𝛼 = 1.0 for 4.0 <
𝑑𝑆ℎ𝑎𝑛𝑘
In case of a steel failure the shear capacity is proportional to the square of the shank-diameter
and the ultimate steel strength. Considering the cross-sectional area and an empirical factor of
1.5 the maximum shear capacity limited by steel properties is
2.045
𝜋 ∙ 𝑑𝑆ℎ𝑎𝑛𝑘
𝑃𝑅𝑚,𝑆𝑡𝑒𝑒𝑙 = 1.5 ∙
∙ 𝑓𝑢
4
The maximum shear capacity can be determined by the minimum of both limits:
2.045
0.4 ∙ 𝛼 ∙ 𝑑𝑆ℎ𝑎𝑛𝑘 ∙ √𝑓𝑐𝑚 ∙ 𝐸𝑐𝑚
𝑃𝑅𝑚 = 𝑚𝑖𝑛 2.045
𝜋 ∙ 𝑑𝑆ℎ𝑎𝑛𝑘
{ 1.5 ∙ ∙ 𝑓𝑢
4
These rules are valid for the two constructive provisions:
𝑑ℎ𝑒𝑎𝑑
1.4 ≤
𝑑𝑠ℎ𝑎𝑛𝑘
𝑎
2.3 ≤
ℎ𝑃𝑖𝑛
CONCLUSIONS AND OUTLOOK

The local static load-bearing and slip capacity due to shear forces of the pin-structures as shear
connectors in steel-concrete-composite structures was investigated in experimental tests and
numerical simulations. The experimental investigations show that pin-structures are able to
transfer the inner forces between the steel and concrete cross sections. Hence, the pin-
structures are usable in composite constructions with thin concrete slabs alternative to headed
studs. On the other hand, the deformability is not sufficient. In most of the component tests the
plastic load-bearing capacity is reached, but afterwards the tests show brittle failure. This is
caused by a current pin-height hpin = 5.1 mm and the brittle shear failure of pin-structures
consisting of structural steel.
In a numerical parametric study the effects of pin-geometry (dimensions of pin-head, pin-foot,
pin-shank), pin-distance, bonding depth and material strengths on the longitudinal shear
resistance were investigated systematically. The determined relations were transferred in an
engineering model, which is based on the Eurocode 4. The engineering model includes the
design equations to determine the longitudinal shear resistance and constructive provisions like
pin-distance. This enables the design of filigree composite structures for potential applications
(Figure 12).
Fig. 12 - Application examples for small-scaled pin connector
ACKNOWLEDGMENT
This research work has been funded by the German research foundation (DFG) and is part of
the research project FE 459/5-1 “Untersuchung der Tragmechanismen neuartiger, kleinskaliger
Pin-Verbundmittel zwischen Stahl und Beton”. The researchers would like to express their
thanks for the support.
REFERENCES
Brockmann, T. (2006). Mechanical and fracture mechanical properties of fine grained concrete
for textile reinforced composites (Doctoral dissertation). RWTH Aachen University,
Aachen, Germany.
Claßen, M., Gallwoszus J., Kopp, M. & Schäfer J. (2015). Kleinskalige Pin-Verbundmittel für
den Stahl-Beton-Verbundbau, Bauingenieur, 90(3), 200-208.
Claßen, M., Hegger, J., Kopp, M., Feldmann, M., Schäfer, J. & Reisgen, U. (2015).
Verbundträger mit kleinskaligen Pin-Verbundmitteln – Experimentelle Untersuchungen
und Finite-Elemente-Simulation. Stahlbau, 84(10), 771-779.
Claßen, M. & Herbrand, M. (2016). Mindestverdübelung von Verbundträgern mit kleinskaligen
Verbundmitteln – Herleitung des Mindestverdübelungsgrades durch systematischen
Einsatz nicht-linearer Finite-Elemente-Berechnungen. Bauingenier, 91(4), 140-151.
Claßen, M., Felber, L., Herbrand, M., Kueres, D. & Hegger, J. (2016). Längsschubtragfähigkeit
von kleinskaligen Pin-Verbundmitteln – Numerische Analyse und Handrechenmodell.
Beton- und Stahlbetonbau, 111(6), 366-376.
EN 1994-1-1 (2010). Eurocode 4: Design of composite steel and concrete structures – Part 1-1:
General rules and rules for buildings. Berlin: Beuth.
Feldmann, M., Hegger, J., Hechler, O. & Rauscher, S. (2007). Forschung für die Praxis: P621 -
Untersuchungen zum Trag- und Verformungsverhalten von Verbundmitteln unter ruhender
und nichtruhender Belastung bei Verwendung hochfester Werkstoffe. Düsseldorf: FOSTA
- Forschungsvereinigung Stahlanwendung e. V.
Feldmann, M., Gündel, M., Kopp, M., Hegger, J., Gallwoszus, J., Heinemeyer, S., Seidl, G. &
Hoyer, O. (2014). Forschung für die Praxis: P804 - Neue Systeme für
Stahlverbundbrücken – Verbundfertigteilträger aus hochfesten Werkstoffen und
innovativen Verbundmitteln. Düsseldorf: FOSTA - Forschungsvereinigung
Stahlanwendung e. V.
Feldmann, M., Kopp, M. & Pak, D. (2016). Composite dowels as shear connectors for
composite beams – background to the German technical approval. Steel Construction,
9(2), 80-88.
Hanswille, G., Schäfer M. & Bergmann M. (2010). Stahlbau-Kalender 2010: Verbundtragwerke
aus Stahl und Beton - Bemessung und Konstruktion. Kommentar zu DIN V 18800-5.
Berlin: Ernst & Sohn.
Heinemeyer, S., Gallwoszus, J. & Hegger, J. (2012). Verbundträger mit Puzzleleisten und
hochfesten Werkstoffen. Stahlbau, 81(8), 595-603.
Kopp, M., Feldmann, M., Claßen, M., Hegger, J., Schäfer, J. & Reisgen, U. (2016).
Untersuchungen zur statischen Schubtragfähigkeit kleinskaliger Pin-Verbundmittel.
Stahlbau, 85(3), 200-206.
Kueres, D., Stark, A., Herbrand, M. & Claßen, M. (2015). Numerische Abbildung von Beton mit
einem plastischen Schädigungsmodell – Grundlegende Untersuchungen zu Normalbeton
und UHPC. Bauingenieur, 90(6), 252-264.
Leutbecher, T. (2008). Rissbildung und Zugtragverhalten von mit Stabstahl und Fasern
bewehrtem Ultrahochfesten Beton (UHPC) (Doctoral dissertation). University of Kassel,
Kassel, Germany.
Schäfer, J., Willms, K., Reisgen, U., Claßen, M., Hegger, J., Kopp, M. & Feldmman, M. (2015).
Small-scale pin-structures: Requirement for a multi material joint. CWA Journal, 11(5), 52-
61.
BEHAVIOR OF A NEW TYPE OF SHEAR CONNECTORS FOR U-
SHAPED STEEL-CONCRETE HYBRID BEAMS
Pisey KEO
LGCGM/Structural Engineering Research Group, INSA de Rennes, 20 avenue des Buttes de
Coësmes, CS 70839, F-35708 Rennes Cedex 7, France
pisey.keo@insa-rennes.fr
Clémence LEPOURRY
INGENOVA, Civil Engineering Office, 5 Rue Louis Jacques Daguerre, 35136 Saint-Jacques-de-
la-Lande, France
clemence.lepourry@bet-ingenova.fr
Hugues SOMJA
hugues.somja@insa-rennes.fr
Frank PALAS
INGENOVA, Civil Engineering Office, 5 Rue Louis Jacques Daguerre, 35136 Saint-Jacques-de-
la-Lande, France
franck.palas@bet-ingenova.fr
ABSTRACT
This paper presents an investigation of the behavior of a new type of shear connectors for U-
shaped steel-concrete hybrid beams. Besides the role in transferring the force between
concrete and steel material, this new type of shear connectors, welded on the upper flange of
the U-section, serves to maintain the shape of the steel cross-section during concrete
encasement. Several forms of shear connectors can be used such as L-shaped or square
cross-sections. The experimental investigation of the behavior of these shear connectors
through asymmetrical push-out tests is presented in this paper. A finite element model has
been developed in order to identify the stress behavior of the connectors and the surrounding
concrete. The FE model is validated by comparing its results against experimental data. Based
on the FEA results, an analytical formula for calculating the force transfer capacity of the
connector is proposed.
INTRODUCTION
In composite construction, profiled steel sheets have been successfully used as permanent and
integral formworks for the underneath of reinforced concrete slab (Wright, Evans, & Harding,
1987). Profiled steel sheets serve not only as the form for the concrete during construction but
also as the principal tensile reinforcement for the bottom fibers of the composite slab, offering
economic design solutions over plywood formwork. Acting compositely with a reinforced
concrete slab, profiled steel sheets produce a considerably stiffer and stronger floor system
than many others. Lately, the profiled steel sheet was introduced as permanent formwork and
integral shuttering for the sides of reinforced concrete beams (Oehlers, Wright, & Burnet,
1994). The collaboration of the profiled steel sheet and RC beams was ensured by bond
strength (Oehlers, Wright, & Burnet, 1994) which can be broken under severe loading. For
classical composite beams, the shear connection between the two materials is usually ensured
by mechanical devices such as headed studs. The behavior of headed stud shear connectors
has been investigated by numerous researchers worldwide by conducting push-out
experimental tests (Viest, 1956) (Chinn, 1965) (Valente & Cruz, 2009). Although the common
type of shear connectors is the headed stud, over the last decade some older generations of
shear connectors such as channel, angle or square bar shear connectors have been
increasingly interested by many researchers (Abe & Hosaka, 2002) (Maleki & Bagheri, 2008)
(Shariati, et al., 2013). The installation of these older generations is not expensive since the
installation procedure is similar to the one used for steel beam stiffeners or steel connection
components, where a specific welding equipment with high voltage is not required. Recently,
angles connectors have been used in steel-concrete composite beams with U-shaped steel
girders (Liu, Guo, Qu, & Zhang, 2017). However, no analytical models were readily available for
calculating the shear force transfer capacity of the angle shear connector used in this new
composite beam configuration as yet.
In this paper, the behavior of a new type of shear connectors used in U-shaped steel-concrete
hybrid beams (USCB) is investigated. Besides the role in transferring the force between the two
materials, this new type of shear connectors, welded on the upper flange of the U-section,
serves to maintain the shape of the steel cross-section during concrete encasement. Several
forms of the connector cross-section can be used such as L-shaped or square cross-sections.
The L-shaped shear connectors, in the absence of top flange, could be cheaper and more
economical in comparison to the channel ones. In general, for classical composite beams a
hoop reinforcement should be provided for the L-shaped shear connector to prevent uplift of the
concrete (ENV, 1994). However, it is not the case for L-shaped shear connector used in USCB,
see Figure 1, in consideration of the connector part fully embedded in the concrete. This new
type of shear connection is not covered in present norms of composite structures and it
requires an investigation on its behavior and on force transfer mechanism. This paper presents
the experimental investigation of the behavior of these shear connectors through asymmetrical
push-out tests. Two different types of the shear connector cross-section are considered in the
experimental test: square and L-shaped section. To get a further insight into force transfer
mechanisms while using L-shaped shear connectors as connector devices in USCB, a finite
element model has been developed. The later has been validated by comparing its results
against four experimental data tests. Based on the FEA results, an analytical formula for
calculating the force transfer capacity of the connector is proposed.
EXPERIMENTAL PROGRAM
An experimental program is developed to quantify the strength and deformation capacities of
shear connectors as well as to gain an insight into force transfer mechanisms in the USCB. A
modification of the typical push-out test setup proposed by Eurocode 4, which is usually
adopted for classic shear stud connectors, is made to represent the real situation on the USCB.
The experimental test setup is presented in the following section.
Test specimens
The specimen, Figure 1, consists of a U-shaped steel beam, two precast-slab, a reinforced
concrete beam encased in U-shaped steel beam and several shear connectors. The internal
surfaces of U-shaped steel beam were greased before concrete encasement to ensure that the
force transfer between the two materials is done only through shear connectors. The geometry
of the specimen is the following. U-shaped steel beam is 500 mm high, 300 mm wide and 6 mm
thick. The precast slabs with 70 mm thick and 450 mm wide are posed on each steel flange.
The connectors are welded on the flanges of the steel beam. Five specimens with different
types of connector are evaluated. The first specimen denoted “PO-S20”, Figure 2.b), has 4
shear connectors with square-shaped section of 20×20 mm. The second specimen denoted
“PO-L40”, Figure 2.a), has 3 shear connectors with L-shaped section of 40×40×4 mm. The third
to fifth specimen denoted respectively “PO-L50a”, “PO-L50b” and “PO-L50c” have 3 shear
connectors with L-shaped section of 50×50×5 mm. The spacing between each connector for all
specimens is 300 mm.
Precast
Stirrup Connector
Φ8@300
Concrete U-shaped steel

encasement
Fig. 1 – Specimen cross-section.
a) PB-POL40/L50 b) PB-POS20
Fig. 2 – Test specimens without concrete encasement.
Materials
The concrete for RC beam-floor has a strength class of C25/30. The concrete characteristics at
the day of test were determined on cylinder samples with dimensions of 11×22 cm. Due to the
different usage of concrete in precast slab and RC beam-floor, the concrete characteristics of
precast slab were measured separately. The steel grades of U-shaped steel beam, of L-shaped
steel profile and of square section steel profile are S355, S235 and S235, respectively. Coupon
samples were taken from the steel sheet and from the shear connector used in the push-out
specimens. Results of material characteristic tests are summarized in Table 1. It is worth
mentioning that PO-L50a,b,c specimens have been tested at an early age of the concrete in
order to have a concrete strength at the day of test equal to 70% of concrete characteristic
strength as prescribed in Eurocode 4 (EC4-1-1, 2005).
Table 1 – Mean material properties.

Specimen Concrete Precast U-shaped Connectors
fcm (MPa) fcm (MPa) fy (MPa) fu (MPa) fy (MPa) fu (MPa)
S20 27.75 42.29 555 693
L40 26.97 43.29 430 550
L50a 20.27 38.15 491 553
L50b 21.29 38.79 325 445
L50c 22.71 38.04
Loading and test procedure

The specimens were tested under monotonic loading. The compressive load was applied
horizontally and monotonically on the specimen via a hydraulic jack with a capacity of 1500 kN.
The actuator was attached to a 40 mm-thick steel plate that serves as a platform for push-out
test specimens. The 40 mm thick steel plate was welded to two 6 mm-thick gussets, which are
connected to the webs of steel beam at one end via twelve 20 mm-diameter high-strength bolts,
see Figure 3 and 4. At the other end of the specimen, the back side surface of the concrete
floor was put in contact with a rigid steel beam. The specimen was positioned horizontally on 2
vertical supporting short steel columns where at the top surface of their end-plate (with slotted
holes), two greased PTFE plates were placed, allowing the horizontal displacement of the steel
beam to occur.
Specimen Rigid beam

40 mm-thick
steel plate
Vertical supports
Fig. 3 – Schematic of pushout test setup.
To measure the relative horizontal displacement (slip) between the concrete and the steel
beam, 6 LVDT displacement sensors were installed at both sides of the beam (3 for each side),
the location where the connectors were placed. The measurement of the vertical separation
between the reinforced concrete and the steel beam (uplift) was done through 4 LVDT
displacement sensors (2 for each size of the beam), see Figure 5 where the sensors with odd
numbering are on the other side of the specimen, not presented in figure. The sensors C1 to C6
and C9 to C12 correspond to slip and uplift measurements, respectively. The force and
displacement readings were recorded by the universal machine and through a data acquisition
system connected to the displacement transducers, respectively.
Fig. 4 – View of pushout test setup.
Fig. 5 – Sensors position on the specimen.
Test results
The relation between the push-out load and the mean slip and uplift between the concrete
beam-floor and the steel beam is illustrated in Figure 6. One can observe that the specimens
with L-shaped shear connectors present a ductile behavior while the one with square bar shear
connectors, PB-S20, exhibits a less ductile behavior. Besides, it can be seen that at 80% of
ultimate load after peak for each specimen, the vertical separation between concrete and steel
beam is less than half of the longitudinal slip. The ultimate load and corresponding slip of each
specimen are presented in Table 2.
Table 2 – Ultimate load and corresponding slip.

Specimen PO-S20 PO-L40 PO-L50a PO-L50b PO-L50c
Pmax (kN) 748.6 637.1 932.1 916.5 975.9
δmax (mm) 2.743 5.054 5.081 3.032 3.341
Fig. 6 – Load-mean slip and uplift response.
For PB-S20 and PB-L40 specimens, no concrete cracking was observed during the test. The
failure of PB-S20 was due to the local concrete crushing. Due to the lack of ductility, which is
required in design codes for partial shear connection design, PB-S20 solution is withdrawn and
no further discussed in this paper. For specimens with L-shaped shear connectors, the concrete
was removed to examine the connectors after testing. The failure of these specimen was
associated with the plastic deformation due to shear of L-shaped shear connector, near the fillet
between the angles of the cross-section and at the leg angle welded on the flange of steel
beam. This mode of failure leaded to a large ductility in the force-slip response. The end of the
plastic yielding was due to the rupture of the connector and the welds, see Figure 7.
Fig. 7 – Fracture of L-shaped shear connector.
FINITE ELEMENT SIMULATIONS

In order to get further information, a numerical simulation has been performed on
ABAQUS/Explicit. Since the use of square bar shear connectors is not a solution for partial
shear connection design in U-shaped steel-concrete hybrid beam, FEA was performed only for
specimens with L-shaped shear connectors. All main components that may affect the behavior
of shear connection were considered in the model. By taking the advantage of symmetry, only a
half of the specimen was simulated.
a) Steel beam b) Concrete beam- c) Connectors & d) Rebars
(S4R) floor welds (C3D8R) (B31)
(C3D8R & C3D6R)
Fig. 8 – FE model.
Finite element type and mesh

The components were modeled in separate parts as presented in Figure 8. The concrete
component was meshed with combine solid element C3D8R and C3D6R available in ABAQUS
library. The former element type is an 8-node brick element and the latter is 6-node linear
triangular prism element. Each node has three translational degrees of freedom. C3D8R
element was also used for meshing shear connectors and welds. The steel beam was meshed
with 4-node shell element (S4R) while the reinforcement bars were meshed with beam element
(B31). An overall mesh size of the concrete component, steel beam, shear connectors, welds
and of the rebars are 20 mm, 20 mm, 1 mm, 2 mm and 20 mm, respectively. For concrete and
steel beam component, a fine mesh was used at zones in contact with shear connectors.
Constraints and contact interactions

Once all parts of the model were positioned together into an assembly, appropriate constraints
were used to describe the interaction between components. The surfaces in contact between
shear connectors and welds were tied together. The bottom surfaces of the latter were tied to
the surface of the beam flange. This is equivalent to the actual push-out specimen where shear
connectors remained tied to the steel beam flange via welding. For rebars which were placed
inside the concrete beam-floor, the embedded constraint was applied. In this constraint, the
translational DOF of the nodes on the rebar elements were constrained to the interpolated
values of the corresponding DOF of the concrete element. The slip and debond of the rebar
was ignored. Contact interactions were applied for the remaining contact surfaces. The
interaction properties of the contact interactions used in the model were defined by the
tangential and normal behavior to the surfaces. Since the interior surfaces of the U-shaped
steel beam were greased, the frictionless for tangential behavior was used at the contact
surfaces between steel beam and other components. The penalty friction formulation was used
for tangential behavior of the contact surfaces between shear connectors/welds and concrete
with a coefficient of friction equal to 0.2. The default normal behavior (hard contact) was
assumed for all contact surfaces. The latter allows a minimum penetration of the slave surface
into the master surface. For all contact pairs, the penalty contact formulation was used.
Material models
Cap plasticity model available in ABAQUS (Systemes, 2012) was adopted to model the
concrete material. The cap parameters used for each specimen (RC-L40 and Precast-L40 for
PO-L40, and RC-L50 and Precast-L50 for PO-L50a,b,c) are listed in Table 3 in which the
envelop curve of the cap model was matched with Mohr-Coulomb surface using simple tensile
and compressive strength.
Table 3 – Concrete cap plasticity parameters.

Cohesion Angle of Cap Initial cap Transition Flow
(MPa) friction eccentricity yield surface surface stress
parameter position radius ratio
RC-L40 3.9548 68.66
RC-L50 3.1405 68.66
0.65 0 0.01 1
Precast-L40 6.0074 68.84
Precast-L50 5.4222 68.78
Table 4 shows the true stress-strain data for the L-shaped shear connectors as input in
ABAQUS. The elastic-perfectly plastic model was adopted for steel beam. The modulus of
elasticity and the Poisson’s ratio of steel used in the model were 210 GPa and 0.3, respectively.
Table 4 – Steel connector properties.

L40×40×4 L50×50×5
Yield stress (MPa) Plastic strain Yield stress (MPa) Plastic strain
430.88 0.0000 325.50 0.0000
438.80 0.0182 330.03 0.1320
660.00 0.1792 534.00 0.1798
Validation of finite element model

Push-out tests presented in previous section were used to verify the developed finite element
models. The shear connection capacity obtained from the tests and finite element analysis as
well as the load-slip and load-uplift behavior of the shear connector, including its failure modes,
have been investigated. Figure 9 and 10 show the comparison of load-slip and load-uplift
obtained from FEA against the experimental results. Due to the same properties of materials
used in PO-L50a,b,c specimens, only one FE simulation was performed for those specimens. It
can be seen that there is a good agreement between the pairs. For PO-L40 specimen, FEA
produced a quite similar load-slip curve up to ultimate load. When the shear connector fully
yields and the fracture begins to occur, the two curves diverge. After the ultimate load, the
experimental load descends progressively while the numerical one drops insignificantly.
The stress pattern in the components is illustrated in Figure 11 and 12 for PO-L50 specimen.
One can observe that there are two plastification plans due to shear in the L-shaped shear
connector and that there is local pressure effect in the concrete block.
Fig. 9 – Comparison of FEA results against experimental test for PO-L40 specimen.
Fig. 10 – Comparison of FEA results against experimental test for PO-L50a,b,c specimen.
FORMULATION OF A NEW DESIGN EQUATION

A new design equation was developed to predict the strength of the L-shaped shear connector
used in USCB. This equation was derived based on the observation on deformation of shear
connectors and the stress pattern obtained from FEA. It showed that the shear connector
resistance may be divided into two parts: concrete contribution and shear resistance of the
connector. Herein, it is assumed that the concrete block in front of shear connector is idealized
by a rectangular stress block. The new design equation for one shear connector is then given
by:
f
Pu  2  As1  As 2   u  2Kc  Ac  fc
3
where As1, As2 and Ac are defined in Figure 13 in which Lc is the welded length of shear
connector to the steel beam flange; fu is the ultimate strength of shear connector; f c is
compressive strength of concrete; and Kc is a calibrated factor on concrete strength due to local
pressure effect. The latter is defined by the following expression obtained from a regression
analysis of a parametric study, which is not presented in this paper:
f L
K c  2.229 u  3.565 c
fc ha
in which ha is the height of the L-shaped cross-section. Table 5 shows the comparison between
the ultimate loads given by proposed design equation against experimental results. Ps and Pc
represent the contribution of L-shaped shear connector and concrete, respectively, to the shear
capacity of the connectors.
Fig. 11 – Stress pattern of shear connectors at ultimate load for PO-L50a,b,c specimen.
Fig. 12 – Stress pattern of surrounding concrete at ultimate load for PO-L50a,b,c specimen.
In the proposed equation, the width of the concrete block was assumed to be equal to the
length of shear connector welded to the steel flange which is extensively larger than what is
observed in numerical results.
Table 5 – Comparison of the ultimate loads given by proposed design equation against
experimental results.
fu fc Lc ha Ps Pc Pu,th Pu,exp
Kc Pu,th/Pu,exp
(MPa) (MPa) (mm) (mm) (kN) (kN) (kN) (kN)
L50a 445 20.27 40 50 7.59 763.8 184.7 948.52 932.1 1.02
L50b 445 21.29 40 50 7.34 763.8 187.5 951.34 916.5 1.04
L50c 445 22.71 40 50 7.01 763.8 191.2 955.02 975.9 0.98
L40 550 26.97 20 40 8.28 517.7 107.2 624.94 637.1 0.98
L-shaped shear connector
Weld
lan
rp
h ea A s1)
S 1(
Lc al
loc
ete (A c)
r
Shear plan 2 nc ure
(A s2) Co ress
tw p
ha
Fig. 13 – Shear failure plans and concrete local pressure zones.
CONCLUSION
Push-out tests of U-shaped steel-concrete hybrid beams with two different shapes of shear
connectors were conducted. It was shown that the beam with L-shaped shear connector
exhibited a ductile behavior while the beam with square bar shear connector presented less
ductility. For that reason, the later solution was withdrawn. Identification of the stress behavior
of the shear connector and the surrounding concrete was performed using FE model. It gave
the key features in developing the new design equation for L-shaped shear connectors used in
U-shaped steel-concrete hybrid beams. It showed that the shear connector resistance may be
divided into two parts: concrete contribution and shear resistance of the connector. The former
was due to the local pressure effect in the concrete block. To take into account this effect, a
calibration factor obtained from a regression analysis has been introduced in the proposed
design equation for the ultimate load of the shear connectors. It showed that the estimated
ultimate loads are in good agreement with experimental results.
ACKNOWLEDGMENTS
The authors gratefully acknowledge financial support by the ANR (Agence Nationale de la
Recherche, France) through the project LabCom ANR B-HYBRID.
REFERENCES
Abe, H., & Hosaka, T. (2002). Flexible shear connectors for railway composite girder bridges.
Composite construction in steel and concrete IV, 71--80.
Chinn, J. (1965). Pushout Tests on Lightweight Composite Slabs. AISC Engineering Journal 2,
129-134.
EC4-1-1. (2005). Calcul des structures mixtes acier-béton. Règles générales et règles pour les
bâtiments.
ENV. (1994). Eurocode 4: design of composite steel and concrete structures—Part 1.1:
General rules and rules for buildings. Brussels: European Committee for
Standardization.
Liu, Y., Guo, L., Qu, B., & Zhang, S. (2017). Experimental investigation on the flexural behavior
of steel-concrete composite beams with U-shaped steel girders and angle connectors.
Engineering Structures, 492-502.
Maleki, S., & Bagheri, S. (2008). Behavior of channel shear connectors, Part I: Experimental
study. Journal of Constructional Steel Research, 1333--1340.
Oehlers, D. J., Wright, H. D., & Burnet, M. J. (1994). Flexural strength of profiled beams.
Journal of Structural Engineering, 378--393.
Shariati, M., Sulong, N. R., Suhatril, M., Shariati, A., Khanouki, M. A., & Sinaei, H. (2013).
Comparison of behaviour between channel and angle shear connectors under
monotonic and fully reversed cyclic loading. Construction and Building Materials, 582--
593.
Systemes, D. (2012). ABAQUS 6.12 Theory manual. Dassault Systemes Simulia Corp.,
Providence, Rhode Island.
Valente, I. B., & Cruz, P. J. (2009). Experimental analysis of shear connection between steel
and lightweight concrete. Journal of Constructional Steel Research 65, 1954-1963.
Viest, I. M. (1956). Investigation of stud shear connectors for composite concrete-steel T-
beams. Journal of the american concrete institute, 875-891.
Wright, H., Evans, H., & Harding, P. (1987). The use of profiled steel sheeting in floor
construction. Journal of Constructional Steel Research, 279--295.
SHEAR PERFORMANCE OF NONPARALLEL TYPE TWIN
PERFOBOND STEEL PLATE SHEAR CONNECTORS: FEM
ANALYSIS AND EXPERIMENTAL VERIFICATION
Hai Chen
Ph.D. Candidate of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: 1300222006@hqu.edu.cn
Yang Liu (corresponding author)

Associated Professor of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: lyliuyang@hqu.edu.cn
Zixiong Guo
Professor of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: guozxcy@hqu.edu.cn
Yong Ye
Lecturer of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: qzyeyong@126.com
Bahram M. Shahrooz
a
Professor of Structural Engineering, Dept. of Civil and Architectural Engineering
Construction Management, Univ. of Cincinnati, Cincinnati, U.S.
b
Distinguished Professor, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: Bahram.shahrooz@uc.edu
ABSTRACT
This paper describes a new type of shear connector, which consists of two pieces of nonparallel
perfobond plate ribs (NPP). Using three-dimensional finite element models, calibrated against
experimental data, were used to conducted parametric studies on the influence of different
parameters such as, rib arrangement and thickness of rib on the shear capacity of the NPP
connectors. A total of 81 push-out specimens were analyzed with different angles and gaps
between the ribs, rib thickness, and concrete strengths. The numerical results show that
increases in the rib thickness and concrete strength enhance the shear capacity of NPP
connectors. The shear strength of NPP connectors is reduced with an increase in the angles
between the ribs. The gap between the ribs does not have a significant influence on the
shear-slip relationships. The NPP connectors are a viable alternative to conventional shear
connectors. Different strength and stiffness requirements can be met by adjusting rib
configuration.
INTRODUCTION
The many benefits of steel and concrete structures rely on sufficient bond transfer between
these two materials to prevent separation of steel and concrete at their interface. This goal is
achieved through the use of shear connectors. Headed studs are the industry standard
developed based on many past studies [e.g., Oehlers and Coughlan 1986]. Another emerging
type is perfobond connectors. These connectors consist of perforated steel plates welded to
steel at the concrete-steel interface. Variations of perfobond connectors have studied, e.g.,
perfobond ribs [Leonhardt et al. 1987], T-perfobond ribs [Vianna et al. 2008], long-hole
perfobond ribs [Zheng et al. 2016], or Y-type perfobond ribs [Kim et al. 2017]. Compared with
the headed stud connector, perfobond connectors have higher shear stiffness and improved
fatigue performance.
A number of previous experimental and numerical simulation studies have examined the shear
capacity and behavior of perfobond connectors [e.g., Medberry and Shahrooz 2002; Valente
and Cruz 2004; Ahn et al. 2010; Oguejiofor and Hosain 2011]. However, previous studies were
limited to single perfobond connectors and the behavior of twin perfobond connectors have not
been evaluated, in particular the influence of the angle between the ribs.
The main objective of this paper is to present a new type of shear connector, which is
composed of two pieces of nonparallel perfobond plate ribs (NPP). The impacts of the rib
arrangement, the rib thickness and rib angle, the concrete strength, and the gap between the
ribs on the shear resistance were investigated. Three-dimensional finite element models were
developed and calibrated using experimental data, and were employed to conduct parametric
studies.
SHEAR RESISTANCE CHARACTERISTICS OF NPP CONNECTOR

Figure 1 illustrates the load carrying mechanism of the NPP connector under in-plane shear.
Four components contribute towards the shear resistance: (A) end bearing resistance of the
steel plate; (B) the resistance of the non-parallel ribs; (C) the dowel resistance of transverse
reinforcing bars, if present; and (D) dowel resistance of concrete studs in the steel plate holes.
tcon1 σcon
Feb1>Feb2
tcon1<tcon2
Rotating rib
tcon2
R.C. R.C.
Feb1 Feb2
Slab Slab
PBL connector NPP connector
(a) Shear resistance mechanism of NPP connector (b) Conventional PBL and NPP connectors
Fig.1 - Mechanical behavior of a NPP connector
The differences between shear resistance mechanism between conventional perfobond shear
connector (Perfobond Leisten, abbreviated as PBL) and NPP connector are shown in Figure
1(b). For the orientation of perfobond rib shown, the end bearing resistance Feb of NPP is
smaller than that of PBL connector. However, the normal stress (con) in NPP increases the
frictional resistance and the interfacial shear (tcon) is larger than that in PBL, i.e., tcon2 > tcon1.
The angle between the ribs is a crucial factor that affects the shear strength of NPP connectors.
FINITE ELEMENT MODEL
To investigate the influence of different parameters three-dimensional finite element models
simulating a typical push-out experiment were developed using finite element pre-processor
ABAQUS/CAE [Hibbit, K. 2014]. Suitable contact interactions, constraints, boundary conditions,
and material models were specified to estimate the shear performance of NPP connectors.
FINITE ELEMENT TYPE AND MESH

A push-out specimen consists of four parts: (A) two concrete components, (B) a structural steel
member, (C) two perfobond ribs, and (D) reinforcing bars, if present. A three-dimensional finite
element modeling of such specimen is shown in Figure 2. Three-dimensional eight-node
reduced integration element (C3D8R) was used to model the perfobond rib, concrete
components, and structural steel member. The reinforcing bars were modelled by
three-dimensional two-node truss elements (T3D2). A fine mesh was used for the concrete
dowels and the regions around the holes in the perfobond rib, to more accurately capture the
high stress levels in these regions. The coarse mesh size was 20 mm and the fine size was
about 3 mm.
Downward displacement
Steel member
Links connecting load (C3D8R)
to steel members Perfobond rib connected
to steel member
Perfobond rib
(C3D8R)
(b) Steel beam and perfobond rib

s
nt Embed reinforcement Concrete component
ne
po (C3D8R)
m
co
ete
cr
on
ofc
ce
rfa
su
m
Y X tto Reinforcement
bo (T3D2)
Z he
a tt
d
xe
Fi
(a) Push-out model (c) Concrete slab and reinforcement rebar
Fig.2 - Finite element model
INTERACTION AND BOUNDARY CONDITIONS

The various components were assembled in accordance with the actual configuration of a
typical push-out specimen. The perfobond ribs were tied to the flanges of the structural steel
members to prevent relative slip, simulating welds that would not fail prior to fully developing the
capacity of the perfobond connection. The contact pair algorithm available in ABAQUS was
used to define the perfobond-flange interface as well as the contact between the perfobond rib
and the surrounding concrete. The default normal behavior was assumed to be a ‘hard’ contact
pressure-overclosure relationship. The penalty frictional formulation was adopted, and the
coefficient of friction between perfobond rib and concrete was taken as 0.4 based on the results
of a sensitivity study. The other contact interactions were assumed to be frictionless. Anchorage
of reinforcing bar was simulated by embedded constraints.
As illustrated in Figure 2(a), the bottom surfaces of the concrete segments were assigned fully
fixed. The downward displacement (i.e., in the Z direction) was applied to a reference that was
linked to the top of the structural steel members. To ensure a quasi-static solution from the
explicit dynamic procedure, the inertia effects were minimized by using a smooth amplitude
function to increase the downward displaced at a rate of 24 mm/min.
CONCRETE MATERIAL MODEL

The concrete damage plasticity model (CDP) was implemented in the reported research. CDP
model needs to be defined in terms of plastic, compressive, and tensile properties. To define the
plastic property, five parameters need to be specified: (A) dilation angle =30°, (B) flow
potential eccentricity ϵ=0.1, (C) ratio of biaxial to uniaxial compressive strength σb0/σc0=1.16, (D)
ratio of second stress invariant on the tensile meridian to that on the compressive meridian
K=2/3, and (E) viscosity parameter μ=0 [Hibbit K. 2014].
The nonlinear behavior of plain concrete material in uniaxial compression was presented by a
uniaxial compressive stress-strain curve shown in Figure 3(a). The stress-strain relationships
(i.e., c –c) were determined from Eq. (1), as suggested in FIB [FIB MC 2010].
 ε
0.4  c  0 < ε c  εc e 
 εc e
σ c  κ  η  η 2

fc 1   κ  2   η
ε ce <εc  εcp  (1)
 ε ε
1  0.15  c cp
 εcu  εcp
ε cp <εc  εcu 
where fc is the cylinder compressive strength of concrete (MPa). The plasticity factor κ=Ec⋅εcp/fc ;
η=εc/εcp; and εce, εcp, εcu are the elastic strain, peak strain, and ultimate strain, respectively with
the following values: εce=0.4fc/Ec, εcp=2.3×10-3, 2.4×10-3 and 2.5×10-3 for C30, C40, C50 concrete
respective, and εcu=α⋅εcp. The coefficient α=13.6 was iteratively calibrated to match push-out test
results from a previous study [Zheng et al. 2016]. The initial Young's modulus (Ec) was
calculated by Eq. (2), according to FIB [FIB MC 2010].
1
Ec  Ec0  αE   fc /10  3
(2)
3
where Ec0=21.5×10 MPa and E=1.0 for quartzite aggregates.
As shown in Figure 3(b), tension plasticity curve was defined as a function of the tensile stress
and crack width, obtained from Eq. (3) [Birtel and Mark 2006].
t   c  w 3   c w w
 1   1
ft   w c  
   exp     1+c13  exp  c2    (3)
   wc  wc
where σt is the tensile stress of concrete (MPa); ft is the tensile strength (MPa); w is the crack
width (mm); wc is the crack width (mm) at the complete loss of tensile strength taken, wc=5.14
GF/ft in which GF is the fracture energy (N/mm) required to create a unit area of stress-free crack,
GF=0.073fc0.18; and the constants are c1=3, and c2 =6.93 according to [Zheng et al. 2016].
σt
ft
T
0 wc w
C
(a) Compression (b) Tension
Fig.3 - Material constitution of concrete
MATERIAL MODEL FOR STRUCTURAL STEEL AND REINFORCEMENT

The stress-strain relationship of structural steel and reinforcement was modeled by a tri-linear
curve, as shown in Figure 4 [Zheng et al. 2016]. The material behavior is initially elastic followed
by strain yielding and then hardening. The tensile yield strength fy and the ultimate tensile
strength fu were selected based on material test data. For structural steel and reinforcement
steel, the Young’s modulus Es was taken as 206GPa and 200GPa, respectively. As shown in
Figure 4, Esp set equal to 0.01Es.
Fig.4 - Stress-strain relationship for structural and reinforcement steel
VERIFICATION OF FINITE ELEMENT MODEL

The finite element model was verified against previous data from a number of tested push-out
specimens [H. Chen 2016; Ahn et al. 2008; Cândido et al. 2010]. These test specimens (A) had
single or twin perfobond connectors, (B) were with and without end bearing, (C) used various
concrete strengths, (D) had thin to relatively thick plates as perfobond ribs, and (E) utilized a
different number of holes in the perfobond connectors.
As evident from Table 1, the results obtained from finite element analysis accurately replicate
the experimental capacities. The mean value of the finite element results over the experimental
ones is 0.98 with a coefficient of variation of 0.038. Moreover, Figure 5 shows a good
agreement between the experimental and numerically generated load–slip curves. Having
demonstrated the accuracy of the finite element modeling (FEM) procedure presented herein, a
similar FEM method was used to perform a series of parametric studies.
a
Table 1 - Comparison of shear connector capacity obtained from experiments and
finite element analysis
fc D t n Ptest PFE
Specimen Ptest /PFE Reference
(MPa) (mm) (mm) (mm) (kN) (kN)
SP1 26 50 12 1 336.1 331.0 1.02
SP2 26 50 12 2 452.4 449.6 1.01 [H. Chen, 2016]
TP1 26 50 12 2 940.9 1005.1 0.94
o
PF-S-A2 28 55 6 4 628.3 650.4 0.97
[Ahn et al. 2010]
PF-S-B2 28 55 6 4 410.0 410.1 1.00
[Cândido-Martins
P1F 31 30 15 1 309.4 331.5 0.93
et al. 2010]
fc is measured concrete strength; D is Holed diameter; t is thickness of perfobond rib; n is
number of holes in perfobond; Ptest is measured capacity; PFEM is calculated capacity from FEM
1200 8 0
6 0
800
Load (kN)
SP1_TEST 4 0
SP1_FEM
SP2_TEST
L
400
SP2_FEM
2 P 0
TP1_TEST
TP1_FEM
P
P
P
0 0
0 10 20 30 40 0 1
Slip (mm) S
(a) [H. Chen, 2016] (b) [Ahn et al. 2010]

450
300
Load (kN)
150
P1F_TEST
P1F_FEM
0
0 10 20 30 40
Slip (mm)
(c) [Cândido-Martins et al. 2010]
Fig.5 - Comparison of experimental and numerical load–slip behavior of push-out test
PARAMETRIC STUDY
A parametric study was conducted to study the influence of the following parameters on the
shear performance of NPP connector: (A) the rib gaps (L), (B) the angle between ribs (α), (C)
the thickness of the rib (t), and (D) the concrete strength. A total of 81 push-out simulations
were conducted. The cases were divided into 27 groups, each group having identical
parameters but with different concrete strengths: 30, 40 and 50 MPa (designated as NO.1, NO.2,
and NO.3, respectively). The details of the cases are shown in Figure 6.
H-250×300×16×14 L= 50, 100 ,150 mm for group

For groups P3 only P1, P2 and P3 respectively.
for all groups
L
50 mm diameter holes
α
Concrete Block:
550×550×300 α was changed
from 0° to 20° by 5°
#10 reinforcing bars @125 mm
Fig.6-Details of finite element model in parametric study
The shear strength of the 81 push-out specimens obtained from the finite element analysis are
summarized in Table 2. As expected, the ultimate load capacity becomes larger as the concrete
strength is increased. The effects of the other variables will be discussed after presenting and
discussing the load-slip relationships.
Figure 7 illustrates the load versus slip curves for specimens in group P2 with a 100mm gap.
The trend of load-slip curves do not change as a function of ribs angle α although the maximum
load becomes larger when the rib angles have negative angles. These curves can be divided
into three stages: (A) elastic, (B) elastoplastic, and (C) plastic. These stages are described with
reference to the load-slip relationship for specimen P2_15 in Figure 7(c). The elastic stage
starts from the origin to Point A. The curve at this stage is almost linear with a very small
slip(<0.5mm). The applied shear load is resisted by the concrete studs in the hole of the rib, as
evident by the inset in Figure 7(c). The peak average stress is concentrated in the vicinity of the
concrete studs, while the average stress in other regions is much smaller than the maximum
concrete stress. The elastoplastic stage starts at Point A and ends at Point B, at which the
displacement is several times larger than that for Point A. A gradual loss of stiffness is evident.
The areas subjected to high stress spread beyond the concrete studs, suggesting good
interaction between the ribs and the surrounding concrete. The contribution to the shear
resistance of end bearing becomes more important. Beyond Point B, the slope of the curve
gradually increases up to a very large displacement, labeled as plastic stage (Point C). A similar
trend may be observed for the other cases. The results indicate that NPP connectors have
sufficient ductility and may be used in composite members where fully plastic rotation is needed.
Table 2 – Calculated capcities
concrete cylinder strength
Group L  t (MPa)
Ultimate load (kN)
a
(mm) ( °) (mm) NO.1 NO.2 NO.3 P1 P2 P3
P1_0 0 12 30 40 50 2740 3295 3570
P1_5 5 12 30 40 50 2674 3164 3392
P1_10 50 10 12 30 40 50 2379 2859 3177
P1_15 15 12 30 40 50 2152 2677 2984
P1_20 20 12 30 40 50 2128 2611 2863
P2_0 0 12 30 40 50 2714 3090 3284
P2_5 5 12 30 40 50 2311 2792 3000
P2_10 10 12 30 40 50 2123 2597 2758
P2_15 15 12 30 40 50 2061 2437 2668
P2_20 20 12 30 40 50 1806 2238 2443
P2_N20 -20 12 30 40 50 2149 2366 2585
P2_N15 -15 12 30 40 50 2273 2660 2883
P2_N10 -10 12 30 40 50 2388 2805 2927
P2_N5 100 -5 12 30 40 50 2559 2955 3140
P2_5A 5 6 30 40 50 1687 1990 2147
P2_10A 10 6 30 40 50 1502 1934 2137
P2_15A 15 6 30 40 50 1530 1960 2135
P2_20A 20 6 30 40 50 1315 1751 1873
P2_5B 5 18 30 40 50 2923 3454 3627
P2_10B 10 18 30 40 50 2630 3090 3314
P2_15B 15 18 30 40 50 2599 3069 3264
P2_20B 20 18 30 40 50 2286 2825 3046
P3_0 0 12 30 40 50 2963 3456 3674
P3_5 5 12 30 40 50 2588 3036 3330
P3_10 150 10 12 30 40 50 2218 2744 3060
P3_15 15 12 30 40 50 2173 2648 2883
P3_20 20 12 30 40 50 2046 2511 2781
a
Negative angle indicates reversed load – see the inset in Figure 7(b)
To understand the influence of the rib angle α, the shear capacity is plotted in Figure 8 versus α
for different concrete strength and rib gaps L. In this figure, N indicates a negative angle of the
NPP connectors, i.e., the load enters the connectors from their wider gap, refer to the inset in
Figure 8(a). The loading direction only affects the shear strength of NPP connectors, with a
higher strength in the negative direction than that in the positive direction. The strength is
increased by 20.8% when α changed from -20° to 0° and reduced by 33.5% when α is changed
from 0° to 20° for concrete with compressive strength of 30MPa. Figure 8 clearly demonstrates
a similar trend for the other concrete strengths and rib gaps. However, as discussed previously
(see Figures 7(a) and 7(b)), the angle does not have an obvious difference on the load-slip
relationships between both directions and the ductility is not adversely affected by the
orientation of rib connectors relative to the loading direction.
a
(a) Positive angle (b) Negative angle
3600
2700
C
Load (kN)
1800
B
o
900
A
0
0 10 20 30 40
Slip (mm)
(c) Specimens P2_15

Fig.7 – Load versus slip curves for specimens of group P2
4 0
3 0
2 L 0
L
1 L C C C 0 =
L C C C =
α L C C C =
0
- 2 5
α
(a) Rib gaps of 100 mm (b) Rib gaps of 50, 100 and 150 mm
Fig.8 - Load versus ribs angle curve for specimens with different rib gaps
The load versus rib thickness (t) for group P2 is shown in Figure 9. The shear strength is
increased almost linearly. The increases for t=18 mm in comparison to t=6mm are summarized
in Table 3.
Table 3 - Influence of rib thickness
on strength (18mm vs. 6mm)
4000
α (°) fc (MPa) % increase
3000 30 73.2%
5 40 73.6%
Load (kN)
2000 L 50 69.0%
30 75.2%
1000 α=5°: C30 C40 C50 10 40 59.2%
α=10°: C30 C40 C50
α α=15°: C30 C40 C50 50 55.1%
α=20°: C30 C40 C50
0 30 69.9%
0 6 12 18 24 15 40 56.6%
Rib thickness (mm)
50 52.9%
Fig.9- Load versus rib thickness for group P2
specimens with different rib angles (α) 30 73.9%
20 40 61.3%
50 62.6%
The load versus the gap between the ribs (L) is plotted in Figure 10. Increasing the gap from 50
to 100mm is seen to reduce the capacity. This reduction is deemed to be because of the out of
plane bending deformation of the flange of the structural steel member onto which the NPP
connectors are welded. This deformation depends on the stiffness of flange extending beyond
the web. The gap between the ribs affects the stiffness. The out-of-plane flange stiffness is
larger for L=50mm than that for L=100mm; hence, the shear capacity is larger for the small gap.
In the case of L=150mm, the capacity is nearly the same as that for L=50mm because two webs
were used. It should be noted that the “span” over which the flange can bend out of plane is
nearly equal for L=50mm with one web and L=150mm with two webs. The results suggest that
the shear capacity of NPP depends not only on the properties of the connector but it also
depends on the stiffness of the flanges onto which they are welded. Additional studies are
needed to quantify the flange stiffness in calculation of shear capacity of NPP connectors.
4000
3000
L=50 mm
L=150 mm
Load (kN)
2000
1000 L=100 mm α=5°: C30 C40 C50

α=10°: C30 C40 C50
α=15°: C30 C40 C50
0
0 50 100 150 200
L (mm)
Fig.10 - Load versus rib gaps curves for different concrete strengths
SUMMARY AND CONCLUSIONS
Refined three-dimensional nonlinear finite element analyses were conducted to investigate the
shear capacity of non-parallel type twin perfobond plate(NPP) connectors. The focus of the
analyses was on modeling of push-out specimens that have been used in past experimental
studies. The models accounted for nonlinear material properties of concrete, structural steel
member, steel reinforcement, and steel ribs; and contact surfaces between steel beam flanges,
steel rib surfaces, and concrete. The shear capacity and load-slip curves of conventional
perfobond shear connector (PBL) from a number of past experimental programs could
successfully be predicted. After validating the nonlinear finite element model, an extensive
parametric study of 81 cases was conducted to examine the influence of various parameters
was. The following conclusions can be drawn based on the reported results:
1) The proposed NPP connector has adequate shear capacity and exhibits excellent ductility.
This type of connector can be used as a practical alternative to meet different connector
performance requirements in composite structures.
2) The load-slip relationships of NPP connectors have three distinct stages: elastic,
elastoplastic, and plastic. The large plastic characteristics of NPP connectors make them a
good choice for composite members where fully plastic rotation is needed.
3) The loading direction relative to the orientation of the ribs in NPP connector affects the
shear strength of NPP connectors, with a higher strength in the negative direction (the load
enters the connector on its wider gap) than that in the positive direction. The load-slip
relationships are, however, not affected. In other words, the ductility of NPP connectors is
not affected by the loading direction.
4) The angle between ribs is a key parameter influencing the shear strength The strength
decreases with an increase in the angle irrespective of the loading direction being positive
or negative.
5) The gap between the ribs has no obvious infect on shear strength.
6) The shear strength becomes higher with an increase in the concrete strength and rib
thickness.
ACKNOWLEDGMENTS
Dr. S. Zheng from Huaqiao University provided significant assistance with the FEM analyses.
The authors sincerely appreciate his help. The research presented in this paper was funded by
the Key Program of Natural Science Foundation for Young Scholar in University of Fujian
Province, China (Grant No. JZ160410) and the Natural Science Foundation of Fujian Province,
China (Grant No. 2014J05061), both of which are greatly acknowledged.
REFERENCES
Oehlers, D. J., & Coughlan, C. G. (1986). The shear stiffness of stud shear connections in
composite beams. Journal of Constructional Steel Research, 6(4), 273–284.
Leonhardt, F., Andrä, W., Andrä, H. P., & Harre, W. (1987). Neues, vorteilhaftes verbundmittel
für stahlverbund-tragwerke mit hoher dauerfestigkeit. Beton- und Stahlbetonbau, 82(12),
325-331.
Vianna, J. D. C., Costa-Neves, L. F., Vellasco, P. C. G. D. S., & Andrade, S. A. L. D. (2008).
Structural behaviour of T-perfobond shear connectors in composite girders: an
experimental approach. Engineering Structures, 30, 2381-2391.
Zheng, S., Liu, Y., Yoda, T., & Lin, W. (2016). Parametric study on shear capacity of
circular-hole and long-hole perfobond shear connector. Journal of Constructional Steel
Research, 117, 64-80
Kim, S. H., Park, S., Kim, K. S., & Jung, C. Y. (2017). Generalized formulation for shear
resistance on y-type perfobond rib shear connectors. Journal of Constructional Steel Research,
128, 245-260.
Medberry, S. B., & Shahrooz, B. M. (2002). Perfobond shear connector for composite
construction. Engineering Journal, 39, 2-12.
Valente, I., & Cruz, P. J. S. (2004). Experimental analysis of perfobond shear connection
between steel and lightweight concrete. Journal of Constructional Steel Research, 60, 465-479.
Ahn, J. H., Lee, C. G., Won, J. H., & Kim, S. H. (2010). Shear resistance of the perfobond-rib
shear connector depending on concrete strength and rib arrangement. Journal of Constructional
Steel Research, 66, 1295-1307.
Oguejiofor, E. C., & Hosain, M. U. (2011). Behavior of perfobond rib shear connectors in
composite beams: full-sized test. Canadian Journal of Civil Engineering, 19(2), 224-235.
Ahn, J. H., Jeong, Y. J., & Kim, S. H. (2008). Shear behaviour of perfobond rib shear connector
under static and cyclic loadings. Magazine of Concrete Research, 60(5), 347-357.
FIB, Model Code 2010 — Final Draft, vol. 1, 2010 (Lausanne).
Birtel, V., & Mark, P. (2006). Parameterised finite element modelling of RC beam shear failure.
2006 ABAQUS Users’ Conference, 95-106.
Nguyen, H. T., & Kim, S. E. (2009). Finite element modeling of push-out tests for large stud
shear connectors. Journal of Constructional Steel Research, 65, 1909-1920.
Hibbit, K. (2014). ABAQUS documentation, Version 6.14.
Cândido-Martins, J. P. S., Costa-Neves, L. F., & Vellasco, P. C. G. D. S. (2010). Experimental
evaluation of the structural response of perfobond shear connectors. Engineering Structures, 32,
1976-1985.
Hai Chen (2016). Experimental and Analytical study on A Novel Perfobond Strip Connector.
Master’s dissertation, Huaqiao University.
180 12
Proceedings of the Eighth International Conference on Composite Construction in Steel and Concrete
EXPERIMENTAL STUDIES OF COMPOSITE SLIM FLOOR BEAMS
Therese Sheehan
University of Bradford
Bradford, United Kingdom
t.sheehan@bradford.ac.uk
Dennis Lam
d.lam1@bradford.ac.uk
Xianghe Dai
x.dai@bradford.ac.uk
ABSTRACT
Composite slim floor beam is characterized by incorporating the steel section into the depth of
the floor which leads to the reduction of the overall depth of the composite beam. Although
Eurocode 4 does not exclude slim floor beams from its scope, no specific requirement or guidance
is given for this form of construction. A series of shear beam tests incorporated dowel
reinforcement as shear connectors had been tested to evaluate the shear connector capacity and
composite action of these beams. The objective of these tests is focused on the degree of shear
connection on the plastic resistance of the composite beam, as well as the effects of clamping
due to eccentric loading. This paper presents the test arrangement and the initial results obtained
from the shear beam tests. Findings from this research will provide fundamental understanding
to the behaviour of this form of composite construction.
INTRODUCTION
Composite beams are widely used in multi-storey buildings and for large, unsupported spans.
Considerable research has been carried out in recent years on conventional types of composite
beams, which consist of a concrete slab on top of a steel section and detailed design guidelines
are provided in Eurocode 4. Slim floor beams are a variant of this type of structure in which the
steel section is embedded in the concrete slab. This set-up offers the advantage of reducing the
depth, but the design rules for conventional beams cannot be applied to slim floor beams, since
the position of the plastic neutral axis in slim floor beams tends to be lower in the section. Thus,
the concrete may fail under large strains before the steel beam has reached its plastic resistance.
Furthermore, Eurocode 4 outlines a minimum degree of shear connection of 40% which is not
economic for slim floor solutions [1-3]. Slim floor systems are also susceptible to punching shear
near inner supports [4], excessive deflection and vibration owing to the shallow construction depth
[5-7].
To date, no one has systematically examined the different mechanisms that contribute to the
shear connection between steel and concrete in this type of configuration, such as shear studs,
friction effects, clamping effects and the effect of holes in the beam web and transverse
reinforcement bars. This paper presents the initial findings of an experimental campaign
comprising 8 shear beam tests conducted in the University of Bradford. This paper reports the
results of the first 5 shear beam tests, which explored different arrangements of reinforcement
and shear connectors. Specimen details are provided in the following sections
TEST SPECIMENS
All test specimens included in this paper employed a HEB200 steel section positioned upon a 400
mm × 15 mm bottom steel plate, encased in a 240 mm deep concrete slab, which the topping
depth was 120 mm. A252 reinforcing mesh was placed 40 mm below the top of the slab and T16
reinforcement bars were placed transverse to the beam length to provide shear connection,
passing through 40 mm diameter holes in the steel beam web at 400 mm spacings. All specimens
were 4300 mm long (4000 mm span between supports). C25/30 concrete was used for each of
the test specimens. Some specimens were loaded concentrically along the middle of the cross-
section and some were loaded at an eccentricity of 300 mm from the centre to investigate the
clamping and friction effect. Cross-sectional and side views of typical specimens under eccentric
loading are shown in Figures 1 and 2. A summary of test specimens is provided in Table 1. All
specimens were designed to have 40% composite action except for SBT3.
(a)
(b)
Fig. 1 – Cross-section view of test specimens: (a) SBT1a (with stirrups); (b) SBT1b
top flange
bottom
transverse bar
flange
bottom plate
Fig. 2 – Side elevation view of steel beam and bottom plate
Table 1 – Test specimen details

Specimen Loading Other comments
SBT1a eccentric reinforcing stirrups
SBT1b eccentric
SBT2 concentric
SBT3 eccentric no transverse bars
SBT5 concentric cellular beam with 80mm holes
SBT1a and SBT1b were identical except for the fact that SBT1a contained extra reinforcement in
the form of stirrups, running from the slab topping into the lower part of the slab. SBT1a, SBT1b
and SBT3 were loaded at an eccentricity of 300 mm from the centre of the cross section, as shown
in Figure 1. SBT3 contained no shear connectors, i.e., no transverse reinforcement bars or holes
in the beam web, and was used to investigate the clamping action as the edges of the concrete
slab were bent downwards under the eccentric loading. SBT2 utilised the same cross-section
and reinforcement as SBT1b but was loaded at the centre of the cross-section. SBT5 was loaded
concentrically and contained larger holes in the beam web (80 mm diameter) than the other
specimens.
TESTING PROCEDURE
The specimens were positioned in the test rig shown in Figure 3. Load was applied at two
locations along the specimen length, 500 mm from the mid-span (applied through either two or
four points in total, depending on whether the loading was concentric or eccentric). Two LVDTs
were used to measure the relative slip between the concrete slab and bottom plate, one placed
at each end of the beam. 9 LVDTS were placed underneath the beam to measure the vertical
deflection – 3 of these were positioned in a line running longitudinally under the centre of the
cross-section whilst the remaining 6 LVDTs were used to measure the vertical deflections of the
edges of the beam. Strain gauges were placed on the top flange of the steel section and the
bottom of the steel plate to measure longitudinal strains at the mid-span. Strain gauges were also
placed on top of the concrete slab at the mid-span for specimens. Strains were monitored in the
reinforcement bars in order to check whether these were activated during the experiment. For
the first test (SBT1a) 20 strain gauges were placed on the bars: 4 at the bars closest to each end
and 3 on the second and third bars from each end. A comparison of strains at different locations
along a single bar, at distances of 50 mm, 300 mm and 600 mm from the centre, is shown in
Figure 4(a) and strains on the first, second and third bar from the end are shown in Figure 4(b).
Most of the measured strains in the first test were below the yield strength of the material. Lower
strains occurred in the bars close to the end than the second/third bars. Larger strains were
measured towards the centres of the bars than at the other locations, which was to be expected,
as the bars were bent as the beam deformed. Upon examining the data after the first test, it was
decided to reduce the number of strain gauges to be used in subsequent tests, and focus more
on areas of interest on the bars where larger strains were likely to occur and so for the remaining
tests, two strain gauges were placed on each bar: one at the centre and the second 50 mm from
the centre.
hydraulic actuator
spreader
beams
specimen
(a)
LVDT
LVDTs
(b) (c)
Fig 3 – Photos of test setup for SBT1a: (a) full specimen view; (b) LVDT at end to measure slip;
(c) LVDTs underneath to measure vertical deflection
(a) (b)
Fig 4 - Comparison on strains in shear connector bars for SBTa: (a) at different locations along
a single bar; (b) on adjacent bars
RESULTS
The tests continued until the specimens could resist no further loading. All tests exhibited
similar load-displacement behaviour, with an initially stiff response. When the first concrete
cracks and initial slippage occurred, the stiffess was suddenly reduced to approximately half of
the initial value, and stayed fairly constant for most of the specimens until the final stage of the
test, when the steel began to yield and the concrete cracks what initiated around the steel
section web and outer edge of the slab, grew deeper and longer and propogated through the
underside of the slab. The concrete cracks during the early and late stages of the test, the end-
slip and deformed shape for typcial specimens, in addition to the load-deflection response for all
specimens, are presented in Figures 5 and 6. . Vertical shear cracks as shown in Figure 5 (a)
became visible early in the test, after initial yielding of the specimen. These cracks grew in size
as the test progressed and extended along the underside of the slab as shown in Figure 5 (b).
In the later stages of the test, the end-slip (Figure 5(c)) also increased significantly. For
specimen SBT1a, one of the loading plates punched through the slab at the end of the test.
However, it was evident from the load deflection relationship that the specimen had already
reached/was close to a maximum load and this was further evidenced by the presence of
visible, deep cracks in the concrete slab and yielding of the steel. A larger loading plate was
employed in subsequent tests to mitigate punching shear.
(a)
(b) (c)
Fig 5 – Specimens during test: (a) vertical shear cracks in side of slab and beam for SBT1a; (b)
deep cracks on underside of slab for SBT3; (c) visible end-slip between steel section and
concrete for SBT3
(a)
(b)
Fig 6 – Deflection of beam during test: (a) Relationship between load and mid-span deflection;
(b) Photo of specimen SBT5 undergoing vertical deflection during test
The maximum loads resisted by SBT1a, SBT1b and SBT2 were 738 kN (corresponding to a
shear force of 39 kN and a bending moment of 554 kNm), 732 kN and 741 kN, showing that the
use of reinforcing stirrups or concentric loading did not significantly affect the load capacity. The
shapes of the load-deflection curves for these specimens presented in Figure 6 were also very
similar, apart from the load at which the initial slippage occurred, but this was similar to that of
SBT3 and was therefore attributed to the concrete properties, since SBT2 and SBT3 had been
cast with the same concrete mix. Specimen SBT3 reached a maximum load of 559 kN.
Comparing this load with SBT1b, it is evident that the presence of the shear connectors
contributed over 170 kN to the load resistance. Specimen SBT5 reached a maximum load of
804 kN. The larger concrete dowel (80 mm diameter) contributed an extra 60 kN to the total
load.
All specimens exhibited reasonably large end-slips prior to the end of the test. SBT1 had
maximum end-slips of 11.7 mm and 13.4 mm recorded during the test. These values were
considerably larger than the minimum slip of 6 mm specified by Clause 6.6.1.1 in Eurocode 4 for
ductile shear connector behaviour. The failure of SBT1b was not symmetric along the length,
and the maximum end-slips were 7.2 mm and 14.1 mm at each end. However, this was not
caused by the presence of the stirrups (which were symmetrical). The end-slips were 9.6 and
11.3 mm respectively for SBT2, 18.6 mm and 1.6 mm for SBT3 and 10.4 mm and 11.7 mm for
SBT5.
The maximum strain measured in the bottom steel plate for SBT1a at the mid-span was 2637μ
and the top steel flange was -9148 μ (positive strains were tensile and negative strains were
compressive), where the estimated yield strains for these were 2167 μ and 2038 μ
respectively. Only one of the reinforcement bars exceeded the yield strain. Similar strains were
measured in the top flange and bottom plate for SBT1b and SBT2, indicating that the steel had
exceeded the yield strain by the end of the test. In SBT2, most of the reinforcement bars
exceeded the yield strain before the end of the test, In SBT3, the maximum strain in the bottom
plate at the mid-span was 1297 μ, less than that associated with yielding, and the strain in the
top plate was only -3003 μ, but larger strains were observed further from the mid-span, with
values of -7511 μ and 3484 μ at 500 mm from the mid-span in the top and bottom steel
respectively. As mentioned in the previous paragraph, the end-slips for SBT3 varied
significantly between each end, suggesting that failure did not propogate from the mid-span. In
SBT5, the top flange only reached a maximum strain of -391 μ, while the bottom plate reached
2195 μ, demonstrating that the use of larger concrete dowels (SBT5), affected the strain
distribution through the cross-section.
All specimens were very ductile, exceeding a maximum mid-span deflection of 80 mm (span/50).
In each of the five tests, the maximum compressive strain measured on the concrete top surface
at the mid-span was less than that associated with failure of the concrete. A summary of the
maximum load Nmax, maximum mid-span deflection δmax, end-slip δu, top steel strain εs,top, bottom
steel strain εs,bot, concrete strain εc and initial stiffness k (applied load/mid-span deflection) for each
specimen is presented in Table 2. Some variation was observed between specimens with regard
to the initial stiffness with specimen SBT5 having a slightly lower stiffness than the other
specimens owing to the enlarged holes in the steel beam web, whereas SBT2, which was
subjected to concentric loading exhibited a slightly higher stiffness than the other specimens.
Table 2 Summary of results
Specimen Nmax δmax δu εs,top εs,bot εc k
(kN) (mm) (mm) (μ) (μ) (μ) (kN/mm)
SBT1a 738 81.5 13.4 -9148 2637 - 32.4
SBT1b 732 87.8 14.1 -8780 2214 -1205 33.5
SBT2 741 87.5 11.3 -9790 2142 -1290 39.2
SBT3 559 80.5 18.6 -3003 1297 -1708 35.1
SBT5 804 89.9 11.7 -391 2195 -1276 28.1
ANALYSIS
For the bare steel section, as shown in Figure 7, the distance of the plastic neutral axis, y, from
the top of the steel section, can be calculated as follows
.
tT = 15 mm
y
tw = 9 mm h = 200 mm Nac
Nat1
Nat2
tP = 15 mm
bT = 200 mm
bP = 400 mm
Fig 7 – Moment capacity of bare steel section
Assuming neutral axis is in bottom flange of HEB

bT×tT + tw×(h-2tT) + bT(y - (h-tT) = bT×(h – y)+ bP×tP (1)
200×15 + 9×(200-30) + 200×(y-(200-15)) = 200×(200-y) + 400×15
y = 196.18 mm
Moment capacity = Asfy,HEB×(h/2 – y) + bp×tp× fy,p (tp/2 + 200-y) (2)

= 7810×428×(200/2 – 196.18) + 400×15×455 (15/2 + 200-196.18)
= 352 kNm
bT is the width of the top flange of the HEB section, h is the depth of the HEB section, As is the
area of the HEB section, bp and tp are the width and thickness of the bottom plate respectively, tw
is the thickness of the web, and fy,HEB and fy,p refer to the yield strengths of the HEB section and
bottom plate respectively.
352 kNm equates to a shear force of 352/1.5 = 234.7 kN and an applied load of 234.7×2 = 469
kN. The maximum load for specimen SBT3 was 559 kN, which is 19% higher, suggesting that
despite the lack of shear connectors, clamping and friction contribute towards the cross-section
resistance. However, overall, the applied load in SBT3 was mostly resisted by the steel section.
Using measured yield strengths of 428 N/mm2 for the HEB section, 455 N/mm2 for the bottom
steel plate and a concrete cube strength of 41.3 N/mm2,and assuming plastic behaviour of the
cross-section, the neutral position and bending moment capacity, Mpl,Rd can be calculated using
the stress-blocks shown in Figure 8, where yc is the depth of the plastic neutral axis below the top
of the slab
yc 53 mm to Nc
centre of hole
Nac
Nat1
40 mm hole
Plastic neutral Nat2
axis
Fig 8 – Stress-blocks for calculating moment capacity
There is a 40 mm hole in the HEB section web, and hence the cross-sectional area As is reduced
from 7810 mm2 to 7810 – 40×9 = 7450 mm2 (As,red). If the section has 100% composite action,
assuming that the neutral axis is located in the HEB upper flange, the depth yc can be calculated
using force equilibrium
Nc + Nac = Nat1 + Nat2 (3)
where Nc is the compressive force in the concrete, Nac is the compressive force in the HEB section
above the neutral axis, Nat1 is the tensile force in the HEB section and Nat2 is the tensile force in
the bottom plate. These forces can be calculated using Equation 2, where beff is the effective
width of the slab, fck is the concrete strength, yc is the depth of the neutral axis.
0.85×beff×fck×yc + (yc–40)×bT×fy,HEB = (As,red-((yc–40)×bT)×fy,HEB + bp×tp×fy,p (4)
Substituting in the values gives:

0.85×2000×41.3×yc + (yc–40)×200×428 = (7450-((yc–40)×200)×428 + 400×15×455
yc = 52.88 mm
Taking moments about line of action of concrete force, Mpl,Rd, is calculated to be:
Mpl,Rd = Nat2× (240 + tp/2 – yc) + (Nat1 – Nac) ×(140-yc) = 726.4kNm (5)
This corresponds to an applied failure load of 726.4/1.5 ×2 = 968.57 kN
For 40% composite action, the maximum bending moment capacity is 564 kNm, corresponding
to an applied load of 752 kN, which is close to the test results.
The area of each concrete dowel in SBT5 (80 mm hole, with 16 mm bar) is equal to π×802/4 -
π×162/4 = 4825 mm2 per hole. The area of each concrete dowel in SBT2 (40 mm hole with 16
mm bar) is equal to π×402/4 - π×162/4 = 1056 mm2 per hole. If the shear strength of the concrete
is taken as the limit of 0.8√fcu = 0.8√(41.3) = 5.14 MPa (Clause 3.4.5.2 in BS 8110) and the dowel
is in double shear, the concrete in each activated connector in SBT5 can resist 4825×2×5.14 =
49.6 kN in shear and each connector in SBT2 can resist 10.9 kN in shear. The strain gauge data
for SBT5 indicated that only 3 of the reinforcement bars exceeded the yield strain whereas for
SBT2, 7 bars reached yield. Considering the active bars in each case, the expected concrete
contribution in SBT2 can be estimated to be 10.9×7 = 76.3 kN and the concrete contribution for
SBT5 is 49.6×3 = 148.8 kN. The expected difference in load is 148.8 – 76.3 = 72.5 kN, which
exceeds the measured difference (63 kN) by 15%.
The neutral axis position was estimated for different specimens using the strains measured in the
top flange and bottom plate. The strains (μ) and neutral axis depth are illustrated for Specimen
SBT1a at four different stages of the test in Fig 9, and the four stages are marked on the load-
displacement curve.
Table 3 summarises the estimated neutral axis positions for the other specimens at similar stages
of their respective tests, based on the strain gauge data. The first point for each specimen is
taken at a low load, when the specimen is in the elastic range. The second point is taken at the
stage when the stiffness first decreases. The third point is taken at the beginning of the final
plateau and the fourth point is taken further on in the plateau (up to half-way between the previous
point and the maximum load. It was not possible to take reliable strain readings during the later
stages of the test in many cases as the strain gauges had become damaged or strained beyond
their working limit. In all of the specimens except for SBT5, a consistent trend can be found in the
strain gauge data, with the neutral axis position appearing to move downwards as the test
progressed and the behaviour became increasingly plastic. The neutral axis position during Stage
4 of the test is close to that calculated previously for the bare steel section, which can be expected
as a significant degree of slippage had occurred between the steel and concrete components at
this stage.
108.6 μ 361.0 μ 3760.3 4874.1 μ

89.8
81.4 149.9 175.8
mm
mm mm mm
178.4 μ 503.1 μ 1634 μ 1973.2 μ
Stage 1 Stage 2 Stage 3 Stage 4
4
3
2
1
Fig 9 - Neutral axis position at different stages of test for SBT1a
Table 3 – Estimated positions of neutral axis (distance below top steel flange) from strain gauge
data
Specimen Point no. Average εs,top Average εs,bot Neutral axis depth
(μ) (μ) (mm)
SBT1b 1 -80 168 69.3

2 -209 462 67.0
3 -3311 1449 149.6
4 -8586 1782 178.0
SBT2 1 -165 429 59.8
2 -375 839 66.4
3 -1782 1355 122.1
4 -6028 1704 167.6
SBT3 1 -228 420 75.7
2 -456 763 80.5
3 -1309 948 124.7
4 -2305 1189 141.8
SBT5 1 -49.6 377 25.0
2 -83.6 771 21.0
3 -153.2 1510 19.8
4 -201.3 1793 21.7
CONCLUSIONS
The initial test results demonstrate the contributions of different components within a slim floor
system. Some initial conclusions can be drawn from these tests:
1. Reinforcing stirrups do not make a significant contribution to the beam performance.
2. The beams exhibited a very ductile response, undergoing a mid-span deflection exceeding
span/50 prior to failure. This was despite being designed for a low degree of shear
connection (40%).
3. The application of eccentric load did not significantly alter the beam behaviour in
comparison with applying the load concentrically.
4. The steel shear connectors have a significant effect on the beam performance. Specimen
SBT3 without shear connectors achieved a maximum load over 20% lower than equivalent
specimens with 40% composite action, and was closer to the response of a steel section
acting alone.
5. The size of the concrete dowel impacts the performance. By increasing the size of the
holes in the steel web by a factor of 2, the load capacity was increased by 8%.
Further tests will be conducted in future, including a specimen without concrete cover to the top
reinforcement, a steel beam with horizontal shear studs on the beam web.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the funding received from the European Community
Research Fund for Coal and Steel under grant agreement number RFSR-CT-2015-00020.
REFERENCES
[1] Årskog, V., Fosssal, S. and Gjørv, O. Methodology and data for calculation of LCE (Life Cycle
Ecology) in repair planning, Lifecon deliverable D 5.3, VTT Building and Transport, Finland, 2003.
[2] Aribert, J. M. Analysis and practical solutions for the influence of steel grade on the minimum
degree of shear connection of a composite beam, Construction Metallique, 40-55, 1997.
[3] Bernuzzi, C., Zandonini, R. Joint action in non-sway frames with steel-concrete composite slim
floor systems, Journal of Singapore Structural Steel Society, 1995, Vol. 6, No. 1, 75-85.
[4] Kuhlmann, U., Hauf, G. Querkrafttragfähigkeit von Slim-Floor Trägem, AiF research project
no. 15639, Institute of Structural Design, University of Stuttgart, Stuttgart, 2011.
[5] Bernuzzi, C., Gadotti, F., Zandonini, R. Semi-continuity in slim floor steel-concrete composite
systems, Proceedings of Eurosteel 95 ed. Kounadis, A. N.; Athens, Greece, 1995, 287-294.
[6] Feldmann, M., Heinmeyer, C., Butz, C., Caetano, E., Cunha, A., Galanti, F., Goldack, A.,
Hechler, O, Hicks, S., Keil, A., Lukic, M, Obiala, R., Schlaich, M., Sedlacek, G., Smith, A, Waarts,
P. Design of floor structures for human induced vibrations, JRC-ECCS, Eds. G. Sedlacek, Ch.
Heinemeyer, Chr, Butz, 2009.
[7] Ranzi, G., Leoni, G., Zandonini, R.. State of the art on the time-dependent behaviour of
composite steel-concrete structures, Journal of Constructional Steel Research, 80, 252-263,
2013.
SLIM-FLOOR CONSTRUCTION
DEFORMATION AND LOAD CARRYING BEHAVIOUR
Johannes Schorr, M.Sc.

University of Stuttgart
Institute of Structural Design
Stuttgart, Germany
johannes.schorr@ke.uni-stuttgart.de

Stuttgart, Germany
ulrike.kuhlmann@ke.uni-stuttgart.de
Prof. Dr.-Ing. Gunter Hauf

DHBW Mosbach
Mosbach, Germany
gunter.hauf@mosbach.dhbw.de
Abstract
For realistic predictions of beam deformations, a good knowledge of the beam´s load carrying
behavior is necessary. For slim-floor constructions in difference to normal height composite
beams, the bending behavior of the concrete slab brings a substantial contribution to the load
carrying behavior. However, the concrete slab of slim-floor constructions cracks under bending
moments already at the serviceability stage. The analytical models presented here allow to make
use of the concrete slabs contribution despite of cracking.
The aim of an ongoing European project (SlimAPP), involving a huge amount of various testing,
is to develop a holistic approach by considering all aspects of efficient design of slim-floor con-
structions for implementation in Eurocode 4. Particular attention lies on the composite shear con-
nection between the steel beam and the concrete floor slab. Push-out tests, beam tests, and long-
term tests are carried out, testing among others the innovative CoSFB connection system.
INTRODUCTION
Composite beams may be differentiated due to their height into three main types: composite
beams of normal height, composite beams of reduced height and slim-floor beams (see Figure
1). In slim-floor constructions a steel beam is embedded into the concrete slab.
Due to high degree of prefabrication, buildings with slim-floor constructions can be erected very
fast and economically. In combination with profiled sheeting, as shown in Figure 2, the production
of the slim-floor can furthermore be achieved by hand and so independently from cranes.
ten ist nur schwer möglich. So besteht die Möglichkeit, die Größe des Baustahlprofils als eine
mögliche Abgrenzung heranzuziehen. Eine zweite und bevorzugte Möglichkeit ist durch das ver-
schiedene Tragverhalten gegeben. Verbundträger werden als !niedrige Verbundträger" bezeichnet,
wenn der Betongurt unter Gebrauchslasten (bei positiven Momenten) Risse aufweist.
Die Abbildung 2.1 zeigt typische Querschnitte eines normal hohen Verbundträgers, eines niedrigen
Verbundträgers und eines Slim-Floor Trägers.
Fig. 1: 2.1:beam
Composite
Abbildung of normal
Normal height, composite
hoher Verbundträger, niedriger of reduced height,
beam Verbundträger, slim-floor
Slim-Floor beam
Träger
(Hauf G. , 2010)
The integration of the steel profile into the floor and the accordingly missing joists enable to lay
buildingSlim-Floor Träger
technology easilywerden übersetzt
at the bottom auch
side alsfloor
of the !de without
ckengleiche" Verbundträger
horizontal breaches. bezeichnet. Bei die-
ser Art von Verbundträger liegt das Baustahlprofil in den Beton eingebettet. Es gibt sehr viele
mögliche Querschnitte, die hierbei zur Anwendung kommen können. Bei der Herstellung wird
zwischen zwei Bauweisen unterschieden [Feldmann 2000]. Bei der sog. Trockenbauweise werden
vorgefertigte Elemente verwendet, die auf den Stahlträgeruntergurt aufgelegt und entsprechend
ausgegossen werden [Bode et al 1997], [Feldmann 1998 ]. Es kann hierbei aber nicht automatisch
4 Trag- und Verformungsverhalten von Slim-Floor Trägern unter Biegebeanspruchung
Fig. 2: Slim-floor system in combination with profiled sheeting

Whereas a neglection of the bending capacity of the concrete slab Mc of normal height composite
beams leads to sufficiently good results in the calculation of realistic deformations, the influence
on the deformation behavior of slim-floor constructions is much higher and not negligible. The
approximation of the neutral axes of the concrete slab and the overall slim-floor beam results into
cracking at the tensile side of the girder under bending. This may occur already in the servicea-
bility state.
Figure 3 shows the strain distribution and the cracking at the bottom of a slim-floor beam under
positive bending.
Fig. 3: Strain distribution of a slim-floor beam (Hauf G. , 2010)

For a realistic calculation of the deformation of slim-floor construction the bending capacity of the
concrete slab and consequently the cracking has to be taken into account. Eurocode 4 (EN 1994-
1-1, 2009) is not providing rules or recommendations for consideration of the concrete slabs bend-
ing capacity and cracking. This leads to an overestimation of the deformation behavior of slim-
floor construction. As the deflection criterion is decisive in many cases for slim-floors, an overes-
timation leads to uneconomically designed constructions.
Within several research projects regarding slim-floor solutions of the Institute of Structural Design
at the University of Stuttgart, Germany, the deformation behavior was investigated. To take all
parameters into account, an analytical model was developed verifying an effective width referring
to deformation. Based on this approach a quasi-elastic calculation method was developed for
single span and two span slim-floor girders.
ANALYTICAL MODELS FOR DEFORMATION CALCULATION OF SLIM-FLOOR CONSTRUC-
TIONS
Introduction
Deformation calculations need to represent reality as close as possible. As a consequence the
bending of the concrete slab in slim-floor constructions has to be taken into account in contrary
to the present practice. Otherwise, the deformations may be overestimated leading to uneconom-
ically designed slim-floor constructions. Among others, the bending capacity Mc of slim-floor con-
structions depends on cracking of the concrete. Starting from a nonlinear analytical model, an
approach with a deformation based effective width and further on a quasi-elastic model have been
developed. The models are outlined in the following. For a more detailed description please refer
to (Hauf G. , 2010) and results on underlying testing are also given in (Kuhlmann & Hauf, 2008).
Effective width
Two different states for the concrete slab may generally be differentiated (Figure 4). In membrane
state, it is assumed that the concrete obtains only compression forces Nc. The slab works as a
shear panel (Figure 4a). bm,S is the corresponding effective width.
In the bending state, it is assumed that the concrete slab has no shear stiffness, but a bending
stiffness Ic. This results in a bending moment Mc in the concrete (Figure 4b)). Calculating the
deformation of normal height composite girders this state can easily be neglected without major
effects. Calculating realistic deformations on slim-floor constructions the bending capacity has to
be considered by an appropriate effective width bm,B (Hauf & Kuhlmann, 2015).
a) b)
Fig. 4: Membrane state a) and bending state b) of the concrete slab (Hauf & Kuhlmann, 2015)
In combination, the above membrane and the bending state lead to an overall stiffness of the
slim-floor girder as indicated in equation (1).
𝑏m,S ⋅ ℎc
𝑏m,B ⋅ ℎc3 𝑛0 ⋅ 𝐴a 2
(1) 𝐼i,0 = 𝐼a + 𝐼c,0 + 𝑆i,0 ⋅ 𝑎st = 𝐼𝑎 + + ⋅ 𝑎st
12 ⋅ 𝑛0 𝑏m,S ⋅ ℎc
𝑛0 + 𝐴a
Where:
Aa Cross sectional area of the steel profile
Ia Inertia of the steel profile
hc Height of the concrete
ast Distance between centroid concrete and steel section
n0 = Ea/Ecm Number of reduction (Ea = young modulus of steel, Ecm = young modulus of
concrete)
This overall stiffness Ii,0 forms the basis for a nonlinear analytical model that is presented in the
following section.
Nonlinear Analytical Model

Based on the assumption made on the effective width and in consideration of cracking an analyt-
ical model was developed to calculate the deflection of slim-floor girders (Hauf G. , 2010)
For determination of the slim-floor beam deflection cracking has to be considered. This is
achieved by dividing the slim-floor girder into a number of elements (Figure 5). Each element is
assigned by an effective width for the membrane state bm,S and by an effective width for the bend-
ing state bm,B.
Fig. 5: Discretization and distribution of mem- Fig. 6: M-к-curve for positive and negative
brane and bending state for a one span girder bending
A simplified M-к curve (Figure 6) under positive bending is defined by four characteristic points:
1. Development of first cracks Mcrack, кcrack
Until first cracks develop on the bottom of the slim-floor beam an elastic calculation is
possible. This includes also a consideration of the tensional concrete strength fct (Figure
7). After Mcrack is reached, a linear-elastic calculation is no longer possible for this element.
Fig. 7: Stress-strain relationship – cracking moment (index: “crack”)

2. First yield of the steel profile Mel, кel (bottom)
This point of the M-к curve (Figure 6) is reached when the steel profile begins to yield at
the bottom (Figure 8). Therefore a nonlinear material model has to be implemented.
Fig. 8: Stress-strain relationship – yield moment (Index: “el”)
3. Steel profile fully plasticized Mpl, кpl
Concrete reaches its maximum stresses fc at a concrete strain of 2 ‰. Furthermore, the
steel profile is fully plasticized and is not able to carry any more loading (Figure 9).
Fig. 9: Stress-strain relationship – plastic moment (Index: “pl”)

4. Concrete reached maximum strains Mu, кu
If the load is inducing a concrete strain over 3.5 ‰, the ultimate moment is reached and
the slim-floor is going to fail (Figure 10).
Fig. 10: Stress-strain relationship – ultimate moment (index: “u”)

For statically determinate systems direct computation of the internal forces (moments) is possible.
The bending can be determined for each element by using the M-к relationship (Figure 6) and
consequently the deflection of the slim-floor beam.
Due to their cracking, statically indeterminate systems, like two span beams (Figure 11) result in
midspan and on the support in a more complex calculation. A corresponding model is presented
in (Hauf G. , 2010).
Fig. 11: Crack development on statically indeterminate two span slim-floor beam
The above developed analytical model showed a good conformity with test series performed in
(Kuhlmann, Hauf, & Rieg, 2006), (Kuhlmann & Fries, 2001) and (Kuhlmann, Konrad, & Hauf,
2010). Parameter studies were performed to determine the influences on the deflection of:
Material parameter: concrete tensional strength fct, concrete compressive strength fc,
young modulus of concrete Ec, young modulus of steel Ea, yield
strength of the steel fy
Geometrical parameter: height of the concrete slab hc, cross sectional area of the steel pro-
file Aa, reinforcement content as,u & as,o, geometrical ratio between
concrete slab width and length of the beam b/L
In general, those parameters are reducing the deflection of the beam, which are increasing the
stiffness of the beam. For example: height of the concrete slab or the used steel profile.
Deformation based effective width

The deformation based effective width comprises for single span slim-floor girders the effective
width of the membrane state bm,S and an elastic value of the deformation based effective width
bm,V,0. The cracking is taken into account by a reduction of the difference between bm,V,0 and bm,S
with factor βm,V, see equation (2).
(2) 𝑏m,V = 𝑏m,S + (𝑏m,V,0 − 𝑏m,S ) ⋅ 𝛽m,V
The reduction factor βm,V was determined by comparing deformations derived from calculations
with the nonlinear analytical model with deformation derived from calculations with an introduced
constant effective width over the length of the beam. The value for the constant effective width
was adapted so that equilibrium of the deformations for both approaches was reached (Hauf G. ,
2010). For single span slim-floor girders the calculation of the reduction factor βm,V is also given
in (Hauf & Kuhlmann, 2015). In (Hauf G. , 2010) an additional formula is provide for βm,V in two
span slim-floor girders.
Quasi-elastic calculation method

A more simple approach has been developed for slim-floor girders with I- or hat-profiles as a
quasi-elastic calculation method. Accordingly, an effective moment of inertia Ii,eff was introduced
in equation (3).
(3) 𝐼i,eff = 𝐼a + 𝛼𝑐 ⋅ (𝐼c,0 + 𝑆i,0 ⋅ 𝑎st )
The adjustment factor αc takes cracking of the concrete into account by introducing a loss of stiff-
ness that is originally provided by the concrete slab. At load beginning, when concrete is
uncracked, a linear elastic deflection behavior is considered: αc = 1.0 and the effective stiffness is
equal to the stiffness of composite beams with normal height. When αc = 0 no stiffness is provided
by the fully cracked concrete slab. The effective stiffness of the slim-floor girder is equal to the
stiffness of the steel profile. 0 < αc < 1 means the concrete is cracked, but still contributing to the
stiffness of the beam.
Adjustment factor αc itself takes material and geometrical influences into account (see equation
(4)). It depends on a factor αV (equation (5)), material factor αMat (equation (6)), factor for the influ-
ences of the cross sectional area αQS (equation (7)) and loading factor αM (equation (8)).
(4) 𝛼c = 𝛼V ⋅ 𝛼Mat ⋅ 𝛼QS ⋅ 𝛼M
(5) 𝛼V = 0.4 for I-Profiles
−0.26 0,46 0,29
𝑓ct 𝑓c 𝑓y
(6) 𝛼Mat= =( ) ⋅( ) ⋅( )
𝑓ct,0 𝑓c,0 𝑓y,0
0,17 3,18 −1,27
𝐴s,u +𝐴a 𝐼 𝑎 𝑏 −0,56
(7) 𝛼QS= = ( ) ⋅ ( 𝐼c,0 ) ⋅ ( ℎst ) ⋅ (𝑡 ) for I-Profiles
𝐴ci,0 i,0 c
𝑀crack 1,22
(8) 𝛼M = ( )
𝑀Ed
The exponential factors in equations (6) to (8) are determined by a parametric study varying ge-
ometrical parameters as well as material parameters. Therefore the here presented exponents
are only valid in range of the varied parameters as indicated in Figure 12 (Hauf G. , 2010).
Span L: 4.0 – 9.0 m
Concrete slab thickness hc: 16 – 34 cm
Concrete slab width b: 2.0 – 8.0 m
Steel-Profile: UPE-, IFB-, SFB-profiles
Concrete compressive strength fc: 20 – 50 N/mm²
Concrete tensional strength fct: 1.0 – 5.5 N/mm²
Steel yield strength fy: 235 – 460 N/mm²
Reinforcement top as,o: 0 – 18 cm²/m
Reinforcement bottom as,u: 0 – 18 cm²/m
Fig. 12: Range of varied parameters for determination of exponential factors in equations (6)
to (8) (Hauf G. , 2010)
The effective stiffness Ii,eff allows to assess the deformation of slim-floor constructions with small
effort taking all geometrical and material parameters into account.
RFCS-PROJECT SLIMAPP
Introduction
The RFCS-Project “Slim-Floor Beams – Preparation of application rules in view of improved
safety, functionality and LCA” (RFCS-SLIMAPP, 2015) is a European project coordinated by the
University of Stuttgart in partnership with the Universities of Bradford (UK) and Trento (Italy), the
Steel Construction Institute (SCI, UK) as well as industrial partners ArcelorMittal Belval & Differ-
dange and LINDAB/Astron (both from Luxembourg).
The aim is to develop a holistic approach considering the design for safety and functionality as
well as a lifecycle assessment. The rules and guidelines, which will be developed should fill the
lack of design rules for slim-floor constructions in Eurocode 4 (EN 1994-1-1, 2009). To achieve
this aim a comprehensive testing program is realized that is expected to widen the knowledge on
the load carrying behavior of slim-floor constructions. Furthermore an inventory of the already
available constructions on the main European markets is made. This forms the basis for the lifecy-
cle assessment (LCA) and the basis for a comparison with pure concrete constructions, where
slim-floor constructions are mainly competing with.
Experimental program
Overview
Push-out tests, beam tests and long term tests are part of the project. The cross sections of all
tests have a strong interrelation. Dimensions of the specimens, material parameters and shear
connection devices are kept the same. This ensures a transferability between all tests. The focus
within the testing is on the innovative shear-connection system CoSFB, presented hereafter.
Composite Slim-Floor Beam

The Composite Slim-Floor Beam (CoSFB) consist of a commercially available H-profile. Holes
are introduced into the web under the upper flange of the profile by drilling or flame cutting. For
proper shear connection, reinforcement bars are conducted through the web holes (Figure 13).
Fig. 13: CoSFB-System: H-profile with holes in the web and reinforcement bars through the
holes (before concreting)
The load capacity of the shear connection is dependent on the load capacities of the pure con-
crete dowel, the reinforcement bar and on friction. The ratio between the shear force of the rein-
forcement bar and the load capacity that the concrete can develop locally influences the overall
load bearing capacity of the shear connection. Concrete with low strength is damaged locally and
allows the reinforcement to bend and activate tensional forces in the bar as indicated in Figure 14
a). High strength concrete may lead to shearing of the reinforcement bar Figure 14 b) (Braun,
Hechler, & Obiala, 2014).
a) b)
Fig. 14: a) Bending and activation of tensional forces in the reinforcement bar, b) pure shear-
ing of the reinforcement bar
The embedment of the shear connector in the concrete slab allows reducing the height of the
construction in comparison with other construction as e.g. headed studs placed on top of the
profile, mostly used for normal composite girders (Figure 15). For a German national approval of
the CoSFB-system (DIBT, 2014) various test series were conducted at the University of Stuttgart.
Within the presented project the already investigated range of parameters is extended.
Fig. 15: Comparison of overall height with identical profiles and different shear connector sys-
tems (Braun, Hechler, & Obiala, 2014)
Basic test configuration

For assurance of good transferability between all test setups (push-out tests, beam tests and long
term beam tests) a basic test configuration has been chosen. The basic test configuration for all
tests consists of:
- HEB 200 with steel grade S355, all profiles are supplied from one batch
- Holes in the web Ø40 mm
- Reinforcement bars Ø16 mm (B500B), supplied from one batch
- Concrete grade C25/30 (fck = 30 N/mm²)
The above described basic test configuration is also indicated in a push-out test setup in Figure
16. The supply of the reinforcement bars as well as HEB 200 profiles from one batch for all tests
ensures a good comparability between all test setups and test series.
Fig. 16: Basic push-out test configuration (HEB200 (S355), Ø40 mm of the holes in the pro-
file web and Ø16 mm reinforcement)
Push-out tests
For determination of the longitudinal shear connection behavior 21 push-out tests in seven test
series were conducted (POT1 – POT6) at the University of Stuttgart. In Figure 16 the basic test
configuration is shown for the push-out test (test specimen half). Aim of the tests is to verify the
influence of the following parameters on the shear connection:
- Diameter of the reinforcement (Ø12 mm, Ø16 mm and Ø20 mm)
- Concrete grade (C25/30 and C60/70)
- Diameter of the hole in the web (Ø40 mm and Ø80 mm)
Each push-out test series differs only in one varied parameter from the other test series. Along
with push-out test series conducted in earlier projects (Braun, Obiala, & Odenbreit, 2015) and
(Braun M. , et al., 2014) the presented testing program widens the knowledge on the longitudinal
shear connection behaviour of the Composite Slim-Floor Beam system.
Table 1 gives an overview on the varied parameters of the push-out tests compared to the basic
test configuration.
Tab. 1: Overview of push-out test series
Push out Parameter Remarks
test no
POT1 Basic push test Basic test configuration
POT2 Reinforcement Ø 12 Influence of bar diameter on longitudinal shear force
POT3a Reinforcement Ø 20 Depending on the modification of the shape
POT3b Reinforcement Ø 20 / Influence of bar diameter on longitudinal shear force
Concrete grade C50/60
POT4 Large circular openings, Behavior of pure concrete dowel
no rebar
POT5 Large circular openings Behavior of “concrete dowel” (different diameter)
Eccentricity of the rebars
POT6 Concrete grade C50/60 C50/60, Influence of bar diameter on longitudinal
shear force
Results
In Table 2 the maximum load per shear connector for each test specimen is presented. For each
test series the lowest value for the shear connection was obtained in the test number one (except
for POT4). In these tests strain gauges were applied on the reinforcement bars. The correct ap-
plication of the strain gauges could only be realized by milling down the specific location, what led
to a reduction of the reinforcement diameter.
Tab. 2: Maximum load per shear connector [kN/shear connector]

Number POT1 POT2 POT3a POT3b POT4 POT5 POT6
1 201 201 344 360 93 261 255
2 237 214 356 411 92 279 267
3 245 206 355 403 86 279 261
The lowest value for the shear connection was obtained for POT4 due to missing reinforcement
bars. The highest value was observed for POT3b as here the higher concrete grade C50/60 was
combined with the biggest used reinforcement diameter within all test series (Ø 20).
During the push-out tests two main failure modes were obtained. The failure of the shear con-
nector (reinforcement bar) was the desired failure mode. It was observed when the higher con-
crete grade (C50/60) was used and when the basic concrete grade was combined with a small
reinforcement diameter (POT2).
The second obtained failure mode was a premature crumbling of the concrete at the support of
the test specimens before the connection itself did fail; see in Figure 17 for POT1-1. This failure
mode was observed when concrete grade C25/30 was combined with Ø ≥ 16 mm reinforcement
bars.
Fig. 17: Failure of the concrete support of POT1-1
Beam tests and Long-term Beam tests

Within the scope of the SlimAPP project, also beam tests are carried out at the University of
Bradford, UK for the investigation of the load carrying behavior of slim-floor constructions under
shear and under bending. At the University of Trento, Italy the long term behavior like creep and
shrinkage of slim-floor constructions is investigated.
CONCLUSION
Deformation calculations as part of serviceability checks should represent the reality as good as
possible. In difference to normal height composite beams the bending capacity of the slim-floors
concrete slab has to be considered for realistic predictions of the deformations. Otherwise, an
overestimation of the deformation may lead to uneconomically designed slim-floor girders. Slim-
floors are due to their shallow geometry subjected to nonlinear effects of cracking, already in the
serviceability state.
Within several projects at the Institute of Structural Design at the University of Stuttgart, Germany
an analytical model was developed and verified. This model allows taking bending capacity of the
concrete slab and consequently the nonlinear cracking into account by an implemented M-к curve.
In a second model an effective width was introduced, that can be assumed constantly over the
length of the girder. A third model, which was oriented for practical application, introduces an
effective stiffness for calculating the deformations. All three models are verified by various tests
conducted within several projects at the Institute of Structural Design at the University of Stuttgart.
Within the ongoing European project SlimAPP (RFCS-SLIMAPP, 2015) the development of a
holistic approach for slim-floor constructions is planned, comprising existing rules and construc-
tion systems as well as lifecycle assessments. During the project four test setups including push-
out tests, shear beam tests, beam tests and long term beam tests are planned. All test series are
correlating to achieve consistent and plausible results. The load capacities of the shear connect-
ors obtained during the push-out tests conducted at the University of Stuttgart are presented.
REFERENCES
Braun, M., Hechler, O., & Obiala, R. (2014). Untersuchungen zur Verbundwirkung von
Betondübeln. In: Stahlbau 83, issue 5, S. pp. 303-308.
Braun, M., Hechler, O., Obiala, R., Kuhlmann, U., Eggert, F., Hauf, G., & Konrad, M. (2014).
Experimentelle Untersuchungen von Slim-Floor-Trägern in Verbundbauweise. Stahlbau
83, Issue 10, pp. 741-749.
Braun, M., Obiala, R., & Odenbreit, C. (2015). Analyses of the loadbearing behaviour of deep-
embedded concrete dowels, CoSFB. IN: Steel Construction, Volume 8, Issue 3, pp. 167-
173.
DIBT. (2014). Z-26.4-59: Allgemeine bauaufsichtliche Zulassung: CoSFB-Betondübel.
EN 1994-1-1. (December 2009). Eurocode 4: Design of composite steel and concrete structures
- Part 1-1: General rules and rules for buildings; EN 1994-1-1:2004 + AC:2009. CEN
(European Commitee for Standardization).
Hauf, G. (2010). Trag- und Verformungsverhalten von Slim-Floor-Trägern unter
Biegebeanspruchungen, Dissertation. Stuttgart: University of Stuttgart, Mitteilung des
Instituts für Konstruktion und Entwurf No. 2010-2.
Hauf, G., & Kuhlmann, U. (2015). Deformation calculation methods for slim floors. In: Steel
Construction, Volume 8, Issue 2, pp. 96-101.
Kuhlmann, U., & Fries, J. (2001). Optimierung der Bemessung von deckengleichen
Verbundträgern in Hutform. Institute of Structural Design, University of Stuttgart:
Schlussberich, AiF-Forschungsvorhaben 12017 N.
Kuhlmann, U., & Hauf, G. (2008). Efficient design for the calculation of the deflection and the
shear capacity of slim-floor girder. Proceedings of Composite Constructions in Steel and
Concrete VI, July 20.-24., Colorado, USA.
Kuhlmann, U., Hauf, G., & Rieg, A. (2006). Effiziente Dimensionierung niedriger Verbundträger.
research project No. 668, funded by Stiftung Industrieforschung, No. 2006-22X, Institute
of Structural Design, Unviersity of Stuttgart.
Kuhlmann, U., Konrad, M., & Hauf, G. (2010). Slim Floor girder with COFRADAL200 deck
elements. Institute of Structural Design, University of Stuttgart: Technical report for
ARCELORMITTAL Belval & Differdange S.A., No. 2010-2X.
RFCS-SLIMAPP. (2015). Grant Agreement RFSR-CT-2015-00020 - Slim-Floor Beams -
Preparation of Application rules in view of improved safety, functionality and LCA.
Coordinator: University of Stuttgart, Partners: Steel Construction Institute, Università
degli Studi Trento, University of Bradford, ArcelorMittal, Lindab Buildings, (running
project).
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support of the research project SlimAPP “Slim-
Floor Beams – Preparation of Application rules in view of improved safety, functionality and LCA”
by the Research Fund for Coal and Steel (RFCS). They also thank the project partners for their
good cooperation and the Material testing Institute at the University of Stuttgart (MPA) for their
help with the realization of the test.
EXPERIMENTAL AND NUMERICAL STUDY ON MECHANICAL
BEHAVIOUR OF COMPOSITE SLABS WITH CLOSED PROFILED
SHEETING
Guochang Li
School of Civil Engineering, Shenyang Jianzhu
University Shenyang, China
liguochang0604@sina.com
Xiao Zhang
syzx024@sina.com
Zhijian Yang
faemail@163.com
Fengwei Guo
807876272@qq.com
ABSTRACT
In this paper, the mechanical behavior of composite slabs with closed profiled steel sheeting
was studied through test and finite element analysis (FEA). Eleven full-scale specimens were
tested under static loading to examine the ultimate capacity of this composite slab. The test
results showed that the failure modes of composite slabs can be divided into three types, which
are longitudinal shear failure, bending-shear failure and flexural failure. The relationship
between load and displacement, development of cross-sectional strain and distribution
regularity of slip were obtained through test, respectively. Based on experimental study, FEA
was conducted with ABAQUS and the accuracy of finite element model (FEM) is also verified
through comparing with test result. With the consideration of different failure modes, a
parameter analysis of ultimate bearing capacity of the slab is carried and the effect of each
design parameter on the ultimate bearing capacity is given.
INTRODUCTION
Composite slab with profiled steel sheeting is one of new steel-concrete composite members in
which profiled steel sheeting and concrete are integrated by using construction measures to
take advantage of the respective properties. Nowadays, the bond behavior between profiled
steel sheeting and concrete has been paid more attention to ensure the effective of composite
action. A good bond behavior in composite slab is the foundation of cooperative work between
two different materials, which can also affect failure mode and ultimate bearing capacity of this
kind of composite slab.
Hicks et al.[1] presented an experiment of static load on composite slabs with profiled steel
sheeting using high-strength steel. The result showed that longitudinal shear behaviour of
composite slab with high-strength steel decreased significantly compared with that of using
plain steel. Meanwhile, an amendment to revise the rules given in Eurocode 4, Annex B.3.1 was
proposed. Rehman et al. [2] conducted 12 full-scale push-off tests, which showed that the
demountable shear connector in composite slab has similar shear capacity and behaviour with
welded shear studs and it also fulfills the requirement of ductility in Eurocode 4. Tremlay et. al
[3] carried out a total of 26 two-point bending slab tests and 50 pullout tests to evaluate the
effect of steel thickness, surface coating, deck position, curing age of the concrete, and the
presence of electrical conduits in the slab on the performance of composite floor deck systems.
The results of the test showed that an increase in the thickness of the steel deck generally
results in a higher overall capacity of the composite slab. Degtyarev [4] presented the results of
parametric studies on steel-deck-reinforced composite slabs with end anchorages provided by
welded shear stud connectors. Effects of the number of studs, longitudinal shear strength, deck
height, steel thickness, deck yield strength, concrete cover depth, concrete compressive
strength, and deck span conditions on strength of the composite slabs and the stud strength
mobilization were investigated. Rana et. al [5] carried out a series of tests on both solid and
profiled composite slabs to investigate the effect of end anchorage on the ultimate load carrying
capacities and failure modes. The test results showed that end anchorage has a positive effect
on the ultimate strength of composite slabs, and the effect of concrete strength and sheeting
thickness on the load-deflection behaviour is significant. Shi et al. and Wu, et. al. [6-7]
conducted series of tests on full scale composite slabs with closed galvanized sheeting plate.
The effects of shear span, sheeting thickness, shear connecting piece on the bond – slip
behavior were analyzed. Wang, et al. [8] investigated the shear bond behavior, failure mode,
deflection, end-slip, and longitudinal shear resistance of the flat-type composite deck slab
Econdek 65-675. The test results showed that the all of the specimens presented significant
end-slip and occurred longitudinal shear failure. Li et al. [9] presented an experimental
investigation of opened profiled steel sheeting and concrete composite slabs, and considered
the influence of design parameters on bond property between profiled steel sheeting and
concrete. Hedaoo et al. [10] did a test study on 18 composite slabs with profiled steel decking.
The comparison of experimental and analytical results of ultimate bearing capacity of composite
slabs revealed that both m-k and PSC methods agreed with test well, and m-k method was
proved to be more conservative than PSC method. Zhang et al. [11] proved that ductility and
ultimate bearing capacity of composite slab with closed profiled steel sheeting is better than
others by conducting the test of composite slabs on static loading. Abdullah et al. [12] studied
the mechanical behavior of steel deck-concrete composite slabs, a new method for modeling
the horizontal shear bond was proposed. Hao et al. [13] reported the test and analysis results of
composite slab with closed profiled steel sheeting, the equation of relationship between shear
bond stress and relative slip was conducted. Crisinel et al. [14] proposed a new design
approach for the prediction of composite slab behaviour by test. Wendel et al. [15] presented a
nonlinear finite element analysis of steel-concrete composite structure, the result showed that
deformation performance and mechanical behavior of composite slabs were simulated well by
considering the crack of concrete. Patrick et al. [16] presented an experiment of composite
slabs under uniform load and concentrated load respectively, the calculation methods of shear
capacity considering two different load forms was proposed and an amendment of partial shear
connection design of composite slabs was also given.
The requirement of long-span composite slabs increases with rapid development of large-span
spatial structures at present. Due to their advantages, more and more researches have been
focusing on the mechanical behaviour of composite slabs with closed profiled steel sheeting,
thus the study about this kind of composite slab in this paper is very meaningful.
2. EXPERIMENTAL INVESTIGATION
2.1 Test design
A total of 11 simply supported composite slab specimens were designed, which decked with
closed profiled steel sheeting, as shown in Figure 1. The parameters of span, depth of the
composite slab and configuration of the end-anchorage stud were considered, the details of
specimens are listed in in Table 1. For all the specimens, two steel plates with size 1110 × 150
× 10 mm were welded at both ends.
(a) Cross-section of the steel sheets (b) End anchorage details of the specimens
Fig.1 Details of test specimens
Table 1 Details of the composite slabs with profiled steel sheeting
B×t×L fy fcu
No. Specimen Studs Failure mode
(mm×mm×mm) (MPa) (MPa)
1 VBI-1 1110×180×2000 D19 348 33.1 Bending-shear
2 VBI-2 1110×180×2000 D19 348 33.1 Bending-shear
3 VBI-3 1110×180×2000 None 348 33.1 Longitudinal shear failure
4 VBII-1 1110×150×3400 D19 348 33.1 Flexural failure
5 VBII-2 1110×150×3400 D19 348 33.1 Flexural failure
6 VBIII-1 1110×160×4800 D19 348 33.1 Flexural failure
7 VBIII-2 1110×200×4800 D19 348 33.1 Flexural failure
8 VBIII-3 1110×160×4800 None 348 33.1 Flexural failure
9 VBIV-1 1110×200×6000 D19 348 33.1 Flexural failure
10 VBIV-2 1110×250×4800 D19 348 33.1 Flexural failure
11 VBIV-3 1110×200×4800 None 348 33.1 Flexural failure
Note: B is the width of the slab. t is the slab thickness. L is the span. D19 means studs’
diameter is 19mm. fy is steel yield strength. fcu is the actual measured 28-day concrete. cube
strength.
2.2 Experimental scheme and layout of measurement points
Two different types of loading devices were used according to the different spans of the
specimens. A four-point loading device was used for specimens with a span 2.0 m, with two
equal point loads applied by a load-distribution beam, as illustrated in Figure 2(a). A six-point
loading device was used for the specimens with spans of 3.4, 4.8, and 6.0 m, with four equal
point loads applied by two levels of load-distribution beams, as displayed in Figure 2(b). The
vertical load was applied by a 500 kN hydraulic jack with load sensor and collected by an
automatic data collection system IMP.
The vertical load was applied with a step increase, which is 10% of theoretical ultimate bearing
capacity of the specimen. At each load level, the vertical load held for 5 min. After the specimen
yielded, the load level was reduced to 1/15 of the theoretical ultimate bearing capacity. As the
deflection of midspan was more than 1/50 of the span, the test was terminated.
The strains of profiled steel sheet and concrete were measured by strain gauges. The
displacement and slip between the steel sheet–concrete were measured using linear variable
differential transformer (LVDT) sensors. Figure 3 shows the configuration of the LVDT sensors.
(a) Four-point bending test (b) Six-point bending test

Fig. 2 Test setup
L/8 L/8 L/4 L/4 L/8 L/8 L/10 L/10 L/5 L/10 L/10 L/5 L/10 L/10
L L
LVDT for vertical LVDT for slip LVDT for vertical
displacement LVDT for slip
displacement
(a) Four-point bending test (b) Six-point bending test

Fig. 3 Arrangement of the LVTDs
2.3 Test results and discussions
2.3.1 Failure mode
The failure modes of test specimens could be divided into three types according to
experimental phenomena, as shown in Figure 4.
The failure mode of specimen VBI-3 is longitudinal shear failure. First vertical crack appeared
near the load point, and then the crack extended through the slab. As the longitudinal crack
appeared, the vertical separation between profiled steel sheeting and concrete occurred
significantly with the increase of load. At the end of test, composite action of specimen lost and
the end slip was 5.13mm.
The failure mode of specimen VBI-1 and VBI-2 is bending-shear failure. First vertical crack
appeared near load point, and became the main crack during failure stage. Longitudinal crack
occurred in shear span, and developed through the slab suddenly when the end studs were
broken. Then, the vertical separation appeared between profiled steel sheeting and concrete. At
the end of failure stage, the composite action of specimen was destroyed and the end slip
increased quickly. The bottom of profiled steel sheeting of mid-span was ruptured when
approaching its ultimate bearing capacity.
The failure mode of specimen VBII, VBIII and VBIV is flexural failure. First vertical crack
appeared near the mid-span, and many small cracks emerged at the begining of failure stage.
Longitudinal crack occurred in shear span but not throughout the slab. The strain of mid-span
section increased rapidly when approaching failure load. Large slip did not happen and
composite action of slab still existed as the specimen failed.
(a) Longitudinal shear failure (b) Bending-shear failure
(c) Flexural failure

Fig. 4 Failure mode of typical specimen
2.3.2 Analysis of the load - deflection curves
The load - deflection curves of mid-span are showed in Figure 5. It can be seen that the load -
deflection curves are similar but the stiffness is quite different. Due to good performance of co-
working between profiled steel sheeting and concrete, specimens show good ductility and
rigidity. The end stud has a great effect on increasing the ultimate bearing capacity of 2.0m
specimen.
700 240
210 VBII-1
600
500 180
VBI-1 VBI-2 150 VBII-2
Load (kN)
Load (kN)
400
120
300 VBI-3
90
200 60
100 30
0 0
0 5 10 15 20 25 30 35 40 45 50 0 10 20 30 40 50 60 70 80 90
Deflection (mm) Deflection (mm)
(a) 2.0m specimen (b) 3.4m specimen
120 120
100 VBIII-2 100 VBIV-2

80 VBIII-1
80
Load (kN)
Load (kN)
VBIV-1
60 VBIII-3 60
VBIV-3
40 40
20 20
0 0
0 30 60 90 120 150 180 210 0 30 60 90 120 150 180
(c) 4.8m specimen (d) 6.0m specimen

Fig. 5 The load-deflection curve of specimens
2.3.3 Analysis of the load - slip curves
The slip between profiled steel sheeting and concrete of 2.0m specimen is larger than other
specimens. The measured points of slip are placed at the end of slab, load point, mid of span.
The load-slip curve of specimen VBI-1 in different parts is showed in Figure 6. The development
of slip in ends is asymmetry during loading. And sequence of slip appearance: first in mid of
shear span, next in load point and last in end of slab. When reaching failure load, the maximum
slippage is in load point and the minimum slippage is in end of slab.
The load-slip curve of specimen VBI-3 in different parts is showed in Figure 7. The sequence of
slip appearance is the same with that of specimen VBI-1 but the final slippage in three different
parts is quite different. It indicates that end studs have a great effect on restraint of end slip but
there is no stud in specimen VBI-3.
600 600 600

The right side of the end
The right load point The right side of mid- span
500 500 500
The left side of mid- span
The left load point
400 The left side of the end 400 400
Load (kN)
Load (kN)
Load/kN
300 300 300
200 200 200
100 100 100
0 0 0
0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 0 1 2 3 4 5 6 7 8
Slip (mm) Slip/mm Slip (mm)
(a) End (b) Load point (c) Shear span

Fig. 6 The load-slip curve of specimen VBI-1
500
Load point
400
The end
Load (kN)
Middle of shear span

300
200
100
0
0 1 2 3 4 5 6 7 8
Slip (mm)
Fig. 7 Load-slip curve of specimen VBI-3
2.3.4 Analysis of the strain-load curve
The load – strain curves of mid-span are showed in Figure 7. As the bottom and web of profiled
steel sheeting yielded, the strain decreases rapidly. The strain at the top of profiled steel
sheeting increases with the increase of span under same loading.
600 250
Top Web Web
500 200 Top
400
Load (kN)
Load (kN)
150 Bottom
300 Bottom
100
200
50
100
0 0
0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500
Strain ( Strain (
(a) VBI-1 (b) VBII-1
100 80
70
80 Top Web Top Web
60
50 Bottom
Load (kN)
60
Load (kN)
40
40 Bottom
30
20
20
10
0 0
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000 3500
Strain ( Strain (
(c) VBIII-1 (d) VBIV-1

Fig. 8 Load-strain curve of profiled steel sheeting in mid span
2.3.5 Strain distribution of mid-span section
Strain distribution of mid-span section is showed in Figure 9. It can be seen that the neutral axis
of specimen VBI-1 moves upward slightly with the increase of load. For specimen VBI-3, the
strain presents nonlinear due to the earlier appearance of slip without end studs, the change of
its neutral axis is similar to that of specimen VBI-1. For specimen VBII-1, large slip and vertical
separation did not appear between profiled steel sheeting and concrete, and the neutral axis
has little changes. The span of VBII-1 is larger than VBI-1 and VBI-3, therefore the neutral axis
is lower at last.
200 200 200
0.2Pu 0.2Pu
150 0.4Pu 150 150 0.4Pu
0.6Pu 0.2Pu 0.6Pu
Height (mm)
Height (mm)
Height (mm)
0.8Pu 0.4Pu 0.8Pu

100 1.0Pu 100 0.6Pu 100 1.0Pu
0.8Pu
50 50 1.0Pu 50
0 0 0
-2000 -1000 0 1000 2000 3000 4000 -2000 -1000 0 1000 2000 3000 -2000 -1000 0 1000 2000 3000
Strain ( Strain ( Strain (
(a) VBI-1 (b) VBI-3 (c) VBII-1

Fig. 9 The strain distribution of mid-span section
3. FINITE ELEMENT ANALYSIS
3.1 Establishment of FEM
The plastic damage model and the constitutive relation of the Code for design of concrete
structures (GB50010-2010) [17] were used for concrete. Bilinear constitutive model was
adopted for steel, which included elastic stage and hardening stage, the module is Es and
0.01Es respectively. Three-dimensional eight-node solid element (C3D8R) was used to simulate
concrete. Profiled steel sheeting was simulated by the four-node reduced integrated shell
element (S4R). Three-dimensional two-node truss element (T3D2) and beam element (B31)
were selected to simulate reinforcement and end studs respectively.
The surface-to-surface contact was used to simulate the interaction between concrete and
profiled steel sheeting, which considered the friction in tangent direction. The friction coefficient
was defined as 0.65 obtained by push-out test. Meanwhile, normal constraint was defined by
“Hard Contact” which allows the appearance of vertical separation between concrete and steel.
Reinforcement was embedded into concrete.
3.2 Comparison between FEA and test results
The comparison between simulation results and the test is shown in Fig. 9. The load –
deflection curve of simulation agreed well with the test. The ultimate bearing capacity of the
simulation results was defined as deflection of mid-span reached L/50. The error of ultimate
bearing capacity between the simulation and test is less than 5%, so the FEM can effectively
simulate the mechanical behaviour of the composite slab.
600 500 250
500 400 200

400
Load (kN)
300 150
Load (kN)
Load (kN)
300
Test 200 100
200 Test Test
FEM FEM FEM
100 100 50
0 0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 0 20 40 60 80 100
Deflection (mm) Deflection (mm) Deflection (mm)
(a) VBI-1 (b) VBI-3 (c) VBII-1

Fig.9 Comparison of load-deflection curve between test and FEM
4. PARAMETERS ANALYSIS
4.1 Longitudinal shear-bond bearing capacity
Parameters analysis is made by control variable method and using specimen (VBI-3) as
example. The effect of different parameters on shear-bond bearing capacity is showed in Fig.
10, where point A, B, C are crack of concrete, yield of the steel, the ultimate bearing capacity of
specimen.
500 500
C4 C4
400 B4 C3 400 C3
C2 B3(B4) C1(C2)
B3 C1
Load (kN)
Load (kN)
300 B2 300 B1(B2)

B1
0.75mm C20
200 0.90mm 200
C25
1.0mm C30
100 1.2mm 100 C40
A1,2,3,4
A1,2,3,4
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) Effect of profiled sheeting thickness (b) Effect of concrete strength
500 600
C4
400 500
B4 C3
B3 400
Load (kN)
Load (kN)
300 C2
B2 C1 300
VBI-3
200 B1 120mm
200 VBI-1
150mm VBI-2
100 180mm 100
A3(A4)
A1(A2)
200mm
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(c) Effect of slab thickness (d) Effect of end studs

Fig.10 Parameters analysis of longitudinal shear-bond bearing capacity
From Figure 10(a), it can observed that point A1 to point A4 is quite close, the thickness of
profiled steel sheeting has little effect on the cracking load but has great impact on the yielding
load (point B) and ultimate bearing load (point C). The yielding load increases with the increase
of thickness. As the thickness increases from 0.75mm to 1.00mm and 1.2mm, the shear-bond
bearing capacity increases by 18.2% and 24.5, respectively. It shows that the bearing capacity
can be improved effectively by increasing the thickness of profiled steel sheeting.
It can be seen from Figure 10(b) that concrete strength has little effect on the ultimate bearing
capacity of the slab. The bearing capacity increases by 7.51% as the concrete strength
increases from C20 to C40. The increasing of concrete strength has impact on the chemical
bond between concrete and profiled steel sheeting, but it is a small part of bond force.
It can be seen from Figure 10(c) that the bearing capacity increases 22.3%, 60.6%, 78.5% as
the thickness of slab increases from 120mm to, 150mm, 180mm, 200mm. It indicates that the
bearing capacity of composite slab can be increased effectively by increasing slab thickness.
However, once shear strength exceeds maximum bond strength between concrete and profiled
steel sheeting, the influence of slab thickness is inexistence.
Figure 10(d) indicates that the bearing capacity of specimen VBI-1 with end studs is 25% bigger
than specimen VBI-3. The end stud has great effect on restraining the development of slip and
improving ductility.
4.2 Parameters analysis of flexural bearing capacity
Specimen VBIII-3 is selected as example to make parameters analysis on flexural bearing
capacity. The effect of different parameters on flexural bearing capacity is showed in Fig. 11.
120 100 100
100 80 80
80
Load (kN)
Load (kN)
60
Load (kN)
60
60 0.75mm C20
VBIII-3 40 0.9mm 40
40 C25
VBIII-1 1.0mm C30
20 VBIII-2 20 20
1.2mm C40
0 0 0
0 30 60 90 120 150 180 210 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120
Deflection (mm) Deflection (mm) Deflection (mm)
(a) Slab thickness (b) Profiled sheeting thickness (c) Concrete strength
Fig.11 Parameters analysis of flexural bearing capacity
It can be seen from Fig. 11(a) that the flexural bearing capacity increases 35% and the rigidity
of slab enhances effectively as the slab thickness increase from 160mm to 200mm. The
bearing capacity of specimen VBIII-1 is 6% bigger than specimen VBIII-3. Thus the end stud
has little effect on long- span composite slabs, because the capacity of the long-span specimen
is influenced by its bending performance. As flexural failure occurs, bonding performance still
keep in good condition and there is no slip in the end, so the end stud can’t be fully used.
It is observed that Figure 11(b) that flexural rigidity and bearing capacity of specimens improves
obviously with the increase of profiled steel sheeting thickness. When the thickness increases
form 0.7mm to 0.9mm, 1.0mm, 1.2mm, the capacity improves 9.8%, 19.7%, 37.7%,
respectively.
Mechanical principle of long-span composite slab is similar with reinforced concrete beam, on
the basis of the equation M=fyAsZ (Z is arm of internal force), the resistance moment of slab can
improve with the increase of profiled steel sheeting thickness effectively.
Figure 11(c) shows that the bearing capacity increases 2.9%, 4.9 %, 13%, respectively as the
concrete strength increases from C20 to C25, C30, C40. The increase of concrete strength has
little impact on the flexural bearing capacity.
5. CONCLUSIONS
Based on the analysis and discussion on the test and FEA results, the following conclusions
can be drawn:
(1) The composite slab with closed profiled steel sheeting under static load has a good ductility
and its bearing capacity is influenced by failure mode.
(2) In the test, there are three different failure modes, which are longitudinal shear failure,
bending- shear failure and flexural failure. In each failure modes, the effect of end studs has
different impact on improving the bearing capacity, only in longitudinal shear failure, they work
well.
(3) The FEA result shows that thickness of slab and profiled steel sheeting has good effect on
the bearing capacity in longitudinal shear failure and flexural failure but the influence of
concrete strength is little.
(4) The failure mode of this kind of composite slabs varies with the change of span. Although
longitudinal shear failure is the main failure mode, the capacity of long-span specimens
depends on their flexural behaviour.
ACKNOWLEDGEMENTS
This project was supported by the Twelfth Five-year National Key Technology Support Program
of China (2011BAJ09B0402), Innovative Research Team of Higher Education in Liaoning
Province (LT2014012), Shenyang Engineering and Techno-logical Research Center for Civil
Steel Construction Industrialization (F16-076-8-00).
REFERENCES
[1] Hicks, S., Jones, A., & Pennington, A. (2013) Performance of Composite Slabs with Profiled
Sheeting Using High-Strength Steel. Proceedings of the 2013 International Conference on
Composite Construction in Steel and Concrete, 744-753.
[2] Rehman, N., Lam, D., Dai, X., & Ashour, A. F. (2016). Experimental study on demountable
shear connectors in composite slabs with profiled decking. Journal of Constructional Steel
Research, 12(2), 178-189. doi: 10.1016/j.jcsr.2016.03.021.
[3] Tremlay. R., Gignac, P., Degrange G, et al. (2002). Variables affecting the shear-bond
resistance of composite floor deck systems. International Specialty Conference on Cold-
Formed Steel Structures, 663-676.
[4] Degtyarev, V. V. (2014). Strength of composite slabs with end anchorages. Part II:
Parametric studies. Journal of Constructional Steel Research, 94, 163-175. doi:
10.1016/j.jcsr.2013.10.004.
[5] Rana, M. M., Uy, B., Mirza O. (2015). Experimental and numerical study of end anchorage in
composite slabs. Journal of Constructional Steel Research, 115: 372-386.
doi.org/10.1016/j.jcsr.2015.08.039.
[6] Shi Q X., Hao J H., Zhang, X H. (2007). Experiments on shear – bond – slip behavior of
steel deck and concrete composite slabs. Journal of Harbin Institute of Technology, 39(S2),
487-490.
[7] Wu, L., Shi, Q. X., Wei, X. (2008). Experimental study on shear-bearing capacity of
composite floor slab with closed profiled steel sheet-concrete. Journal of Water Resources
and Architectural Engineering, 6(4), 137-138.
[8] Wang Y. Q., Cheng, Z. T., Shi, Y. J. (2010). Experimental analysis on the longitudinal shear
resistance of flat-type composite slab. Journal of Civil, Architectural &Environmental
Engineering, 32(6), 1-6.
[9] Li, G. C., Wang, Y. K., Yang, Z. J., & Xu, W. (2015). Study on bond property between
opened profiled steel sheet and concrete composite slabs. Journal of Building Structures,
S1, 100-106. doi: 10.14006/j.jzjgxb.2015.S1.016.
[10] Hedaoo, N. A., Gupta, L. M, & Ronghe, G. N. (2012). Design of composite slabs with
profiled steel decking: a comparison between experimental and analytical studies.
International Journal of Advanced Structural Engineering (IJASE), 4(1), 1-15.
[11] Zhang, J., & Liu, Z. H. (2013). Experimental study on longitudinal bond behaviors of closed
profiled steel sheet-concrete composite slabs. Structural Engineers, 1(13), 124-129.
[12] Abdullah, R., Easterling, W. S. (2009). New evaluation and modeling procedure for
horizontal shear bond in composite slabs. Journal of Constructional Steel Research, 65(4),
891-899. doi: 10.1016/j.jcsr.2008.10.009.
[13] Hao, J. H. (2007). Experimental study on shear bond-slip behavior of profiled sheeting
concrete composite slabs. Xi’an University of Architecture and Technology, Master's Thesis.
[14] Crisinel, M., & Marimon, F. (2004). A new simplified method for the design of composite
slabs. Journal of Constructional Steel Research, 60(3), 481-491. doi: 10.1016/S0143-
974X(03)00125-1.
[15] Wendel, M., & McConnell, J. (2001). Nonlinear finite element analysis of steel-concrete
composite structures. Journal of Structural Engineering, 126(6), 662-674.
[16] Patrick, M., & Bridge, R. Q. (1994). Partial shear connection design of composite slabs[J].
Engineering Structures, 16(5), 348-362. doi: 10.1016/0141-0296(94)90028-0
[17] Standards China. (2010). Code for design of concrete structures (GB 50010-2010). Beijing:
China Architecture & Building Press
ROLE OF THE FLOOR SYSTEM IN THE CYCLIC RESPONSE OF
COMPOSITE STEEL GRAVITY FRAMING
Sean Donahue, Michael Engelhardt, Patricia Clayton, Eric Williamson, Todd Helwig
University of Texas at Austin, Austin, TX, USA
Contact: sdonahue@utexas.edu
ABSTRACT
In typical U.S. design practice for steel buildings, lateral loads are assumed to be resisted by
a small number of lateral-force-resisting frames, while the rest of the structure consists of
gravity framing, assumed to have no flexural resistance. It has long been recognized that the
simple shear connections used in gravity framing do possess some flexural strength; however,
there is little experimental data on the cyclic response of these connections, particularly
double-angle connections. This paper details ongoing experimental research into the
response of such connections. Full-scale cruciform composite beam-column subassemblies
were tested cyclically under large inter-story drifts. These tests have shown that simple shear
connections can exhibit significant flexural resistance under large drifts, largely due to the
large tensile capacity of the metal decking in the floor system. Given the large number of shear
connections in most structures, this capacity could significantly enhance the seismic collapse
resistance of a structure.
Keywords: seismic behavior; composite connections; steel buildings; cyclic loads; gravity
framing.
1 INTRODUCTION
Seismic resistance is typically provided in steel structures by a small number of moment
frames, braced frames, or shear walls. The remainder of the structure is composed of gravity
framing (beams framing into columns with “simple shear” connections, topped by a composite
slab). These simple shear connections are typically designed as perfect pins, assumed for
design purposes to have no flexural strength or stiffness. This assumption implies that these
connections cannot contribute to the seismic resistance of a structure. In reality, these
connections do provide non-negligible flexural strength, particularly when they act compositely
with the floor slab. While this strength may be relatively small, the large number of gravity
frames present in a typical steel structure means their overall contribution can be significant.
After the 1994 Northridge earthquake, many steel structures experienced significant fractures
in the moment-resisting connections of their lateral frames, yet none of those structures
exhibited full collapse [1]. During the investigations that followed that event, analytical studies
suggested that the gravity framing can play a key role in the survival of steel buildings in
earthquakes, in some cases being the difference between survival and collapse [2,3]. Though
these studies suggest that gravity framing can play a significant role in the seismic response
of steel buildings, their behavior is not yet fully understood. Current models of shear
connection behavior are based on a limited number of experimental studies, primarily testing
of shear tab connections done by Liu and Astaneh [4], but also tests of other semi-rigid
connection details conducted by Leon [5], Azizinamini and Radziminski [6], and others. While
those studies have shown that simple shear connections can exhibit significant flexural
capacity, especially in the presence of a composite floor slab, the strength, stiffness and
ductility of those connections can vary significantly depending on connection and floor system
details. Given the large number of different details and configurations used in gravity framing
systems, research is necessary to better characterize the seismic behavior of the wide range
of such systems. This paper details large-scale experimental research intended to improve
our understanding of the behavior of simple shear connections, particularly the behavior of
double-angle shear connections with a variety of different floor slab details.
2 TEST SETUP
The experimental study includes testing of full-scale cruciform beam-column subassemblies
subjected to large cyclic column rotations. The sub-assemblies were designed as components
of a prototype building, with W12x96 columns at 8.99 m (29.5 ft.) spacing and with W 14x22
floor beams framing in to W21x55 girders at 2.21 m (7.25 ft.) spacing. The prototype structure
was assumed to have 3.51 m (11.5 ft.) floor heights.
The cruciform subassemblies and boundary conditions of the test set-up shown in Figure 1
were designed to be representative of a section of the prototype building, cut out at mid span
of the girders and mid height of the column. The floor slab was constructed with an assumed
tributary width of one-fourth the girder span (i.e., the slab was 2.21 m (7.25 ft.) wide). Through
the use of a series of hydraulic actuators, concentrated forces and moments were applied to
the test girders to simulate the gravity loads recommended by ASCE for a seismic load
scenario (1.2 times dead load and 0.5 times live load) over the full length of the specimen,
which was maintained throughout the test. Once the gravity load was applied, horizontal
actuators at the top and bottom of the column applied equal and opposite displacements,
rotating the column to simulate the inter-story drifts expected during a seismic event. Drifts
were applied according to the SAC load protocol [7].
Figure 1-Test setup (plan view)

The double-angle shear connections used in all the test specimens consisted of 2L4x4x1/4
angles (also called web cleats), each with four 19mm (3/4”) A325 bolts connected to both the
column and girder. The floor system consisted of a 165mm (6-1/2”) thick normal-weight
concrete slab on 51mm (2”) tall, 1mm thick (20GA) corrugated metal decking. Welded wire
reinforcement (WWR 6x6x2.9) was placed in the slab for shrinkage control, and 19 mm (3/4”)
shear studs were welded through the decking to the girder to achieve approximately 25%
composite action with the slab as shown in the connection detail in Figure 2.
This paper will discuss five test specimens completed as part of this research to date, though
more tests are planned. The first specimen (designated 1B) consisted of only the bare steel
beam-column subassembly, with no composite slab system included. The next two specimens
consisted of an identical double-angle connection, with the composite concrete slab included.
The second specimen (designated 1W) had the ribs of the corrugated decking running
perpendicular to the girders (referred to as “weak axis” decking in this document), while the
third specimen (designated 1S) had the ribs running parallel to the girders (referred to as
“strong axis” decking in this document). The fourth specimen (designated 2W) was similar to
specimen 1W, however, additional 1.52 m (5’) lengths of reinforcement were placed over the
secondary beam (Figure 3). The fifth specimen (designated 2S) was similar to specimen 1S,
with the addition of a transverse seam in the decking at the column location, instead of a
continuous sheet. One end of the metal decking was attached to the secondary beam with
5/8” puddle welds at each low rib, while the other end was attached through the weld in the
shear stud, also at each low rib (Figure 4).
Figure 2-Connection detail (a) specimen 1S (b) specimen 1W
ADDITIONAL #4 X 5’-0” @6”

SLAB STEEL OVER BEAMS
CENTER STEEL OVER
BEAMS
Figure 3-Additional reinforcement in specimen 2W
Figure 4-Deck connection at longitudinal seam in specimen 2S
3 TEST RESULTS
The presence of a floor slab is known to significantly increase the flexural capacity of simple
shear connections under seismic loading [4,5]; however, this increase in strength is highly
dependent on the detailing of the composite floor system. Additionally, the response of the
connection is very different under positive and negative moment. Thus, although the test
specimens each consist of two shear connections, the discussions below will examine the
connections independently, to highlight the different mechanisms of the connection’s positive
and negative moment behavior. Similarly, the moment-rotation response plots shown below
represent the response of one connection, rather than the response of the full, two-connection
system. The flexural response of the single connections was obtained from consideration of
the specimen free-body diagram and actuator load cells, and the rotation response was
obtained from string potentiometers located at the top and bottom of the test column. Full
details of the data analysis can be found in [10].
3.1 Response of Bare Steel Connection (Specimen 1B)

The flexural capacity of the bare steel double-angle connection was very small relative to the
flexural capacity of the beam. At small inter-story drifts, the flexibility of the angles prevented
them from generating significant moment resistance as they pried away from the column. This
observation is relatively consistent with the perfect pin assumption typically used in design. At
approximately 3.5% drift, however, the bottom flange of the girder began to bear on the flange
of the column, creating a force couple between the beam bottom flange and the angles,
significantly increasing the flexural stiffness and negative moment capacity of the connection
(Figure 5). At approximately 5% drift, tearing at the top of the angles resulted in a rapid
degradation in negative moment capacity of the connections, with the capacity peaking at
approximately 10% of the plastic moment capacity of the girder (Figure 6). While the flexural
strength of the connection was relatively small, the shear strength of the connection was
maintained up to very high drift levels. Despite the tears in the angles propagating nearly the
entire length of the angles (as shown in Figure 7), the angles continued to carry the full gravity
load of the girder up to 10% drift, the maximum rotation that could be accommodated by the
test setup.
Figure 5-Response of specimen 1B at high drifts
Figure 6-Moment-rotation response of connection (specimen 1B)
Figure 7-Fracture in bare steel double-angle connection
3.2 Response of Connection with Perpendicular Deck Ribs (Specimen 1W)

The inclusion of a floor slab significantly increased the positive moment capacity of the
connection, as suggested by previous researchers [8]. As the column rotated, the concrete
floor slab bore on the column flange. This compression in the slab was opposed by tension in
the angle as the column pulled away from the girder as shown schematically in Figure 8. This
force couple produced a positive moment in the connection that reached approximately 20%
of the plastic moment capacity of the girder. This capacity began to drop off quickly after 5%
drift, as tearing initiated at the bottom of the angles and quickly grew as shown in Figure 9.
Figure 8-Response of specimen 1W at low drifts

Figure 9-Moment rotation response of connection (specimen 1W vs. 1B)
When the column rotated in the opposite direction, causing the column flange to pull away
from the slab, the negative moment connection capacity was limited. Due to the cracking in
the concrete and fracture of the slab welded wire reinforcement, the floor slab had minimal
tensile capacity, and the connection’s negative moment capacity was limited to the moment
capacity of the clip angle itself, similar to what was observed in Specimen 1B. This negative
moment-resisting mechanism also had limited robustness, reaching its peak strength at 1.5%
drift, and decaying significantly thereafter.
Due to the much greater stiffness of the compressive elements of the connection (primarily
bearing of the concrete slab on the column face) compared to the tensile elements (prying in
the angles) and due to the roller support at the end of one of the girders, a phenomenon known
as frame expansion was evident over the course of the test, with the girders pulling further
away from the column as drift demands increased. This frame expansion delayed the onset
of flange binding, which occurred at 5.5% drift, rather than at 3.5% drift as in specimen 1B (
Figure 10). As tearing in the angles had already begun by this point, the increase in flexural
capacity associated with bottom flange binding was limited, reaching approximately 12% of
the girder’s plastic moment capacity before degrading almost immediately.
Figure 10-Response of specimen 1W at high drifts

Additionally, due to the greater tensile demand placed on the angles when opposing the
compression in the concrete slab, tearing initiated at a smaller drift demand compared to the
bare steel specimen (tearing was first observed in Specimen 1W at 5% drift, rather than 6%
in Specimen 1B) and occurred only at the bottom of the angles, rather than the top and bottom.
These increased tensile demands caused the tears to propagate through the entire depth of
the angle, leading to loss of gravity load-carrying capacity of the connection at 8% drift (Figure
11). These finding suggest that the inclusion of a composite concrete slab can reduce the
overall rotational capacity of the double-angle connection.
Figure 11-Complete tearing of connection in specimen 1W at end of testing
3.3 Response of Connection with Parallel Deck Ribs (Specimen 1S)

The positive moment response of specimen 1S was similar to the response of specimen 1W.
A force couple formed between the compressive force from the concrete slab bearing on the
column, and tension in the angles; however, the “strong axis” orientation increased the tensile
stiffness and strength of the metal decking, increasing the positive moment capacity of the
connection to approximately 25% of the plastic moment capacity of the girder. This increase
in strength also altered the failure mode of the connection, with the positive moment response
governed by concrete crushing rather than tearing in the angles. This concrete crushing
behavior reduced the strength of the connection at larger drifts (compared to specimen 1W),
with the positive moment capacity peaking at 3% drift, and degrading significantly afterwards
as shown in Figure 12.
Figure 12-Moment-rotation response of connection (specimen 1S vs. 1W)

The presence of a tensile component in the slab (i.e. the decking oriented in its “strong axis”)
had a much greater effect on the negative moment capacity of the connection. At small drifts,
a force couple was formed between the tension in the metal decking and compression in the
clip angle, significantly increasing the negative moment capacity compared to specimen 1W,
reaching approximately 20% of the girder’s plastic moment capacity (Figure 13). Additionally,
because the strong-axis orientation of the metal decking increased the resistance to frame
expansion, the bottom flange of the beam began to bind on the column at approximately 2.5%
drift, creating a new force couple between the tension in the metal decking and this bearing
force (Figure 14). This new force transfer mechanism in the bottom beam flange significantly
increased the stiffness and strength of the connection, reaching moments in excess of 50% of
the girder’s plastic moment capacity by 4% drift.
Figure 13-Response of specimen 1S at low drifts
Figure 14-Response of specimen 1S at high drifts

The flexural resistance developed by beam bottom flange binding and tension in the metal
decking showed little degradation up to very large drift demands. Even after multiple cycles of
9% inter-story drift (the largest drift accommodated by the test setup for composite
specimens), the connection still possessed a negative moment capacity of approximately 33%
of the plastic moment capacity of the girder, unlike the previous tests, which exhibited
negligible moment capacity at similar drifts. The additional tensile resistance in the metal deck
also may have limited the tension demands on the angles, delaying the onset of tearing in the
angles compared to specimen 1W. As in the bare steel case, despite nearly full depth tearing
in the angles, the connection continued to support the full gravity load even at the maximum
drift accommodated by the test setup. Although the connection does retain its strength at high
drifts, there is significant pinching of the hysteretic loops, as the force couple between the
flange binding and metal decking only demonstrates significant flexural resistance after
closing any gaps formed by plastic deformation of the connection in previous cycles of loading.
3.4 Response of Connection with Additional Reinforcement (Specimen 2W)

The tensile capacity in the floor system in specimen 1S, where the metal decking was oriented
in its “strong axis” direction, significantly increased the overall strength of the connection
compared to comparable specimens with metal decking oriented in the “weak axis” direction.
In current building practice, short lengths of rebar are often placed over flexural members
perpendicular to the weak axis of the steel decking to limit cracking in the floor slab during
service. It was hypothesized that this additional crack control reinforcement might provide
additional tensile resistance to increase the flexural capacity of composite connections with
metal decking in the “weak axis” orientation. To test this hypothesis, a specimen with metal
decking oriented in its “weak axis” was tested with this crack control reinforcement included.
As the amount of crack control reinforcement in a typical building can vary significantly, the
reinforcement in this specimen was designed to have approximately the same cross-sectional
area of steel as the metal decking oriented in its strong axis.
The presence of reinforcement in the floor slab did result in moderate tensile forces
developing in the composite floor slab system, producing a significant increase in the negative
moment capacity of the connection under low drifts compared to specimen 1W, reaching
approximately 20% of the girder’s plastic moment capacity (Figure 15). By 1% drift, however,
cracks developed in the floor slab at the ends of the reinforcement. These cracks grew quickly,
leading to a complete shear failure in the slab at 3% drift as shown in Figure 16. This slab
failure severely limited the ability of the reinforcement to carry tensile load, and the
connection’s negative moment capacity quickly degraded. The reinforcement was still able to
develop small tensile forces through the few shear studs connecting the undamaged portion
of the slab and the girder, resulting in a slight increase in negative moment strength upon the
onset of flange binding, but the connection never regains its initial peak strength.
Figure 15-Moment-rotation response of connection (specimen 2W vs. 1W)
Figure 16-Fracture in slab at end of supplementary reinforcement
3.5 Response of Connection with Transverse Deck Seam (Specimen 2S)

While the metal decking used in composite construction has the potential to generate
significant tensile forces, particularly if oriented in its “strong axis,” this capacity is not
considered in typical design. Thus, the connections between decking elements are intended
only for construction purposes, and typically can only develop a fraction of the deck’s tensile
capacity [9]. If the metal decking is discontinuous at the connection (i.e. if there is a seam
between decking sheets), the ability of the floor system to generate tensile capacity (and form
a force couple with the flange binding), will be limited by these weak, brittle connections.
Despite the low capacity of the deck-to-beam puddle welds in specimen 2S, there was some
development of tensile resistance in the composite floor system at low drifts, resulting in a
slight increase in negative moment capacity compared to specimen 1W (Figure 17). However,
at approximately 2.5% drift, the puddle welds began to fracture, causing a significant loss in
the deck’s tensile capacity, and a corresponding drop in negative moment capacity of the
connection. By 5% drift, the connection’s negative moment capacity was approximately equal
to that of specimen 1W, implying a near complete loss in the floor system’s tensile capacity,
similar to that of metal decking oriented in its “weak axis.” The connection did also experience
an increase in positive moment capacity compared to specimen 1W, but this is believed to
primarily be due to an unexpectedly high concrete strength in the specimen, and not a virtue
of the connection detailing.
Figure 17-Moment-rotation response of connection (specimen 2S vs. 1W)
4 DISCUSSION AND CONCLUSIONS

This paper discussed preliminary results from a series of ongoing tests on the response of
double-angle shear connections with and without a composite floor slab under cyclic loading.
These results indicate that the inclusion of the composite floor slab significantly alters the
response mechanism of such connections, increasing their stiffness and strength. If the floor
system possesses significant tensile capacity, as in the case with continuous metal decking
with ribbing oriented parallel to the direction of lateral drift (i.e., “strong axis” orientation), this
increase in strength can be very large, resulting in negative moment capacities greater than
50% of the girder’s plastic moment capacity, approximately four times the capacity of a
comparable connection with decking oriented in the orthogonal direction. However, as the
composite slab and decking system is not explicitly designed to carry tension, such
configurations may be rare in many structures.
There are inherent limitations in examining the response of a subassembly specimen rather
than a full structure. While every effort was made to match the boundary conditions present in
a full building, practical limitations and uncertainty about the response of a real structure meant
some aspects of the test setup may not adequately represent the conditions in an actual
building. For example, in specimen 1S, our testing suggests that the metal decking developed
significant tensile force along the entire width of our test specimen. In a full structure, with a
much greater width of decking present, it is possible a greater tensile force could develop,
increasing the flexural capacity of the connection. The increased width could also serve to
delay the shear cracking seen in specimen 2W; the additional puddle welds in a wider deck
could also improve the ability of the deck to transmit tension across a seam, improving the
capacity of both details.
Additionally, in a full structure, the presence of neighboring bays of gravity framing would
provide some degree of restraint to the type of frame expansion observed in these tests that
were free to expand due to the pin-roller support conditions at the ends of the specimen.
Accurately simulating this restraint to frame expansion in a subassembly test is difficult and
was not attempted in this test program or in other cruciform subassembly test programs. In
specimen 1S, the tensile resistance of the decking provided significant restraint against
expansion of the system, and thus the lack of external restraint to frame expansion is believed
to have had less impact. In the other specimens, however, as the compressive elements of
the connection (the bearing of the concrete slab on the column face and the binding of the
bottom girder flange on the column) had much greater stiffness and strength than the tensile
ones (primarily the angles), significant frame expansion occurred, with the girders pulling
further away from the column as the test progressed. In a full structure, the neighboring bays
would likely limit this frame expansion, potentially improving the flexural capacity of the
connection. The effect of adding such restraint to the connection will be examined in future
tests.
5 ACKNOWLEDGEMENTS
Funding for this test program was provided by the National Science Foundation (Award
#CMMI 1344592) through the Network for Earthquake Engineering Simulation. Any opinions,
findings, conclusions, and recommendations presented in this paper are those of the authors
and do not necessarily reflect the views of the sponsors.
6 REFERENCES
[1] Anderson, J.C. et al. (1995) Case Studies of Steel Moment Frame Performance in the
Northridge Earthquake of January 17th, 1994. Report No. SAC 95-07. SAC Joint
Venture.
[2] Foutch, D.A. and Yun, S.Y. (2002). “Modeling of steel moment frames for seismic loads,”
Journal of Constructional Steel Research, Vol. 58, pp. 529-564.
[3] Ji, X., Kato, M., Wang, T., Hitaka, T., Nakashima, M. (2009). “Effect of gravity columns
on mitigation of drift concentration for braced frames,” Journal of Constructional Steel
Research, Vol. 65, pp. 2148-2156.
[4] Liu, J. and Astaneh-Asl, A. (2000). “Cyclic Testing of Simple Connections Including
Effects of Slab.” Journal of Structural Engineering, Vol. 126, No. 1, pp. 32-39.
[5] Leon, R. T. (1990). “Semi-rigid Composite Construction.” Journal of Constructional Steel
Research, Vol. 15, No. 1-2, pp. 99-120.
[6] Azizinamini, A and Radziminski, J. (1989). “Static and Cyclic Performance of Semirigid
Steel Beam-to-Column Connections,” Journal of Structural Engineering, Vol. 115, No.
12, pp. 2979-2999.
[7] Clark, P., Frank, K. Krawinkler, H., and Shaw, R., 1997. “Protocol for Fabrication,
Inspection, Testing and Documentation of Beam-Column Connection Tests and Other
Experimental Specimens,” SAC Steel Project Background Document. October, Report
No. SAC/BD-97/02
[8] Liu, J. and Astaneh-Asl, A. “Moment-Rotation Parameters for Composite Shear Tab
Connections,” Journal of Structural Engineering, Vol. 130, No. 9, pp. 1371-1380.
[9] Rogers, C. A., and Tremblay, R. (2003). “Inelastic Seismic Response of Frame
Fasteners for Steel Roof Deck Diaphragms.” Journal of Structural Engineering, Vol. 129,
No. 12 pp. 1647-57.
[10] Donahue, S., Engelhardt, M., Helwig, T., Clayton, P., Williamson, E. (2018) “NEES
Planning–Role of Gravity Framing in Seismic Performance of Steel Buildings,”
DesignSafe-CI [publisher], Dataset, doi:10.17603/DS2M38H.
A DESIGN APPROACH FOR THE SERVICEABILITY LIMIT STATE
OF COMPOSITE STEEL-CONCRETE SLABS
Gianluca Ranzi
School of Civil Engineering
The University of Sydney
Sydney, Australia
gianluca.ranzi@sydney.edu.au
ABSTRACT
Composite floors are a well‐established cost‐efficient and effective form of construction. These
flooring systems include thin‐walled profiled steel sheeting, concrete and steel reinforcement.
This paper presents a design model for the evaluation of the service deflections of composite
floor slabs that account for the deformations induced by instantaneous loads, creep and
shrinkage effects. In the latter case, the model considers the occurrence of a non‐uniform
shrinkage profile that develops through the slab thickness due to the inability of the concrete to
dry from its underside because of the presence of the profiled sheeting. The proposed
approach is validated against experimental measurements available in the literature for full‐
scale long‐term tests of composite slab samples.
INTRODUCTION
Composite steel-concrete slabs are widely used for building floors and include thin-walled
profiled steel sheeting, concrete and steel reinforcement (Figure 1). Their design is usually
governed by serviceability limit state requirements associated with deflections (Ranzi et al,
2013a). At present, available design codes provide different design recommendations for this
limit state condition and, in some cases, deflection checks can be omitted if the slabs fall within
specific requirements, such as specified span-to-depth ratios (e.g. EN 1994-1-1, 2004). Usually,
engineers have accounted for the time-dependent behavior of the composite slabs based on
design guidelines available for concrete floors, e.g. (EN 1992-1-1, 2004; AS3600, 2009). In the
latter case, the shrinkage strain to be used in design is determined assuming a hypothetical
thickness th equal to about twice the slab thickness to account for the sealing action provided by
the presence of the profiled sheeting (i.e. equal to th = 2Ag/u, where Ag represents the gross
area of the concrete component and u depicts the exposed perimeter) and the calculated
shrinkage is then assumed to remain constant over the slab thickness for the service analysis
and calculations.
In this context, this paper presents a design model for the serviceability limit state of composite
steel-concrete slabs that has been included in the final draft of the Australian composite steel-
concrete standard expected to come out in 2017 (AS2327, 2017). The model builds on recent
experimental observations that observed a non-uniform shrinkage profile to occur through the
thickness of a composite slab due to the sealing action provided by the presence of the profiled
steel sheeting that prevents moisture egress from the slab underside. The first part of the paper
provides an overview of the time-dependent behavior of concrete relevant to composite floor
systems and it is followed by a presentation of the key features of the design model. The design
approach is then validated against the experimental measurements performed on the long-term
tests of composite steel-concrete slabs reported in (Gholamhoseini et al 2014, Gholamhoseini
2014).
Concrete slab
Steel reinforcement
Profiled steel sheeting
Fig. 1 – Typical composite steel-concrete slab
TIME-DEPENDENT BEHAVIOUR OF THE CONCRETE

The time-dependent behaviour of concrete, such as creep and shrinkage, produces variations in
deformations and stresses over time. In particular, when concrete is subjected to a sustained
stress it undergoes deformations which increase with time. There are two limit conditions which
are usually referred to when dealing with this behaviour, namely the creep and relaxation
problems. The former relates to the case in which a concrete member is loaded with a constant
sustained load and, being unreinforced and unrestrained, is free to deform with time. The
consequent time-dependent deformations are accompanied by no changes in stress. The
relaxation problem defines the condition in which a concrete component is subjected over time
to a constant deformation (i.e. constant strain) and, due to creep, reduces its time-dependent
stress state over time. Shrinkage is a stress-independent effect and leads to significant volume
changes which, if restrained, can induce the development of internal stresses and possible
occurrence of cracking (CEB, 1984; Gilbert and Ranzi, 2011). It is usually convenient to express
the total deformation exhibited in an uncracked uniaxially loaded concrete element at an instant
in time t as the sum of an instantaneous strain component e(t0), which depicts the elastic strain
occurred at the time of loading t0, a creep strain cr(t,t0) and a shrinkage strain sh(t). Based on
this, the concrete strain developed in a point of a structure at constant temperature can be
calculated using:
 t    e t0    cr t , t0    sh t  (1)
in which it is assumed that these strain contributions can be conveniently treated independently
from each other, even if this might not be strictly the case in reality.
The extent of the time-dependent behaviour of concrete depends on its mix design, dimensions
and geometry of the component considered, as well as the environmental conditions. A
convenient parameter used to capture the tendency of concrete to creep is referred to as the
creep coefficient and is defined as the ratio between the creep and instantaneous strains
induced by a sustained stress at time t first applied at time t0:
 cr t , t0 
 t , t0   (2)
 e t0 
For compressive stress levels smaller than half of the compressive concrete strength, creep
deformations are approximately proportional to the stress and this representation is commonly
referred to as linear creep. For higher stress levels creep increases at a faster rate and
becomes nonlinear with respect to stress (CEB, 1984; Gilbert and Ranzi, 2011).
There are different components that contribute to shrinkage effects, such as plastic shrinkage,
autogenous shrinkage, thermal shrinkage and drying shrinkage. Among these, the drying
component is of particular interest to composite floor behaviour considered in this paper and it
represents the reduction in volume caused principally by the loss of water during the drying
process. Figure 2a shows the shrinkage strains through the thickness of an unloaded and
unrestrained plain concrete slab that is drying on both its top and bottom surfaces. The mean
shrinkage strain εsh represents the average contraction. The strain marked Δεsh is the portion of
the shrinkage strain that causes the internal stresses required to restore strain compatibility
(Figure 2b). These self-equilibrating internal stresses occur in all concrete structures and are
tensile near the drying surface and compressive in the interior of the member. Because the
shrinkage-induced internal stresses develop gradually with time, they are relieved by creep.
Nevertheless, soon after the commencement of drying, the tensile stresses near the drying
surfaces may overcome the tensile strength of the concrete and result in surface cracking. The
total strain distribution, obtained by summing the elastic, creep and shrinkage components, is
linear, as shown in Figure 2c (e.g. Gilbert and Ranzi, 2011).
Recent work carried out on composite floor systems has highlighted the occurrence of a
shrinkage gradient through the depth of the slab. In particular, this behavior has been first noted
in long-term experiments performed on composite steel-concrete beams prepared with a
composite slab (Al-Deen et al 2011). Subsequent work has dealt with the service behaviour of
composite floor slabs and post-tensioned composite slabs, e.g. (Shayan et al 2010, Ranzi et al
2013b, Al-deen et al 2015, Gholamhoseini et al 2014). Modelling work in this area has also
been reported in the literature and focused on the inclusion of shrinkage gradients in the service
representation of the composite floors, e.g. (Ranzi and Vrcelj 2009, Bradford 2010, Ranzi et al
2013b, Gilbert et al 2012, Al-Deen and Ranzi 2015).
εsh εsh
top
tensile compressive
bottom
(a) Shrinkage strain of an unreinforced, (b) Elastic and creep strain caused by internal
unloaded and unrestrained concrete sample stresses
εsh
(c) Total strain caused by shrinkage.

Fig. 2 – Development of shrinkage through slab thickness.
The experimental characterization of the non-uniform shrinkage profile has been carried out by
monitoring the total deformations that take place over time in isolated unreinforced concrete
samples thought the thickness of the slab, e.g. (Al-Deen et al 2011, Gilbert et al 2012, Ranzi et
al 2013b). In this work, companion samples with different exposure conditions have been
prepared and monitored, i.e. with one exposure condition allowing the concrete to dry from both
sides of the slab sample (Figure 3a) and a second exposure condition exposing only one side
for drying and sealing the opposite slab surface (Figure 4a). Qualitative representations of the
total deformations recorded during these tests for the two exposure conditions are outlined in
Figures 3b and 4b. In particular, Figure 3b depicts constant deformations taking place through a
slab exposed on both sides and varying over time. Under the simplifying assumptions of linear
shrinkage profiles (commonly adopted for reinforced concrete design), the total deformations
depicted in Figure 3b can be assumed to represent the shrinkage deformations. In reality, the
shrinkage distribution is non-uniform and produces self-equilibrating stresses, e.g. (Gilbert and
Ranzi 2011). In the case of a slab exposed on one side only (Figure 4), a strain gradient
develops through the cross-section over time. Relying on the simplifying assumption of linear
shrinkage distributions (acceptable for design purposes), it is possible to assume that the
measured total deformations correspond to the shrinkage profile. Also in this case, the real
shrinkage distribution is more complex. Obviously, higher order polynomials, or other functions,
could be considered for refined service calculations while it is assumed that a linear shrinkage
distribution is sufficient for routine design purposes.
Moisture egress
Moisture egress
(a) Drying conditions of typical concrete slab exposed on both sides.
Slab
height
Time
Shrinkage
(b) Qualitative shrinkage distribution of typical concrete slab exposed on both sides
(plotted assuming a linear shrinkage profile)
Fig. 3 – Typical concrete slab exposed on both sides.
DESIGN MODEL
The main features of the model proposed for the serviceability limit state design of composite
slabs associated with the deflection requirements is outlined in this section and further details
can be found in references (AS2327, 2017; Ranzi, 2017). The proposed procedure is suitable
for composite floors whose service behavior is not affected by the partial interaction between
the profiled steel sheeting and the concrete slab.
The total deflection of a composite slab  can be by combining the contributions of the
components associated with the instantaneous deflection 0 and the deflections associated with
creep (cc) and shrinkage (cs) effects:
   0   cc   cs (3)
Moisture egress
No moisture egress
(a) Drying conditions of typical concrete slab exposed on one side only.
Slab
height
Time
Shrinkage
(b) Qualitative shrinkage distribution of typical concrete slab exposed on one side only
(plotted assuming a linear shrinkage profile)
Fig. 4 – Typical concrete slab exposed on one side only.
The instantaneous deflection 0 is calculated considering an effective flexural rigidity Ec Ief, in

which Ec depicts the modulus of elasticity of the concrete and Ief represents the effective second
moment of area usually available in concrete design standards. For the purpose of this paper,
the following expressions is provided: (AS2327, 2017; Ranzi, 2017)
3
M 
I ef  I cr  I uncr  I cr  cr   I ef . max (4)
 Ms 
where Iuncr and Icr are the second moment of area of the uncracked and cracked composite slab
sections, respectively, with the steel reinforcement and steel sheeting transformed to an
equivalent area of concrete, Ms is the maximum in-service moment resisted by the cross-
section, Ief.max represents the maximum effective second moment of area corresponding to the
uncracked properties, and Mcr denotes the cracking moment and is determined as:
M cr  Z  f ct . f   cs   0 (5)
where f ct . f defines the characteristic flexural tensile strength of concrete, Z is the section
modulus of the uncracked section referred to the extreme tensile fibre, and cs is the maximum
shrinkage-induced tensile stress at the extreme fibre of the uncracked section.
The creep deflection component cc is determined based on the effective modulus method, e.g.
(Gilbert and Ranzi 2011), and is calculated considering the creep deformations produced by the
sustained part of the service load. For this purpose, the instantaneous deflection produced by
the sustained service load is denoted as 0.sus and the creep deflection cc can then be
determined as follows:
 cc   0. sus cc (6)
and
Ec I ef
 cc  1 (7)
Eef.cc I ef.cc
where Eef.cc equals Ec / (1+cc), Ief.cc represents the effective second moment of area (calculated
using the format of Equation 4 with the concrete effective modulus Eef.cc) and cc is the concrete
creep coefficient.
The deflection component due to shrinkage cs is evaluated considering the effects of the
following shrinkage-induced curvature cs on the overall member response:
 cs  1   cs  cs ,cr   cs cs ,uncr (8)
and
2
M 
 cs   cr   1 (9)
 Ms 
where cs.uncr is the shrinkage-induced curvature over an uncracked section, and cs.cr depicts
the shrinkage-induced curvature over a cracked section. The cross-sectional properties used for
the calculation of Equation 8 are based on an effective modulus for the concrete Eef.cs (= Ec /
(1+0.55cc)).
For design purposes, the shrinkage gradient to be used for routine design can be expressed
(under the simplifying assumptions of a linear shrinkage profile as discussed in the previous
section) as a function of a reference shrinkage εsh calculated in accordance with reinforced
concrete guidelines, e.g. (EN 1992-1-1, 2004; AS3600, 2009), considering a hypothetical
thickness equal to the thickness of the composite slab and assuming both sides of the slab to
be exposed. The shrinkage gradient is then assumed, for the purpose of the design model, to
vary from a value of 0.2 εsh at the base of the slab to 1.2 εsh at the top surface of the slab. The
selection of the reference strain εsh is a consequence of the fact that the top and bottom values
of the non-uniform shrinkage profile have been calibrated experimentally against those exhibited
by companion concrete slab samples exposed on both sides with same thickness as the
composite specimens, e.g. (Al-deen et al 2011, Al-deen and Ranzi 2015). The adopted
shrinkage distribution is a simplification of the real shrinkage behaviour that takes place in the
slab over time. More refined representations could be implemented to describe the shrinkage
distribution.
VALIDATION
The model described in this paper is validated in the following against the experimental long-
term tests on composite slabs reported in (Gholamhoseini et al 2014, Gholamhoseini 2014). For
this purpose, five representative test results have been considered in the following of which two
samples (referred to in the following as 1LT-70-0 and 1LT-70-6) were cast on KF70 trapezoidal
profiled steel sheeting (Fielders 2008) with a trough height of about 70 mm and the remaining
three specimens (denoted as 6LT-40-0, 9LT-40-6 and 10LT-40-6) were prepared using KF40
trapezoidal profiled steel sheeting (Fielders 2008) with a 40 mm trough height. These samples
were subjected to different loading histories. In particular, specimen 1LT-70-0 was placed in a
simply-supported configuration at 7 days from casting and maintained unloaded for the entire
duration of the long-term test (i.e. 247 days), as shown in Figure 5a. Sample 4LT-70-6 followed
the same loading history of specimen 1LT-70-0 till 64 days, after which it was subjected to an
external load of 6.1 kPa and increased to 7.9 kPa at 197 days from casting (Figure 5b).
Specimen 6LT-40-0 was kept unloaded for the entire duration of the long-term experiment (i.e.
223 days) as depicted in Figure 6a. Samples 6LT-40-9 and 6LT-40-9 were loaded at 28 days
from casting with an external sustained load of 6.4 kPa as shown in Figure 6b.
Figures 5 and 6 illustrate the comparisons carried out between the measured deflections and
the values calculated with the proposed approach. In all comparisons, good predictions have
been observed when considering the numerical values obtained based on the non-uniform
shrinkage profile while the use of the uniform shrinkage underestimated the experimental
deflections.
CONCLUSIONS
This paper presented a model for the serviceability limit state design of composite steel-
concrete slabs associated with the deflection calculations. The proposed approach builds on
recent experimental observations that identified the occurrence of a shrinkage gradient through
the thickness of composite slabs due to the inability of the slab to dry from its underside
because of the presence of the profiled steel sheeting. The main features of the design model
have been presented placing particular attention at the occurrence shrinkage effects. The
proposed approach has then been validated against representative long-term measurements
reported in the literature on composite steel-concrete slabs. The comparisons presented
between experimental deflections and numerical values highlighted the need to account for the
non-uniform shrinkage profile for a good representation of the long-term response, while the use
of the constant shrinkage profile underestimated the experimental results.
Experiment (1LT‐70‐0)
Design model with shrinkage gradient
Design model with uniform shrinkage
5
 (mm)
4
3
2
1
Time (days)
0
0 50 100 150 200 250
(a) Sample 1LT-70-0
8
 (mm)
6
2
Time (days)
0
0 50 100 150 200 250
(b) Sample 4LT-70-6
Figure 5. Comparisons between experimental measurements and calculated values.
ACKNOWLEDGEMENTS
The contribution of the work reported in this paper was supported by the Australian Research
Council through its Future Fellowship scheme (FT140100130). Support from the University of
Sydney (Materials & Structures Research Cluster) is also gratefully acknowledged.
6
 (mm)
5
4
3
2
1
Time (days)
0
0 50 100 150 200 250
(a) Sample 6LT-40-0
Experiment (9LT‐40‐6, 10LT‐40‐6)

10
 (mm)
8
6
4
2
Time (days)
0
0 50 100 150 200 250
(b) Samples 9LT-40-6 and 10LT-40-6
Figure 6. Comparisons between experimental measurements and calculated values.
REFERENCES
Al‐Deen S and Ranzi G (2015) Effects of non‐uniform shrinkage on the long‐term behaviour of composite
steel‐concrete slabs. International Journal of Steel Structures 15(2): 415‐432.
Al‐Deen S, Ranzi G and Uy B (2015) Non‐uniform shrinkage in simply‐supported composite steel‐
concrete slabs, Steel and Composite Structures 18(2): 375‐394.
Al‐deen S, Ranzi G and Vrcelj Z (2011) Full‐scale long‐term and ultimate experiments of simply‐supported
composite beams with steel deck. Journal of Constructional Steel Research 67(10): 1658‐1676.
AS2327 (2017) Australian Standard for Composite Structures AS32327‐2017. Final draft. Standards
Australia.
AS3600 (2009) Australian Standard for Concrete Structures AS3600‐2009. Standards Australia.
Bradford MA (2010) Generic modelling of composite steel‐concrete slabs subjected to shrinkage, creep
and initial thermal strains including partial interaction. Engineering Structures 32: 1459‐1465.
CEB (Comité Euro‐International du Béton) (1984) CEB Design manual on structural effects of time‐
dependent behaviour of concrete. Edited by Chiorino MA, Napoli P, Mola F, Koprna M. Georgi
Publishing.
EN 1992‐1‐1 (2004) Eurocode 2: Design of concrete structures ‐ Part 1‐1: General rules and rules for
buildings, European Committee for Standardization.
EN 1994‐1‐1 (2004) Eurocode 4: Design of composite steel and concrete structures – Part 1.1: General
rules and rules for buildings, European Committee for Standardization.
Gholamhoseini A (2014) Time‐dependent behaviour of composite concrete slabs. PhD Thesis, The
University of New South Wales, Australia.
Gholamhoseini A, Gilbert RI, Bradford MA and Chang ZT (2014) Time‐dependent deflection of composite
concrete slabs. ACI Structural Journal 111(4): 765‐776.
Gilbert RI, Bradford MA, Gholamhoseini A and Chang Z‐T (2012) Effects of Shrinkage on the Long‐Term
Stresses and Deformations of Composite Concrete Slabs. Engineering Structures 40: 9‐19.
Gilbert RI, Ranzi G. (2011) Time‐dependent behaviour of concrete structures. Spon Press.
Ranzi G (2017) Service design approach for composite steel‐concrete floors. Proceedings of the
Institution of Civil Engineers, Structures and Buildings, http://dx.doi.org/10/1680/jstbu.16.00196.
Ranzi G, Leoni G and Zandonini R (2013a) State of the art on the time‐dependent behaviour of composite
steel‐concrete structures. Journal of Constructional Steel Research 80: 252‐263.
Ranzi G, Al‐Deen S, Ambrogi L and Uy B (2013b) Long‐term behaviour of simply‐supported post‐
tensioned composite slabs. Journal of Constructional Steel Research 88: 172‐180.
Ranzi G and Vrcelj Z (2009) Closed form solutions for the long‐term analysis of composite steel‐concrete
members subjected to non‐uniform shrinkage distributions. Proceedings of the 5th International
Conference on Advances in Steel Structures (ICASS’09), Hong Kong 16‐18 December 2009.
Shayan S, Al‐Deen S, Ranzi G and Vrcelj Z (2010) Long‐term behaviour of composite concrete slabs: an
experimental study. Proceedings of the 4th International Conference on Steel & Composite Structures
(ICSCS’10) Sydney 21‐23 July 2010.
Fielders Australia (2008) Specifying Fielders – Kingflor – Composite steel formwork system design
manual, Fielders Australia Pty Ltd.
LONGITUDINAL SHEAR RESISTANCE OF COMFLOR 210
Roland Abspoel
Delft University of Technology, Faculty of Civil Engineering and Geosciences
Delft, The Netherlands
r.abspoel@tudelft.nl
Jan Stark
Delft University of Technology, Faculty of Civil Engineering and Geosciences
janstark@planet.nl
ABSTRACT
The longitudinal shear resistance of steel-concrete floors can only be determined with full scale
experiments since no theoretical model is available. The properties of the steel decking may
vary over time due to wear and maintenance of the rollers. Retesting of the composite slabs is
therefore necessary at regular intervals.
The longitudinal shear resistance can be determined by using two methods, namely the m-k
method and the τu.Rd method. Although the deep deck ComFlor 210 is not within the scope of
EN1994-1-1 the test procedure in Annex B is used. Two series of three specimens were tested.
Three additional tests were performed to get information about the effective cross-sectional area
of the sheeting, the influence of the rib reinforcement and the influence of end anchorage.
The results of laboratory tests by O’Leary and Duffy (1992), the Salford experiments, are re-
evaluated in line with the Delft Experiments to compare both results.
INTRODUCTION
Full scale tests to investigate longitudinal shear resistance of ComFlor 210 were conducted in
the nineties of the twentieth century. Because of the large time lap and because of possible
variations of the rolling equipment the manufacturer, Tata Steel UK, wanted to confirm the
design information. Therefore new tests were conducted in the Stevin II laboratory of the Delft
University of Technology.
In EN1994 a subdivision of composite slabs is given based on the spacing of the webs. EN1994
covers only narrow spaced webs which are defined by the ratio br / bs, where br is the width of
the upper flange and bs is the distance between centers of adjacent ribs. The recommended
value for the upper limit of this ratio is br / bs ≤ 0.6. ComFlor 210 has a ratio br / bs = 0.71 > 0.6
and is consequently not covered in EN1994. Although ComFlor 210 is not within the scope of
EN1994-1-1 the test procedure in Annex B of EN1994-1-1 is used as far as possible. In Figure 1
the ComFlor 210 decking system is shown.
Fig. 1 – ComFlor 210
According EN1994-1-1 (Eurocode 4) the longitudinal shear resistance can be defined by using
one of two different methods, the m-k method, developed in the USA, and the u.Rd method,
developed in Europe. The u.Rd method is only suitable for slabs with ductile connection. The m-
k method can be used for slabs with either ductile or non-ductile connections. The u.Rd method
can take into account both end anchorage and rib reinforcement. Two series of three tests with
a different span were performed, series A with a nominal span of 3.60 m and Series B with a
span of 5.4 m.
The methods result in properties of the longitudinal shear for the specific tested slabs, but also
for slabs consisting of the same steel sheet, but with higher material properties or larger
dimensions. The longitudinal shear is based on interconnection of the steel sheeting and the
concrete topping, so without reinforcement in the ribs. Both methods are covered in EN1994
and each method has advantages and disadvantages. The largest advantage of the τu.Rd
method is that it is designed to take into account the influence of end-anchorage and rib
reinforcement. A minimum of six test specimens are required for the m-k method, subdivided
into two groups of three specimens with different lengths, and for the τu.Rd method a limited
number of four specimens is required and so these tests are combined.
Three additional test specimens were tested to verify the influence of the rib reinforcement and
the effect of end-anchorage on the longitudinal shear resistance. ComFlor 210 in practice is
always with rib reinforcement. This paper shows the test specimens, the test rig and the test
results of the nine test specimens. Particular attention is given to the evaluation procedure for
the m-k method in case of rib reinforcement and the influence of large areas of embossments
and indentations. In this paper the results are presented of a test program carried out in the
Stevin II laboratory of the Delft University of Technology for the re-evaluation of the longitudinal
shear resistance of ComFlor 210 deep decking profile.
TEST SETUP
The test set-up was designed according to Annex B of EN1994-1-1. Two line loads, located at
L/4 and 3L/4 on the span, were applied to the specimen by using a spreader beam. Load cells
were used for every support of the three ribs instead of the line support according Annex B.
The test loading procedure consisted of an initial cyclic test and a subsequent deformation
controlled loading test in accordance with EN1994-1-1, Annex B clause B.3.4.
TEST SPECIMENS
To determine the m and k values at least 6 tests have to be performed and for the u.Rd a
minimum of 4 tests are necessary. In the Delft experiments, the 6 tests were used for the
evaluation of both methods. With the 3 additional specimens a total of 9 specimens were
prepared and tested, see Figure 2:
 6 for the determination of the m and the k values and the u.Rd values
o 3 specimens with L = 3.6 m
 3 additional specimens
o 1 specimen with L = 7.2 m intended to determine the sagging moment resistance.
 1 with a practical support detail consisting of concrete end beams at both
supports of the specimen
 1 without rib reinforcement
Fig. 2 - Test rig including test specimen during testing
It was expected that specimen 7 with a span of 7.2 m and specimen 8 with the concrete beams
as supports would fail in flexure, while the other test specimens would fail in longitudinal shear.
All test specimens have a width of 1.80 m with 3 profile ribs, so consisting of 3 steel sheets of
600 mm width. One of the sheets is cut half and both halves are used for the longitudinal edges.
Load introducers are used at the places of load introduction, so the shear length is well defined.
The nominal thickness of the sheet is 0.96 mm without a zinc layer on both sides. The average
actual thickness measured is 0.954 mm. The sheeting has transverse embossments in the top
flange and the webs (see figure 1). The total length of the embossments per sheet is 577 mm.
According to EN1994-1-1 clause 9.7.2 the length of the embossments perpendicular to the rib
should be completely neglected for the effective cross-sectional area. Using this rule the
effective cross-sectional area is 403.54 mm2/rib or 672.57mm2/m.
The strains were measured during testing by 21 strain gauges onto one profiled sheet of
specimens 1 and 9 (without rib reinforcement) of the Delft experiments with a span of 3.6 m and
7 strain gauges for specimen 7 with a span of 7.2 m. Based on the measurements of the sheets
with 21 strain gauges, Van Erp (2016) concluded that the effective area is close to 560 mm2/rib
or 933 mm2/m.
The mesh reinforcement is #8-150 for all specimens and the rib reinforcement is Ø12 for all
specimens, except specimen 9, which is tested without rib reinforcement. The test specimens
were assembled in Zoeterwoude and after hardening of the concrete, transported to the Stevin
II laboratory in Delft. The geometry of the specimens and the instrumentation are presented in
Figure 3.
0.3 m
0.6 m
b = 1.8 m
0.6 m
0.3 m
Fig. 3 - Plan and cross-section of the test specimens, including the instrumentation
To determine the actual material properties coupon tests were carried out for the profiled
sheeting, the mesh reinforcement and the reinforcement bars in the ribs. The sheeting is roll
formed from one coil material and 7 coupons were tested. The coupons were cut from the
effective areas of the sheeting. The average lower yield stress is fy.p = 407.10 MPa and the
average ultimate stress fu.p = 471.21 MPa. The average thickness of the test specimens is
t  0.954 mm without taken into account both zinc layers. The mesh reinforcement in the
concrete topping has a diameter of #8-150 with an average yield stress fy.r8 = 606.69 MPa and
an average ultimate stress fu.r8 = 612.41 MPa. The rib reinforcement is Ø12-600 with an average
yield stress fy.r12 = 564.32 MPa and an average ultimate strength fu.r12 = 600.21 MPa. The
average diameter is 11.85 mm. The concrete properties were determined by testing 3 or 4
cubes of a total of 40 at the day the slabs were tested. The average strength of these 3 a 4
cubes tested are shown in Table 1.
RESULTS OF THE DELFT EXPERIMENTS

The test specimens were tested according to the requirements in Annex B of EN1994-1-4.
Table 1 shows the specimens and their average dimensions, the span L, the width b and the
height hav. Next to this the average concrete strength and the measured reaction forces,
respectively the total reaction force under loading Rtot, the total self-weight including, Rsw.tot, and
Rsw excluding the weight of the spreader beam used in all tests.
Table 1 – Test results of the Delft experiments specimens 1 to 9
Test L b hav fc Rtot Rsw.tot Rsw
[mm] [mm] [mm] [MPa] [kN] [kN] [kN]
1 3600 1831 274.0 26.42 263.05 30.81 20.13
2 3614 1827 269.0 24.41 256.91 27.94 17.26
3 3603 1831 266.5 23.44 260.07 28.44 17.76
4 5390 1836 269.3 27.92 182.81 37.39 26.71
5 5388 1829 267.5 28.61 178.21 37.59 26.91
6 5394 1826 267.5 31.05 185.03 36.44 25.76
7 7183 1838 271.0 31.75 136.80 46.60 35.92
8 3597 1815 280.0 31.35 306.18 30.86 20.18
9 3599 1824 271.0 29.97 157.61 29.04 18.36
LONGITUDINAL SHEAR RESISTANCE

The longitudinal shear resistance of a composite slab is based on the composite behavior of a
part of the concrete and the steel sheeting. The influence of the rib reinforcement is not taken
into account and so this influence during testing has to be deduced from the results. For the
Vt
ratio as used in this m-k method, the lever arm zp is not used, but instead the height
b  dp
measure dp, see Figure 4, the measure distance of the centroid of the steel sheets and the
extreme fiber of the composite slab in compression.
Fig. 4 - Stress distribution composite slab Fig. 5 - Loading scheme of the slab
It seems obviously to deduce the influence of the rib reinforcement on the reaction force Vt by
using the measure ds, the distance of the centroid of the rib reinforcement and the extreme fiber
of the composite slab in compression. The contribution of the rib reinforcement bar with a
diameter of 12 mm to the bending moment resistance is in this case:
(1) M r 12  Nr 12  d s
The force in the rib reinforcement bars is Nr12 = 186.35 kN, based on an average actual
diameter of 11.85 mm and an average yield strength of 564.32 MPa. The reaction force Vt per
support can to be determined as follows:
Rtot M r 12
(2) Vt  
2 L
4
For the partial shear interaction method it is necessary to determine the longitudinal shear Nc ,
but this force is not measured during the experiments. Nevertheless, the reduced bending
moment resistance of the slab, based on partial longitudinal shear, but without the influence of
the rib reinforcement, is based on:
 Nc 
(3) Mu.Rd  Nc  z  1.25  M pl .se   1  
 Ape  fy . p
 
This reduced bending moment resistance M u .Rd is also determined by the measured total
reaction force Rtot times the distance Ls from the support to the load introduction minus the
moment by the dead load of the slab, see Figure 5, in which L0 = 50 mm for all specimens:
Rtot 1
 Ls   qDL   Ls  L0 
2
(4) Mu.Rd 
2 2
With qDL the dead load of the slab, based on the reaction force of the total self-weight Rsw.tot
minus the weight of the spreader beam Rsw.sb:
Rsw
(5) qDL 
L  2  L0
The lever arm z between the effective height of the concrete and the centroid of the longitudinal
shear force is based on EN1994-1-1:
x Nc
(6) zh  ep   ep  e  
2 Ape  fy . p
With e the centroid of the steel sheeting, ep the plastic neutral axis of the effective cross-
sectional area Ape and the centroid of the effective height x of the concrete. This formula is a
simplification, because the lever arm will be in between the centroid e and the plastic neutral
axis ep. Many researchers calculated the effective height of the concrete x based on the
reinforcement force, but this has to be based on the longitudinal shear Nc and the force Nr12 in
the reinforcement with nominal diameter 12 mm:
Nc  Nr 12
(7) x
b  0.85  fc
The plastic bending moment resistance of the effective areas of the steel sheeting is:
(8) M pl . pe .Rd  W y . pe . pl  fy . p
Based on above equations the longitudinal shear force Nc has to be determined, in this case by
using a solver in the Excel spreadsheet program. For the nominal span of 3.6 m, the longitudinal
shear force Nc is solved for the effective area Ape  1210.63 mm 2 over the width of 1.8 m or
maybe better, for the three ribs used. The plastic section modulus is W pl . pe  94663 mm 3 over
the width of the slab or W pl . pe  31554 mm 3 per rib.
The plastic bending moment resistance of the composite slab without reinforcement is based on
yielding of the effective cross-sectional area of the sheeting by taken into account the nominal
effective cross-sectional area Ape = 406.08 mm2 per rib, so:
(9) Nc  Ape  fyp  406.08  407.1 103  165.32 kN
The effective height x of the concrete is based on only axial resistance Nc and by taken into
account the average strength fc.av of the concrete:
Nc 165.32  103
(10) x   12.0 mm
b  0.85  fc 600  0.85  27.0
The lever arm zs is calculated as follows:
x Nc 12.0
(11) zs  h   ep   ep  e    270   95  169 mm
2 Ape  fyp 2
And so:
(12) M pl .Rd  Nc  zs  165.32  0.169  27.94 kNm per rib
In contradiction to the m-k method in which the gross cross-sectional area Ap is taken into
account, in this partial shear resistance the effective cross-sectional area Ape is used. EN1994-
1-1 requires to neglect the areas with the embossments. This means that the effective cross-
sectional area for this deep deck ComFlor 210 with relative large embossments is rather small,
compared with the gross cross-sectional area. In shallow deck profiles other types of
embossments are commonly used and so their effective area is rather large, when compared to
their gross cross-sectional area.
DETERMINATION OF THE m- AND k-VALUES

DELFT EXPERIMENTS
Based on the results of the laboratory tests the value of m and k have to be determined. The
vertical shear as effect of the load is checked to the requirement as given in EN1994-1-1:
b  d p  m  Ap 
(13) VEd  Vt .Rd   k
 vs  b  Ls 
Ap
The characteristic m- and k-values are determined from two parameters, which relates
b  Ls
the gross cross-sectional area Ap of the sheeting to the dimensions of the test specimens and
Vt
which relates the reaction force Vt to the dimensions of the test specimen. The m- and k-
b  dp
values are based on the regression line through 90% of the minimum values of the two series of
test specimens for the scenarios that the deviation from the average value is less than 10%.
The specimens of series A have a nominal span of 3.6 m and the specimens of series B have a
nominal span of 5.4 m. For the six mentioned specimens of the Delft experiments these results
of the calculations are summarized in Table 2.
Table 2 – Results for the m-k method of the Delft experiments of specimens 1 to 6
Test ds Mr12 dp Vt Ap Vt
[mm] [kNm] [mm] [kN]
b  Ls b  dp
[-] [MPa]
1 219.0 40.88 179.0 86.10 0.00174 0.263
2 214.0 39.95 174.0 84.24 0.00174 0.265
3 211.5 39.48 171.5 86.21 0.00175 0.275
4 214.3 40.00 174.3 61.72 0.00116 0.193
5 212.5 39.67 172.5 59.66 0.00117 0.189
6 212.5 39.67 172.5 63.10 0.00117 0.200
Based on these results the value for m and k are determined, using the smallest values of the
first series of three test specimens, series A, and the second series of three test specimens,
series B, are used as underlined in the table:
 V 
  0.9  t 

(14) m 
b  dp   0.9   0.263  0.189   116.71 MPa
 A  0.00174  0.00117
 p 
 b  Ls 
The value k is determine by using the value m in Equation (14) in Equation (15):
 V   Ap 
(15) k  0.9   t
 bd   m     0.9   0.263   116.71  0.00174   0.0337 MPa
 p   b  Ls 
For the special test specimens 7 to 9 the results are shown in Table 3.
Table 3 - Results for the m-k method of the Delft experiments of specimens 7 to 9
Test ds Mr12 dp Vt Ap Vt
[mm] [kNm] [mm] [kN]
b  Ls b  dp
[-] [MPa]
7 216.0 40.32 176.0 45.95 0.00088 0.143
8 225.0 42.00 185.0 106.4 0.00125 0.315
9 - - 176.0 78.81 0.00175 0.245
SALFORD EXPERIMENTS
O’Leary and Duffy (1992) conducted similar tests on ComFlor 210 at Salford University. Series
A have a span of 5.0 m, total height 310 mm and reinforcement bar in the ribs of Ø12-600 and
series B have a span of 7.5 m, total height 300 mm and reinforcement bar in the ribs of Ø20-
600. Both reinforcement bars used are located at 70 mm from the bottom of the sheeting. The
width of all test specimens were 1.20 m. Figure 6 shows the geometry of the specimens.
L = 5.0 m / 7.5 m
L/8 L/4 L/4 L/4 L/8
0.6 m 0.032 m
b = 1.2 m
0.032 m 0.6 m
Fig. 6 - Plan and cross-section of the Salford test specimens
The test specimens are loaded according to BS 5950, which means that the load is applied with
four line loads and two spreader beams. According O’Leary and Duffy (1992), the shear length
is L/4, but this has to be changed into 3L/8, the location with the largest bending moment. The
values for m and k are determined similarly to previous calculations for the Delft experiments.
 V 
  0.9  t 
b  d p  0.9   0.108  0.063 
(16) m    128.3 MPa
 Ap  0.00085  0.00057
 
 b  Ls 
 Vt   Ap 
(17) k  0.9     m   0.9   0.108   128.3   0.00085   0.0167
 b  dp   b  Ls 
The value for m according to the Delft and Salford experiments are very similar, see Figure 7.
PARTIAL SHEAR METHOD

DELFT EXPERIMENTS
The second method given in EN1994-1-1 to determine the longitudinal shear is the partial shear
method, the τu.Rd method. For this method, the minimum number of specimens required is four.
One test specimen with a relative small span and three specimens with a relative large span.
The first test is to check that the slab behaves ductile and the three others are used to
determine τu.Rd according to EN1994:
  Nc .f
(18) u 
b   Ls  L0 
0,300
0,275
0,250
0,225
0,200
0,175
0,150
Delft experiments EN1994-1-1
Vt / b dp [MPa]
0,125 Salford experiments EN1994-1-1

Delft experiment
0,100
Salford experiments
0,075
0,050
0,025
0,000
0,0000
0,0001
0,0003
0,0004
0,0005
0,0006
0,0008
0,0009
0,0010
0,0011
0,0013
0,0014
0,0015
0,0016
0,0018
0,0019
0,0020
-0,025
-0,050
Ap / b Ls [-]
Fig. 7 - Results of the Delft and Salford experiments for the m-k method
EN1994-1-1 gives also the possibility to take into account the friction in longitudinal direction
due to the reaction force Vt:
  Nc .f    Vt
(19)  u. 
b   Ls  L0 
This shear stress  u.  according to Equation (19) will be smaller than  u according to Equation
(18). For the friction coefficient   0.5 is used. The characteristic value for τu.Rk is determined
according to EN1994-1-1 which refers to EN1990 Annex D:
(20)  u .Rk   u .Rm  1  k n  V 
With  u .Rk , the characteristic value of the longitudinal shear,  u.Rm the average longitudinal shear
related on the laboratory tests, k n is the coefficient that relates the average value to
characteristic based on the number of tests performed (with n = 3 is kn  3.37 ) and V is the
coefficient of variation for the longitudinal shear. The results for the partial shear resistance is
shown in Table 4. The design value for the longitudinal shear is determined with:
 u .Rk
(21)  u .Rd 
v
Table 4 - Results for the partial shear method of the Delft experiments of specimens 1 to 6
Sp qDL Mtot Nc x zs Mr12 zp Mu τu τu.
[kN/m] [kNm] [kN] [mm] [mm] [kNm] [mm] [kNm] [MPa] [MPa]
1 5.44 115.9 291.7 11.6 213.2 39.80 193.0 76.1 0.168 0.143
2 4.65 114.0 303.8 12.9 207.5 38.74 186.1 75.2 0.174 0.150
3 4.80 113.3 318.5 13.9 204.6 38.19 181.8 75.1 0.183 0.158
4 4.87 118.4 389.0 13.2 207.7 38.77 178.0 79.6 0.152 0.140
5 4.90 115.2 334.9 11.7 206.6 38.57 182.2 76.7 0.131 0.119
6 4.69 120.2 481.6 13.9 205.6 38.37 167.0 81.8 0.189 0.176
Additional to the above 6 test specimens as used for the m-k method, 3 more specimens are
tested, namely specimen 7 with a span of 7.2 m, specimen 8 with a concrete beam at the
supports as commonly used in practice and specimen 9 without rib reinforcement. The results of
these tests are shown in Table 5.
Table 5 - Results for the partial shear method of the Delft experiments of specimens 7 to 9
Sp qDL Mtot Nc x zs Mr12 zp Mu τu τu.
[kN/m] [kNm] [kN] [mm] [mm] [kNm] [mm] [kNm] [MPa] [MPa]
7 4.93 114.4 283.3 9.5 211.2 39.43 191.8 75.0 0.084 0.077
8 5.46 135.2 496.0 14.0 218.0 40.70 178.0 94.5 0.286 0.254
9 4.96 68.7 197.9 8.3 - - 200.7 68.7 0.114 0.091
The ultimate moment Mu of specimen 7 is of the same magnitude of this moment of specimens
1 to 6. This means that the contribution of the sheeting on the resistance is rather high, which is,
related to the height of the sheeting, expected.
Specimen 8, the one with concrete beams at the ends, an end anchorages of the deck, shows
that Mu is larger than the plastic bending moment resistance as determined in Equation (12),
based on nominal values. This means that the effective area Ape of the sheet has to be larger
than the effective area as defined in EN1994-1-1. Assuming that Mu = 94.5 kNm for specimen 8
is the plastic bending moment resistance, an effective cross-sectional area can be determined,
namely Ape = 436.05 mm2 per rib. Although the effective cross-sectional area increases less
than 10%, the longitudinal force Nc decreases much more than this 10% and so the shear stress
reduces as well. Table 6 shows the adapted longitudinal force Nc and the both shear stresses.
The characteristic shear stress is based on an average shear stress  u.Rm  0.127 MPa of
specimens 4 to 6, and a variation coefficient of V  0.015 MPa and kn  3.37 :
(22)  u .Rk   u .Rm  1  k n  V   0.127  1  3.37  0.015   0.121 MPa
Table 6 - Adapted results of the Delft experiments specimens 1 to 9

Sp Nc τu τu.µ Sp Nc τu τu.µ
[kN] [MPa] [MPa] [kN] [MPa] [MPa]
1 258.5 0.149 0.124 6 365.6 0.143 0.131
2 264.8 0.152 0.128 7 251.5 0.075 0.068
3 273.7 0.157 0.132 8 532.6 0.307 0.277
4 323.7 0.126 0.114 9 180.9 0.104 0.082
5 287.4 0.112 0.101
SALFORD EXPERIMENTS
To compare the partial shear resistance of the Delft and Salford experiments, the differences in
especially the yield stress, but also the concrete strength is too big.
CONCLUSIONS
The following can be concluded:
 The rather high difference in especially m in different researches is caused by using an
effective cross-sectional area Ape instead of the required gross cross-sectional area Ap
 For shallow decks the ratio Ape / Ap is much closer to 1 than for a deep deck like the
ComFlor 210, which means that by using Ape instead of Ap the difference in m is much
larger for those deep decks than for shallow decks
 The influence of the rib reinforcement on the bending moment has to be reduced on the
bending moment by using the distance ds and not by using the effective height x of the
concrete based on only the resistance of this rib bar
 The effective cross-sectional area Ape of the sheeting is larger than the cross-sectional
area without the areas of the embossments
REFERENCES
Van Erp (2016). The horizontal shear resistance of ComFlor 210.
O’Leary, D. and Duffy, C. (1993). The Behaviour of ComFlor 210 Long-Span Composite Floor
Slabs.
NEN-EN 1994-1-1 (2004). Eurocode 4: Design of composite steel and concrete structures –
Part 1-1: General rules and rules for buildings.
SCI Steel Knowledge (2010). Assessment of CF225 Composite Decking Profile. Berkshire,
United Kingdom: Smith, A.
Behavior of a sustainable composite floor system with
deconstructable clamping connectors
Lizhong Wang
Department of Civil and Environmental Engineering, Northeastern University
Boston, MA, USA
wang.l@husky.neu.edu
Mark D. Webster
Simpson Gumpertz & Heger Inc.
Waltham, MA, USA
MDWebster@sgh.com
Jerome F.Hajjar
Department of Civil and Environmental Engineering, Northeastern University
Boston, MA, USA
JF.Hajjar@northeastern.edu
ABSTRACT
This paper presents the behavior of a sustainable composite steel/concrete floor system under
gravity and seismic loading. In this system, precast concrete planks are attached to steel beams
using deconstructable clamping connectors, enabling reuse of the structural components and
reducing the energy consumption related to material fabrication and waste disposal. The results
of several composite beam tests are presented along with the companion pushout tests. In the
pushout tests, both monotonic and cyclic load-slip curves were established for the clamping
connectors. Full-scale composite beams were then designed and tested to investigate the flexural
behavior of the system under gravity loading.
INTRODUCTION
In 2012, buildings are responsible for approximately 47.6% of the energy consumption and 44.6%
of the CO2 emission in the U.S. (Energy Information Administration 2012; Architecture 2030,
2013). To preserve the natural environment, the building industry has to be transformed from the
major contributor to the solution to climate changes and global warming. In addition to exploiting
renewable energy, reduction of energy consumption and emission of greenhouse gases is most
effectively achieved by implementing sustainable design strategies, such as selecting materials
and products with lower embodied energy, utilizing energy efficient operating systems, adopting
deconstructable structural systems to maximize material reuse, etc. Great progress has been
made to optimize material and energy use, for example, replacing cement in concrete with fly ash
(Bilodeau et al. 2000), designing green roofs to reduce the solar radiation reaching the structures
below (Castleton et al. 2010), etc. However, limited research is focused on new structural systems
to facilitate design for deconstruction, particularly for composite construction (Lam et al. 2013;
Lee et al. 2013).
As the most ubiquitous type of structural steel framing for commercial and institutional buildings,
steel-concrete composite floor systems make efficient use of the two materials, with concrete
being subjected to compression and steel resisting tension. However, the integration of steel
beams and concrete slabs via shear studs inhibits the separation of the two materials, making
impossible the deconstruction of the composite flooring systems and reuse of the structural
components. Steel beams and shear studs can be recycled after being extracted from demolition
debris, while concrete slabs are crushed for fill or making aggregates for new concrete.
Conventional composite floor systems are therefore not the best choice for reducing the long-
term environmental impacts of building materials.
In this paper, a sustainable composite floor system is presented. Deconstructable clamping
connectors are utilized to attach the precast concrete planks to the steel beams, and the planks
are connected in-plane with post-tensioned rods. To investigate the behavior of the floor system
under gravity and seismic loading, both pushout tests and composite beam tests were conducted.
In the pushout tests, both monotonic and cyclic load-slip curves were established for the clamping
connectors. Full-scale beam tests were then performed to study the strength, stiffness, and
ductility of the composite beam specimens.
DECONSTRUCTABLE COMPOSITE FLOOR SYSTEM

The deconstructable composite prototype is illustrated in Figure 1; this concept was first
introduced in Webster et al. (2007).This system is designed to maintain the benefits of steel-
concrete composite construction, such as enhanced flexural strength and stiffness, reduced steel
beam size and weight, and ease of construction, and to enable sustainable design of composite
floor systems in steel building structures, components disassembly and reuse of the structural
components.
In this system, high-strength T-bolts, which are inserted into the cast-in channels embedded in
precast concrete planks, are pretensioned to firmly clamp the top flanges of the steel beams with
the underside of the concrete planks. Composite action is thus achieved by utilizing the friction
generated at the steel-concrete interface and the steel-clamp interface. In addition to
deconstructability, the proposed system also offers adaptability and flexibility in that no predrilled
holes are required, and the embedded channels allow for beams with different flange widths.
Grouting precast concrete panels and placing a cast-in-place topping are common in conventional
precast concrete construction, but they may impede the deconstruction of the system and are
thus not recommended. Therefore, load transfer under gravity loading is achieved with tongue
and groove joints that also facilitate alignment during construction. Shown in Figure 2, unbonded
threaded rods are post-tensioned to clamp adjacent concrete planks, and the resulting friction
resists plank joint sliding under shear and plank joint opening under flexure during earthquakes.
The planks are staggered to assist diaphragm load transfer in the perpendicular direction and to
enhance reuse flexibility. After the end-of-life of the structural system, the precast concrete planks
and steel beams can be easily disassembled and reconfigured in future projects by loosening the
bolts and rods.
Precast concrete plank
Steel beam
Cast-in channels
Tongue and groove joint
Threaded rods
Bolt
Clamp
Fig. 1 - Deconstructable composite beam

Fig. 2 - Precast concrete plank connections
prototype
PUSHOUT TESTS
Pushout tests were conducted under both monotonic and cyclic loading to establish the behavior
of the clamping connections for both gravity applications and for use as fasteners to accomplish
load transfer between diaphragms and lateral load-resisting systems in seismic applications.
Pretension tests
For the T-bolt connections, as the nut is turned, both the T-bolt heads and the channel lips deform.
Consequently, the nut rotation calibrated for regular bolted connections, which is given in Table
8.2 in RCSC Specification (2009), does not apply to the pretensioning of the T-bolts. Pretension
tests were thus performed to decide the number of turns of nut. The pretension test configuration
is given in Figure 3. Three bolts were snug-tightened to restrain the movement of the beam,
whereas the tested bolt was first snug-tightened and then torqued until fracture.
Bolt tested
M24 bolts
Snug-tightened bolts
M20 bolts
Fig. 3 – Pretension test setup Fig. 4 – Fractured bolts
As shown in Figure 4, three M24 and M20 bolts were tested, and the axial strain variation was
tracked for each bolt throughout the test using uniaxial strain gages attached on the bolt shank.
The bolt axial stress was then converted from the strain measurements utilizing the stress-strain
curves obtained from tensile testing of round coupons machined from the bolts. Because bolt
yielding is clearly shown in the plots in Figure 5, 2.0 turns and 1.5 turns after a snug-tight condition
are recommended for pretensioning the M24 and M20 bolts, respectively.
1/2 turn 1 turn 1-1/2 turns 2 turns 1/2 turn 1 turn 1-1/2 turns 2 turns
1000 1000
140 140
800 800
105 105
Stress (MPa)
Stress (MPa)
Stress (ksi)
Stress (ksi)
600 600
70 70
400 400
35 35
200 200
0 0 0 0
0 0.004 0.008 0.012 0.016 0.020 0 0.003 0.006 0.009 0.012 0.015
Strain Strain
a) M24 bolt 1 b) M24 bolt 2
1/2 turn 1 turn 1-1/2 turns 2 turns 1/2 turn 1 turn 1-1/2 turns
1000 1000
140 140
800 800
105 105
Stress (MPa)
Stress (MPa)
Stress (ksi)
Stress (ksi)
600 600
70 70
400 400
35 35
200 200
0 0 0 0
0 0.003 0.006 0.009 0.012 0.015 0 0.006 0.012 0.018 0.024 0.030
Strain Strain
c) M24 bolt 3 d) M20 bolt 1
1/2 turn 1 turn 1-1/2 turns 1/2 turn 1 turn 1-1/2 turns
1000 1000
140 140
800 800
105 105
Stress (MPa)
Stress (MPa)
Stress (ksi)
Stress (ksi)
600 600
70 70
400 400
35 35
200 200
0 0 0 0
0 0.006 0.012 0.018 0.024 0.030 0 0.006 0.012 0.018 0.024 0.030
Strain Strain
e) M20 bolt 2 f) M20 bolt 3
Fig. 5 – Bolt axial stress and strain variation in pretension tests
Pushout tests
As shown in Figure 6, each pushout specimen consisted of a 4 ft. x 2 ft. x 6 in. (1219 mm x 610
mm x 152 mm) concrete plank connected with a WT5x30 or WT4x15.5 section using either M24
or M20 clamps. The specimen was mounted upside down to view the motion of the clamps and
steel beam. The pushout test matrix is given in Table 1. The light reinforcement configuration was
designed only for gravity loading, while additional supplementary reinforcement was placed
around the channel anchors to prevent anchor-related concrete failure modes. Since the flanges
of the WT4x15.5 sections were very thin, testing of the WT4x15.5 sections with the M24 clamps
required shims placed between the clamps and the steel beam flanges. The specimen naming
convention is explained using Specimen 3-M24-T4-RH-S, with M describing monotonic loading,
24 describing M24 bolts, T4 describing two-channel specimens, RH describing heavy
reinforcement configuration, and S describing shims. More details about the test configuration
and test matrix can be found in Wang et al. (2015).
Actuators
WT5x30
M24 clamp
Concrete plank
Specimen
Reaction frame
a) Overall view b) Zoom-in view
Fig. 6 – Typical pushout test specimen
Table 1 - Pushout test matrix

Test parameters
Number of
Series Specimen Bolt Number Reinforcement
Shim turns
diameter of T bolts configuration
M 2-M24-T4-RH M24 4 Heavy No 3 turns
M 3-M24-T4-RH-S M24 4 Heavy Yes 3 turns
M 4-M24-T6-RH M24 6 Heavy No 2 turns
M 5-M20-T4-RH M20 4 Heavy No 1.5 turns
C 6-C24-T4-RH M24 4 Heavy No 2 turns
C 7-C24-T4-RL M24 4 Light No 2 turns
C 8-C24-T4-RH-S M24 4 Heavy Yes 2 turns
C 9-C24-T6-RH M24 6 Heavy No 2 turns
C 10-C20-T4-RH M20 4 Heavy No 1.5 turns
The pushout test results are illustrated in Figure 7 and Figure 8. The load-slip curves of
Specimens 2-M24-T4-RH and 4-M24-T6-RH exhibit ductile behavior and excellent slip capacity.
Provided the limit state is fracture in the stud shank, the peak shear strength of a 19 mm (3/4 in.)
diameter shear stud embedded in a solid concrete slab is 95.6 kN (21.5 kips), close to 98.3 kN
(22.1 kips) which is the average shear strength of an M24 clamp. Both specimens retained almost
80% of their peak strengths at a slip of 127 mm (5 in.). In contrast, according to the statistical
analysis conducted by Oehlers and Coughlan (1986), the mean ultimate slip of shear studs is 7.4
mm (0.29 in.). Initially, three complete turns of nut were applied to the bolts in the first two tests
(Tests 2 and 3) before it was determined in subsequent pushout tests to use 2.0 turns of nut for
the M24 clamps, as discussed above. The head of one of these bolts fractured in Test 3-M24-T4-
RH-S because of the excessive rotation, as is demonstrated by the sharp strength drop at a slip
around 25.4 mm (1 in.). Shortly after the fracture, load oscillation occurred, which could be
attributed to a stick-slip mechanism exaggerated by the shims. The stick-slip behavior was also
confirmed in prior research by Grigorian et al. (1994) on the cyclic behavior of clamped bolted
connections with steel-steel sliding surfaces. Compared to the M24 clamps, the post-peak
strength of the M20 clamps degraded more quickly because the M20 clamps were smaller and
they were prone to rotate when significant displacements occurred along the steel beam, as
shown in Figure 9. This was due to the channel lips (which were the same size for all tests) not
being adequately large to support the M20 clamps as fully as the M24 clamps were supported, or
due to the contact of the clamp teeth with the steel flange having too small an area compared to
the M24 clamp. Consequently, the bolt pretension decreased as the clamps rotated. It might be
advised that the M20 clamping connections be redesigned to delay rotation, e.g., a design where
the embedded channel restrains the rotation of the clamp using an interlocking connection, in
which case it is anticipated that the behavior of the clamps will be comparable to the M24 clamps
in this work.
Slip (in.)
0 1 2 3 4 5 6 7 8 9 10 11
25
100
20
80
Load (kips)
15
Load (kN)
60 2-M24-T4-RH
3-M24-T4-RH-S
4-M24-T6-RH 10
40 5-M20-T4-RH
20 5
0 0
0 30 60 90 120 150 180 210 240 270
Slip (mm)
Fig. 7 – Load-slip curves of monotonic specimens (per connector)
Similar to the strength reduction seen in shear studs under cyclic loading (Pallarés et al 2009),
the cyclic load-slip curves of the clamping connectors also exhibit decreasing shear strength as
the cyclic behavior continues. In the clamping system, the abrasion between the concrete plank
and the steel beam and between the clamp and the steel beam smoothed the contact surfaces
and damaged the clamp teeth and the steel flanges, lowering the frictional coefficients and
releasing the bolt pretension, as shown in Figure 10. This strength reduction can be accounted
for in design, thus enabling the clamping connectors to be designed reliably to connect composite
diaphragms and lateral load-resisting systems due to the excellent energy dissipation. The lightly
reinforced concrete plank did not display any anchor-related concrete failure modes, and the
comparison in Figure 8 also indicates that the light reinforcement configuration had negligible
effects on the strength of Specimen 7-C24-T4-RL. Also, as seen when comparing monotonic
specimens 2-M24-T4-RH and 3-M24-T4-RH-S, and cyclic specimens 6-C24-T4-RH and 8-C24-
T4-RH-S, adding a shim between the clamp and steel flange produced oscillations due to stick-
slip behavior and may not be recommended for design.
All the tests were terminated when the stroke of the linear potentiometers was reached or when
all the clamps detached from the steel beam. No other specific limit states were observed.
Slip (in.)
-4 -3 -2 -1 0 1 2 3 4
20
80
60 15
40 10
Load (kips)
Load (kN)
20 5
0 0
-20 -5
-40 -10
-60 -15
-80
-20
-100 -75 -50 -25 0 25 50 75 100
Slip (mm)
a) Tests 6-C24-T4-RH and 7-C24-T4-RL b) Test 8-C24-T4-RH-S
Slip (in.) Slip (in.)
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
20 16
80 60
60 15 45 12
40 10 30 8
Load (kips)
Load (kips)
Load (kN)
Load (kN)
20 5 15 4
0 0 0 0
-20 -5 -15 -4
-40 -10 -30 -8
-60 -15 -45 -12
-80 -60
-20 -16
-150-120 -90 -60 -30 0 30 60 90 120 150 -150-120 -90 -60 -30 0 30 60 90 120 150
Slip (mm) Slip (mm)
c) Test 9-C24-T6-RH d) Test 10-C20-T4-RH
Fig. 8 – Load-slip curves of cyclic specimens (per connector)
Scraped steel flange
Fig. 9 – Large rotation of M20 Fig. 10 – Damage of the steel flange in Test 6
clamp at 1.28 in. (32.5 mm) slip
COMPOSITE BEAM TESTS

After completing the pushout tests, four full-scale composite beams were designed and tested.
As shown in Figure 11, each beam specimen consisted of a 30-foot (9144 mm) long beam
attached with fifteen 2-ft.-wide (609.6 mm) planks using clamping connectors. The actuator loads
were spread using spreader beams at four points along the length (six-point bending) to mimic a
secondary beam under approximately uniform loading. A pin support and a roller support were
placed at the ends of the beams to permit horizontal movement as well as end rotation. Braces
were utilized to prevent lateral deformation of the system due to accidental eccentricity existing in
the test setup and load application. To simplify specimen construction, tongue and groove joints
at the plank edges shown in Figure 1 were eliminated. Grade A36 5/8 in. (16 mm) diameter fully
threaded rods were utilized to connect adjacent concrete planks to resist in-plane diaphragm
forces. One full turn of nut from the snug-tight position was determined for pretensioning the rods
after performing calibration tests in which rods passing through two planks were torqued until
fracture. The concrete planks were 8 ft. (2438 mm) wide, which is sufficient to avoid any premature
concrete failure in narrow slabs (Grant et al. 1977). The composite beam test matrix is given in
Table 2. The naming convention of the specimens is explained using Test 2-M24-1C-RL, with
M24 describing M24 bolts, 1C describing one channel embedded in each concrete plank, RL
describing light reinforcement pattern. The heavy reinforcement pattern contained not only bars
required to resist negative bending of the planks under concentrated loads but also
supplementary rebar placed around channel anchors.
Spreader beams
M20 clamp
Concrete plank Reaction frames

Steel beam
Fig. 11 – Composite beam test setup
Table 2 – Composite beam test matrix
Composite Bolt Beam Reinforcement # of Degree of
beam # size size configuration clamps composite action
1-M24-2C-RH M24 W14x38 Heavy 56 82.7%
2-M24-1C-RL M24 W14x38 Light 30 45.1%
3-M20-3C-RL M20 W14x26 Light 90 137.8%
4-M20-1C-RL M20 W14x26 Light 30 43.8%
The specimens were loaded to 40% of their expected flexural strength and then reloaded three
times. Two more cycles were then undertaken, with one cycle at 60% and the other one at 80%
of the estimated flexural strength. These cycles were intended to mimic serviceability conditions.
After completing the loading/unloading cycles, the beams were then loaded until the deflections
were excessive, surpassing L/25, where L is the beam span. All the beams were shored during
construction, and the load-deflection curves plotted in Figure 12 are shifted from the origins to
account for the bending moment and deflection under the self-weight of the composite beam and
loading structures. All the beams demonstrated ductile behavior. Major events are identified on
the curves, including slip of the clamps, yielding of the steel beam, concrete crushing, and first
bang heard during the test. Slip is identified when the maximum relative movement between the
steel beam and the concrete planks is larger than 0.02 in. Bangs were heard when abrupt slips
occurred between the steel beam and the concrete planks. All the tests were terminated because
of excessive deflection.
Deflection (in.) Deflection (in.)

0 3 6 9 12 15 0 3 6 9 12 15
600 140 120
500
120 100
500
100 400
80
Load (kips)
Load (kips)
400
Load (kN)
Load (kN)
80 300 (11.43 mm, 96.39 kN)

300 (9.65 mm, 96.39 kN) 60
60 Slip
Beam yielding 200 Beam yielding
200 40
Slip 40 Concrete crushing at east side
100 First bang 100 First bang 20
20
Concrete crushing Concrete crushing at west side
0 0 0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
Test 1-M24-2C-RH Test 2-M24-1C-RL
Deflection (in.) Deflection (in.)
0 3 6 9 12 15 18 0 3 6 9 12 15
400 90 80
350 300
75
300 250 60
60
Load (kips)
Load (kips)
Load (kN)
Load (kN)
250 200
(22.61 mm, 96.39 kN)
200 45 40
150 Slip
150 (20.07 mm, 96.39 kN) Beam yielding
30 100
100 Beam yielding Concrete crushing at east side 20
Concrete crushing at west side 15 50 First bang
50
Concrete crushing at east side Concrete crushing at west side
0 0 0 0
0 50 100 150 200 250 300 350 400 450 0 50 100 150 200 250 300 350
Test 3-M20-3C-RL Test 4-M20-1C-RL
Fig. 12 – Load-deflection curves of composite beam specimens
The key results are summarized in Table 3 for all the specimens. For the specimens using the
same steel section, the stiffness increases as the percentage of composite action of the beam
increases. The stiffness calculated using a lower bound moment of inertia [ILB from AISC (2016)]
is below the tested stiffness of the deconstructable composite beam specimens and establishes
that ILB remains an appropriate lower bound estimate for these composite beams. The reason
could be the different mechanisms of achieving composite action for the two types of shear
connectors, with clamping connectors relying on the friction between the structural components
and shear studs bearing against concrete slabs. The ultimate flexural strengths of the beams are
also predicted using AISC design equations (AISC 2016), and the tested strengths are close to
the predictions. It is seen that the degree of shear connection is proportional to the degree of
composite action, with the maximum and minimum amount of slip occurring in beams with the
smallest and largest degree of composite action, respectively. In the composite beam tests, the
maximum slip demand on the clamping connectors was much smaller than the clamp slip demand
during the pushout tests. After completing the tests, the composite beams were disassembled
and a deconstructed steel beam is shown in Figure 14. The beam is intact except for the
impressions on the top flange under the clamp teeth. In typical applications where a beam would
not be subjected to ultimate loads, it is anticipated the steel beam would be in its elastic state
when deconstructed.
Table 3 – Composite beam test results

Stiffness Moment Max slip
Test AISC Test/AISC Test AISC Test/AISC mm (in.)
Specimen #
West East
kN/mm (kips/in.) kN-m (ft-kips)
end end
9.24 8.67 777 767 5.94 6.43
1-M24-2C-RH 1.07 1.01
(52.8) (49.5) (571) (566) (0.234) (0.253)
7.76 6.81 634 632 8.18 6.45
2-M24-1C-RL 1.14 1.00
(44.3) (38.9) (469) (466) (0.322) (0.254)
6.46 5.99 494 510 0.46 0.23
3-M20-3C-RL 1.08 0.97
(36.9) (34.2) (364) (376) (0.018) (0.009)
6.08 4.43 476 400 8.79 8.08
4-M20-1C-RL 1.37 1.19
(34.7) (25.3) (351) (295) (0.346) (0.318)
Fig. 13 – Concrete crushing between planks Fig. 14 – Deconstructed steel beam
CONCLUSIONS
A new deconstructable composite floor system is proposed to promote sustainable design of
composite floor systems within bolted steel building construction through comprehensive reuse
of all key structural components.
Based on the pretension tests, 2.0 turns and 1.5 turns after a snug-tight condition were
recommended for pretensioning the M24 and M20 bolts in the deconstructable composite floor
system. Under monotonic loading, the pushout specimens using M24 clamps were ductile, with
almost 80% of their peak strengths retained at a slip of 127 mm (5 in.). In contrast, the post-peak
strength of the M20 clamps declined more quickly because the clamps were prone to rotate as
the beam moved. Nonetheless, the slip at which the curve started to descend was much larger
than the slip demand on shear connectors in composite beams. If the rotation were restrained or
an interlocking component design were developed, the M20 clamps would behave similarly as
the M24 clamps. Because of the abrasion between the steel flange and the concrete plank and
between the steel flange and the clamps, the strengths of the cyclic pushout specimens were
lower than the strengths of the corresponding monotonic specimens; this reduction should be
accounted for in design. Adding a shim between the clamp and steel flange produced oscillations
due to stick-slip behavior and may not be recommended for design. The hysteresis load-slip loops
demonstrate the potential of the clamping connectors to transfer in-plane diaphragm forces.
Four composite beams of different levels of composite action were designed and deflected to
about L/25. Although some localized concrete crushing occurred along the top edges of the
precast planks at large deflections, all the beams behaved in a ductile manner with little or no
strength degradation. Compared to the stiffness calculated using a lower bound moment of inertia,
the actual stiffness of the deconstructable composite beam specimens was slightly larger. The
tested flexural strengths of the beams were close to those predicted by the AISC (2016) design
equations. The specimens were readily deconstructed after the testing was completed.
The channel, T-bolt, and clamp are commercially available components. The components are not
designed to work together in the proposed configuration, which resulted in certain behavior
limitations that could be addressed by the development of modified components tailored to this
particular application.
ACKNOWLEDGMENTS
This material is based upon work supported by the National Science Foundation under Grant No.
CMMI-1200820 and Grant No. IIS-1328816, the American Institute of Steel Construction,
Northeastern University, and Simpson Gumpertz and Heger. In-kind support is provided by
Benevento Companies, Capone Iron Corporation, Fastenal, Halfen, Lehigh Cement Company,
Lindapter, Meadow Burke, S&F Concrete, and Souza Concrete. This support is gratefully
acknowledged. The authors would like to thank Kyle Coleman, Michael McNeil, Kurt Braun,
Corinne Bowers, Edward Myers, Majed Alnaji, Madeline Augustine, Ian Carver, Morgan Foster,
Michael Bangert-Drowns, Kara Peterman, Angelina Jay, Justin Kordas, David Padilla-Llano, and
Yujie Yan for their assistance with the experiments. Any opinions, findings, and conclusions
expressed in this material are those of the authors and do not necessarily reflect the views of the
National Science Foundation or other sponsors.
REFERENCES
AISC (2016). Specification for Structural Steel Buildings, American Institute of Steel Construction,
Chicago, Illinois.
Architecture 2030 (2013). Available at: http://architecture2030.org/buildings_problem_why/
(accessed April 2017).
Bilodeau, A. and Malhotra, V. M. (2000). “High-Volume Fly Ash System: Concrete Solution for
Sustainable Development,” ACI Materials Journal, Vol. 97, No. 1, pp. 41-50.
Castleton, H. F., Stovin, V., Beck, S. B., and Davison, J. B. (2010). “Green roofs; building energy
savings and the potential for retrofit”. Energy and buildings, Vol. 42, No. 10, pp. 1582-1591.
Energy Information Administration (2012), Annual Energy Review, U.S. Energy Information
Administration, Washingtong, D.C.
Grant, J. A., Fisher, J. W., and Slutter, R. G. (1977). “Composite beams with formed steel
deck,” Engineering Journal, Vol. 14, No.1, pp. 24-43.
Grigorian, C. E., and Popov, E. P. (1994), "Energy Dissipation with Slotted Bolted Connections,”
Report UCB/EERC-94/02, Earthquake Engineering Research Center, College of Engineering,
University of California at Berkeley, Berkeley, California.
Lam, D., Dai, X., and Saveri, E. (2013), "Behavior of Demountable Shear Connectors in Steel-
Concrete Composite Beams," Composite Construction in Steel and Concrete VII, American
Society of Civil Engineers, July 2013, Queensland, Australia, pp. 618-631.
Lee, S. S. M., and Bradford, M. A. (2013), “Sustainable composite beam behavior with
deconstructable bolted shear connectors,” Composite Construction in Steel and Concrete VII,
American Society of Civil Engineers, July 2013, Queensland, Australia, pp. 445-455.
Oehlers, D. J., and Coughlan, C. G. (1986). "The shear stiffness of stud shear connections in
composite beams," Journal of Constructional Steel Research, Vol. 6, No. 4, pp. 273-284.
Pallarés, L. and Hajjar, J. F. (2009). “Headed Steel Stud Anchors in Composite Structures: Part
I. Shear,” Report No. NSEL-013, Newmark Structural Laboratory Report Series (ISSN 1940-
9826), Department of Civil and Environmental Engineering, University of Illinois at Urbana-
Champaign, Urbana, Illinois, April.
RCSC (2009). Specification for Structural Joints Using High-Strength Bolts, Research Council on
Structural Connections, Chicago, Illinois.
Wang, L., Webster, M. D., and Hajjar, J. F. (2015). “Behavior of Deconstructable Steel-Concrete
Shear Connections in Composite Beams,” Proceedings of the 2015 SEI Structures Congress,
Portland, Oregon, April 23-25, 2015, ASCE, Reston, Virginia.
Webster, M., Kestner, D., Parker, J., Johnson, M. (2007) “Deconstructable and Reusable
Composite Slab,” Winners in the Building Category: Component – Professional Unbuilt, Lifecycle
Building Challenge http://www.lifecyclebuilding.org/2007.php
Preliminary Assessment of a Composite Flooring System for Reuse
Martin Nijgh
Delft University of Technology
M.P.Nijgh@tudelft.nl
Mareike von Arnim

Karlsruhe Institute of Technology
Karlsruhe, Germany
Marko Pavlović
Milan Veljković
ABSTRACT
Composite flooring systems consisting of concrete slabs and steel beams are used extensively,
mainly because of competitive design of the cross-section. Shear interaction is often achieved
by welded shear studs, which inherently prohibit demountability of the connection and
reusability of the individual components. Demountability of composite structures can be
achieved by using bolted shear connectors embedded in the concrete. The goal of this paper is
to show how such demountable shear connections can be implemented, which allows for easy
execution and the reuse of all structural components. A key element in the suggested
longitudinal shear device is injection bolts, used to fill the hole clearance with epoxy resin,
increasing stiffness and ductility compared to traditional demountable composite structures.
Push-out tests and beam tests are planned and pre-analysis is made for these tests using
extensive Finite Element modelling. Based on current experience, recommendations for design
and execution are made.
INTRODUCTION
Composite flooring systems consisting of a concrete slab and steel beams are used extensively
in structures, mainly because of the reduced material demand to achieve a certain strength and
stiffness. In composite flooring systems, the different elements are loaded in their most
favourable way (i.e. concrete in compression, steel in tension) and interaction is often achieved
by welded shear studs, which prohibit the demountability of the connection and reusability of the
individual structural components. An alternative to welded shear studs are bolted shear
connectors. Depending on the way such bolted shear connectors are used, demountability of
the connection and reusability of the structural members is possible. The goal of this paper is to
show how to achieve demountable shear connections with appropriate stiffness and ductility,
allowing for the fast execution and accompanying requirements for the reuse of a composite
structure.
This paper focuses on the pre-analysis of dry flooring systems, which have a new type of shear
connector embedded in the concrete slab. Common solution of the embedded bolted shear
connector is that the bolt sticks out from the concrete deck. The possibility of damaging one of
the bolts during transporation, execution and demounting is very realistic and will cause that the
deck segment is not suitable for (re)use. In order to avoid such problems, an embedded coupler
is used (Figure 1). The goal of the coupler is to separate the internal (embedded) part of the
shear connector from the externally in-situ assembled bolt.
Embedded bolt
(external) injection nozzle
Embedded
coupler
Beam flange
Assembly bolt,
suitable for injection
Figure 1 - Embedded coupler including Figure 2 - Geometry of an injection bolt,

embedded bolt and assembly bolt showing a distinct injection channel through
(suitable for injection) (Kozma & its head (EN 1090-2)
Odenbreit, 2017)
Fabrication tolerances and the requirement for easy demounting influence the needs for the
hole clearances in the beam’s top flange. Within the EU-research project REDUCE (Reuse and
Demountability using Steel Structures and the Circular Economy), the magnitude of the hole
clearance is investigated. Preliminary results from ABAQUS simulation indicate that a 6 mm
hole clearance (3 mm bolt-to-hole gap) would be sufficient for a typical span of 16 m consisting
of large prefabricated concrete decks of 8.00 x 2.60 metres, since no bearing occurs due to self-
load. Excessive hole clearances (i.e. hole clearances that are not closed by the self-weight of
the concrete slabs) increase the speed of erection and accomplish the demountability of the
structure. However, if excessive hole clearances are used, the bending stiffness of the
composite floor decreases as a result of initial slip between both members.
In order to overcome the reduction in stiffness, the clearance between the bolt and hole may be
injected with an epoxy resin, which mitigates the slip between slab and beam as well as the
vertical displacement of composite deck. Injecting the hole clearance with epoxy resin can be
done using bolts with a small hole through the bolt head, which is known as an injection bolt
(Figure 2). Such injection bolts have been used since the 1970s in The Netherlands, initially for
refurbishment of railway bridges (Gresnigt & Stark, 1995) (Gresnigt et al., 2000). According to
the resin (RenGel SW 404 + HY 2404) manufacturer, the resin is fully cured after 24 hours, but
earlier investigations for this resin at TU Delft have shown that there is no significant increase in
strength and stiffness after 6 hours of curing. In order to prevent the resin from escaping
through the beam-slab interface during injection, the injection bolt must be slightly preloaded in
order to close any gaps in this interface (e.g. as a result of welding, shrinkage, etc.). Since the
small preload is only to facilitate the injection procedure and is not of any structural importance,
it is not a problem that this preload is lost in time as a result of (through-thickness) concrete
creep. A recent TU Delft innovation in the field of injection bolts is the strengthening by (steel)
shot prior to injection of the resin (Nijgh, 2017). This way, a stiff load bearing skeleton of steel
particles is obtained, and the resin functions as a binder rather than a load bearer.
Although epoxy resins are generally highly adhesive, the adhesion of the epoxy to the steel and
concrete substrate can be prevented by the use of a wax-based release agent, which allows for
demountability of the shear connection (Figure 3). Nijgh (2017) has proven that all common
release agents (wax-, silicon- and PVA-based) work, but that a wax-based release agent has
the preference given that it provides a non-porous resin structure and is not known to degrade
the epoxy material properties.
Figure 3 - Prevention of adhesion between steel and epoxy resin using wax-
based release agent. (a): steel plate with resin infill after removal of the bolt,
and (b): resin infill pressed out of the steel plate (Nijgh, 2017)
Previous research by Pavlović (2013) has shown that bolted shear connectors may not provide
enough ductility (e.g. minimum of 6 mm of slip demanded by EN 1994-1-1) to allow for a plastic
design of a composite structure. The use of injection bolts may provide additional ductility
compared to non-injected connections by deformation of this epoxy layer. The largest
contribution to the ductility is mostly observed when the connection is close to its ultimate
resistance, because of plastic deformation (e.g. in the steel section) the confinement of the
epoxy resin is significantly reduced and crushing occurs (Figure 4). The ultimate resistance of
shear connector is independent of the injection material; after failure of the resin in compression
the behaviour is defined by the bolted shear connector resistance.
Figure 4 - Crushing of the resin Figure 5 - HE260B beam used in push-out

resulting from plastic plate tests, featuring a special groove (Ø 5mm)
deformation approaching the failure which indicates successful injection if
load of the plate (bearing failure) resin exits at side of flange
(Nijgh, 2017)
DEMOUNTABLE SHEAR CONNECTOR
The (demountable) bolted shear connector system under consideration in the push-out tests is
shown in Figure 6. The edge angle, 70 mm bolt and coupler are embedded within the concrete
slab with ℎ = 150 mm. On the assembly site, the concrete slab is placed on the steel beam and
connected using an injection bolt. The allowable deviation between the centre of the coupler
and the centre of the injection bolt is 3 mm. After installation of the injection bolt, all clearances
are filled using epoxy resin (RenGel SW 404 + HY 2404) and adhesion is prevented by applying
a wax-layer to all components a priori. An air vent is foreseen in the beam flange in order to
allow air to escape during injection and to confirm that the injection procedure is successful
(Figure 5). The geometry of the push-out tests is not conform EN 1994-1-1, but has (1) larger
shear connector spacing (300 vs 250 mm) and (2) more reinforcement to more closely resemble
the final application.
The potential of injection bolts in composite structures is assessed based on numerical analysis
and laboratory experiments. The numerical analysis is largely based on the work of Pavlović
(2013) (including damage models for concrete and steel) and combined with the results from
Nijgh (2017) to include resin behaviour. Experiments will be carried out at the University of
Luxembourg (push-out tests) and Delft University of Technology (additional push-out tests and
feasibility test of the composite floor system).
Figure 6 - Bolted shear connector system, dimensions in mm (Kozma & Odenbreit, 2017)
CASE STUDY: COMPOSITE STRUCTURE WITH DEMOUNTABLE FLOORING SYSTEM
The results of the push-out tests on the system described above are used to predict the
behaviour of a composite flooring system, as in Figure 8, with the same bolted shear connector.
Beams with a 16 meter span and a 2.60 m centre-to-centre distance are assumed. The beam
(grade S355, modelled elastic-plastic), is assumed to be tapered, with height 590 mm at the
supports and 740 mm at midspan. The web is 4.5 mm thick with a stiffener is foreseen at
midspan. The flanges are 300 mm wide, with thickness 10 and 12 mm for the bottom and top
flange, respectively. The concrete slab is assumed to be grade C35/45 with thickness ℎ =
120 mm (modelled elastic, = 34000 MPa). According to hand calculation, 44 shear connectors
on each flange-side are required which are uniformly distributed in Reference Spacing, as
illustrated in Figure 7. Since shear force reduces towards the middle of the girder, a decreased
number of couplers (32) with wider spacing only in the middle of the girders is examined as well
(Spacing 1). Moreover, influence of partly injected structure for both spacing arrangements is
considered where only outer ten bolts on each beam-side are injected with resin while middle
shear connectors have blank bolt-to-hole clearance. Fewer shear connectors and less injection
bolts reduce overall costs of structure since less material, and more important, fewer workers
are required. The shear connectors are assumed to be centrically in all their respective bolt
holes. Loads on the floor are adopted as: self-weight of concrete elements (2.83 kN/m2),
additional flooring load (0,5 kN/m2) and live load (2.5 kN/m2). During installation of concrete
decks, only self-load in the amount of 2.83 kN/m2 is applied. After completed installation
imposed design load on structure is 1.35 0.5 1.5 2.5 = 4.43 kN/m . Hence, total load on
structure is 2.83 4.43 2.6 8 4 = 604 kN.
Figure 7 - Spacing arrangements of shear connectors (mm)
FE PREDICTION OF PUSH-OUT TEST AND BEAM TEST EXPERIMENTS
The ABAQUS/Explicit finite element models of Pavlović (2013), including material and damage
models for concrete (C35/45) and steel (grade 8.8), have been used as a basis to model the
push-out tests and obtain an indication on the initial stiffness of the demountable shear
connector system. The finite element model is shown in Figure 9. Based on earlier numerical
work by Nijgh (2017), a Young’s Modulus of 4250 MPa and Poisson’s ratio of 0,3 are assumed
for the resin. No damage or plasticity models are available yet for this specific resin meaning
that only until a certain load level (90 kN for this case) the results are reliable. Under the same
conditions, also resin reinforced with steel shot is modelled ( = 9500 MPa as sensitivity study.
The connection elements causing most of the slip are identified by assuming one connection
element to be infinitely stiff per analysis. A friction coefficient of 0,14 is assumed for all
tangential interactions, whereas for the normal contact the ‘hard contact’ formulation is used.
The results obtained from the numerical analysis of the push-out tests are used to predict the
behaviour of the composite floor tests. Finite element analysis was made using
ABAQUS/Standard to examine the behaviour during and after installation of concrete slabs on
the tapered steel beam. Therefore, a model with three steel girders was made where four
concrete decks are put in place step by step, as Figure 8 shows, before loading the structure.
Results indicate that the installation-order of concrete decks has small influence (5%) on
deflection. The work is further extended by investigating what the effects are if only part of the
total amount of bolted shear connectors is injected.
Spring elements with non-linear behaviour in longitudinal direction represent bolted shear
connectors between concrete decks (solid elements) and steel girders (shell elements) in
ABAQUS model. To prevent twisting of girders during step by step installation of concrete slabs
L-shaped diagonals (beam elements) are used to stiffen the structure. They are fixed crosswise
at bottom flange of steel beams. Furthermore, concrete decks are connected with “Tie
constraint” in longitudinal direction while they do not interact in transversal direction. As well as
in finite element model of push-out test friction coefficient of 0.14 is assumed for tangential
interaction and hard contact for normal interaction.
4
3
1
2
Figure 8 - Model for beam test with installation order of concrete

decks, order of installation is most unfavourable for deflection
Injection bolt Coupler Embedded bolt
Beam flange L-profile Concrete Reinforcement
Figure 9 - Finite element model of push-out test. Double symmetry is used to reduce computational
cost.
RESULTS OF PRELIMINARY INVESTIGATIONS
The load-slip relationship derived for the bolted shear connector is shown in Figure 10 for resin,
reinforced resin (as developed by Nijgh, 2017) and steel infill in the 6 mm bolt-to-hole clearance
(M20 bolt centrically in Ø26 mm hole). The preliminary investigation focuses on the execution
process and behaviour at the Serviceability Limit State. Based on the FE-analysis it is predicted
that the failure load per shear connector is approximately 140 kN. The resin reduces the initial
shear connector stiffness by 30% compared to a connection injected with steel. The reinforced
resin expects to reduce the stiffness by 20%.
120
Force per shear connector (kN)
100
100%
80
80% Steel injection
60
70% Resin injection
40
Reinforced resin
20 injection
0
0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40
Slip (mm)
Figure 10 – Load-slip diagram derived using FE model for connections injected

with steel, resin and a combination of resin and shot. Above 90 kN the resin starts
to crush (not included in model) meaning the prediction is less accurate.
Figure 11 illustrates the increase of the initial connection stiffness (i.e. until 1 mm of slip)
depending on which element is modelled infinitely stiff. The connection elements responsible for
most of the slip are the resin, injection bolt and concrete, whereas the coupler and embedded
bolt do not provide a significant contribution to the total deformation.
The effects of the decrease in bolted shear connector stiffness on the bending stiffness of the
composite floor are shown through Figure 12. The effect of the rather deformable resin on the
deflection at midspan is rather small: only a 5% increase of deflection is expected based on the
SLS load level compared to the situation in which the bolt is directly bearing to the beam flange.
In Figure 12 the horizontal line at 235 kN shows completed installation of concrete decks on
steel beams. Deflection curves become steeper during installation-process of concrete decks
since stiffness of structure increases with further installed decks. Difference in deflection with
imposed design load (604 kN) between both spacing arrangements is rather small with 5%.
Injecting only outer ten bolts with resin results in increased deflection of 8% for Reference
Spacing and 6% for Spacing 1. In this respect, reducing shear connectors and injecting less
bolt-to-hole gaps are considerations to decrease costs at the expanse of slightly increased
deflection under imposed design load.
In all illustrated cases deflection at design load level is less than serviceability limit state of l/250
= 16000/250 = 64 mm; hence at SLS load level the deformation limit is also not exceeded.
120
Load per shear connector (kN)
Element modelled
100
infinitely stiff:
80 Resin
Coupler
60
Embedded bolt
40 Injection bolt
Concrete
20
All elements deformable
0
0 0,2 0,4 0,6 0,8 1
Slip (mm)
Figure 11 - Load-slip diagram in initial phase in case one of the elements is modelled infinitely stiff. A
large increase of stiffness with respect to the case in which all elements are deformable indicates a
large slip contribution.
700
Design Load
600
500
Total Load [kN]
400
300
Self-weight decks
200 Fitted holes (Reference Spacing)

3 mm gap with resin (Reference Spacing)
3 mm gap with resin (Spacing 1)
100
3 mm gap with resin in 10 outer bolts (Reference Spacing)
3 mm gap with resin in 10 outer bolts (Spacing 1)
0
0 5 10 15 20 25 30 35 40 45 50
Deflection [mm]
Figure 12 - Deflection of tapered beam as a function of injection material and shear connector spacing
EXECUTION OF A COMPOSITE FLOOR
The results show that the use of injection bolts in combination with oversize holes in the beam
flange does not significantly affect the bending stiffness of the composite floor. It must be noted
that the results are preliminary, given that (1) no effect of epoxy creep is taken into account and
(2) the (reinforced) resin is modelled as elastic. The beam flange thickness is small compared to
the diameter of the bolt, meaning the entire flange thickness can be assumed to be effective in
load transfer, as also suggested in the European Recommendations for Bolted Connections
with Injection Bolts (1994).
CONCLUSION
The pre-study demonstrates that a demountable composite dry flooring system with sufficient
hole clearance to aid both execution and demounting can be achieved using injection bolts. The
following conclusions are made based on the preliminary calculations:
1. Initial composite floor bending stiffness is largely unaffected (increased deflection of 5%
at midspan) in case of 3 mm injected gap around the bolts for the considered case study
of the composite flooring system.
2. The injected epoxy resin has the potential to increase ductility of the shear connector
system by (plastic) deformation of the injected resin. A 30% increase in ductility is
observed in the FE analysis compared to direct bolt-to-hole interaction in the linear
branch of the load-slip relationship (slip < 1 mm).
3. The effect of epoxy creep leads to a time-dependent reduction in bending stiffness of the
flooring system, and will be experimentally investigated in the ongoing project.
4. Compared to reference spacing, partial interaction of the composite structure and wider
spacing between shear connections result in larger deflection (up to 10%). Partial
interaction in the considered composite structure leads to saving of in-situ labour.
Quality of FE prediction of the push-out test behaviour will be evaluated after results from the
experiments are obtained.
ACKNOWLEDGEMENT
The authors acknowledge financial support of the Research Fund for Coal and Steel (RFCS) for
funding the research project ”Reuse and Demountability using Steel Structures and the Circular
Economy” REDUCE (RFCS-02-2015). Cooperation of the project partners (SCI, University of
Luxembourg, University of Bradford, Lindab A/S, Tata Steel, Bouwen met Staal and AEC3) is
gratefully acknowledged.
REFERENCES
European Convention for Constructional Steelwork. (1994). European Recommendations for

Bolted Connections with Injection Bolts. ECCS.
Gresnigt, A. M., & Stark, J. W. (1995). Design of Bolted Connections with Injection Bolts.
Proceedings of the Third International Workshop on Connections in Steel Structures (pp.
77-87). Trento: Elsevier.
Gresnigt, A. M., Sedlacek, G., & Paschen, M. (2000). Injection bolts to repair old bridges.
Proceedings of the Fourth International Workshop on Connections in Steel Structures.
Roanoke, VA, USA: Delft University of Technology.
Inter-Composite. (n.d.). Gelcoat Resin - Araldite (R) SW 404 Resin with HY 2404 Hardener.
Retrieved from Inter-Composite: http://inter-composite.com/wp-
content/uploads/2012/12/Araldite-SW-404-a-HY-2404.pdf
Kozma, A., & Odenbreit, C. (2017). Cross-section of demountable bolted shear connector
system. Background document of REDUCE project. University of Luxembourg.
Nederlands Normalisatie-Instituut. (2011). Eurocode 3: Design of steel structures - Part 1-8:

Design of joints. Delft: Nederlands Normalisatie-Insituut.
Nijgh, M. P. (2017). New Materials for Injected Bolted Connections (MSc. thesis). TU Delft.
Pavlovic, M. S. (2013). Resistance of Bolted Shear Connectors in Prefabricated Steel-Concrete

Composite Decks (PhD thesis). University of Belgrade.
Seismic Collectors in Composite Steel Deck Diaphragms
Robert B. Fleischman
University of Arizona
Tucson, U.S.A.
rfleisch@email.arizona.edu
Anshul Agarwal, Haitham Ayyad

University of Arizona
Tucson, U.S.A.
anshul@email.arizona.edu, haithamayyad@email.arizona.edu
Richard Sause, James Ricles

Lehigh University
Bethlehem, U.S.A.
rs0c@lehigh.edu, jmr5@lehigh.edu
Chi-Ming Uang
University of California, San Diego
San Diego, U.S.A.
cmu@ucsd.edu
ABSTRACT
Seismic collectors are elements that bring inertial forces to the Seismic Force Resisting System
elements. In steel composite structures, collectors are typically part of the underlying frame,
modified to carry collector forces as needed. Due to the reversing nature of earthquake loads,
collectors are designed both for tension (with a focus on connections) and compression (beam-
columns). The actual seismic demands on the collector elements are poorly defined. The steel
composite deck floor system is a complicated indeterminate assemblage that must maintain
gravity load resistance while transferring the collector forces. An analytical research program is
presented to study of the seismic performance of collectors in composite steel deck diaphragms,
including load paths, characteristics under compression and tension, and the impact of the
composite slab in load participation and bracing. Design recommendations are sought for
appropriate strength and details of steel composite collectors.
1. INTRODUCTION
Seismic collectors are key elements of a building’s Seismic Force Resisting System (SFRS).
Current design code provisions for collectors recognize their critical role through special load
combinations (ASCE 7, 2010) that include the System Overstrength Factor Ωo (varying from 2.0
to 3.0), resulting in large design forces for collectors. This stringent prescriptive design approach
is an attempt to ensure that critical collector elements remain elastic. Loss of collector elements
can be catastrophic, as has been shown by failures of concrete structures, including the collapse
of the CTV building in the 2011 Christchurch earthquake, in which 115 lives were lost (Royal
Commission, 2012), or in the collapse of the Northridge Fashion Center parking structure (EERI,
1994) in which the shear walls were undamaged while the floor system detached and collapsed.
Collector behavior remains one of the less defined aspects of building response. In an
earthquake, inertial forces must be carried through the floor or roof diaphragm to the primary
elements of the SFRS. The collector path involves complicated indeterminate assemblages of
different materials and geometries, acting at different elevations, and connected by elements
which may not be intended for collector action. Complex load paths develop in the floor system
as a result not only of collector action, but also interaction with lateral load and the gravity load
resisting system (GLRS). Due to the building’s nonlinear dynamic response, peak inertial force
magnitudes during an earthquake may significantly exceed the levels anticipated in past codes.
With new larger inertial forces reflected in upcoming design codes, it is crucial to better understand
how the collectors carry these forces to permit efficient and economical designs.
This paper presents the details of an ongoing analytical study aimed at examining collector paths
in steel floor systems for the purposes of better describing these systems. The analytical study is
based on models constructed and calibrated using previous testing of steel composite deck
components and bays. Two-dimensional (2D, horizontal and vertical plane) have been developed
to: (1) Characterize collector load paths within composite steel deck floor systems; (2) Quantify
properties and performance of steel collectors; and (3) Examine capacity-based design solutions
for steel collectors. In the future, 3D models will be used to investigate collector seismic demands.
2. SEISMIC COLLECTORS IN STEEL STRUCTURES

The SRFS of a steel structure has three primary components, which, acting in series, provide a
complete load path between the seismic mass and the foundation: (1) the composite floor
diaphragms, in which inertial forces from the floor mass and attached elements are transferred
through the in-plane shear stiffness of the floor deck/slab system; (2) the vertical-plane elements
that provide a laterally stiff vertical load path for these forces to the foundation; and (3) the seismic
collectors, which serve as the critical link between these other two components.
2.1 STEEL COMPOSITE FLOOR SYSTEMS

Steel composite deck is the typical construction for floor systems in steel building structures. In
this floor system, a concrete-filled metal deck is placed on top of the underlying steel gravity
framing: girders span the columns; floor beams span the girders; and the metal deck (thin-gage
cold-formed corrugated sheets attached by fasteners) spans the floor beams (See Figure 1a).
The concrete floor slab is cast-in-place using the deck as formwork, and may be reinforced with
bars or welded wire fabric. In composite floor systems, steel shear studs are welded to the
underlying framing and project into the slab (See Figure 1b). Composite action is attained in floor
systems through a combination of chemical bond between the slab and deck, mechanical
interlock by embossments in the deck profile, and the shear studs.
(Rogers
& (Hedaoo,
Tremblay, et al.,
2003) 2012).
Fig. 1. Steel Deck Diaphragms: (a) Bare Metal-Deck; (b) Composite Steel Deck.
2.2 STEEL COLLECTOR DESIGN

Seismic collectors bring the inertial forces that develop in the floor system during earthquakes to
the vertical plane SFRS elements, e.g. braced frames (See Figure 2a). Collector demands
depend on seismic hazard (ASCE 7, 2010) and collector length. In modern structures, SFRS
elements have become isolated within the floor plan, resulting in significant collector runs.
Due to the reversing nature of earthquake loads, collectors must be designed both as tension and
compression members. Tension member design focuses on the collector connections (Astaneh-
Asl, 2005), e.g. TFW etc. (See Figure 2b). In compression, the collector is considered a beam-
column, since the member is under combined flexure (due to gravity load) and compression (due
to collector action). Collector design involves the special seismic load combination (ASCE 7,
2010) using the System Overstrength Factor Ωo, resulting in amplified design forces.
(b) (m-1)nL
Collector Force Diagram
nL
1 2 3
m bays 4 5 6
Deck braces Floor beams brace

the collector the collector
AFW TFW MST ST

(all flange welded) (top flange welded) (multiple bolt row) (shear tab)
for GLRS
L
Fig. 2. Steel Collector Design Forces (AISC, 2010): (a) Plan View; (b) Collector Profile.
Under low seismic demands, collector forces are often sufficiently small that existing floor system
elements serving as part of the GLRS can also counted as collectors, e.g. shear tabs (ST) with
slip-critical bolted connections (AISC, 2010). In cases for wind or more modest seismic loads, the
collector may be provided by reinforcing bars placed within the concrete slab (AISC, 2010). For
moderate seismic demand, gravity system details may be modified to carry the loads, e.g. adding
extra rows of bolts, MST (Figure 3a). For high seismic collector demands, however, the outcome
of current code provisions is that expensive special elements must be introduced into the floor
system specifically to carry the amplified forces. Members significantly larger than the surrounding
gravity system elements are required to prevent instability under compressive collector forces;
expensive full-penetration welded connections are often required for tension transfer, applied to
only the top flange (TFW) if possible (Figure 3b) to minimize moment, but also to both flanges
(AFW) when necessary. Due to the way collector forces build up (Figure 2b), different collector
connection types may be used along a single collector run. Depending on the gravity framing
layout, collectors in one direction might act through the column weak axis (e.g. Figure 3a).
CL COL.
STL. COL. DCW CJP,
BU BAR TO 5/16 PL TO COL
Plate REMAIN 5/16 WEB, TYP.
DCW CJP,
STL. BM. 5/16
SEE NOTE
CJP
'A'
'A'
PLATE
WEB PLATE
TO COL. 7/16
PLAN (SEISMIC DRAG)
FLANGES 7/16
NOTE- "DCW"=DEMAND CRITICAL WELD AT SEISMIC DRAG SPECS
BEAM TO STRONG-WAY COLUMN
Weak-Way column CONN., TOP-FLANGE WELDED
Fig. 3. Collector Details: (a) MST; (b) TFW; (c) Constrained Flange FTB (AISC, 2011)
Limit states in compression include flexural, torsional, flexural torsional, and lateral torsional
buckling (AISC, 2011). The lateral and torsional bracing inherent in the floor framing and deck is
an important consideration in design. Deck orientation (both are shown in Figure 2b), level of
composite action, and the presence of openings all can have a significant effect (AISC, 2011). In
general, lateral bracing is ignored for parallel deck ribs. Torsional bracing, often ignored for a bare
steel deck, is considered continuous for a composite steel deck (AISC, 2011). The unique
boundary conditions for collectors, in comparison to most beam-column situations, leads to a
constrained-flange flexural torsional buckling (FTB) mode about the top flange, where shear studs
anchor the member (See Figure 3c). Designers are permitted to follow approximate methods
based on design equations for other conditions (Helwig & Yura, 1999), as modified using
simplified assumptions (AISC, 2011). Collectors are treated in several technical documents
(Sabelli, et al., 2011) including for new construction (FEMA, 2009) and retrofit (ASCE, 2013).
New diaphragm seismic design forces are being adopted in the code (BSSC, 2014) in recognition
of large peak inertial forces that can develop during a seismic event. These forces can be
substantially larger than the prescribed current code (ASCE 7, 2010) equivalent lateral forces
(ELF). Figure 4b compares the current and newly proposed ASCE 7 diaphragm design forces,
indicating an impact on collector design. Collector overstrength factors are also treated differently.
Fig. 4. Seismic design forces: (a) ELF; (b) Fpx Comparison.
2.3 DISCUSSION: SEISMIC COLLECTOR ACTION IN COMPOSITE STEEL FLOORS

Seismic collectors must be able to transfer the seismic forces as axial load into the SFRS through
a dependable, stiff load path. The collector forces are accumulated through shear studs on the
elements, or to a lesser extent transverse framing members. Cowie, et. al. (2013) emphasize the
need for sufficient shear strength in the composite slab along a collector to ensure inertial force
transfer. MacRae & Clifton (2015) point to the influence of pre-existing shrinkage/temperature
cracks on collector transfer strength. Codes do not formally address these conditions, with
exceptions, e.g. shear strength design equations in (NZS 3404, 2007) based on rib orientation.
Collector compression forces may be strongly carried through the concrete slab (See Figure 5a).
Little evidence exists to distinguish strut action (Bull, 1997) in the composite slab relative to the
transfer of the underlying frame. Tension collector action depends on relative stiffness of the steel
frame, deck and slab, the fixity of collector connections, and deck orientation (Cowie, et al., 2013).
Experiments with composite steel decks show the slab contributing to frame action (Kasai &
Matsuda, 2016) and laterally bracing members, except when early cracking of concrete ribs
occurs (Zhang & Ricles, 2006). Large-scale tests of MRFs with composite slabs found collector
forces influenced by the level of composite action (Herrera, et al., 2008). Easterling & Porter
(1994) identified controlling limit states in composite decks, including slab X-cracking and
longitudinal cracking in thin slabs; with higher strength for loading perpendicular to ribs. A
compression collector reverses and becomes a tension collector with each earthquake oscillation.
Relative stiffness changes during the seismic event as elements soften, yield, slip, crack, or crush.
The floor system profile also involves significant vertical eccentricity (See Figure 5b). The
concrete slab detail at columns affects collector force developing due to floor slab inertial effects
(Chaudhari, et al., 2014). Force transfer between the slab and column face can occur through
bearing or strut and tie mechanisms (Braconi, et al., 2010). Moment in a collector due to the
eccentricity of the slab inertial force transfer is assumed to be completely compensated by vertical
end shear force, and is ignored in design (Cowie, et al., 2013). The same elements providing
collector action also provide gravity load resistance (Sabelli, et al., 2011). These interactions are
typically ignored (Cowie, et al., 2013) as is gravity load interaction on shear studs (Burmeister &
Jacobs, 2008). MacRae & Clifton (2015) suggest studs added to non-composite gravity beams
for collector action may be overstressed unwittingly due to unavoidable gravity action.
Shear studs have a significant effect on cyclic stiffness (Zhang & Ricles, 2006). Tests on the shear
stud-concrete interface found localized splitting behavior at low loads (Lowe, et al., 2014) and
low-cycle fatigue failures (Bursi & Gramola, 1999). Adhesion between the slab and deck (Souici,
et al., 2013) was the key transfer mechanism in non-composite decks (Majdi, et al., 2014). Tests
on deck provide shear stiffness (Davies, 1986) and orthotropic properties (Wennberg, et al., 2011)
The collector load path must cross gravity columns. Collector connections at the columns are
often detailed to minimize moment (e.g., Figure 3b). Nonetheless, these connections have partial
fixity, especially when slab contributions are considered, unless special articulating details are
introduced (MacRae, et al., 2010). At the building perimeter, gravity load transfer induces torsion
into the collector, requiring proper spandrel beam slab reinforcement (Clifton & El Sarraf, 2005).
Ubiquitous slab openings (e.g. Figure 5a), can create torsion and an unbraced collector condition.
Figure 5. Composite Diaphragms Load Paths: (a) Horizontal-Plane; (b) Vertical Profile.
3 ANALYTICAL RESEARCH PROGRAM

The analytical research program involves three phases (See Table 1):
(1) Evaluation of isolated floor systems to determine collector force paths in composite steel deck
diaphragms, including: (a) the horizontal-plane spatial distribution of collector forces; (b) the
relative participation of the slab, deck, and underlying framing in the collector force transfer; and
(c) the limit-state sequence of elements within the collector load path. This phase is achieved
through 2D (horizontal plane) nonlinear pushover (NP) analyses of isolated floor diaphragms
(2) Evaluation of isolated collector frames to determine the properties (strength, stiffness and
ductility) of the collector system in composite steel deck diaphragms for different: (a) Collector
connections; (b) deck orientation; and (c) floor slab details. Strength and stability limit states are
considered. This phase is achieved through vertical plane models of the collector frame with: 2D
elements for strength (tension) limit states and 3D elements for stability (compression) limit states.
(3) Evaluation of seismic demands acting on collectors in composite steel deck diaphragms for
different design approaches. This phase will be achieved through nonlinear time history analyses
(NTHA) of full building structures under design and maximum considered earthquakes. This
phase is future work and will not be considered further in the paper.
Three specific design approaches will be evaluated: (1) current code designs (ASCE 7, 2010)
based on overstrength, Ω0 Fpx; (2) collectors designed using emerging code provisions (BSSC,
2014); and, (3) new design concepts in which the collector overstrength requirement is relaxed,
and instead different relative capacities of the collector and collector connections are considered.
Table 1. Analytical Research Phases.
Phase Objective Structure Analytical Model Elements Loading Analysis
1 Collector Force Path Isolated Floor 2D (Horizontal Plane) 1D (Truss) Body Force NP
Collector 2D (Strength)
2 Collector Properties 2D (Vertical Plane) Body Force NP
Frame 3D (Stability)
3 Collector Demands Building Whole Structure 1D Earthquake NTHA
3.1 EVALUATION STRUCTURE

An evaluation structure, a 4-story office building representing typical construction, has been
designed for the study (See Figure 6a). The building footprint is 54.9m x 36.6m (See Figure 6b).
The SFRS is a Special Concentric Braced Frame (SCBF). A baseline design is performed for
current code (ASCE 7, 2010) SDC E (R=6, Cd=5,  o=2.0) and typical live loads (LL = 2.4kN/m2).
Other designs used in the study are modified from this design (See Table 2).
Table 2. ELF Design Forces: Recent vs. New Code.

Level Floor hi (m) Wi (kN) Cvi Vi (kN) Mi (kN-m) ASCE 7-10 Fpx (kN) ASCE 7-16Fpx (kN)
4 Roof 16.9 3132 0.261 1696 6979 1671 2626
3 4 12.8 9757 0.465 1061 19427 3574 5357
2 3 8.7 9757 0.214 1375 25158 2634 5308
1 2 4.6 9757 0.059 1461 29718 2602 5259
The floor gravity system framing is designed as composite members: 9.1m x 9.1m bays with floor
beams at 3m on center (o.c.). The composite floor deck is a 63.5mm (2.5”) concrete 20MPa (3ksi)
slab over 50mm (2”) LOK 18 gage steel deck with single 19mm (¾”) diameter shear studs at
300mm o.c. All members are A992; detail material is A36.
The 2nd floor baseline design is presented in this paper and described here: The collector is shown
in Figure 6b; collector connection schematics are shown in the inset and described in Table 3.
Note that the braced frame on the east end of the floor system is removed to create a significant
collector condition. The collector is designed as non-composite member as per current code
guidelines (AISC, 2010). The floor beams are assumed to brace the collector (ribs parallel). Shear
studs along the collector are in pairs at 280mm o.c. Gravity framing connections are standard
shear tabs with 19mm (¾”) diameter A-325 bolts (See Table 4).
Table 3. 2nd Floor Collector Baseline Design (without East End Braced Frame).
Column Line B C C D D E E F F G
Axial Force Nu (kN) 2169 1735 1301 867 434
ΩoNu (kN) 4337 3470 2602 1735 867
Shear Vu (kN) 262 262 262 262 262
Moment Mu (kN-m) 761 761 761 761 761
Collector W24x131 W21x122 W21x111 W21x111 W21x101
Collector Connection AFW TFW TFW TFW TFW MST MST ST ST ST
Table 4. Connection Design.

Connection Type Collector Gravity
Location on Plan (Figure 6b) A B C D E F
Plate Thickness (mm) 22.2 19.05 31.75 22.2 9.5 12.7
# of bolts* per row / # of rows 21 / 3 14 / 2 14 / 2 7/1 3/1 4/1
Connection Detail AFW TFW MST ST ST ST
* 1” (25.4mm) Dia A490 bolts are used for collector connections; ¾” (19 mm) Dia A325 bolts for gravity connections.
Fig. 6. Evaluation Structure: (a) collector connections; (b) 2nd Floor plan (baseline design).
3.2 ANALYTICAL MODELING

The analytical models are created using the general purpose finite element program ANSYS 1.
The models are constructed using existing data from tests of components within the composite
floor system. The models are calibrated using results from past subassemblage testing of
composite frames and bays. The models used in each phase are described in detail here:
(1) Isolated Floor System Models: The isolated floor system models encompass one floor and
associated vertical elements (gravity columns, SCBFs) of the evaluation structure (See Figure
7a). The floor system was initially modeled in the horizontal plane (with slab and framing at the
same elevation), and was subsequently extended to account for the vertical offset of the slab (See
Figure 7b). The models are subjected to NP (body force) analyses creating collector tension or
compression, as well as cyclic load protocols. It is noted that reduced degree-of-freedom versions
of this model, calibrated by the results of Phase 2, are needed for the 3D NTHA structure models.
The model consists of nonlinear 1D beam elements for framing members (including collectors);
non-linear degrading spring elements for collector and gravity connections (Main & Sadek, 2012),
and shear studs (Hwang & Kwak, 2013); 2D nonlinear orthotropic shell elements for the metal
deck, and a nonlinear 2D truss model for the concrete slab (See Figure 7c). Concrete bearing is
modeled with contact elements; slab-to-deck slip is modeled using interface elements as needed.
(a) Concrete
(See Fig. 8a)
Body force direction
Brace members
Shear stud
Rigid offset to top of flange (b) (c)
Fig. 7: Isolated Floor models: (a) Isometric View; (b) Elevation View; (c) Close-up at SFRS.
1
ANSYS Inc., Canonsburg, PA.
Collector yielding is captured in the beam elements; yielding, degradation and failure of
connections and shear studs are included directly in the nonlinear springs. The 2D truss model
used for the concrete slab is shown schematically in Figure 8a. The model, based on (Lu &
Panagiotou, 2014) (Scarry, 2014) is capable of capturing cracking and crushing of the concrete.
The model was calibrated vs. test results (Easterling & Porter, 1994) as shown in Figure 8b.
Fig.8. Concrete Truss model: (a) Schematic; (b) Calibration using test data.
(2) Collector Frame Models: The evaluation structure composite collector frame (collector,
connections, deck, slab, studs and columns) is modeled using a vertical plane 2D model. Different
models are used to investigate tension and compression limit states.
(a)
3D Collector
Participating slab
(b) (c)
Fig. 9. Collector Frame Model: (a) 2D Elements; (b) Test Calibration; (c) 3D Elements.
For tension (strength), 2D plane stress elements are used (See Figure 9a). The model is
calibrated (Figure 9b) using test results of shear tabs with (Liu & Astaneh-Asl, 2000) and without
(Main & Sadek, 2012) the composite slab. For compression (stability), the frame is represented
in the vertical plane by 3D elements with nonlinear geometry capabilities (Figure 9c). The collector
frame models are subjected to monotonic NP and cyclic body force analyses. The results from
this phase provides information on the collector characteristics for different connection and slab
bracing details, and will be used to update the isolated floor models.
3.3 ANALYTICAL STUDY MATRIX

The analytical research first focuses on existing structures with collector systems based on the
current/recent design codes. The Phase 1 and 2 models will be used to investigate: (1) axial force
distribution along the collector; (2) the limit-state sequence as the floor-system reaches the
capacity; (3) relative strength and ductility of the collector elements in comparison to other
elements in the floor system; (4) the difference in tension and compression load paths as the
collector response reverses; (5) slab effects; (6) collector stability limit states, and; (7) the
evolution of the collector load path as the collector and surrounding region takes on damage. The
parameters varied in this step will include: (1) Collector connection details; (2) Slab thickness; (3)
Deck depth and orientation; and (4) Shear stud layout. The evaluation structure will be considered
for other live load conditions (e.g. heavy manufacturing) to examine the sensitivity of collector
response to gravity load design. Bounding cases (top story, new code, etc.) will be evaluated.
The analytical research will then focus on the relative strength of the collector elements. The
design overstrength factors applied to the collectors (coll) and collector connections (conn) will
be reduced from current levels (See Table 5). The goal of the (Phase 1 and 2) analyses will be to
examine inelastic deformation demands in these intentionally “weaker” systems to determine
optimum relative strengths for the collector with respect to collector connections. The need for
special details in the collector system or slab (e.g. seismic joints) will be assessed through
deformation compatibility. In Phase 3, seismic demand reduction vs. collector inelastic demands
will be assessed for different building heights to determine optimum strength relative to the SFRS.
Table 5. Analytical Study Matrix: Collector Design Overstrength.

Case ΩColl ΩConn Objective Case ΩColl ΩConn Objective
1.0 1.0 No Overstrength Design
2.0 Current Code Design
2.0
Collector
Collector
Distributed Inelastic
Strong
Weak
1.5 Concentrated inelastic 1.5

2.0 Deformation demand
deformation at 0.7
1.0 1.0 along collector elements
collector connections
0.7 Fuse in connections 0.7 Weak collector system
4 ANALYTICAL RESULTS
Preliminary analysis results are briefly summarized here for the baseline collector design shown
in Figure 6b. Figure 10a shows the pushover curve for total diaphragm force vs. collector drift as
the inertial force increases creating compression in the collector. The loads transferred to the
SFRS by key components of the system (collector, slab frame) are shown in the figure. Individual
limit state occurrences are indicated as markers on the global response curve (offset for clarity),
providing the limit state sequence (stability is not considered for now). Diaphragm response is
seen to remain essentially linear to the design strength (ΩoNu/ = 4800kN), though local cracking
of concrete, and some yielding of shear studs and gravity system connections (in the vicinity of
the SFRS) is observed. The concrete transfers over half of the collector forces in this range; this
mechanism is essentially lost after reaching the strength limit, primarily due to shear stud failure,
followed by concrete crushing. Figure 10b shows the axial force profile in collector along the run
at the design force (Point A) and at maximum collector force (Point B). Also shown is the typical
internal force profile assumed in design. As seen, the axial force profile in collector is not linear at
design force level (prior to slab damage), contrary to design assumptions, but matches at ultimate.
(a) (b)
Fig. 10. Compression NP Results: (a) Global response; (b) Collector axial force profile.
The nonlinear profile shown in Figure 10b leads to overstressing of the shear studs near the
SFRS, which causes the shear stud failures indicated in Figure 10a. Note that slab strength and
collector strength are developed at different deformation demands, meaning that their strengths
are not additive. Figure 11 shows contour plots of concrete total strain for the compression body
force. Compression struts are shown in (a,b); tension struts in (c,d). Result before and after
damage (Points A,B) are shown in both cases. “Hotter” colors (and thicker lines) indicate
crushing/cracking. Both pairs of contours show the domination of strut actions in the linear
response range, and the concentration of demands in the slab at the SFRS in the nonlinear range.
Comp. Struts Tension. Cracking
(a) (b) (c) (d)

Fig. 11. Comp. NP Strain Contours: (a,b) before, after crushing; (c,d) before, after cracking.
Figure 12a shows the corresponding pushover curve for inertial force creating tension in the
collector. The initial stiffness of the collector, Ki, is essentially the same in tension as it is in
compression. However, as seen the global stiffness decreases noticeably during the early stages
of response due to cracking of concrete (adjacent to the SFRS). The collector deformation at the
design force is approximately 45% larger in tension than in compression. The collector connection
yields next to the SFRS at 11mm collector deformation. The collector limit state sequence under
tension is also indicated in Figure 12a. Initially, the concrete cracks. Forces redistribute to the
frame, which acts as a tie; leading to some force transfer through forwardly-inclined compression
struts (shown subsequently in Figure 13a,b), causing yielding of shear studs on SFRS and the
adjacent collector. Shear stud failure does not occur from reduction in demands due to concrete
cracking. Since the collector element is designed for stability, the collector connection yields first.
As seen in Figure 12b, the axial force profile in the collector before the concrete cracks is similar
to the profile in compression. The profile is nonlinear, with the collector element adjacent to the
SFRS transferring more force. The gradient of the profile near the SFRS explains why the shear
studs on the collector adjacent to the LFRS yielded early in the response. As seen in Figure 12b,
the axial force profile in the collector after concrete cracking remains nonlinear.
(a) (b)
Fig. 12. Tension NP Results: (a) Global response; (b) Collector axial force profile.
Figures 13a,b show the compression strut action in the concrete slab; Figures 13c,d show the
tension action. The first plot of each pair is prior to concrete cracking; the second plot of each
pair is after concrete cracking. As seen, after cracking, forces redistribute to the frame which acts
as a tie; leading to some force transfer through forwardly-inclined compression struts.
Comp. Struts Tension. Cracking
(a) (b) (c) (d)

Fig. 13. Tension NP Strain Contours: (a,b) compression; (c,d) tension (before, after cracking).
5 CONCLUSIONS AND FUTURE WORK

Nonlinear models of steel composite deck collector systems have been developed. These models
include isolated floor models (2D horizontal plane) and collector frame models (2D vertical plane).
The isolated floor models are being used to examine collector force paths in tension and
compression as various limit states are reached for different slab conditions. The collector frame
models with 2D plane stress elements are being used to determine collector strength limit states
and inelastic deformation demands for different collector connection details. Collector frame
models with 3D nonlinear geometric elements are being used to examine collector stability states
for different bracing schemes and deck orientations. The calibrated model will be used to create
reduced degree of freedom full structure models for earthquake simulations to examine collector
seismic demands. Recommendations will be made on appropriate design force patterns, levels
and details for steel composite deck collectors and their connections.
6 REFERENCES
AISC. (2010). Seismic Design Manual 2nd Edition. American Institute of Steel Construction.
AISC. (2011). Steel Construction Manual 14th edition.
ASCE. (2013). Seismic Evaluation and Retrofit of Existing Buildings (ASCE/SEI 41-13).
ASCE 7. (2010). Minimum Design Loads for Buildings and Other Structures (ASCE/SEI 7-10).
Astaneh-Asl, A. (2005).Steel Tips: Design of Shear Tab Connections for Gravity & Seismic Loads.
Braconi, A., Elamary, A., &Salvatore, W. (2010). Seismic Behavior of beam-to-column partial-
strength joints for steel-concrete composite frames. Jrnl. of Const. Steel Research, 65, 1431-44.
BSSC. (2014). Diaphragm Design Force Level. Proposal IT06-001 - Revise ASCE/DEI 7-10
Chapters 11&12. Building Seismic Safety Council, Committee IT6.
Bull, D. (1997). "Diaphragms", Seismic Design of Reinforced Concrete Structures. Technical
report No 20, New Zealand Concrete Society.
Burmeister, S., & Jacobs, W. P. (2008, December). Horizontal floor diaphragm load effects on
composite beam design. Modern Steel Construction.
Bursi, O., & Gramola, G. (1999). Behaviour of headed stud shear connectors under low-cycle high
amplitude displacements. Materials and Structures, 32, 290-297.
Chaudhari,T.,MacRae, G.,Bull, D.,Chase,G.,Hobbs,M.,Clifton,C.,& Hicks,S. (2014). Composite
slab effects on beam-column subassemblies: Further development. 2014 NZSEE Conference.
Clifton, G., & El Sarraf, R. (2005). Composite Floor Construction Handbook. Namukau City, New
Zealand: HERA Report R4-107. N.Z. HERA.
Cowie,K., Hicks,S., MacRae,G., Clifton,G., & Fussell,A. (2013).Seismic Design of Composite
Metal Deck & Concrete-filled Diaphragms-a discussion paper. Steel Innovation Conference 2013.
Davies, J. (1986). A General Solution for the Shear Flexibility of Profiled Sheets II: Application of
the Method. Thin-Walled Structures, 4, 151-161.
Easterling, W., & Porter, M. (1994). Steel Deck Reinforced Concrete Diaphragms I. Journal of
Structural Engineering, 120(2), 560-576.
EERI. (1994). Preliminary Report Northridge, California, Earthquake of January 17, 1994.
www.eeri.org/products/reconnaissance-reports/preliniary-report-northridge-eq-of-jan-17-1994-2/.
FEMA. (2009). NEHRP Recommended Seismic Provisions for New Buildings & Other Structures.
Hedaoo, N.A., Gupta, L.M. & Ronghe, G.N. (2012). Design of composite slabs with profiled steel
decking: A comparison between experimental & analytical studies. Int. Jrnl. of Adv. Str. Eng.
Helwig, T., & Yura, J. (1999). Torsional Bracing of Columns. ASCE STJ., 125(5), 547-555.
Herrera, R., Ricles, J., & Sause, R. (2008). Seismic Performance Evaluation of a Large-Scale
Composite MRF using Pseudo-Dynamic Testing. ASCE STJ., 134(2), 279-288.
Hwang, J.-W., & Kwak, H.-G. (2013). Improved FE model to simulate interfacial bond-slip
behavior in composite beams under cyclic loading. Computers and Structures, 164-176.
Kasai,K., & Matsuda,Y.(2016). Cyclic Behavior of Comp. Beam, Column, & Gusset Pl. Connection
under Multiple Loads from Frame & Damper Actions. AISC Connections Workshop VIII. Boston.
Liu, J., & Astaneh-Asl, A. (2000). Cyclic Testing of Simple Connections Including Effects of Slab.
Journal of Structural Engineering, 32-39.
Lowe,D., Das,R., & Clifton,C. (2014). Characterization of the Splitting Behavior of Steel-Concrete
Composite Beams with Shear Stud Connection. Procedia Material Science, 3, 2174-2179.
Lu, Y., & Panagiotou, M. (2014). Three-Dimensional Cyclic Beam-Truss Model for Nonplanar
Reinforced Concrete Walls. Journal of Structural Engineering.
MacRae, G., & Clifton, G. (2015). Research on Seismic Performance of Steel Structures. Steel
Innovations 2015 Conference. Auckland, NZ.
MacRae,G., Clifton,G., Mackinven,H., Mago,N., Butterworth,J.,& Pampanin,S. (2010).The Sliding
Hinge Joint Moment Connection. Bulletin of New Zealand Society for EQ Eng., 43(3), 202-212.
Maffei, J. (2008). Floor diaphragms, collectors, and podium and backstay effects in tall buildings.
http://peer.berkeley.edu/tbi/wp-content/ /2010/09/Maffei_LATBSDC_9may08_without_extra-.pdf
Main, J. A., & Sadek, F. (2012). Robustness of Steel Gravity Frame Systems with Single-Plate
Shear Connections. U.S. Dept. of Commerce. National Institute of Standards and Technology.
Majdi, Y., Hsu, C.-T. T., & Punurai, S. (2014). Local bond-slip behavior between cold-formed
metal and concrete. Engineering Structures, 69, 271-284.
Royal Commission. (2012), Canterbury Earthquakes Royal Commission, Christchurch NZ.
http://canterbury.royalcommission.govt.nz/Final-Report---Volumes-1-2-and-3
NZS 3404. (2007). Steel Structures Standard, Standards New Zealand. Wellington, New Zealand.
Rogers, C. A., & Tremblay, R. (2003). Inelastic Seismic Response of Side Lap Fasteners for Steel
Roof Deck Diaphragms. Journal of Structural Engineering, 129(12), 1637-46.
Sabelli, R., Sabol, T. A., & Easterling, W. (2011). Seismic Design of Composite Steel Deck &
Concrete-filled Diaphragms. National Institute of Standards and Technology.
Scarry, J. (2014). Floor diaphragms– Seismic bulwark or Achilles’ heel. 2014 NZSEE Conference.
Souici, A., Berthet, J., Li, A., & Rahal, N. (2013). Behavior of both mechanically connected and
bonded steel-concrete composite beams. Engineering Structures, 49, 11-23.
Wennberg, D., Wennhage, P., & Stichel, S. (2011). Orthotropic Models of Corrugated Sheets in
Finite Element Analysis. International Scholarly Research Network, 2011.
Zhang, X., & Ricles, J. (2006). Experimental Evaluation of Reduced Beam Section Connections
to Deep-Columns. Journal of Structural Engineering, 132(3), 346-357.
EVALUATION OF STRENGTH AND STIFFNESS PREDICTIONS FOR
STEEL DECK DIAPHRAGMS WITH STRUCTURAL CONCRETE FILL
Patrick O’Brien
Virginia Tech, M.S. Student
Blacksburg, VA, USA
pato91@vt.edu
W. Samuel Easterling
Virginia Tech, Montague-Betts Professor of Structural Steel Design and Department Head
Blacksburg, VA, USA
seaster@vt.edu
Matthew R. Eatherton
Virginia Tech, Associate Professor
Blacksburg, VA, USA
meather@vt.edu
ABSTRACT
Steel deck floor and roof systems, common to steel framed buildings, distribute lateral forces to
vertical frames through diaphragm action. Floor diaphragms offer significant shear resistance,
primarily due to the concrete fill placed on the steel deck. Current diaphragm design documents
(AISI, 2013a; Luttrell et al., 2015) offer shear strength and stiffness expressions for steel deck
diaphragms with structural concrete fill, hereby referred to as SDDCFs. In this paper, a shear
strength expression, as proposed in Porter and Easterling (1988), and an alternative shear
stiffness equation are compared to current design procedures and test results from the extensive
Iowa State University experimental program on SDDCFs (Porter and Greimann, 1980; Porter and
Easterling, 1988; Easterling and Porter, 1994a, 1994b). A resistance factor is also recommended
based on available test data. It is shown that the proposed equations result in a 46% and 52%
improvement in predicting experimental strength and stiffness, respectively.
INTRODUCTION
Steel deck is used for its many advantages, including light weight, low cost, and ease of
installation, among others. In addition to a chemical bond, embossments rolled into composite
steel deck improve the mechanical bond with the concrete placed on top, thus creating a
composite system. In this case, steel deck acts as the flexural reinforcement for gravity loads,
negating need for flexural reinforcing bars. Steel headed stud anchors used for composite beams
also promote composite action. When properly constructed, these systems exhibit significant
strength and stiffness when loaded in-plane. A schematic for a typical SDDCF in a steel framed
building is shown in Figure 1. Adjacent steel sheets are attached to each other at panel edges
Endlap fastener to
supports Sidelap
Pour strip fastsener
or edge Concrete fill
angle
Composite
beam
Shear studs for

collectors
Brace member of vertical
lateral force resisting
system
Figure 1 – Concrete filled diaphragm components
through sidelap connections. Panels are usually butted at their ends creating an ‘endlap’
condition, and fastened to the supporting frame with deck-to-support fasteners, hereby referred
to as structural fasteners, ensuring there is no gap below for the concrete to fall through when
placed. Steel headed stud anchors, hereby referred to as shear studs, are fastened through the
deck directly to underlying composite beams. Temperature and shrinkage steel (e.g. welded wire
fabric) for the concrete slab may also be added.
Research initially focused on this system’s ability to resist gravity loads before examining in-plane
resistance. Due to the proprietary nature of many diaphragm testing programs, and the extensive
cost and time needed for construction, concrete curing, and disassembly, only a very limited
amount of experimental research is available on SDDCFs.
A series of eleven monotonic cantilever tests, nine with insulating concrete fill utilizing welded
structural fasteners, was carried out by Luttrell (1971). Concrete compressive strengths (f’c)
ranged from 0.7 MPa to 1.2 MPa. Ultimate experimental shear strengths ranged from 8 – 35 kN/m
with shear stiffness values ranging from 6 – 185 kN/mm for the nine specimens with concrete fill.
Davies and Fisher (1979) tested four full scale cantilever SDDCF specimens with nominal f’c = 25
MPa. Three of four specimens incorporated re-entrant profile steel deck types (uncharacteristic
of North American SDDCF construction), with the final specimen using the more traditional
trapezoidal steel deck profile. Ultimate diaphragm shear strengths, predominantly limited by
failure of the structural fasteners, which were screws, ranged from 15 to 29 kN/m.
A series of fourteen diaphragm tests were performed by ABK (1981), with only one specimen
including structural concrete fill (nominal f’c = 20.7 MPa). The specimen was tested in a three
span, simply supported condition with welded structural fasteners and subjected to displacement
histories at a dynamic rate. The specimen, however, was not loaded to failure; the maximum
recorded shear strength was 41.6 kN/m. In combination with the Iowa State University
experimental research program on SDDCFs, as discussed in detail in the following section, this
test represents the only other SDDCF specimen subject to reversed-cyclic loading, which is a
more representative loading protocol to simulate seismic demands. Fukuda et al. (1991) also
tested similar composite-type specimens in reversed cyclic loading, but included geometric
irregularities, thus dismissing experimental results from consideration in this paper. A substantial
amount of private, proprietary testing on SDDCFS was also performed in the 1960s through 1980s
by S.B. Barnes and Associates, as identified by Porter and Easterling (1988).
The findings of these research programs and proprietary testing data led to the development of
SDDCF design equations provided in the Steel Deck Institute’s Diaphragm Design Manual, 4th
edition (Luttrell et al., 2015), widely considered as the principal document for diaphragm design
in North America and hereby referred to as DDM04. The Seismic Design for Buildings Manual,
commonly referred to as the Tri-Services Manual, is a less utilized design guide that also
incorporates findings from earlier experimental research (Army, Navy and Air Force, 1982). The
American Iron and Steel Institute (AISI) produced a consensus document summarizing these
design guides into AISI S310-13, North American Standard for the Design of Profiled Steel
Diaphragm Panels (AISI, 2013a), hereby referred to as AISI S310.
This paper evaluates current strength and stiffness equations available in AISI S310 and DDM04,
and recommends alternative strength and stiffness equations that better represent the mechanics
of shear resistance in a SDDCF. Strength and stiffness predictions are calculated for the SDDCF
specimens tested in the Iowa State University experimental program (Porter and Greimann, 1980;
Porter and Easterling 1988). Furthermore, resistance factors per AISI S310 are also calculated
for the recommended shear strength prediction equation.
IOWA STATE UNIVERSITY EXPERIMENTAL PROGRAM

A series of 32 full scale SDDCF specimens were tested at Iowa State University’s (ISU) Structural
Engineering Research Laboratory in two phases. The first phase tested specimens 1 – 9, reported
by Porter and Greimann (1980), and identified a need for further testing. The second phase
included specimens 10 – 32 and are reported in Porter and Easterling (1988). Six of these
specimens included a combination of in-plane and vertical loads, as examined by Neilson (1984).
Figure 2 shows a schematic of the diaphragm test frame setup. Table 1 gives general information
about each test specimen setup. All specimens were cantilever tested and subjected to a reversed
cyclic loading protocol applied at a quasistatic rate. Plan dimensions of specimens 1 - 21 were
4.57 m x 4.57 m. Specimens 22 – 32 plan dimensions were 3.66 m x 4.57 m. The 3.66 m
dimension was the span dimension, oriented perpendicular to the applied load. A concrete
reaction block, post tensioned with anchor rods, comprised the fixed depth dimension (parallel to
applied load) of the cantilevered diaphragm test setup. In order to fasten the steel deck to the test
frame, a steel plate was embedded in the concrete reaction block. Shored construction negated
the need for intermediate support test frame members. All specimens used arc spot welds, shear
W24 x 76 except spec. 32
Depth = 4.57 m Hydraulic actuators.

N 890 kN Load-displacement

Span = 4.57 m or 3.66 m
load cell recorded at N

Direction of deck span corners
for all specimens except
spec. 4 (oriented at 90° Note: Spec. 32
to direction indicated) used W14 x 22
Reinforced concrete perimeter framing
anchor block with post- sections
tensioned anchor rods Embedded steel
to the strong floor plate for diaphragm Flexible
attachment Tee
Figure 2 – Test frame for Iowa State University SDDCF program
Table 1 – ISU Test Setup Variables
Spec. Measured Steel Deck Properties Measured Conc. Properties Connections
ID Thickness (mm) Fy (MPa) Fu (MPa) Thicknessa (mm) f’c (MPa) per side
1 0.86 288 368 137 38.8 30 studs
2 0.86 288 368 140 36.2 30 studs
3 0.86 288 368 144 28.0 60 welds
4 0.86 288 368 134 26.5 60 welds
5 1.57 332 419 90 20.5 30 welds
6 1.57 332 419 189 31.4 60 welds
7 1.47 343 421 137 37.5 60 welds
8 0.89 288 368 139 23.1 4 studs (N,S)
6 studs (E,W)
9b 1.47/1.45 357/361 436/447 139 37.3 60 welds
10 1.57 279 367 140 22.8 60 welds
11 1.19 619 646 145 24.3 60 welds
12c 1.57 279 367 142 23.5 60 welds
13b,c 1.47/1.45 357/361 436/447 140 42.7 60 welds
14c 1.57 279 367 208 25.5 60 welds
15 1.19 619 646 107 19.6 60 welds
16c 1.19 619 646 106 20.4 60 welds
17c 1.57 317 375 189 29.4 60 welds
18c 1.57 279 367 141 21.0 60 welds
19 1.57 341 383 146 18.5 60 welds
20 0.94 335 388 141 27.4 40 welds
21 1.57 279 367 144 25.1 15 welds
22 1.57 279 367 144 22.8 60 welds (N,S)
48 welds (E,W)
23 0.94 335 388 146 24.1 40 welds (N,S)
34 welds (E,W)
24 1.57 341 383 143 27.9 48 welds
25 1.57 279 367 145 32.2 16 studs (N,S)
8 studs (E,W)
26 0.91 640 645 120 23.9 8 studs 15 welds (N,S)
7 studs (W) 11 studs (E)
27 0.94 335 388 144 19.9 8 studs 15 welds (N,S)
9 welds (E,W)
28 0.94 335 388 142 24.9 8 studs 15 welds (N,S)
6 studs (E,W)
29 0.89 599 619 141 19.9 16 studs (N,S)
11 studs (E,W)
30 0.89 599 619 144 24.6 12 studs 4 welds (N,S)
7 studs (E,W)
31 0.89 599 619 146 23.0 23 welds (N,S)
13 welds (E,W)
32 0.89 599 619 144 16.9 30 welds (N,S)
23 welds (E,W)
a = thickness measured from top of concrete fill to bottom of steel deck panel
b = cellular deck; steel deck properties reported as flat sheet/corrugated sheet
c = combination of gravity and in-plane loads
studs, or a combination of both as structural fasteners. Normal weight structural concrete fill was
placed on top of steel deck for all specimens except specimen 26, which used lightweight
concrete. Steel deck profile dimension details for the different deck types can be found in
Easterling and Porter (1988).
The reversed cyclic loading protocol specified increasing displacement steps after a minimum of
three cycles. Due to cyclic strength degradation, loads at a specific displacement step dropped
during successive cycles. A second criteria for progressing to the next displacement step was for
the peak load of a cycle to stabilize within a 5% difference from the previous cycle.
FAILURE MODES
The restraining effects that concrete imposes on steel deck dismisses several bare deck
diaphragm limit states (i.e. steel deck warping, plate-like shear buckling and out-of-plane
movements are no longer of concern). Three major limit states were identified during the ISU
research program for SDDCFs, only two of which are deemed practical to typical construction.
Shear transfer failure is the least probable of the three limit states, as a set of unusual construction
conditions must apply. Shear transfer failure describes the failure of a SDDCF due to separation
of the concrete from steel deck. To create strong composite action, embossments are rolled into
the steel deck. Additionally, shear studs are the standard fasteners used in this system, and allow
for a direct load path between the structural frame and concrete, promoting composite behavior
at composite beam locations. Aside from these mechanical bonds, the chemical bond between
the concrete and deck allows for additional shear transfer strength.
For this failure mode to occur, an unusually large amount of structural fasteners must be used
without any shear studs, or other structural fasteners that allow for a direct mechanical connection
between steel frame and concrete fill (e.g. standoff screws). This ensures a strong connection
between deck and frame, but a weaker connection between deck and concrete. Test setup
configurations likely to exhibit this failure mode included specimens with 13 welds/m. Specimens
also subjected to gravity loads were less likely to fail in this fashion, due to normal loads imposing
additional shear resistance from increased deck-concrete friction (Neilsen, 1984). Although this
failure mode was investigated in the ISU program, it does not represent practical construction and
is excluded from the limit state equations of a SDDCF further examined in this paper.
Perimeter fastener failure occurs when there is a fracture of structural fasteners that transfer
forces between the structural frame and the deck-concrete system. This failure results in relative
motion between the steel deck-concrete and the supporting steel beams. Steel deck sidelap
connection strength is of little importance since the majority of the shear is transferred through
the concrete and the deck-concrete bond prevents slip at sidelaps. The ISU program identifies a
diaphragm ‘edge zone’ over which shear forces are transferred to and from the steel frame (Porter
and Easterling, 1988). Structural fasteners to intermediate beams oriented perpendicular to
collectors can lie within this edge zone and contribute to the collector’s perimeter fastener shear
strength. The effects of connector force distributions at SDDCF perimeters are further analyzed
in Widjaja (1993). For the ISU program, which did not employ intermediate supports, perimeter
fastener limit shear strength is calculated by summing the total strength of the structural fasteners
in both the depth and span dimension of the SDDCF, with the lesser value controlling.
Diagonal tension failure of a SDDCF occurs when the concrete placed on the steel deck behaves
as a deep concrete beam and fails through shear cracking. Adequate concrete cover must also
be provided to eliminate possible longitudinal cracks. Fire ratings typically govern the minimum
depth of the concrete and provide enough concrete cover to alleviate this concern. Failure of the
concrete will occur at a 45° angle to the diaphragm edge, initiating a diagonal crack. This behavior
is analogous to shear failure of concrete beams. Although steel reinforcement (e.g. welded wire
fabric for shrinkage and thermal effects) in the slab may impact shear strength, it is not explicitly
considered in strength predictions for SDDCFs.
CURRENT STRENGTH EQUATION

AISI S310 and DDM04 provide a strength equation for SDDCFs, as shown in eq. (1). The
diaphragm shear strength, Sn, is the summation of a steel deck connector contribution and
concrete fill contribution. The concrete fill contribution (second term of summation in eq. (1)) is
the product of four variables. The product of unit width, b (1000 mm), and concrete cover (between
top of steel deck and top of concrete fill), dc (mm), represents the concrete cross sectional shear
area per unit width, with specified concrete compressive strength, f’c (MPa) The structural
concrete strength factor, k, as shown in eq. (4), is dependent on the concrete mix unit weight, w2.
The steel deck contribution to a SDDCF’s shear strength (first term of summation in eq. (1)) is a
statics based expression that considers the shear strength and location of each fastener along a
panel. The term β is a fastener distribution factor that when multiplied by the fastener shear
strength, Pnf, and divided by the panel’s length, L, gives a shear strength per unit length for the
steel deck’s contribution towards a SDDCF’s diaphragm strength. Figure 3 shows a free body
diagram of a steel panel restrained to in-plane deformations only. The deck contribution is limited
to 25% of total shear strength of SDDCF. This rational limit, as described in AISI S310
commentary, is to ensure that strength contribution from the panels for thicker decks and heavier
fastener configurations is not overstated. Note that Pnf for shear studs is specified to be calculated
as an arc spot weld for application in eq. (1). Conversely, AISC 360 equations for shear studs is
to be used when checking diaphragm strength limited by perimeter fastener. DDM04 specifies
the perimeter fastener strength be checked to ensure development of Sn, as given in eq. (1).
The fastener distribution factor, β (see eq. (2)), considers the location of fasteners along a panel,
and effectively reduces the diaphragm strength contribution of a fastener dependent on its
location with respect to the panel width centerline. DDM04 acknowledges that yielding can first
occur at the edge-most fasteners. The progression of yielding, assuming adequate ductility, can
allow for the fasteners towards the panel centerline to approach their capacity. However, brittle
behavior is assumed and capacities for fasteners towards the panel centerline are scaled down
l
Exterior Pnf
Support Pnf Pns Pnf Pns
Sidelap
Pnf1 Pnf1 w/2
Pnf1
Sn w Panel Span Xp1 Pnf2
Sn w
Pnf2 Pnf2 Xp2
w
Xe2 Pnf2 Pnf2
Pnf2 Xe1 One deck Interior
Pnf1 panel Pnf1 Support Pnf1
Sidelap
Pnf Pns Pnf Pns Pnf

Structural Fastener
Sidelap Fastener
Figure 3 – Shear forces on a steel deck panel without concrete fill
(1)
(2) β 2 4
∑ ∑
(3) α ,
.
(4) , w2 in kg/m3
proportional to the distance between the fastener and the panel centerline (i.e. Pnf1 = Pnf[Xe1/(w/2)]
in Figure 3). The term ns in eq. (2) gives the amount of sidelap fasteners along a single panel
edge (ns = 2 for Figure 3); αs is the sidelap shear strength, Pns, divided by structural fastener shear
strength, Pnf. Expressions for sidelap and structural fastener shear strengths are available in AISI
S310 and DDM04. The number of intermediate supports is represented by np (np = 1 in Figure 3).
The term α1 and α2 in eq. (3) are structural fastener distribution factors (xe, xp, and w are further
defined in Figure 3).
PROPOSED STRENGTH EQUATION

The method described in the previous section uses the same approach to calculate diaphragm
strength for SDDCFs as bare deck diaphragms. However, this implies assumptions that don’t
accurately represent the mechanics of shear load resistance in a SDDCF. The strength
contribution from the deck for a SDDCF is dependent on the deck thickness and interfacial deck-
concrete bond, and is not dependent on fastener strength. Thus, fastener strengths need only to
be calculated when checking the perimeter fastener failure mode. The concrete strength
contribution in eq. (1) characterizes the shear strength of the concrete placed above the top flute.
However, diagonal tensions cracks can pass through the concrete in the flutes therefore
contributing towards diaphragm strength, especially when deeper panels with gradual changes in
panel profile dimensions between top and bottom flutes are used.
To address these issues, a method for predicting SDDCF strength limited by diagonal tension
failure, as proposed in Porter and Easterling (1988), is recommended here. ACI 318 (Section
22.5.5.1) defines a range of concrete shear strengths for beams without axial force, as shown in
eq. (5) (ACI, 2014). The product of beam width, bw (mm), and depth, d (mm), in eq. (5) represents
the beam shear area, with λ serving as a lightweight concrete factor (f’c is as previously defined).
Porter and Easterling (1988) take a similar approach by idealizing a SDDCF as a deep, horizontal
beam and calculating an equivalent cross sectional shear area.
Concrete Embossments dc
Fill
Note: average concrete fill

tc thickness, tc, assumes area of
concrete fill in flute is equal to
Dd area in adjacent ‘empty’
Dd / 2 corrugation such that A1 = A2
A1 A2
Figure 4 – SDDCF profile dimensions
Equation (6) is the SDDCF shear strength equation recommended by this work as reported by
Porter and Easterling (1988). A value of 0.26 is given for the concrete strength factor, k. Concrete
compressive strength, f’c, is as previously defined. The product of the average concrete thickness,
(5) ,k 0.17 to 0.29 per ACI‐318
(6) _ ,k 0.26
(7)
te (mm) and diaphragm width, b (mm), represent the cross sectional shear area of the SDDCF,
as defined in the following. Note that using a unit width, b, will result in SDDCF shear strength per
unit length. Equation 7 gives an equivalent average concrete thickness, te, defined as the
summation of the average depth of concrete, tc (as shown in Figure 4), and an equivalent
transformed concrete thickness for the steel deck. The steel deck’s shear strength contribution is
accounted for by converting the steel deck thickness, tcd, into an equivalent concrete thickness
through a shear modular ratio, nsc. The d / s term is a ratio that reduces the shear strength of a
panel the more its profile geometry diverges from its equivalent flat plate-like condition, with d
equal to a panel corrugation’s pitch and s equivalent to a developed flute width. DDM04 contains
calculation procedures for s. Note that the recommended eq. (6) is independent of fastener
strengths.
Table 2 compares test and predicted strength values for the ISU SDDCF specimens that failed in
diagonal tensions cracking using the DDM04 equation and Easterling (1988) equation. Test to
predicted ratios are calculated by dividing the maximum experimental shear strength by the
predicted strength. Test to predicted ratios using DDM04 predictions (eq. (5)) and the
recommended prediction equation (eq. (6)) are also given in Table 2. Specimens are grouped by
whether studs were included. Average values are given for both groups individually, and a total
average for all specimens, regardless of fastener type, is presented at the bottom of Table 2.
DDM04 prediction equations can give overly conservative values, as shown by an average test
to predicted ratio for of 1.54 for all specimens. This ratio reduces to a more accurate value of 1.08
when using eq. (6), marking a 46% improvement over the DDM04 prediction methods.
It is also worth noting that the variability in predicted strengths using eq. (6) reduces in comparison
to DDM04 values, as shown by a standard deviation of 0.10 for eq. (6) vs. 0.20 for DDM04. When
using eq. (6), the test / predicted ratios closer to unity in combination with lower standard
deviations indicate a more reliable prediction method than DDM04’s procedure. It is important to
note that these expressions are calibrated to the ISU experimental data, with a specific range of
measured compressive strengths ranging from 16.9 - 42.7 MPa. Table 2 and Figure 5 compare
test vs. predicted values for strength and stiffness. Strength is only reported for diagonal tension
failure (DT). For the remaining specimens, the failure mode S refers to shear transfer failure mode
and P refers to the perimeter fastener failure mode. Specimens listed with multiple failure modes
indicates specimens where a clear distinction between the failure modes was not possible (Porter
and Easterling, 1988).
S Test vs Predicted G' Test vs Predicted

250 1400
DDM04 DDM04
G' Predicted (kN/mm)
1200
S Predicted (kN/m)
200
Proposed Eq. (6) 1000 Proposed Eq. (11)
150 800
100 600
400
50
200
0 0
0 50 100 150 200 250 0 100 200 300 400
S Test (kN/m) G' Test (kN/mm)
Figure 5 – Strength and stiffness: test to predicted comparison
RESISITANCE AND SAFETY FACTORS

The current resistance factor specified in AISI S310 for all steel deck diaphragms utilizing
structural or insulating concrete fill, regardless of the structural fastener type used, is Φ = 0.5 (for
LRFD). If the DDM04 strength equation is coupled with this resistance factor, design strengths for
SDDCFs are dramatically reduced in comparison to their experimental strengths. This relationship
is indicated by an average test/predicted ratio of 3.02 for all specimens failing in diagonal tension
when using the DDM04 equation with Φ = 0.5. Considering the accuracy and low variability in test
to predicted ratios for eq. (6), a need to calculate a new Φ factor was identified. AISI S310 uses
eq. (8) to calculate resistance factors for a rational analytical model when corresponding large-
scale diaphragm tests are available.
Table 2 – Strength and stiffness test to predicted comparisons for ISU specimens
Spec. Sn Testa Test / Predicted G’ Test Test / Predicted
ID (kN/m) DDM04 Eq. (6) (kN/mm) DDM04 Eq. (11)
Welded
3 S 198 0.34 0.75
4 S 236 0.49 1.00
5 113 1.46 1.15 299 0.80 1.95
6 S 331 0.24 0.75
7 S 234 0.36 0.80
9 214 1.68 1.11 236 0.35 0.79
10 157 1.75 1.14 282 0.64 1.20
11 S 310 0.57 1.21
12 175 1.88 1.24 300 0.65 1.24
13 243 1.75 1.18 354 0.48 1.10
14 S 322 0.33 0.81
15 S and DT 198 0.42 1.04
16 121 1.19 1.05 161 0.34 0.84
17 S 279 0.21 0.65
18 157 1.81 1.17 277 0.65 1.22
19 143 1.44 1.05 163 0.35 0.70
20 S and P 228 0.35 0.86
21 S and P 152 0.31 0.65
22 164 1.73 1.15 290 0.61 1.17
23 S and P 240 0.38 0.93
24 163 1.40 1.02 290 0.49 1.07
31 S and P 236 0.46 1.00
32 S and P 161 0.39 0.80
Average 165 1.61 1.13 251 0.45 0.98
Std. dev 37.3 0.21 0.06 57.8 0.15 0.28
Including Shear Studs
1 163 1.35 0.96 266 0.42 0.94
2 181 1.48 1.07 343 0.54 1.22
8 P 118 0.25 0.54
25 175 1.54 1.05 302 0.52 1.13
26b 84.6 1.26 0.81 279 0.50 1.27
27 P 307 0.57 1.34
28 P 277 0.45 1.10
29 133 1.40 1.03 330 0.75 1.57
30 P 269 0.51 1.12
Avg. 147 1.41 0.98 277 0.50 1.14
Std. Dev. 35.5 0.10 0.09 61.7 0.13 0.27
Avg. Total 159 1.54 1.08 258 0.46 1.02
Std. Dev. Total 37.6 0.20 0.10 60.0 0.14 0.29
bDT = diagonal tension failure, P = perimeter fasteners failure, S = shear transfer failure
aUsing eq. (6) with lightweight concrete reduction factor of λ = 0.75
Equation (8) utilizes calibration methods and probability analysis concepts considering several
sources of variability. Calibrations factors consider construction variability (Fm, Vf), material
variability (Mm, Vm), test data scatter (Pm, Vp), load effects (CΦ, VQ) and more. AISI S310 further
defines the variables in eq. (8). Using prescribed AISI S310 calibration factors, as given in Table
3, results in a calculated resistance factor (using eq. (8)) of Φ = 0.58 with ASD safety factor, Ω =
2.74. Because calibration factors considered for SDDCFs are limited by both the fasteners and
the concrete slab, a worst case target reliability index, β0 = 3.5, is assigned to the current
resistance factor in AISI S310. However, for the proposed shear strength equation constrained to
the diagonal tension failure mode, concrete failure will be the limiting condition. Therefore, the
conservative β0 = 3.5 may no longer be appropriate. Instead, diagonal tension failure is treated
as a member limit state rather than a connection limit state. As such, AISI specifies β0 = 2.5 and
an ‘adapted’ resistance factor of Φ = 0.79 (Ω = 2.74) is calculated as shown in Table 3. Further
reliability studies are recommended to evaluate whether calibration factors as given by AISI S310
are appropriate, before conclusively proposing a resistance.
(8) ∅
Table 3 – Calibration factors and calculated resistance factors for diagonal tension limit state
Calibration
CΦ VQ Mm VM Fm VF VPa na CPa Pma β0 Φ
Method
AISI S310 3.5 0.58
1.6 0.25 1.10 0.10 0.90 0.10 0.095 15 1.24 1.08
Adapted 2.5 0.79
aCalculated using ISU test data
CURRENT STIFFNESS EQUATION

Equation (9) is AISI S310’s shear stiffness equation, G’ (kN/mm), for SDDCFs. Like the strength
equation, eq. (9) is the summation of a steel deck term and concrete term. The steel deck term is
equivalent to DDM04’s G’ equation for steel deck diaphragms without concrete, with the only
modification being that a warping term is no longer applicable, due to the restraining effects of the
concrete fill on the steel deck. The first term in the denominator of Eq. (9) is a mechanics based
expression that accounts for the stiffness of the panel in pure shear (Poisson’s ratio for steel of υ
= 0.3) with an applied factor of s / d (as previously discussed) effectively decreasing G’ for
corrugated sheets. The second term in the denominator, C, is a fastener slip term that accounts
for movement of a steel deck panel with width, w, due to both structural and sidelap fastener
flexibility, as shown in eq. (10). AISI S310 provides expressions for structural fastener (Sf) and
sidelap fastener (Ss) flexibility. However, no expression is available for shear studs. Instead, shear
stud flexibility is calculated as an arc spot weld. Equations (9) and (10) E and t terms correspond
to the steel deck’s elastic modulus (MPa) and thickness (mm), respectively. Equation (10) uses a
concrete cover thickness, dc, as shown in Figure 4. The concrete term follows classical mechanics
stiffness derivations for concrete material in shear and is a semi-empirical expression adjusted to
fit test data, hence the 0.7 exponent and 0.786 coefficient (Mattingly, 2017).
.
(9) 0.786
ʋ
(10)
PROPOSED STIFFNESS
A few modifications to the current G’ equation are proposed here. There are several sources of
flexibility that may contribute to an SDDCF’s shear stiffness. First, flexural stiffness of the beams
parallel to the diaphragm’s span may contribute to G’, but the effects are often considered
negligible. In experiments, support movement of the test frame may also contribute to flexibility,
but published load-deformation data is often corrected to account for these movements. The most
significant contribution to G’ is due to the composite deck/concrete system. There is also the
potential for slip between the deck and the supporting steel frame due to the structural fasteners’
flexibility, as captured in the fastener slip term, C, in eq. (10). This effect may be more prevalent
for SDDCFs utilizing a light structural fastening configuration. The sidelap fastener flexibility,
however, will not contribute due to the deck-concrete bond and its restraining effects on the deck.
As such, a minor modification to C to exclude sidelap flexibility term is suggested, as shown in
eq. (12).
A less usual source of flexibility may occur from the deck-concrete interface de-bonding at a
diaphragm’s ‘edge zone’. This behavior is only observed for diaphragms with an unusually large
amount of structural fasteners, but no shear studs which is atypical for SDDCFs, and is therefore
not considered further. The concrete term in eq. (9) only considers the concrete cover, dc. Instead,
an average concrete thickness, te, as previously defined and used in the proposed eq. (6) is
recommended to account for the stiffness contributions of the concrete in the steel deck flutes
and the steel deck itself. The coefficient and exponent in the concrete term was calibrated to give
best fit predictions to experimental results, as shown in Table 2, by an average test/predicted ratio
of 1.02 for all 32 specimens, in comparison to DDM04’s average value of 0.46, giving a 52% more
accurate prediction. Prediction values are compared to experimentally obtained G’ calculated as
a secant stiffness at 40% Smax, in keeping with AISI specified test standards and as shown in
Figure 6 (AISI, 2013b).
(11) 0.40
(12)
CONCLUSIONS
SDDCF strength and stiffness predictions were evaluated and compared to experimental results
currently available from the ISU testing program. Current methods for predicting strength were
found not to accurately represent the mechanics of a SDDCF subjected to shear. Alternative
equations to predict SDDCF strength and stiffness were proposed. The resulting expressions give
average test to predicted ratios of 1.08 and 1.02 for strength and stiffness, respectively. This
marks 46% and 52% improvements when compared to DDM04 strength and stiffness methods,
respectively. Reduced variability when using the recommended prediction equations is also
noted. A resistance factor of Φ = 0.58 (LRFD) for the proposed strength equation is calculated
using prescribed calibration factors given in AISI S310. Furthermore, an ‘adapted’ resistance
b = Depth δ (Specimen 12)

Shear Strength, S/Smax
Smax
G’
P
a = Span
γ=δ /a
S=P/b S40%
γ
0 0.002 0.004 0.006

Shear Angle, γ (rad)
a.) Cantilever test schematic b.) 1st quadrant of load-deformation data
Figure 6 – Determination of G’ from ISU experimental data
factor of Φ = 0.79 (LRFD) is calculated when using a target reliability index β0 = 3.5, appropriate
for member limit states as applicable to diagonal tension failure investigated in this paper.
ACKNOWLEDGEMENTS
This work was part of the Steel Diaphragm Innovation Initiative (SDII) which is funded by AISI,
AISC, SDI, SJI, and MBMA. Input from Ben Schafer and Jerry Hajjar was instrumental. This
material is based upon work supported by the National Science Foundation under Grant No.
(CMMI-1562669). Any opinions, findings, and conclusions or recommendations expressed in this
material are those of the authors and do not necessarily reflect the views of the National Science
Foundation or other sponsors.
REFERENCES
ACI Committee 318. (2014) Building code requirements for structural concrete: (ACI 318-14).
Farmington Hills, Michigan.
ABK, A Joint Venture. (1981). Methodology for mitigation of seismic hazards in existing
unreinforced masonry buildings: diaphragm testing. El Segundo, California.
AISC 360 (2010). Specification for structural steel buildings. American Institute of Steel
Construction.
AISI (2013a). North American standard for the design of profiled steel diaphragm panels, AISI
S310-13. American Iron and Steel Institute.
AISI (2013b). Test Standard for Cantilever Test Method for Cold-Formed Steel Diaphragms,
AISI S907-13. American Iron and Steel Institute.
Davies, J., & Fisher, J. (1979). The diaphragm action of composite slabs. Proceedings of the
Institution of Civil Engineers. DOI: http://dx.doi.org/10.1680/iicep.1979.2780
Departments of Army, Navy and Air Force. (1982). Seismic design for buildings (Tri-services
manual). Government Printing Office.
Easterling, W. S., and Porter, M. (1994a, 1994b). Steel deck reinforced concrete diaphragms.
I and II. Journal of Structural Engineering, 120(2), pp. 560–576, and 577-596.
Luttrell, L. (1971). Shear Diaphragms with lightweight concrete fill. Paper presented at the 1st
International Specialty Conference on Cold-Formed Steel Structures. Paper 2.
Luttrell, L., Mattingly, J., Schultz, W., & Sputo, T. (2015). Diaphragm design manual, fourth edition.
Steel Deck Institute.
Mattingly, J. (2016), Personal communication, August 29, 2016.
Neilsen, M. (1984). Effects of gravity load on composite floor diaphragm behavior (Master’s
thesis). Iowa State University, Ames, Iowa.
Porter, M., & Easterling, W. (1988). Behavior, analysis, and design of steel-deck-reinforced
concrete diaphragms (Report no. 88305). Iowa State University, Ames, Iowa.
Porter, M., & Greimann, L. (1980). Seismic resistance of composite floor diaphragms (Report no.
800133). Iowa State University, Ames, Iowa.
Widjaja, B. (1993). Analytical investigation of composite floor diaphragms strength and behavior
(Master’s thesis). Virginia Tech, Blacksburg, Virginia.
Review of plastic moment resistance for composite beams
Development of second generation of Eurocode 4
Univ.-Doz. Dr.-Ing. Markus Schäfer

University of Luxemburg
Luxemburg
markus.schaefer@uni.lu
ABSTRACT
For composite cross-sections of classes 1 and 2 the moment resistance can be determined by
plastic design methods according to Eurocode 4. Thereby it is still assumed, that each cross-
sectional fibre can plastify without any limitation of strains. For normal composite beams in case
of sagging moments and a high-lying plastic neutral axis, the real moment resistance is quite
greater than the plastic design. This is based on the large strains at the bottom side of the
section, so that the lower steel flange reaches the solidification range. In case of sections with a
large compression zone height xpl, a concrete failure in the compression zone can happen,
before reaching the plastic moment resistance Mpl,Rd. EN 1994-1-1 provides only for sections
with steel grades S420 and S460 a limitation of plastic design. Hence, the plastic design can
lead to unsafe results.
Analyses of plastic and strain limited moment resistance for different composite cross sections
underline the requirement of additional design rules for plastic bending design.
1 Composite beam sections and classification in cross-sections classes

While classic composite beams are often characterized by a double-symmetrical steel profile
with an upper concrete chord; today even slimmer, more compact and more individual beam
systems become important. The trend in the last years often shows sections which are
integrated into the concrete slab (shallow-floor or slim-floor), partially encased composite beams
as well as composite beams with a low construction height.
Fig. 1 - Forms of composite beam sections

Especially in case of welded slim-floor profiles, it has to be differentiated between open-sections
and box-sections. These sections are used in combination with solid concrete slabs, pre-cast
elements or composite slabs constructed by ribbed metal sheets with in-situ concrete.
Depending on the slab-system, the upper concrete flange can increase the moment resistance
if it is connected by shear connectors to the steel section (Fig. 1). The thickness hct of the upper
concrete above the steel flange has to be greater than 1/6 of the steel flange width bft, but in
each case more than 50 mm. This regulation is defined according to the requirements of
composite columns, consisting of fully encased steel profiles [1]. The effective width beff is also
assigned by EN 1994-1-1. Based on the b/t-ratio of the flanges and web a differentiation in four
cross-section classes is possible. This indirectly reflects the rotation capacity and provides local
buckling effects. For the one-side attached bottom flange (bfb,2) the identification follows the
requirements for flanges of partially-encased steel sections [1]. The middle part of the bottom
flange (bfb,1) of a concrete filled box-section can be classified based on the criterions for
concrete-filled rectangular hollow steel sections. For full encased sections there is no local
stability threat to the encased web plates, a steel web of class 3 encased in concrete, is
represented by an effective web of the same cross-section of class 2 [1]. If a classification into
the classes 1 and 2 is possible, the plastic moment resistance can be taken into account.
In case of a large cross-section rotation, there is a risk of losing the stabilizing effect by the
concrete due to exceeding the concrete limit strains. Thereby the rotation capability is limited.
Actually, there is not sufficient experimental experience to ensure the application of plastic
methods for the calculation of action effects for partial or fully encased cross-sections. So that
the action effects are to be determined by the theory of elasticity, also in case of class 1
sections. Nerveless it needs to be pointed out, that the rotation capacity depends not only on
the limits due to buckling of the steel parts, but also on the effects of the concrete part due to
the limitation of strains in the external section fibres. Especially for composite beams with a
massive concrete chord, partially encased sections with a high degree of reinforcement, slim-
floor sections and systems whose effective width in the concrete chord is limited due openings
or using precast elements, the strain limit design becomes often decisive, because of the
reduced rotation capacity. For such sections, the plastic design can lead to an overestimation of
moment resistance. Therefore, further verifications are necessary.
2 Review of plastic bending design for composite beam sections

2.1 Confusion in the approach of bending design of composite sections according to EC 4
In Eurocode 2 [2] the concrete compression strength is defined as fcd = αcc fck/c. Where the
coefficient αcc takes into account long-term effects on the concrete compressive strength and
unfavorable effects resulting from the way the load is applied to the test specimen. This
reduction factor considers mainly the difference of the concrete compressive strength under
long-term loading and short-term resistance. EN 1992-1-1 suggests a coefficient cc between
0.8 and 1.0, often defined as 0.85.
Fig. 2 - Difference in designation of design approaches in Eurocode 2 and Eurocode 4
The plastic design of composite sections according to Eurocode 4 is based on concrete stress
block, whereby the concrete compression strength is defined as fcd = fck/c. While simultaneously
the concrete compressive stresses over the total stress block xpl needs to be reduced by a
constant factor 0.85.
Currently, there are several discussions, whether the reduction of the design value of concrete
strength in EN 1994 by the coefficient 0.85 results either from a rectangular compression block
over the total height of its compression zone [4], or if it is reducible to long-term effects (αcc) of
concrete. In order to explore this, the historical development of Eurocodes has to be
considered. In the previous version ENV 1992-1-1:1992 [3] the concrete compression strength
was still defined as fcd = fck/c and the coefficient α taking into account the long-term effects on
the compression strength and other unfavorable effects was added to the stress-strain curve (α
fcd). So that the stress ordinate of the idealized stress-strain diagram was defined by the
characteristic value of the concrete compression strength fck multiplied with the factor α/c. Only
in later versions of EN 1992 the definition of fcd was changed to fcd = αcc fck/c, used until now.
However, EN 1994 has been developed in the same time and today it is not comprehensible, if
the change from ENV- to EN-Version of concrete code was also followed for the design of
composite structures. Based on the background information it appears reasonable, that the
added coefficient 0.85 (Fig. 2) for the design of composite beams corresponds to αcc. Therefore,
the design level seems to be the same. In parallel, the comment to EN 1994 in [4] underlines
the relevance of the approximation made by the shape of the concrete stress-block, leading to
an overestimation of the moment resistance. Though, the illumination in [7] leads to the result
that for beam sections the plastic design based on rectangular stress block does not need any
reduction, because of the high rotation capacity of composite beams. But concurrently, an
additional reduction factor of 0.8 is claimed for the concrete stress α fck/c when using the
rectangular concrete stress bloc for the design of composite slabs. Assuming, the consideration
of the long-term effects by factor α = 0.85, results in a design value of concrete compression
stress about 0.68 fck/c. The following chapter will clarify the significance of rotation capacity and
the impact off stress block onto plastic bending for composite beams, considering also new
forms of composite beam sections, representing often a structural element between classical
beam and slab.
Fig. 3 - Stress-strain diagram and design value of concrete compression strength according to
ENV 1992-1-1: 1992 [3] and EN 1992-1-1:2004 [1]
2.2 Comparison of parabola-rectangular and rectangular shape of concrete stresses

Of course, there is an impact of the stress block’s shape on the bending design. According to
EN 1992 mainly curved shapes are used for the stress-strain curves. Therefore, the approach of
a simplified rectangular stress-block for the structural use of concrete can lead to an
overestimation of the concrete compression force in the concrete chord if the height of the
compression stress block is defined by the strain distribution. Therefore, in case of a strain-
limited design, the height of the stress-block has to be reduced by factor  (EN 1992-1-1,
section 3.1.7). However, also in this case there remains a divergence in the position of the
resulting compression force with its impact on the bending resistance. The concrete
compression force Ncd results by integration of the parabola-rectangular stress-strain function
over its surface, respectively above the compression zone height x(z), Fig. 5.
To compare the parabola curve with a rectangular equivalent stress block, a dimensionless
coefficient can be derived by dividing Ncd by the compression force bxfcd resulting from a
rectangular stress block. This coefficient describes the solidity ratio αr of the surface of parabola
curve.
x
b   c ( x ) dx
0 (1)
r 
b  x  fcd
Fig. 4 - Shapes of stress-strain curves used in EN 1992 and comparison to rectangular stress-
block according to EN 1994
The solidity ratio αr depends on the concrete strains at the top fibre of the concrete compression
zone. For the pure parabola shape with a strain limitation of εc equal -2‰ the solidity ration
results to 2/3, which represents the area of a parabola compared to a rectangle. Assuming a
concrete strain at the top fibre of -3.5‰, which will be reached for ductile concrete or composite
sections with sufficient rotation capacity, the solidity range arises to a value of 0.809.
Consequently, a rectangular stress block with the same resulting concrete compression force
Ncd has always a height of 0.809 times the height of a parabola rectangular shaped stress
curve. Therefore, only in case of strain limited design and use of a simplified rectangular stress
block according to EN 1992, the height of concrete compression zone has to be reduced by
factor  compared to the strain zero-point.
Fig. 5 - Comparison of shape form for parable-rectangular stress-strain curve and rectangular
stress block
However, in case of plastic design the height of the plastic neutral axis is not determined by the
strain distribution over the section! In this case, the compression force is defined by the
equilibrium of the internal normal forces (Ni = 0: Npl.a.Rd + Ncd = 0). This means, the
compression force Nc in the concrete chord corresponds to the tension force Na in steel section.
Finally, for common composite beam sections the height of the plastic concrete compression
block in the chord is 19.1% (1-0.809) lower in contrast to a parabola-rectangular stress block,
based on the hypothesis that the full plastic internal normal force Npl.a.Rd of the steel section is
achieved. An additional reduction coefficient established by the scale of the resulting
compression force is not necessary in case of plastic design. Hence, the assumption that in EN
1994 the reduction factor 0.85 for the concrete stresses in plastic design results from the
approach of the stress block over the total compression zone height is invalidated. This is only
valid for a strain-limited design in combination with a pure rectangular stress block; note the
reference of application of factor  in EN 1992!
Even if there is no impact on the magnitude of the resulting concrete compressive force of the
concrete slab, there remains a difference in the position of this force compared to parable-
rectangular shape form. The centre of the stress-shape of a parable-rectangular relation results
by equation (2):
xc xc
  c ( x )  x dA b   c ( x )  x dx
xs  0  0
(2)
xc xc
  c ( x ) dA b   c ( x ) dx
0 0
Fig. 6 - Difference  related to position of resulting concrete compression force
Fig. 6 shows the deviation of the position of the resulting concrete compression force relative to
its compression zone height xc based on the hypothesis of reaching εc2u for normal weight
concrete of -3.5‰ at the outer section fibre. The difference of the position of resulting
compression force Ncd is about 2.7%. For sections with non-ductile behaviour where e.g. the
strains are limited by the ductility of reinforcement in tension zone, lower strains are reached in
the outer section fibre. Of course, in this case the difference in position of the resulting force Ncd
becomes more important. This is also valid for composite columns where in case of N-M-
interaction the edge strains are between -2.0‰ and -3.5‰. Fig. 7 illustrates the increasing
difference in the position of resulting compression force by decreasing of concrete strains.
Indeed, comparison in Fig. 7 presents only ductile composite beam sections of cross-section
classes 1 and 2 with plastic neutral axis in the concrete chord. The graphic demonstrates an
impact of the divergences in position of resulting compression forces for the two mentioned
design methods up to 3% on the resulting bending moment for common composite beam
sections. It should be pointed out, that the importance increases with the height of concrete
compression zone and with decrease of the lever arm for the internal forces. For composite
slabs the discrepancy for the design based on a stress block will be more important than for
composite beam sections [7]. At least, for classical composite beam sections with a high
rotation capacity the difference between parabola-rectangular and rectangular stress block can
normally be neglected. For slabs, shallow-floor beams and composite beams with a low height
further investigation needs to be provided.
Fig. 7 - Difference in position of resulting concrete compression force and for bending
resistance
2.3 Comparison of strain limited and plastic design

Furthermore, the question is whether the steel part of the section reaches its plastic strain level
with its full plastic stress bloc! If not, there will be a significant difference between plastic and
strain limited design. Fig. 8 demonstrates the design methods for an IPE 500-S355 with its
difference between both stress shapes. On one hand a strain limited calculation according to
EN 1992-1-1 (Fig. B, MRd = 1554 kNm) without activation of strain hardening range for steel, on
the other hand plastic design according to EN 1994-1-1 (Fig. C, Mpl.Rd = 1557 kNm). To ensure
a comparable result, the concrete compression strength in all cases is defined according to EN
1992 considering the long-term effects (αcc = 0.85). The result shows clearly, a difference
between compression zone height of B and C by factor 0.809 and nearly the same moment
resistance. The strain distribution in A considers the strain hardening range of steel stress-strain
relationship.
Fig. 8 - Comparison of plastic and strain limited of bending resistance for common composite
beam-section
This example, presenting a typical composite beam, e.g. for car parks, shows that the bending
moment resistance MRd (Fig. A) due to a strain limited design is at least higher than the full
plastic moment resistance Mpl.Rd. (Fig. C). Changing the steel grade to S235, would result in a
strain limited moment resistance 7% higher than the plastic resistance. In comparison, the
impact of reducing the stress block by an additional factor 0.85 would be negligible (< 2%). This
points out, that a general reduction of the stress block in case of plastic design is not justifiable,
even though the impact on this kind of sections is very low.
Fig. 9 - Comparison of strain-limited and plastic bending resistance for a slim-floor section
The approach of reducing the plastic moment resistance becomes more relevant in case of
composite sections with a small depth and inner lever arm z or for composite slabs. Especially
for slim-floor sections this phenomenon is also exposed in [9] - [11]. The example of a slim-floor
section with a small effective width in Fig. 9 demonstrates this effect. It results in a high value of
0.81 for the relationship ξpl = xpl/h. The moment resistance based on strain-limited design (A)
becomes decisive; it represents only 90% of the plastic moment resistance (C). Even the
approach of an additional reduction coefficient of 0.85 for the concrete stress block
overestimates the moment resistance, Fig. 9 (D).
If the stresses in the section are limited by the strains, the plastic bending design can lead to an
overestimated bending resistance. This occurs, if the rotation capacity of the cross section is
limited.
3 Development of additional design rules for plastic bending resistance
3.1 Impact of concrete part on rotation capacity

In case of classical composite beams a classification in four cross-section classes based on the
b/t-ratio of the flanges and web is possible. For partially or fully encased sections the rotation
capacity depends not only on the limits due to buckling of the steel parts, but also the effects of
the concrete part influence the limitation of strains in the external section fibres. Especially for
composite beams with a massive concrete chord, partially encased sections with a high degree
of reinforcement, slim-floor sections and systems whose effective width in the concrete chord is
limited due to openings or using precast elements, the strain limited design becomes often
decisive, because of the reduced rotation capacity.
For the plastic design of the moment resistance it is still assumed, that each cross-sectional
fibre can plastify without any limitation of the strains. For normal composite beams in case of
sagging moments and a high-lying plastic neutral axis, the real moment resistance is
considerably greater than the plastic design moment. This is based on the large strains at the
bottom side of the section, so that for the lower steel flange strain hardening can be taken into
account.
In case of sections with a large compression zone height xpl, a concrete failure in the
compression zone can happen before reaching the plastic moment resistance Mpl,Rd. Therefore,
the plastic moment resistance has to be reduced. Nevertheless, this is not only a question of the
shape of the concrete stresses, moreover a combination of rotation capacity, stress-strain curve
and maximum strains in the external fibre of cross-section. A general reduction of the concrete
stresses or the plastic moment due to these effects is not appropriate for the design.
Fig. 10 - Rotation capacity of a concrete section

The reduction of full plastic moment resistance depends among others on the relation of the
concrete compression zone height to the total section height. This indirectly reflects the
limitation of rotation capacity for the section. Fig. 10 illustrates the principle for rotation capacity
of concrete sections in hogging zone, depending on the compression zone height. EN 1992-1-1,
section 5.5 allows for the moment redistribution a simplified estimation of rotation capacity
through the relationship x/d. For a limited rotation capacity, also the strains in the steel sections
will be reduced and therefore the full plastic force Npl.a.Rd cannot be reached. The comparison of
plastic and strain limited moment resistances for a multiplicity of cross-sections will lead to an
adequate reduction factor, chapter 3.2. EN 1994-1-1 still provides a limitation of plastic design
only for sections with steel grades S420 and S460. Thereby the plastic ultimate moment
resistance, according to section 6.2.1.2 (2), is limited by the coefficient , taking into account
that for large compression zones the design moment capacity is restricted by attaining the
ultimate strain in the concrete chord [8]. The verification of the coefficient  for S420 and S460
is based on analysis in [12].
3.2 Implementation of a reduction factor for plastic moment resistance

The in Eurocode 4 currently used design value for concrete compression strength in
combination with the coefficient 0.85 at the plastic concrete compression block displays an
inadmissible mixing of two different phenomena. On one hand there is the reduction due to
long-term effects; on the other hand, this non-variable coefficient for the reduction of the stress
block (Fig. 2). The problem of the strain limitation should not be mixed with the long-term
coefficient according to Eurocode 2! It would be more correct to introduce a correction
coefficient for the moment design resistance in composite cross sections and to consider the
long-term effects according to EN 1992. Therefore, the concrete compressive strength in
Eurocode 4 should be defined by reference to Eurocode 2. This will lead to consistence in
design and harmonization of Eurocode 4 and Eurocode 2. Actually, the evaluation of Eurocode
2 should be attained. Anyway, the long-term effect has to be considered in design of composite
structures.
To respect the effects of a limited rotation capacity for sections with an important compression
zone height due to strain limitation, an additional reduction coefficient  for the plastic moment
resistance is to be introduced. This correction coefficient would capture the difference between
a strain limited design and a full plastic design. Furthermore it would consider, through the
relationship xpl/h the rotation capacity considering the concrete part. Eurocode 4 provides this
already for high strength steels. This correction coefficient will always be 1.0 for typical
composite cross sections. Additionally, it could take better into account untypical cross-sections,
where the plastic zero line lies very deep.
MRd    M pl ,Rd (3)
The deviation of the reduction functions of the plastic moment resistance for different steel
grades are based on extensive comparisons of plastic and strain limited design in [9], [13] and
[14]. About 35.000 combinations of different cross sections and materials have been
considered. The results of different analysis are consolidated in [15], to reach a general design
approach.
The analyses are based on stress-strain relationship for concrete according to EN 1992-1-1,
whereby the parabola-rectangular shape was used. Fig. 11 visualizes the used stress-strain
relationship for concrete.
Fig. 11 - Stress-strain relationship for concrete, structural steel and reinforcement
For structural, steel the stress-strain relationship according EN 1993-1-5, Annex C6, with strain
hardening of tan-1(E/10000) was applied. Only for the analysis of slim-floor section in [9] a
horizontal plateau for yielding was considered. Reinforcement steel follows the rules of EN
1992-1-1, also for this steel the strain hardening was applied behind the elastic strains.
Fig. 12 to Fig. 14 show the comparison and analysis of strain limited moment resistance and
plastic moment resistance for concrete class up to C50/60. Thereby it needs to be pointed point
out, that with a relationship of xpl/h about 0.15 a reduction of plastic moment resistance
becomes important for steel grades S420 and S460. For grades of S355 the plastic moment
resistance is to be reduced if the relationship xpl/h is more than 0.20 and for steel grade S235 if
the relationship is more than 0.3. The definition of these limits prevent an inefficient reduction of
plastic moment resistance for common composite sections with a high lying plastic neutral axis.
Concrete strength of C60/70 and higher are not considered in these figures. Because for this
concrete strengths, the variety is very high.
Fig. 12 - Reduction coefficient  for concrete C20/25 to C50/60 and steel grades S420/S460
Fig. 13 - Reduction coefficient  for concrete C20/25 to C50/60 and steel grades S355
Fig. 14 - Reduction coefficient  for concrete C20/25 to C50/60 and steel grades S235
4 Summary
The analysis of plastic moment resistance points out, that the plastic design can lead to an
overestimation of the bending resistance, if the rotation capacity is limited due to the ultimate
strains of concrete. In the current version of Eurocode 4 for the design of composite structures
in steel and concrete, the impact of concrete part due to rotation capacity is not considered. If
no more exact design method will be applied, the introduction of an additional reduction factor
for the plastic moment resistance, based on the relationship xpl/h, leads to a reliable design
level. For classic composite beams with a double-symmetrical steel profile with an upper
concrete chord and a slight compression zone height, there will be no impact on the bending
design by the reduction factor . However, the risk in case of compact and encased sections will
be covered. Actually, this design approach is proposed for the development of the second
generation of Eurocode 4.
5 References
[1] DIN EN 1994-1-1: Eurocode 4: Design of composite steel and concrete
structures - Part 1-1: General rules and rules for buildings; German version
EN 1994-1-1:2004 + AC:2009
[2] DIN EN 1992-1-1:2011-01: Eurocode 2: Design of concrete structures -
Part 1-1: General rules and rules for buildings; German version EN 1992-1-
1:2004 + AC:2010
[3] ENV 1992-1-1: EuroCode 2: Design of concrete structures, Part 1. General
rules for buildings (1992)
[4] Johnson, R., P., Anderson, D.: Designers' Guide to EN 1994-1-1, Eurocode 4:
Design of composite steel and concrete structures: Part 1.1: General Rules
and Rules for Buildings, Thomas Telford (2004)
[5] Rüsch, H.: Researches Toward a General Flexural Theory for Structural
Concrete, In: ACI Journal 57, (1960)
[6] Rüsch, H., Sell, R., Rasch C., Grasser, E., Hummel, A., Wesche, K., Flatten,
H.: Festigkeit und Verformung von unbewehrtem Beton unter konstanter
Dauerlast. DAfStb-Heft, Ernst & Sohn, Berlin, (1968)
[7] Bode, H.: Euro-Verbundbau, 2nd Edition, Werner Verlag (1998)
[8] Hanswille G., Schäfer M., Bergmann M.: Verbundtragwerke aus Stahl und
Beton, Bemessung und Konstruktion - Kommentar zu DIN 18800-5 Ausgabe
März 2007, Stahlbau-Kalender 2010, Ernst & Sohn Verlag für Architektur und
technische Wissenschaften GmbH, Berlin (2010)
[9] Schäfer M. „Zum Tragverhalten von Flachdecken mit integrierten
hohlkastenförmigen Stahlprofilen“, Dissertation, Institut für Konstruktiven
Ingenieurbau, Heft 8, Bergische Universität Wuppertal, 2007
[10] Schäfer, M.: Design rules for slim-floor girders considering the composite
behaviour, Nordic Steel Construction Conference 2015, Tampere, Finland
(2015)
[11] Schäfer, M.: Zur Biegebemessung von Flachdecken in Verbundbauweise;
Stahlbau, 84. Jahrgang, Heft 4, Ernst & Sohn Verlag, Berlin (2015)
[12] Hanswille, G., Sedlacek, G., Anderson, D.: The Use of Steel Grades S460
and S420 in Composite Structures, ECCS-Eurofer, Improvements by TC 11 to
Eurocode 4, University of Wuppertal, RWTH Aachen, Ernst & Sohn Verlag,
Berlin (1996)
[13] Banfi, M.: The next generation of Eurocode 4, EUROSTEEL conference,
September 13–15, 2017, Copenhagen, Denmark
[14] Bahovic, A.: Comparison of strain limited an plastic design for composite
beam sections, bachelor thesis supervised by M. Schäfer, University of
Luxemburg, in progress
[15] Schäfer, M., Banfi, M.: Plastic moment resistance of composite beams -
harmonization of Eurocode 4 with Eurocode 2, Background Report to EN
1994, Development of Second generation of Eurocode 4, European
Committee for Standardization CEN, Project Team CEN/TC250/SC4.T1,
Luxemburg/London (2017)
INFLUENCE OF THE COMPOSITE ACTION ON THE LOAD BEARING AND DEFORMATION
BEHAVIOR OF COMPOSITE BEAMS WITH PROFILED STEEL SHEETING
Dr.-Ing. Florian Eggert

PERI GmbH
Weißenhorn, Germany
florian.eggert@peri.de
Former: University of Stuttgart, Institute of Structural Design

Stuttgart, Germany
ulrike.kuhlmann@ke.uni-stuttgart.de
ABSTRACT
In the frame of the European research project DISCCO (Aggelopoulos et al., 2016), composite
beams with additional profiled steel decking were investigated. Under certain conditions, current
Eurocode 4 (EN 1994-1-1, 2004) leads to inefficient dimensions of composite beams especially
in view of the minimum degree of shear connection, above all when using modern profiled steel
decking. Furthermore, the rib distance of the steel decking profiles limits the connection of steel
beam and concrete slab. To investigate the realistic behavior of composite beams with low de-
grees of shear connection, a comprehensive study on push-out specimens and short composite
beams was carried out. New findings gained by these tests lead to new rules for the minimum
degree of shear connection and for the load-bearing capacity of headed studs in combination with
additional profiled steel decking.
1 GENERAL
The usage of composite beams with composite slabs in modern building applications has a lot of
advantages. However, the existing rules for the minimum degree of shear connection, in some
cases, make the design of composite beams impossible, particularly with the use of modern deck
profiles with high, narrow but widely spaced ribs and hence with widely spaced shear connectors.
Recent research (Konrad, 2011; Nellinger, 2015) showed that shear connector resistances ac-
cording to Eurocode 4 (EN 1994-1-1, 2004) insufficiently predict the observed concrete failure
modes and resistances in the presence of modern forms of steel decking. New approaches to the
shear connector resistance (Konrad, 2011; Nellinger, 2015) lead to more accurate predictions of
the shear resistance of headed studs.
This paper presents the transfer and the influence of these results to composite beams. The re-
sults of a parametric study on eight short-span composite beam tests with small degrees of shear
connection between 16% and 38% are presented. The beam tests were conducted using compo-
site slabs with a modern form of profiled steel sheeting. Important parameters of the shear con-
nection, such as stud diameter, number of studs per rib, reinforcement pattern, welding procedure
and concrete strength, were varied. All the test configurations were selected in accordance with
a previously conducted study on the behavior of shear connectors in corresponding push-out
specimens. A comparison of the bending resistance of the beam tests and the calculated plastic
bending resistance is presented.
1 SHEAR RESISTANCE OF HEADED STUDS

1.1 PROCEDURE OF EN 1994-1-1
The currently used Eurocode 4 (EN 1994-1-1, 2004) uses empirically derived formulae to deter-
mine the shear connector resistance in solid slabs. The mean shear connector resistance (Roik
et al., 1988) is calculated as the minimum of Equations (1) and (2). To consider the influence of
the deck profiles, a reduction factor according to Equation (3) is applied.
Steel failure: 𝑃𝑡1 = 𝑘𝑡 ∙ 1.00 ∙ 𝑓𝑢 ∙ 𝜋 ∙ 𝑑2 /4 (1)
Concrete failure: 𝑃𝑡2 = 𝑘𝑡 ∙ 0.374 ∙ 𝛼 ∙ 𝑑2 √𝑓𝑐𝑚 ∙ 𝐸𝑐𝑚 (2)

0.7 𝑏𝑚 ℎ𝑠𝑐
Reduction: 𝑘𝑡 = ∙ ∙( − 1) ≤ 𝑘𝑡,𝑚𝑎𝑥 (3)
√𝑛𝑟 ℎ𝑝 ℎ𝑝
With d: Diameter of the shear stud [mm]
fu: Tensile strength of the stud [N/mm²]
α: = 1 for hsc/d > 4, d and hsc in [mm]
fcm: Concrete compressive strength [N/mm²]
Ecm: Modulus of elasticity of concrete, using 𝐸𝑐𝑚 = 9500 ∙ 𝑓𝑐
1/3
[N/mm²]
nr: Number of studs per rib (max. nr = 2)
1.2 NEW DESIGN APPROACHES

Evaluations by (Kuhlmann and
Raichle, 2005) already showed that
the current formulae of Eurocode 4
(EN 1994-1-1, 2004) overestimate
shear stud resistances when using
modern decking. This is due to the
value kt,max, cf. Equation (3), that is
decisive in most of the cases. This
means that important geometrical pa-
rameters of the deck profiles as well
as the position of the stud are not
considered for the determination the
reduction factor.
Recent tests and investigations (Kon- Fig. 1 – Geometrical parameters of the profiled decking
rad, 2011; Kuhlmann and Konrad,
2009; Kuhlmann and Konrad, 2010; Rambo-Roddenberry, 2002; Nellinger, 2015) consider the
position of the headed stud as well as the geometry of the decking. It could be shown (Konrad,
2011) that a shear connector is in so-called “unfavourable position” if the distance e between the
web of the rib and the shear stud is less than 55 mm. In the now performed tests this distance
was clearly smaller than 55 mm. New developed formulae (Konrad, 2011) determine the failure
load of a shear stud as the minimum of Equations (4) and (5) in combination with the reduction
factors given by Equation (6) for open sheeting geometry .
According to (Konrad, 2011), the reduction factor is controlled by the distance e between the web
of the rib and the shear stud. Formally, the procedure is the same as in Eurocode 4 (EN 1994-1-1,
2004): the shear connector resistance in a solid slab is reduced by a reduction factor. However,
the dimension of the reduction depends on the position of the stud in the rib, see Equation (6):
• e > 100 mm: favourable position
• 55 mm < e ≤ 100 mm: mid position
• e ≤ 55 mm: unfavourable position
2
Steel failure: 3
𝑃𝑚,𝑠 = 𝑘𝑡 ∙ 39.85 ∙ 𝐴𝑊𝑢𝑙𝑠𝑡,𝑒𝑓𝑓 ∙ 𝑓𝑐,𝑐𝑦𝑙 + 0.59 ∙ 𝑓𝑢 ∙ 𝑑2 (4)
2
Concrete failure: 3
𝑃𝑚,𝑐 = 𝑘𝑡 ∙ 39.85 ∙ 𝐴𝑊𝑢𝑙𝑠𝑡,𝑒𝑓𝑓 ∙ 𝑓𝑐,𝑐𝑦𝑙
1/3 1/2
+ 3.75 ∙ 𝑑2 ∙ 𝑓𝑐,𝑐𝑦𝑙 ∙ 𝑓𝑢 (5)
𝑏
𝑘𝑡 = 𝑘𝑛 ∙ (0.084 ∙ ( ℎ𝑚) + 0.663) ≤ 1.00 („favourable position“)
𝑝
𝑏
Reduction: 𝑘𝑡 = 𝑘𝑛 ∙ (0.042 ∙ ( ℎ𝑚) + 0.663) ≤ 1.00 („mid position“) (6)
𝑝
𝑏
𝑘𝑡 = 𝑘𝑛 ∙ (0.317 ∙ ( ℎ𝑚) + 0.060) ≤ 0.80 („unfavourable position“)
𝑝
With d Diameter of the shear stud [mm]

fc,cyl: Concrete compressive strength [N/mm²]
nr: Number of studs per rib (max. nr = 2)
fu: Tensile strength of the stud [N/mm²]
AWulst,eff: Bead weld area of the shear stud [mm²]; cf. Table 1
kn: 1 if nr = 1, 0.80 if nr = 2
Table 1 – Bead weld area of the shear stud of eq. (4) and (5) according to (Konrad, 2011)
Diameter of Bead weld area
the stud [mm] AWulst,eff [mm²]
16 47.3
19 63.0
22 87.0
25 140.0
1.3 PUSH-OUT TEST RESULTS

Several reports (Aggelopoulos et al., 2016; Kuhlmann, Eggert et al., 2014) present results of nu-
merous push-out tests performed within the DISCCO project. Most tests used ArcelorMittal
Cofraplus60 decking representing the open decking geometry. Figures 2a and 2b show exempla-
rily a push-out specimen at the end of the test as well as its typical failure modes. Beside the
delamination of the sheeting, concrete pull-out failure could be observed (top half of Figure 2b).
Larger deflections led to an additional rib punch-through failure (bottom half of Figure 2b).
Figure 3 shows the comparison of the push-out test results and the calculated values according
to Eurocode 4 (EN 1994-1-1, 2004; Roik et al., 1988) and (Konrad, 2011). The results prove that
there is a wide difference between the two calculation approaches when using modern deck pro-
files with a narrow rib. (Konrad, 2011) slightly underestimates the test results which is safe and
conservative. On the other hand, the recalculation using Eurocode 4 (EN 1994-1-1, 2004) over-
estimates the test results considerably. None of the performed tests is conservative for this com-
parison, but some tests are extremely unconservative nearly by factor 2 which means that the
calculated stud capacity is nearly twice as high as the measured value.
Fig. 2a – Push-out specimen 1-03 after testing Fig. 2b – Typical push-out failure modes (top:
(Kuhlmann, Eggert et al., 2014) concrete pull-out; bottom: rib punch-through)
(Kuhlmann, Eggert et al., 2014)
The currently used approach of Eurocode 4 (EN 1994-1-1, 2004) for the reduction factor kt was
mainly calibrated using undercut decking profiles or open profiles with a wide rib. Therefore, spe-
cial effects of open decking profiles with narrow ribs – like a reduced bearing capacity caused by
a missing concrete strut at the weld bead – are not considered – although the Cofraplus60 decking
profile complies with the geometrical specifications of Eurocode 4 (EN 1994-1-1, 2004). Consid-
ering these new findings, the authors propose to revise the respecting rules in a future edition of
Eurocode 4.
1.3 INFLUENCE OF SPECIAL PARAMETERS ON THE STUD BEARING CAPACITY

1.3.1 WELDING METHOD
According to Eurocode 4 (EN 1994-1-1, 2004), through-deck welding of the studs leads to an
increase of the stud bearing capacity of about 13% in contrast to composite slabs using pre-
punched decking. (Konrad, 2011) estimates an increase of about 11% based on a numerical
parametric study. The performed tests showed an increase of about 22% when studs were
through-deck welded compared with stud welded directly on the flange of the steel beam.
Fig. 3 – Comparison of the push-out test results to different calculation approaches
1.3.1 NUMBER OF STUDS PER RIB

If shear studs are placed too close together, the influencing concrete zones might overlap which
may lead to a decrease of the load bearing capacity. Therefore, Eurocode 4 reduces the capacity
per stud by the factor 1/√𝑛𝑟 = 1/√2 = 0.71 (cf. Equation (3)) for a minimum stud distance of 4*d.
(Konrad, 2011) improved this factor to 0.80 – based on numerical studies. The now performed
tests with 5*d showed, that the reduction per stud was only about 2.5%. This proves, that the
improved reduction factor of Konrad is more realistic for open decking geometries.
1.3.2 STUD DIAMETER

Both Eurocode 4 (EN 1994-1-1, 2004) and (Konrad, 2011) assume that the load bearing capacity
of a stud is proportional to the square of its diameter. This would lead to an increase of about 34%
when using a diameter 22 mm stud instead of a 19 mm one. Although the tests were performed
with narrow ribs, an increase of about 65% could be observed. Therefore, the rules of (EN 1994-
1-1, 2004) as well as (Konrad, 2011) are conservative and on the safe side for narrow open deck-
ing geometries.
1.3.3 POSITION OF REINFORCEMENT LAYERS

Current standards do not provide any information about the influence of the reinforcement on the
ductility behavior of composite beams. Therefore, some push-out tests were carried out only with
a single reinforcement layer on top of the slab and some with an addition layer positioned directly
at the top of the metal deck sheeting as required in (EN 1994-1-1, 2004). The load slip curves
showed that two layers of reinforcement increase the ductility behavior nearly by factor 2 but do
not influence the bearing capacity itself significantly.
1.3.4 CONCRETE QUALITY

According to Eurocode 4 (EN 1994-1-1, 2004), the load bearing capacity is in proportion to the
square root of the concrete compressive strength. Numerical investigations of (Konrad, 2011)
2/3
showed a depending on √𝑓𝑐 . Applied on the actual measured concrete compressive strengths
40 N/mm² and 46 N/mm², Eurocode 4 predicts an increase of 7%; Konrad’s equations led to an
increase of 11% which fits very well with the measured increase of about 11%.
1.3.5 INFLUENCE OF CONCRETE RELAXATION

Although push-out tests are performed quasi-static, the influence of the loading speed
(0.01 mm/s) on concrete relaxation behavior shouldn’t be neglected. Numerous relaxation load
drops were provoked by keeping the deflection constant for 5 minutes. During this time, the con-
crete relaxed and the load decreased. The German National Annex of Eurocode 4 (DIN EN 1994-
1-1/NA, 2010) considers this effect by increasing the partial factor γv for Equation (2) from 1.25 to
1.50 which implies a reduction of the failure load of about 16%. The now determined relaxation
was in the range of 10%.
2 BEAM TESTS
2.1 TEST SETUP
Within the European research project DISCCO (Aggelopoulos et al., 2016) 8 beam tests (ST 1 –
ST 8) were performed using open Cofraplus60 decking. The beam design was derived from cor-
responding push-out tests. The beams had a length of 5 m and the tests were performed as four-
point bending tests. Therefore, the shear studs of the outer – 1.80 m long – areas were predicted
to take shear force at the same level (cf. Figure 4). The rib distance of the decking limited the
number of shear studs. 8 (single positioning) or 16 (pairwise positioning) shear studs per shear
stressed area led to very low degrees of shear connection. In addition, the beams were designed
to fail first by shear failure in the composite joint before bending failure occurred. To connect
directly to the push-out tests, the same parameters were investigated in the beam tests, see
Chapter 1.3.
Fig. 4 – Structural system of the DISCCO composite beam tests (Aggelopoulos et al., 2016)
2.2 TEST PROCEDURE
The beam tests were performed
following the Eurocode 4 (EN
1994-1-1, 2004) guidelines for
push-out tests. Especially, 25
load cycles were carried out,
before the actual test started, to
eliminate influences of friction
as best as possible. All tests
were performed deflection-con-
trolled running 1 mm/min. In ad-
dition, the tests were stopped in
intervals of about 5 mm vertical
deflection to capture the influ-
ence of concrete relaxation. Be- Fig. 5 – Load-deflection graph of the beam test, exemplarily
side the vertical deflection, the shown for test ST 1 (Eggert and Kuhlmann, 2016)
development of slip was meas-
ured as well as strains of concrete and steel beam. Figure 5 exemplarily shows the load-deflection
graph of beam test ST 1. The pre-loading cycles can be identified as well as the load drops caused
by concrete relaxation. The influence of concrete relaxation was determined to 4-6% in average.
After having reached the maximum load, the load of all beams decreased very slowly which shows
the very ductile behavior of the investigated beams. Figure 6 exemplarily shows beam test ST 1
during testing.
Fig. 6 – Composite beam ST 1 during test realization (Aggelopoulos et al., 2016)
2.3 TEST RESULTS AND TEST EVALUATION

Shear studs need to be sufficiently ductile to resist the occurring shear forces when performing a
plastic analysis of a composite beam. According to Eurocode 4, clause 6.6.1.2 (EN 1994-1-1,
2004), shear studs can be regarded as ductile if the degree of shear connection satisfies Equation
(7). It is known that shear studs within a rib have a larger deformation capacity as studs in solid
slabs. In this case, shear studs can also be regarded as ductile, if Equation (8) can be satisfied
which is a bit more optimistic. As a lower limit, Eurocode 4 (EN 1994-1-1, 2004) gives a minimum
degree of shear connection of 40%.
Using Equation (8) is only possible in compliance with the following criteria of Eurocode 4 (EN
1994-1-1, 2010):
• the studs have an overall length after welding not less than 76 mm, and a shank of nominal
diameter of 19 mm,
• the steel section is a rolled or welded I or H with equal flanges,
• the concrete slab is composite with profiled steel sheeting that spans perpendicular to the
beam and the concrete ribs are continuous across it,
• there is one stud per rib of sheeting, placed either centrally within the rib or alternately on
the left side and on the right side of the trough throughout the length of the span,
• for the sheeting bm / hp ≥ 2 and hp ≤ 60 mm (see Figure 1),
• the force Nc is calculated in accordance with the simplified method of Eurocode 4.
355
𝜂 ≥ 1 − ( 𝑓 ) (0.75 − 0.03𝐿𝑒 ) and 𝜂 ≥ 0.4 (7)
𝑦
355
𝜂 ≥ 1 − ( 𝑓 ) (1.00 − 0.04𝐿𝑒 ) and 𝜂 ≥ 0.4 (8)
𝑦
With: Le = Equivalent span [m] and fy = Yield strength of structural steel [N/mm²]
Using Cofraplus60 decking it was not possible to satisfy the geometrical criterion, because the
ratio bm / hp is only 1.4 and therefore clearly below the given limit. In addition, some of the tests
used pairs of shear connectors per rib or studs with a diameter of 22 mm.
Because of this according to Eurocode 4, Equation (7) should be used instead of Equation (8).
According to Equation (7), the minimum degree of shear connection is 44.4%. Equation (8) leads
to a minimum degree of shear connection of only 25.8%. But Equation (8) is also limited to a
minimum degree of shear connection of 0.4 = 40%.
The degree of shear connection of the beam tests were determined to 16% up to 37.9%. Thereby,
all tests showed a degree of shear connection below the Eurocode 4 (EN 1994-1-1, 2010) limit.
All performed tests didn’t satisfy the Eurocode 4 rules and therefore, none of the tests may be
calculated using the plastic analysis including partial interaction, as described in Eurocode 4 (EN
1994-1-1, 2004).
Although the beams didn’t satisfy the guidelines, all beam tests were recalculated using Eurocode
4 plastic analysis method for partial interaction. The stud bearing capacity is one of the most
influencing parameters of this method. Therefore, this parameter was considered in three ver-
sions:
• stud capacities of own corresponding push-out tests (“WP1”),
• according to Equations (1) – (3) of Eurocode 4 (EN 1994-1-1, 2004)
• according to Equations (4) – (6) of (Konrad, 2011).
Table 2 – Results of the recalculation of the beam tests: Mtest/Mcalculation and belonging degrees of
shear connection in [%]
ST 1 ST 2 ST 3 ST 4 ST 5 ST 6 ST 7 ST 8 Average
„WP 1“ 0.96 1.02 0.83 0.98 0.92 0.96 0.94 1.06 0.98
(23.4) (19.4) (46.2) (41.1) (32.6) (31.2) (38.1) (19.7)
Konrad 0.94 0.95 0.85 0.98 0.91 0.95 0.93 0.96 0.95
(25.9) (25.7) (41.5) (41.1) (33.6) (33.6) (40.2) (29.7)
Roik / EC4 0.85 0.86 0.78 0.90 0.82 0.85 0.87 0.90 0.86
(41.3) (41.3) (58.5) (58.5) (55.4) (55.4) (55.4) (41.3)
Table 2 shows the results of the
determined moment bearing ca-
pacities related to the calculated
values Mtest/Mcalculation as well as the
belonging degrees of shear con-
nection. Mtest means the plastic
bending moment considering the
partial interaction including meas-
ured material strengths. Figure 7
compares the results in a graphical
presentation. Values directly on
the bisecting line would mean a full
accordance of test result and cal-
culation.
The mean values of Table 2 are
determined neglecting the results
of beam test ST 3. This composite
beam used pairwise through-deck
welded shear studs. Because of a
bad quality of the welding the
shear studs couldn’t achieve their
full load-bearing capacity. It should Fig. 7 – Graphical comparison of the results of the recalcu-
be noted that the quality of lation of the beam tests (Eggert and Kuhlmann, 2016)
through-deck welding is very sen-
sitive to several influencing factors
– especially if studs are not
welded centric on the beam – as it
is the case for pairs of studs that
are placed eccentric on the flange
of the beam. The eccentric weld-
ing position and incomplete con-
tact of decking and steel beam
may cause an incorrect magnetic
field, which may lead to imperfec-
tions of the welding and therefore
reduced stud capacities.
Comparisons using “WP1” stud
capacities lead to mean value of
Mtest/Mcalculation = 0.98. Using equa-
tions of (Konrad, 2011) leads to a
mean value of 0.95. This means a
very good accordance of test and
calculation. As 100% of the calcu-
lated plastic moment bearing ca-
pacity comes along with very high
strains in the concrete, it is hard to Fig. 8 – Welding imperfections of eccentric positioned
reach this in a test. Therefore, through-deck welded studs (Aggelopoulos et al., 2016)
Mtest/Mcalculation = 0.95 is considered
sufficient for the assessment of Mpl. The results of the recalculation (cf. Figure 7 and Table 2)
clearly show that the “95%-criterion” is reached or even exceeded. However, predictions using
stud capacities of Eurocode 4 on the mean level (Roik et al., 1988) clearly overestimate the test
results. None of the tests reaches the “95%-criterion”. Therefore, this calculation approach should
be improved, though of course, the comparison is not made with design values. It may be as-
sumed that the safety margin, included in the existing statistical evaluation of (Roik et al., 1988),
is not satisfying for the presented problems.
2.4 FURTHER INVESTIGATIONS

Based on the test results, a numerical model was created using SOFiSTiK finite element software
(SOFISTIK, 2014) to investigate further parameters and especially have a look at the develop-
ment of slip depending on several parameters.
One of the main results of this study describes the relation of slip and vertical mid-deflection of
the beam. The lower the degree of shear connection the larger is the vertical deflection of the
beam. Therefore, in most of the cases the design of the beam is controlled by SLS and not ULS.
The parametric study on single span composite beams (Eggert, 2019; Mahler, 2016) showed that
the development of slip can be divided into three main parts. First, all elements of the composite
beam behave elastically. Then, the outermost studs close to the support of the beam reach their
bearing capacity which leads to a “kink” in the load-slip diagram (cf. Figure 9). When all studs
have reached their maximum load, the steel beam starts to yield and the concrete reaches the
compression strength which finally leads to the failure (plastic hinge) of the composite beam.
The current Eurocode 4 (EN 1994-1-1, 2004) gives no precise deflection limit for composite
beams. Eurocode 2 (EN 1992-1-1, 2004) and a pre-standard of Eurocode 3 (ENV 1993-1-1, 1992)
limit the deflection of slabs and beams to l/250. Most of the available shear connectors (e.g.
headed studs, Hilti X-HVB shear connectors, Arcelor Mittal COSFB concrete dowels) have a de-
formation capacity of at least 6 mm.
To understand better the load-slip development of a composite beam, several parameters were
identified that have a large influence on the development of slip. Based on previous work of (John-
son and Molenstra, 1991), the length to depth ratio was identified as one of the most influencing
parameters on the load-slip behavior. Figure 9 exemplarily shows an excerpt of the parametric
study showing the results of several composite beam configurations having a length/depth ratio
of 18 and a degree of shear connection of 30%, which is significantly below the restrictions of
Eurocode 4.
The results of this partial result of the study clearly shows that the deflection criterion l/250 is
reached at an end slip of only 2.5 mm. So, there is no danger of a premature shear connector
failure when designing the composite beam in compliance with the deflection criterion. Even the
steel beam is still in an elastic state at this point which is the same for most of the investigated
cases.
3 CONCLUSIONS
Within the European research project DISCCO (Aggelopoulos et al., 2016) the realistic behavior
of some versions of modern composite beams with profiled sheeting was investigated. The results
of the conducted beam tests showed that in some cases the failure load was clearly lower than
expected by the equations given in Eurocode 4 (EN 1994-1-1, 2004). Although the minimum de-
gree of shear connection in Eurocode 4 (EN 1994-1-1, 2004) is defined as a limit of the occurring
slip, it could be shown that going below this limit is possible using sufficiently ductile shear con-
nectors.
The exact knowledge of the stud bearing capacity is a basic requirement for a good quality of the
prediction of the moment bearing capacity of composite beams with partial interaction. The current
approach of Eurocode 4 (EN 1994-1-1, 2004) is not a satisfying solution because of an incorrect
reduction factor kt when using the new generation of open decking. New design approaches or
the use of stud capacities gained from corresponding push-out tests lead to very good results.
Fig. 9 – Numerical parametric study: Load ratio q/qmax vs. end slip (Eggert, 2019)
Furthermore, it could be shown that a plastic analysis of composite beams with low degrees of
shear connection is possible, although the requirements of the code concerning the minimum
degree of shear connection were not satisfied. The evaluation of the tests showed that the current
restrictions – concerning decking geometry as well as the limit of the minimum degree of shear
connection – are very conservative which makes it hard to realize composite structures in an
economical way, although a sufficient beam capacity could be proved by the tests. Further inves-
tigations regarding the occurring slip are used in (Eggert, 2019) to show new design approaches.
4 ACKNOWLEDGEMENTS
The research leading to these results was part of a common project of the Steel Construction
Institute, University of Stuttgart, University of Luxembourg, University of Bradford, and ArcelorMit-
tal. We thank all involved colleagues for the good cooperation and the professional exchange.
The project received funding from the European Community’s Research Fund for Coal and Steel
(RFCS) under grant agreement RFSR-CT-2012-00030.
5 REFERENCES
Aggelopoulos, E. S., Lawson, R. M., Kuhlmann, U., Eggert, F., Odenbreit, C., Nellinger, S., Lam,
D., Dai, X., Sheehan, T. and Obiala, R. (2016). Development of Improved Shear Connection
Rules in Composite Beams (DISCCO), Final Report, RFCS Project No. RFSR CT 2012-00030.
Eggert, F. (2019): Influence of the composite action on the load-bearing and deflection behavior
of composite beams with and without additional steel sheeting, PhD Thesis; Institute of Structural
Design, University of Stuttgart, Issue 2019-1, ISSN 1439-3751.
Eggert, F., Kuhlmann, U. (2016). Influence of the degree of shear connection on the load-bearing
behavior of composite beams with additional profiled steel decking. In: Stahlbau, 85, 7. pp. 459–
465 (2016).
DIN EN 1994-1-1/NA (2010). National Annex – Eurocode 4: Design of composite steel and con-
crete structures – Part 1-1: General rules and rules for buildings.
EN 1992-1-1 (2004). Eurocode 2: Design of concrete structures – Part 1-1: General rules and
rules for buildings; German version: DIN EN 1992-1-1:2004 + AC:2010.
ENV 1993-1-1 (1992). Eurocode 3: Design of steel structures; Part 1.1 – General rules and rules
for buildings; German version: DIN ENV 1993-1-1:1993.
EN 1994-1-1 (2004). Eurocode 4: Design of composite steel and concrete structures – Part 1-1:
General rules and rules for buildings; German version DIN EN 1994-1-1:2004 + AC:2009.
Johnson, R. P., Molenstra, N. (1991). Partial shear connection in composite beams for buildings.
In: Proceedings of the Institution of Civil Engineers - Structures and Buildings, Part 2, Bd. 91. pp.
679–704 (1991).
Konrad, M. (2011). Load-bearing behavior of headed studs in combination with composite slabs
with additional steel sheeting spanning transverse to the supporting, PhD Thesis; Institute of
Structural Design, University of Stuttgart, Issue 2011-1, ISSN 1439-3751.
Kuhlmann, U., Eggert, F., Odenbreit, C., Nellinger, S. (2014): Push-out tests with modern deck
sheeting to evaluate shear connector resistances. In: Eurosteel 2014, 7th European Conference
on Steel and Composite Structures, Naples, Italy, pp. 519–520.
Kuhlmann, U., Konrad, M. (2009). Bearing capacity of headed studs when using profiled sheeting.
University of Stuttgart, Institute of Structural Design, No. 2009-1X, DIBt research project ZP 52-
5-17.20-1260/07, Fraunhofer IRB-Verlag.
Kuhlmann, U., Konrad, M. (2010). Guarantee of sufficient load-bearing capacity of headed studs
when using profiled sheeting. University of Stuttgart, Institute of Structural Design, No. 2010-3,
DIBt research project ZP 52-5-17.20-1287/08, Fraunhofer IRB-Verlag.
Kuhlmann, U., Raichle, J. (2006). Shear strength of composite beams when using additional steel
sheeting according to Eurocode 4, Part 1.1, Research report; University of Stuttgart, Institute of
Structural Design, No. 2006-9X.
Mahler, F. (2016). Numerical investigations on the load-slip behavior of composite beams with
low degree of shear connection, Master Thesis; University of Stuttgart, Institute of Structural De-
sign, No. 2016-44X.
Nellinger, S. (2015). On the Behavior of Shear Stud Connections in Composite Beams with Deep
Decking, Dissertation; The Faculty of Sciences, Technology and Communication, Université du
Luxembourg.
Rambo-Roddenberry, M.D. (2002). Behavior and strength of welded shear stud connectors. Vir-
ginia, Virginia Polytechnic Institute and State University, PhD Thesis.
Roik, K., Hanswille, G., Cunze, A., Lanna, O. (1988). Background document on Eurocode 4 -
Clause 6.3.2: Stud connectors, Ruhr-University Bochum. Research report.
SOFiSTiK AG: SOFiSTiK Version 2014, Finite Element Software.
EFFECT OF POST-WELDING RESIDUAL STRESSES
ON DISTORTIONAL BUCKLING OF COMPOSITE BEAMS
Marian A. Gizejowski
Warsaw University of Technology, Faculty of Civil Engineering, Warsaw, Poland
m.gizejowski@il.pw.edu.pl
Radoslaw Szczerba
r.szczerba@il.pw.edu.pl
Marcin Gajewski
m.gajewski@il.pw.edu.pl
Wioleta Barcewicz
w.barcewicz@il.pw.edu.pl
ABSTRACT
Welded steel-concrete composite beams with large openings are nowadays more frequently
used for commercial applications in multi-storey buildings. The openings may be placed in a
regular spacing along the beam length or only at certain locations leaving the full webbed
sections in zones where they are not requested from the structural or functional purposes (for
services passing through the beam depth). With regard to the optimum integration of load
bearing structures and services, continuous or semi-continuous composite web perforated
beams may be used. Such beams demonstrate at the hogging moment much weaker behavior
against distortional buckling in comparison with their plain-webbed counterparts. This paper
discusses different techniques for finite element modeling of the effect of post-welding residual
stresses on distortional buckling resistance of steel-concrete composite beams in reference to a
different arrangement of web openings. The results of verification and validation exercise of
proposed modeling technique are also summarized hereafter.
INTRODUCTION
Thin-walled I-section steel-concrete composite beams are in hogging bending zones susceptible
to lateral, lateral-torsional or torsional distortional buckling of the structural steel section flange
that is not restrained by the reinforced slab. The buckling resistance of I-section rolled plain-
webbed composite beams subjected to hogging bending had been well researched so that
corresponding design procedures were proposed and codified (Eurocode 4, 2004). However, in
modern applications of composite beams, they are made of welded thin-walled I-sections with
larger openings than those in castellated beams made out of steel rolled sections. Welded thin-
walled sections inherit material and geometric imperfections as a result of their fabrication
process. The effect of such imperfections on the distortional buckling of composite beams may
be modeled with use of the finite element method and available commercial software. In
research carried out at the Warsaw University of Technology, the effect of residual stresses was
considered implicitly by using a single parametric equivalent constitutive model of post-welded
steel (Gizejowski, Salah and Barcewicz, 2008c). The shape of adopted equivalent curvilinear
relationship σ-ε is dependent upon the model parameter n [cf. eqn. (1) in this paper]. Its value
may be approximated by using the inverse modeling technique, either by fitting the experimental
equilibrium path of the welded full scale specimen or by fitting the experimental results from the
stub test of the short length plate girder of the same welded section size as the full scale
specimen, depending upon the problem being under consideration. This model was successfully
applied for the numerical evaluation of buckling resistance of composite beams (Gizejowski and
Salah, 2008a). Geometric imperfections were considered independently adopting their profile to
be the lowest buckling mode from LBA with the amplitude scaled to conform with the maximum
web bulging deflection of fabricated steel section. Reinforced concrete was modeled using
smeared crack model adopted in ABAQUS software (ABAQUS, 2011a and 2011b). The
developed finite element model of the composite beam behavior was validated with use of
results of experimental investigations on long- and short-length composite beams (Gizejowski
and Salah, 2008b). The parameter of steel equivalent constitutive model was calibrated in such
a way that the validated finite element model represented the most accurately the equilibrium
paths of experimentally tested beams. The entire investigations considering the other aspects of
the composite beam behavior were then presented as a Ph.D. thesis (Salah, 2009).
There are several ways of modeling the contributing effects of post-welding residual stresses as
well as the reinforced concrete slab stiffness degradation and inelastic stress redistribution
processes on the buckling resistance of composite beams. In this paper, an alternative finite
element modeling technique to that of (Salah, 2009) is proposed in which geometric
imperfections are introduced first and then the residual stresses are superimposed and modeled
through a direct application of the residual stress block to a geometrically imperfect beam
model. The shape of stress block is predefined and based on the experimental evidence. The
goal of above mentioned two stages of analysis is to identify the initial self-equilibrated state of
the structural steel product in order to determine its initial configuration prior to assembling it for
a full-scale composite beam to be analyzed for the mechanical response to applied loads. Finite
element analysis deals with the response of two specimens of composite beams tested
experimentally (Gizejowski and Salah, 2008b) and (Salah, 2009). Investigations were
concerned with distortional buckling and post-buckling behavior of specimens consisting of the
plain-webbed span and the overhanging part with a different number of circular openings. The
created numerical models are then validated using the results recorded in experiments.
Different ways of modeling the effect of residual stresses on the beam response are discussed.
As the reference, the other four specimens were analyzed, two with plain webbed girder and
two girders with openings in both the span and the overhang. The summary of predicted
buckling strength is presented and conclusions for engineering practice are drawn.
SCOPE OF INVESTIGATIONS
Figure 1 shows the composite beams tested experimentally, C2S355 and C4S355, that are
considered hereafter for numerical investigations and validation of the proposed finite element
modeling technique. Beam specimens consisted of steel welded girders that were compositely
connected to the reinforced concrete slab with a use of commercial steel studs welded to the
beam upper flange. The steel girders were assembled in one product by MAG automatic
welding of three plates, the prefabricated web plate of 4 mm in thickness the openings of which
had been made by laser cutting and two flange plates of 6 mm in thickness. Prior to welding, the
maximum geometric out-of-flatness imperfections of the perforated web plate were measured by
the manufacturer and used in numerical modeling for scaling the amplitude of geometric
imperfection profile. The beam section was reinforced in the three locations corresponding to
the points of concentrated load transfer, i.e. at the load application and support points. The
welded steel girder was then equipped with shear studs M12 and the reinforced slab of a
constant width as shown in Figure 1a). The side view of specimens is also shown in Figure 1,
for the short span specimen in Figure 1b) and the long span one in Figure 1c). The steel grades
applied were S 355 for the structural steel, A-III (A400) for the rebars, S 520 for the studs, their
measured average mechanical properties are given in Table 1. Concrete strength class C35/45
was used for the composite beam slab the actual compressive and tensile strength properties of
which were derived from standard material tests. The average concrete properties are given in
Table 2. The Young modulus for steel and concrete are taken hereafter at their nominal values.
a) 1000 b) A
25 25
100
100
68
8
R1
72
6
480
f10 (c/c 135 mm) 100´6
R1
f6 (c/c 182 mm)
thick 4 mm M12
336
468´48´6 100 1058 264 529 265 100

1158 1158
100´6 A
72
c) A
100
68
8
68
68
480
R1
6
R1
R1
R1
100 2116 264 529 529 529 265 100
2216 2216
A
Fig.1 – Composite beams tested experimentally, a) cross section, b) short span beam,
c) long span beams
Table 1 - Measured steel average properties of composite beam

Ultimate strength Hardening strain Ultimate strain
Steel for Yield stress fy *)
fu *) εhar *) εu *)
Beam plates 391.4 506.8 1.42 16.0
Shear studs 611.3 662.5 - -
Main rebars 459.6 608.0 0.22 15.7
*)
in N/mm2, **) in %
Table 2 - Measured concrete average properties of composite beam slab

Concrete for Age on the test day Cube compression strength*) Tension strength*)
Long span beam 46 56.5 4.7
Short span beam 40 46.6 3.8
*)
in N/mm2
The sketch of specimen placement in the test stand is shown in Figure 2. The load is distributed
over an area of the slab and the slab is additionally reinforced there in order to avoid the local
concrete failure. The support closer to the applied load ensures a linear contact between the
beam bottom flange and the roller. The support placed away from the applied load was uplifted
so that an additional frame-to-floor resisting structure had been designed that elastically
supported the specimen in order to transfer the resultant uplifting reactive force. In numerical
modeling, the translational springs in the direction of gravity force are applied covering the area
of contact between the upper flange and the rafter of the frame-to-floor resisting structure.
Hydraulic jack
Support beam Detail (1)
floor
Fig. 2 – Arrangement of testing stand and beam specimen
NUMERICAL MODEL
Structural steel, rebars and studs
Typical multi-linear uniaxial stress-strain relationships for the girder steel grade of S 355 are
given in Figure 3, IEP with nominal values and EXP with actual values of mechanical properties.
IEP model is used also for rebars and studs.
Fig. 3 – Standard stress-strain relationships for parent steel
The constitutive model of steel in the beam girder is dependent upon the modeling technique of
the effect of residual stresses. In the case of implicit residual stress modeling, the concept of
equivalent isotropic hardening constitutive model (CUR) of post-welded steel used in
(Gizejowski and Salah, 2008a) and (Salah, 2009) is followed hereafter. The parametric stress-
strain relationship for such a model is given by:
1 n
 1 n  1  
n
 
(1)        
 E   f y ,red  Est  
 
where:
fu  f y
(2a,b) f y ,red  f y  Est st , Est 
 u   st
Figure 4 presents the graphical representation of eqn. (1) for different values of n parameter.
The parameter n is to be calibrated in order to represent the experimental equilibrium path in
numerical simulations as close as possible. The simulations presented in (Salah, 2009) were
conducted using such a model together with the smeared crack model available in ABAQUS.
The values of n between 2 to 4 were shown for representing the best fit of numerical equilibrium
paths to those obtained from experimental investigations.
Fig. 4 – Equivalent isotropic hardening constitutive model
The other modeling of the residual stress effect on the structural behavior of composite beams
is based on the explicit modeling of residual stress distribution. A number of investigations have
been conducted for welded sections in order to come up with a standard section residual stress
distribution diagram for plain webbed steel girders (Gizejowski, Szczerba and Gajewski, 2017).
Such a parametric residual stress block being dependent upon the type of welding process
through the parameter cw is given in Figure 5. Tensile residual stresses are taken as equal to
the nominal yield stress of parent steel while the compressive residual stresses are equal to the
factored yield stress in which the flange and web factors are given by:
(3) 1 
2 c2
, 2 

2  c fw  0.5 t f 
b  2 c2 
h w t f  2  c fw  0.5 t f 
a) b)
Fig. 5 – Standard residual stress distribution, a) parametric stress block, b) diagram for cw
Aw 12000  p Aw
For the weld thickness aw=3.5 mm, there are  0.8 mm, cw   (p = 0.8 for
t fy t
MAG) and c fw  cw , c2  cw  0.5 t w . The stress block geometric parameters calculated are
cfw+0.5tf=23.6 mm, c2=22.7 mm, ψ1=0.83 and ψ2=0.11. The application of residual stresses
applied together with the tri-linear parent steel constitutive model given in Figure 3 implies
dealing with the material anisotropy.
Concrete
Numerical models of reinforced concrete using ABAQUS software are usually based on the
models of Concrete Smeared Crack (CSC) or the Concrete Damage Plasticity (CDP). The
former one was used in (Gizejowski and Salah, 2008b) and (Salah, 2009). Smeared Crack
model represents cracked concrete as an elastic orthotropic material with reduced elastic
modulus in the direction normal to the crack plane. With this continuum approach, the local
displacement discontinuities at cracks are distributed over some tributary area within the finite
element and the behavior of cracked concrete can be represented by average stress-strain
relations. In contrast to the discrete crack model, the continuity of the displacement field
remains intact in CSC model. Adopting such an approximation of discrete nature of concrete
cracking, some model parameters need to be calibrated as not having the physical meaning in
order to get a better reproduction of the experimental equilibrium path of composite beams with
openings (Salah, 2009). The CDP model is therefore claimed nowadays of being more
practically related since its major parameters may be directly derived from material testing while
the secondary parameters may be taken as default values for certain concrete classes
(Kaminska and Szwed, 2015). The CDP constitutive model is described in ABAQUS manuals
(ABAQUS, 2011a and 2011b) and discussed in many papers. The most crucial point for CDP is
the parameter determination, cf. (Jankowiak and Lodygowski, 2005) and (Szczecina and
Winnicki, 2015). Here some of the data is assumed on the basis of the tests conducted on
concrete samples like compression test and Brazilian test (see Table 2), while the rest is
evaluated on the literature basis.
In Table 3, the parameters assumed for concrete in case of the tested long span beam are fully
presented. The physical data is always extended for abnormally large strains to provide
convergence in case of local strain concentrations. It is worth to underline here that the
convergence of full composite beam models is dependent mainly upon the concrete behavior.
The concrete slab is locally heavily loaded through the studs what causes the stress and strain
concentrations and activates the yield condition just from the beginning of the load
incrementation.
Table 3 - Assumed parameters of CDP (long span beam), cf. (Kaminska and Szwed, 2015)
and (Jankowiak and Lodygowski, 2005)
Material’s Yield condition shape
Concrete
parameters m 0.1
Elasticity  31 o
E [GPa] 36.0 fb 0 / f c 1.16
 0.167  0.66
Compression hardening Compression damage
Stress [MPa] Crushing strain [-] d c [-] Crushing strain [-]
55.0 0.00 0.00 0.00
56.5 0.001 0.70 0.01
40.0 0.01 0.95 10.0
20.0 0.1
5.0 10.0
Tension stiffening Tension damage
Stress [MPa] Cracking strain [-] d t [-] Cracking strain [-]
4.7 0.00 0.00 0.00
3.9 0.01 0.70 0.01
1.7 0.1 0.80 0.1
0.1 10.0 0.95 10.0
Finite elements, discretization and mesh size

Numerical models of tested composite beams are constructed with the use of ABAQUS
commercial software. In order to get the simplest representation of the tested composite beam
specimens, numerical models developed hereafter do not involve the brick elements and the
discrete representation of the crack and reinforcement bond effects. The reinforced slab and
wall elements of the structural steel section are modeled with use of shell elements S4R. The
shear studs are modeled with use of one beam element B31 per stud that connects the mid-
surfaces of the slab and the upper flange of structural steel welded section by using the
ABAQUS *Tie option. Rebars are modeled with use of *Rebar Layer option placed at the slab
mid-surface.
The boundary and loading conditions refer to the physical conditions of tested specimens. The
uplifted support is modeled through directional uz springs of the stiffness kz= 100 N/mm, applied
at each node of the contact surface, using the ABAQUS *Spring option. Such a modeling
reproduces the actual behavior of the frame-to-floor resisting structure (ux allowed while uy=0).
At the pressed down support, the beam web is reinforced in the out-plane direction by the
stiffener and the flange is constrained along the stiffener width (ux=0, uy=0, uz=0). In slab
modeling at the load application, the edge node translations are constrained in the direction
perpendicular to the specimen axis (uy=0). All the nodes on the intersection line of the web and
flanges are constrained against translations in the y-direction and against twist rotation about
the specimen axis (uy=0, Rx=0). The load is distributed over the contact area of 150x200 mm2.
For short span beams, the mesh at which the slab and beam section are modeled consisting of
7030 shell elements of dimensions close to 23x23 mm on average. In the case of long span
beams, there are as much as 13298 elements of the similar in-plane dimensions. The fact that
the reinforced concrete slab is modeled as a shell 2D object, the studs are modeled by beam
elements joining the nodes of steel beam and reinforced concrete slab mid-surface.
Modeling of material imperfections
Imperfections for the prediction of the composite beam initial configuration are modeled using
two approaches. Approach 1 conforms with that of (Salah, 2009) where the post-welded steel is
represented by an equivalent constitutive model (see Figure 4) and the geometric imperfection
profile corresponds to the lowest buckling mode from LBA. The amplitude of 4 mm web out of
flatness is used. Approach 2 developed in this paper requires evaluating of geometric
imperfection profile as in the Approach 1 and then superimposing the residual stress state.
Predefined numerical model of the imperfect welded steel product is then linked compositely
with the reinforced concrete slab model in such a way that stud elements connect vertically the
respective nodes of the top flange of welded section and the reinforced concrete slab.
Method of analysis
The multistep analysis is carried out for the evaluation of composite imperfect beam response.
In the case of Approach 1 three steps of analysis carried out with the use of ABAQUS software
are required:
1. Analysis of LBA type using *Buckle option in order to get the eigenmodes the combination of
which are used for the representation of imperfection profile.
2. Analysis using *Static General option in order to find the stress state and beam configuration
of the initially imperfect beam due to self-weigh acting prior to the load application.
3. Incremental analysis of GMNIA type using *Nlgeom option and Riks algorithm for the
evaluation of pre-limit branch, limit point and post-limit branch of the equilibrium path.
In the case of Approach 2 four steps of analysis are required. The first one is identical to that of
Approach 1. In the second step, the material imperfections are introduced using the standard
residual stress block of Figure 5a) and *Predefined Field, Mechanical, Stress ABAQUS option.
This step is followed by steps 3 and 4 identical to 2 and 3 of Approach 1. Figure 6 shows the
magnified residual deformation state of the C4S355 beam as a result of the welding
imperfections.
Fig. 6 – Magnified post-welding initial deformation profile of steel girder

(deformation scale factor: 1500)
VALIDATION OF NUMERICAL MODEL
Table 4 presents the results from LBA as to the lowest eigenmode and eigenvalue. Comparing
the eigenmodes and the eigenvalues from the present simulation in comparison of those in
(Salah, 2009) (given in round brackets), one may conclude that the eigenmode profiles seem to
be identical while slight differences may be seen for eigenvalues.
Table 4 – Lowest eigenmodes and eigenvalues

Beam Lowest eigenmode Eigenvalue [kN]
C4S355 63.4 (62.7)
C2S355 76.1 (76.0)
Figures 7 presents the results of numerical equilibrium path response carried out for C4S355
composite beam using Approach 1 (with the implicit residual stress modeling by using an
equivalent constitutive relationship with the parameter n=2) and Approach 2 (with the explicit
residual stress modeling developed in this study). The numerical results are validated with use
of experimental equilibrium path presented in (Gizejowski and Salah, 2008a) and (Salah, 2009).
Fig. 7 – Equilibrium path response for C4S355 composite beam
Approach 1 is able to reproduce the experimental equilibrium path quite accurately in the pre-
limit range and as to the evaluation of limit point. The post-limit range of the behavior is however
traced less accurately. Approach 2 gives in this case less accurate results and only in the range
of advanced post-limiting displacements allows for a safe estimation of the equilibrium path.
Figure 8 presents the similar comparison for the C2S355 composite beam. One can draw
similar conclusions to those derived from the validation of the C4S355 beam numerical results.
Fig. 8 - Equilibrium path response for C2S355 composite beam
Table 5 shows the validation of deformed shapes of considered beams obtained from numerical
simulations EXP,MAG-355 against those recorded during experimental investigations.
Table 5 – Deflected profiles of composite beams

Beam Numerical Experimental
C4S355
C2S355
NUMERICAL RESULTS FOR REFERENCE COMPOSITE BEAMS

The numerical model developed in the previous section is used hereafter for the evaluation of
beam response of composite full plain webbed girders (LPS355 and SPS355) and girders with
circular openings along the full specimen length (LFS355 and SFS355) that correspond to the
analyzed earlier beams C4S355 and C2S355, respectively.
Figures 9 presents the numerical equilibrium path responses of the LPS355 beam in Figure 9a)
and the SPS355 beam in Figure 9b). The results are compared with the experimental curves for
C4S355 and C2S355, respectively. Figure 10 shows the similar comparison but for the LFS355
beam in Figure 10a) and the SFS355 beam in Figure 10b).
a) b)
Fig. 9 - Equilibrium path response of plain webbed composite beams
a) b)
Fig. 10 - Equilibrium path response of composite beams with openings along the full length
From the observation of results presented in Figures 9 and 10 one may come up with the
conclusion that composite beams with large openings are much more prone to distortional
buckling than their plain webbed counterparts. The differences between the beams with
openings are however meaningless for composite beams with openings only on one side of the
internal support and with openings arranged symmetrically on both sides of the internal support.
COCLUDING REMARKS
Aspects of two approaches to the residual stress modeling and its effect on the equilibrium path
response using finite element numerical simulations were the primary objectives of the present
study. Numerical models were developed using the implicit and explicit modeling of the section
residual stress distribution. Both approaches were validated with use of experimental results for
the beams with the overhang having openings. The validation exercise allows for the following
conclusions to be drawn:
1. Approach 1 of implicit residual stress modeling with the equivalent constitutive relationship
and model parameter n=2 allows for more accurate estimation of the experimental equilibrium
paths for both long span and short span composite beams taking account for both the beam
stiffness in the pre-limit range and the limit point at the beam ultimate state.
2. Approach 2 is less accurate and predicting safely the beam stiffness in the pre-limit range but
overestimating the limit point at the beam ultimate state.
3. The distortional failure modes are predicted accurately in numerical simulations with regard to
both imperfection modeling approaches. The distortional failure load is very sensitive to the
number of openings in the regions of the hogging moment.
4. Numerical simulations of the reference composite beams, plain webbed and with the
openings along the full length, clearly show that the openings weaken the distortional buckling
resistance. It is however proven that the distortional buckling resistance for the beams having
the openings only on one side of the internal support is practically at the same level as for the
beams having the openings on both sides of the internal support. Taking the resistance of the
full webbed composite beams as 100%, the resistance reduction of the perforated beams is as
follows:
- by approx. 1/3 for the long span beams,
- by approx. 2/5 for the short span beams,
regardless whether the openings are arranged only on one side of the internal support or on
both sides.
REFERENCES
ABAQUS. (2011a). Theory manual (Version 6.11). Dassault Systèmes.
ABAQUS (2011b). Standard User’s manual (Version 6.11). Dassault Systèmes.
Eurocode 4. (2004). Design of Composite Steel and Concrete Structures. Part 1-1: General
rules and rules for buildings. Brussels: CEN.
Gizejowski, M., & Salah, W. (2008a). Experimental investigation of the stability behaviour of
slender section steel-concrete composite beams with web openings. In J. Obrebski (Ed.), XIV
Lightweight Structures in Civil Engineering – Contemporary Problems: Proceedings of Polish
IASS Chapter Seminar, Warszawa, 2008 (68-75). Warszawa: Micro-Publisher.
Gizejowski, M., & Salah, W. (2008b). Numerical finite element modelling of the stability
behaviour of slender section steel-concrete composite beams with web openings. In J.
Obrebski (Ed.), XIV Lightweight Structures in Civil Engineering – Contemporary Problems:
Proceedings of Polish IASS Chapter Seminar, Warszawa, 2008 (76-86). Warszawa: Micro-
Publisher.
Gizejowski, M., Salah, W., & Barcewicz, W. (2008c). Finite element modeling of the behaviour
of steel end-plate beam-to-column joints. Archives of Civil Engineering, LIV(4), 693-733.
Gizejowski, M., Szczerba, R., & Gajewski, M. (2017). Resistance of mono-axially bent beams of
welded I-sections. In: Proceedings of the 8th European Conference on Steel and Composite
Structures, Copenhagen, 2017 (submitted).
Jankowiak, T., & Lodygowski, T. (2005). Identification of parameters of concrete damage
plasticity constitutive model. Foundation of Civil and Engineering, (6), 53-69.
Kaminska, I., & Szwed, A. (2015). Calibration of concrete constitutive model and experimental
tests used for it. In E. Szmigiera, P. Łukowski & S. Jemioło (Eds.), Concrete and concrete
structures – experimental tests (pp. 93-110), Warsaw: Warsaw University of Technology
Printing House (in Polish).
Salah, W. (2009). Modelling of Instability Behaviour in Hogging Moment Regions of Steel-
Concrete Composite Beams. Ph.D. thesis, Warsaw University of Technology.
Szczecina, M. & Winnicki, A. (2015). Calibration of the CDP model parameters in Abaqus. In:
Proceedings of the 2015 World Congress on Advances in Structural Engineering and
Mechanics (ASEM15), Incheon, Korea, 2015.
STRENGTH OF LATERAL–TORSIONAL BUCKLING OF A COMPOSITE
STEEL BEAM SUBJECTED TO REVERSE CURVATURE BENDING
Satoshi KITAOKA Ryoichi KANNO

Nippon Steel Corporation Nippon Steel Corporation
Futtsu, Japan Futtsu, Japan
kitaoka.7pn.satoshi@jp. nipponsteel.com kanno.kx4.ryoichi@ jp.nipponsteel.com
Satoru HIROSHIMA Koji HANYA

Nippon Steel Corporation Nippon Steel Corporation
Futtsu, Japan Futtsu, Japan
hiroshima.kp9.satoru@jp. nipponsteel.com hanya.rs9.koji@jp. nipponsteel.com
Keiichi TAKADA Fumihisa YOSHIDA

Retired from Nippon Steel Corporation Daiwa House Industry Co., Ltd.
Tokyo, Japan Nara, Japan
n1000835kt@feel.ocn.ne.jp fumihisa@daiwahouse.jp
ABSTRACT
Design guidelines for lateral–torsional buckling of I-shaped steel beams have been well
established, but those of a composite beam consisting of an I-shaped beam and a floor slab
have not been sufficiently understood. It is especially true for a beam subjected to reverse
curvature bending, where the top flange is continuously restrained with a slab but the bottom
flange may still buckle laterally. Although several strength formulae have been derived, as far as
the authors know, the restraining effect is not widely considered as a common practice, partly
due to lack of behavioral data available. Therefore, in this study, three sub-assemblage
specimens consisting of two box steel columns, an I-shaped beam, and a composite floor slab
were tested to clarify the behavior. Then a closed-form buckling strength formula was derived
for the beam based on an energy method, and the results were compared with those of finite
element analyses and the test results. Overall, the restraining effects were clearly observed,
and the proposed formula was found to provide reasonable strength estimates.
INTRODUCTION
There has been a substantial amount of research work on lateral–torsional buckling (called LTB
hereafter) of I-shaped steel beams in steel building frames (Ziemian, 2010). Design guidelines
for the strength evaluation and its bracing have been well established. However, the strength
evaluation for a composite beam consisting of an I-shaped beam and a floor slab is not well
understood yet. It is especially true for a composite beam subjected to reverse curvature
bending, where the top flange is continuously restrained with a floor slab but the bottom flange
may still buckle laterally due to the compression stress. There would be an obvious benefit
especially in seismic design for eliminating bracing members when such a restraining effect of
the floor is considered in design. As a result, some strength evaluation formulae have been
proposed, such as those by Wakabayashi and Nakamura (1973), Bradford and Gao (1992),
Yura (1995), and Kimura and Yoshino (2011).
However, as far as the authors know, the restraining effect by the floor slab is not widely
considered as a common practice in design. This is partly because of the fact that enough
behavioral data on the interaction between the I-shaped beam and the floor slab are not
available, resulting in insufficient resources on the strength formulae. From this background, in
this study, three sub-assemblage specimens consisting of two box steel columns, an I-shaped
beam, and a reinforced concrete floor slab were first tested under horizontal loading applied to
the columns. Such a loading condition is typically seen in seismic design. Then, a formula for
closed-form elastic buckling strength for the restrained I-shaped steel beam under various
moment gradients was developed based on an energy method using newly proposed
displacement functions. Comparisons of results of elastic finite element buckling analyses, the
proposed formula, and a typical existing formula were made, and finally, the accuracy of the
proposed formula was examined with the experimental data.
OUTLINE OF EXPERIMENTS
Sub-assemblage specimens shown in Figure 1 were fabricated and tested under a horizontal
loading condition typically encountered in strong seismic and/or wind regions. The specimens
were designed as a part of a moment-resisting frame seen in Japan, consisting of two square
box columns, an I-shaped beam, and a reinforced concrete floor slab. The box columns were
made of cold-formed (roll-formed) steel and have a length of 1620 mm and a square section
with 300 mm width and 16 mm thickness. An I-shaped beam with a length of 5700 mm was
welded rigidly to the columns at both its ends. At the beam–column connections, reinforcement
was made with continuity plates between the beam flanges and the columns so that rigid
connection can be realized. The floor slab above the beam was of reinforced concrete with 70
(a) Side view (A-A' section)
(b) Plan view

Fig. 1 – Specimen overview
mm thickness and 2000 mm width. A Table 1 – Specimens and their beam sizes
flat-type form deck was set beneath the
floor for concrete casting. Headed steel Web Flange
Specimen no. Depth (mm) Width (mm) thickness thickness
studs with 10 mm in diameter and 50 (mm) (mm)
mm in length were welded at the
No.1 500 150 9 12
middle of the top flange. Following
No.2 500 116 9 12
Japanese practice, the studs were
No.3 500 65 9 12
arranged with a spacing of 200 mm in a
row along the entire steel beam, which
provides about one-half of the full composite action required by the AISC specification. From
strong restraints by the continuity plates, columns and floor slab, it can be assumed that both
twisting and warping were substantially restrained at the beam ends.
There were three specimens (Specimens No. 1 through No. 3). The experimental variable was
straightforward and was only the width of the beam that significantly affects the LTB behavior.
The width changed from 65 mm to 150 mm, while the height, flange thickness, and web
thickness remained constant in all specimens. Geometry information for the specimens is
shown in Table 1. As understood from information in the table, all the sections were categorized
into a compact section. The beams were built-up sections, and the yield strengths of the steel
plates used for the flange and web were 360 N/mm2 and 382 N/mm2, respectively. Based on a
widely used Japanese design standard, strength estimates of the steel beams without the floor
slab were 0.64M p , 0.41 M p , and 0.17 M p for Specimens No. 1 through No. 3, respectively, where
M p is the plastic moment. As seen, no specimens would reach M p if the slab restraining effect
was not taken into account.
The test set-up is shown in Figure 2. The top and bottom parts of the two columns were pin-
supported and two horizontal actuators were located at the top of the columns, which gave an
equal displacement in the same direction. The loading provided by the actuators was controlled
by displacement and the following protocol was applied: two cycles of loading at 0.5% drift angle
and then monotonically increasing loading until an obvious buckling was observed.
EXPERIMENTAL RESULTS AND OBSERVATIONS

Load and displacement relationships are shown in Figure 3. The load is represented by bending
moment M that acts at the ends of the beam, and the displacement is represented by the story
drift angle θ , which is defined by an inclination between the upper and lower pin-supports of the
columns. Since composite beams behaved differently in positive and negative bending, both
Fig. 2 – Test set-up and loading condition
relationships at positive and negative sides are shown in Figure 3. The average relationships
are also indicated in the same figures with finer broken lines. From the average relationships, it
can be said that the specimens were quite ductile and relatively similar to each other. They
started with elastic behavior, reached the maximum strength, and then showed yielding without
much strength deterioration. The figure also indicates that both stiffness and strength were
apparently larger in positive bending side than in negative side because of the composite action.
In the three specimens, no buckling and cracking were observed in the slabs at the cyclic
loading of 0.5% drift angle.
M (kNm) M (kNm) M (kNm)

800 800 800
600 600 600

Mp
400 Mp 400 400
Mp
200 200 200
0 0 0
-0.02 0.00 0.02 0.04 0.06 -0.02 0.00 0.02 0.04 0.06 -0.02 0.00 0.02 0.04 0.06
-200 θ (rad) -200 θ (rad) -200 θ (rad)
Positive bending side Positive bending side Positive bending side
-400 Average -400 Average -400 Average
Negative bending side Negative bending side Negative bending side
-600 -600 -600
(a) Specimen No. 1 (b) Specimen No. 2 (c) Specimen No. 3

Fig. 3 – Load–displacement relationships
(a) Specimen No. 1 (b) Specimen No. 2 (c) Specimen No. 3

Fig. 4 –Buckling deformation observed at the negative bending side
Each specimen’s behavior is described as follows: Specimen No. 1 having the widest flange
showed an obvious load drop at a drift angle of around 1.9% due to concrete crushing of the
slab around the column face at the positive side. Subsequently, flange local buckling occurred
at a drift angle of about 2.5% as seen in Figure 4 (a), but little strength deterioration was
observed until the loading was terminated. LTB did not occur in this case, instead, local flange
buckling was identified as in Figure 4 (a). Specimen No. 2 showed a similar load–displacement
relationship to that of Specimen No. 1. After concrete crushing was observed, buckling occurred
at a drift angle of around 2.4% with LTB and slight flange local buckling as shown in Figure 4 (b).
The last specimen, Specimen No. 3 also showed concrete crushing at the positive bending side
and a clear LTB at a smaller drift angle of around 1.9% than those of the previous two
specimens. Although concrete crushing of the floor slabs was observed in all the specimens,
obvious damage was not identified around the stud connections, as seen in Figure 5.
Through the behaviors of the three specimens shown in Figure 4, it is clear that the top flange of
the steel beam was completely restrained at least laterally. This demonstrated that the floor
slabs had enough capability to restrain the lateral movement of the steel beams, even though
the beams were not fully composite. Plastic moment strengths M p are shown with horizontal
broken lines in Figure 3, indicating that for all the specimens, the maximum strengths of the
beams in the negative bending side were found around their plastic moment strengths. The
strengths were 1.06 M p , 1.00 M p , and 0.91 M p for Specimens No. 1 through No. 3, respectively.
Compared with the previously indicated strength estimates for those without a floor slab,
substantial strength increases were found especially in Specimens No. 2. and No. 3. These
observations indicate that floor slabs possess considerably strong restraining effects on the LTB
of the I-shaped steel beams.
(a) Concrete crushing around the columns (b) Appearance around the studs
Fig. 5 – Damage and appearance observed in and around reinforced concrete slab
EXISTING STRENGTH FORMULA

Various design formulae for LTB of I-shaped beams have been proposed. All the formulae are
established primarily as a function of plastic moment strength M p and elastic bucking strength
M cr . For the cases where the top flange is continuously restrained with a floor slab, as seen in
the previous experiments, several LTB formulae have been proposed; for example by
Wakabayashi and Nakamura (1973), Bradford and Gao (1992), Yura (1995), and Kimura and
Yoshino (2011). Most of the formulae are developed based on governing differential equations
or the equivalent energy equation, but their derivation is so complex that the closed-from
solution is rarely possible. Therefore, existing formulae are rather approximate, established
based on numerical analysis.
Among the formulae, widely recognized is the one adopted in the AISC specification (AISC,
2010), which is originally proposed by Yura (Yura, 1995). It provides the following elastic
buckling strength M cr :
M cr = C b M cr 0 (1)
where M cr 0 is the elastic buckling strength for uniform bending moment, and C b is the LTB
modification factor for non-uniform moment including reverse curvature bending:
Cb = 3.0 − 2  M1  − 8  M CL  (2)
   
3  Mo  3  (Mo + M1)* 
where M o is the moment at the end of the unbraced length, which gives the largest compressive
stress in the bottom flange, M1 is the moment at the other end of the length, and M CL is the
moment at the middle of the length. Note that in Eqn. (2) that (M o + M1 ) * = M o , if M1 is positive.
To obtain the design strength M n , the elastic buckling strength M cr is applied to design
equations given in the AISC specification.
PROPOSED CLOSED FORM STRENGTH FORMULA

Since the existing formulae for the buckling are approximate, there might be some errors in the
strength estimation. As far as the authors know, no closed-form solution is currently available for
the elastic LTB strength of an I-shaped steel beam restrained by a floor slab. This is because
the equation becomes so complex that only numerical solutions can solve the problem.
A closed-form solution is derived in this study based on an energy method (Rayleigh-Ritz
method) by assuming rather untraditional displacement functions. Figure 6 shows a stability
problem of simply supported doubly symmetric I-shaped beam whose top flange is continuously
restrained laterally (a sign “X” means the lateral restraint). The beam is subjected to arbitrary
end moments, where β is introduced as a parameter to express various moment gradients. At
the middle of the top flange, the lateral displacement is continuously restrained, while the twist
rotation is free to move. Two types of end restraint conditions are considered: torsional simple
support and fixed support conditions. Composite action with floor slab is neglected.
Note on this model in Figure 6 that Bradford and Gao (1992) assumed a laterally and torsionally
fixed condition at the top flange, resulting in what they called “lateral-distortional buckling.” Since
it is still uncertain from a practical
viewpoint whether the torsionally
fixed condition is assumed, this
study employs only a laterally fixed
condition, which is considered
conservative for strength evaluation.
Back to the formulation, the total
potential energy is described as
follows (Trahair, 1993): Fig. 6 – Restrained I-shaped beam with end moments
1 Lb 1 Lb 1 Lb Lb
Π = ∫ EI y u′′2 dz + ∫ GJφ ′2 dz + ECw φ ′′2 dz +
∫ ∫
M z φ u′′dz (3)
2 0 2 0 2 0 0
where E is Young’s modulus, G is the shear modulus, I y is the moment of inertia about the
weak axes, J is the torsional constant, Cw is the warping constant, L b is the unbraced length
as shown in Figure 6, M z is the bending moment along the beam, and u and φ are the
horizontal displacement and twist rotation at the centroid of the beam, respectively. Neglecting
the sectional distortion of the beam, the following relationship can be assumed:
db
u= φ (4)
2
where db is the distance between the centers of the flanges. Referring to Figure 6, M z can be
expressed at a location of z as follows:
 z 
M z = 1 − b + b M
 (5)
 L b
Substituting Eqns. (4) and (5) into Eqn. (3), the total potential energy becomes as follows:
2
1 db  2 π E I yf db GJ 
Π = 
2
B + A − M {(1 − b )A + b C} (6)
2 Lb  L db 
 b 
where I yf is the moment of inertia for one flange. Note that an approximate relationship of
I y = 2 I yf is used in the derivation from Eqn. (3). A , B and C in Eqn. (6) are functionals of φ
shown as follows:
Lb L3b Lb Lb
A = Lb ∫0 φ ′ 2 dy , B= ∫ φ ′′ 2 dy , C= ∫0 zφ ′ 2 dy (7), (8), (9)
2π 2 0
As seen in Figure 6, φ is the only variable to describe the buckled displacement field. Here, φ is
expressed with the following finite series such that the given boundary conditions are satisfied:
n
φ = a0 + ∑ an φn (10)
n =1
where a 0 and an are arbitrary constants, φn is the base function, and n is the number of series.
Generally, the Fourier series have been applied to Eqn. (10), but in this study, rather special
functions are applied to make the closed-form solutions possible. The functions were found
through try and error, and are shown for two end restraint conditions against LTB as follows:
 n
  z  
φ n = sin π 
L

  for torsional simple support (no twist, free to warp) (11)
  b  
 n
  z  
φn = cos 2π  
L   for torsional fixed support (no twist and no warping) (12)
  b 
With the above functions, enough accuracy can be obtained using only three series, i.e., n = 3 ,
whereas the number of series is considerably larger with traditional Fourier series, especially for
the cases with various moment gradients. This large number of series has made the closed-
form solutions quite difficult. The proposed displacement functions can describe the buckled
shape more efficiently. Substituting Eqn. (10) into Eqn. (6) and applying a stationary condition in
terms of an to Eqn. (6), a cubic equation in terms of M can be obtained as an eigenvalue
problem. By solving the equation, the following closed-form elastic buckling strength M cr* can be
found:
*
M cr {
= min M 0 , M 1, M 2 } (13)
where
2 3 2 3 A2
k 3 q q  p 3 q q  p
M =ωk − +   +   + ω 3− k − −   +  − , k = 0, 1, 2 (14)
2 2  3  2 2 3 3
−1+ 3i −1− 3i
ω 0 = ω 3 = 1, ω 1 = , ω2 = (15), (16), (17)
2 2
1 1 2 αj
p = A1 − A 2 2 , q = A 0 − A1 A 2 + A2 2 , A j = (18), (19), (20)
3 3 27 α3
h 11 h 12 h 13
α 0 = h 21 h 22 h 23 (21)
h 31 h 32 h 33
g11 g12 g13 h 11 h 12 h 13 h 11 h 12 h 13
α 1 = h 21 h 22 h 23 + g 21 g 22 g 23 + h 21 h 22 h 23 (22)
h 31 h 32 h 33 h 31 h 32 h 33 g31 g32 g33
h 11 h 12 h 13 g11 g12 g13 g11 g12 g13

α 2 = g 21 g 22 g 23 + h 21 h 22 h 23 + g 21 g 22 g 23 (23)
g31 g32 g33 g31 g32 g33 h 31 h 32 h 33
g11 g12 g13

α 3 = g 21 g 22 g 23 (24)
g31 g32 g33
g nm = (1 − β )L nm + β N nm (25)
2 π 2 E I yf GJ
h nm = − M nm
2
d b − L nm (26)
Lb db
Lb L b3 Lb
L nm = Lb ∫0 ′ dz ,
φ n′ φ m M nm =
2π 2
∫0 φn′′ φm
′′ dz (27), (28)
Lb
N nm = ∫0 z φ n′ φ m
′ dz (29)
Applying the displacement functions in Eqns. (11) and (12), L nm , M nm , and N nm become
constants by utilizing partial integration technique and numerical integration. Tables 2 and 3
show the values of L nm , M nm , and N nm for simple support case and fixed support case,
respectively.
Table 2 – Values of L nm , M nm , and N nm for torsional simple support
n =1, m=1 n=2, m=2 n=3, m=3 n=1, m=2 n=1, m=3 n=2, m=3
Lnm 4.935 6.310 8.589 3.799 2.750 6.667
Mnm 2.467 9.329 26.33 1.899 1.375 13.49
Nnm 2.467 4.935 7.402 2.802 2.489 5.604
Table 3 – Values of L nm , M nm , and N nm for torsional fixed support

n=1, m=1 n=2, m=2 n=3, m=3 n=1, m=2 n=1, m=3 n=2, m=3
Lnm 19.74 26.75 35.96 9.224 -0.2315 23.93
Mnm 39.48 128.8 362.8 18.45 -0.4629 165.7
Nnm 9.870 19.74 29.61 6.763 1.077 19.21
Eqn. (14) is known as Cardano’s formula, which provides three roots. They may take real
numbers and/or complex numbers, depending on the value of the discriminant
D = ( q / 2 ) 2 + ( p / 3 ) 3 . Since the real roots are meaningful, their minimum value becomes the
solution for Eqn. (13), which is the buckling strength M cr* . Although the solution is expressed in a
closed-form, rather bothersome calculation work is needed.
ACCURACY OF PROPOSED FORMULA
The accuracies of the proposed MFEM
strength formula and Yura’s formula
were examined by comparing elastic
finite element buckling analysis (i.e.,
eigenvalue analysis). Note that the
comparisons made here is on elastic
buckling strength, and the inelastic Sway restreined
strength will be discussed later in
this paper. Figure 7 shows a finite
element (FE) model of simply
supported I-shaped beam with end
moments and shearing forces with
the top flange continuously
restrained horizontally at the middle
of the flange. Variables considered in (1- β )MFEM
the analyses included the beam end Fig. 7 – Finite element analysis model
support condition for LTB, beam size,
beam length Lb , and moment Table 4 – Beam sizes considered in the analyses
gradient defined by β . There were
two end restraint conditions: one is Web Flange
torsional simple support (only sway is Beam type Depth (mm) Width (mm) thickness thickness
(mm) (mm)
restrained) and another is torsional
fixed support (both sway and warping Case 1 400 400 13 21
are prevented). These end restraint Case 2 588 300 12 20
conditions were made possible by Case 3 600 200 11 17
discrete restraints given at the
centroids of the sections and rigid-bar addition along the sections at the beam ends. Note that
as mentioned previously, the fixed support case may correspond to the experiments conducted
in this study. The beam sizes considered here is shown in Table 4. Three types of beams with
different widths and depths were taken into consideration. The beam length Lb was varied such
that Lb / H ( H is the beam depth) changes from 6 to 100 at an interval of 2. Four types of
moment gradients were also considered by taking β equal to 0, 1, 2, and 3. In these moment
gradients, β = 2 corresponds to an equal end moment and a double curvature bending case
that is typically seen in seismic design. By combining all the variables, more than 500 cases of
buckling analyses were carried out to obtain a database for the critical buckling strengths M FEM .
Figure 8 shows comparisons between the FE analyses and the proposed formula given by Eqn.
(13). Both simple and fixed cases are shown in Figure 8 (a) and (b), respectively. In the figure,
the horizontal axis represents the non-dimensional slenderness ratio defined by M p / M cr* ,
where M p is the plastic moment strength and M cr* is the calculated buckling strength given by
Eqn. (13). Yield strength of the steel beams was assumed to be Fy = 325 N/mm2. On one hand,
the vertical axis shows the ratio of calculated strength to plastic moment strength for the FE
analysis and the proposed formula. The ratios for the FE results and the proposed formula are
expressed as M FEM / M p and M cr* / M p and are shown in the graphs with a small circles and a
broken line, respectively. When both circles and line are closely situated, it indicates that the
proposed formula has good accuracy. Keeping in mind this basis for evaluation, the proposed
formula provides good overall strength estimates. Note that in Figure 8 (a), there is some scatter
in the smaller M p / M cr* region. This is possibly caused by LTB and web shear buckling
interaction, which was not directly considered in the proposed formula.
(a) Torsional simple support (b) Torsional fixed support
Fig. 8 – Comparisons between proposed formula and FE analyses
(a) Torsional simple support (b) Torsional fixed support

Fig. 9 – Comparison between AISC formula (Yura’s formula) and FE analyses
On the other hand, regarding the AISC formula (Yura’s formula) given in Eqn. (1), comparisons
with the FE analyses are made in Figure 9. Note that Yura’s formula was provided primarily for
torsional simple support condition, but in this comparison, an effective length factor of 0.5 was
applied as an attempt to the torsional fixed support case. Overall, Yura’s formula provided
conservative strength estimates, but relatively wider scatter was observed for some cases.
Although relatively good approximation was provided for uniform bending moment cases, larger
scatter was found especially in the cases where the moment gradient was larger (i.e., β was
larger), the depth of the steel beam became smaller, and the length of the beam was longer. It
is interesting to note that smaller depth and larger length conditions corresponded to the cases
where uniform torsion resistant became dominant in comparison with warping torsion resistance.
The larger scatter might be caused by the fact that Yura’s formula was approximate based on
limited numerical solutions. It should be noted that Yura’s formula slightly overestimated the
cases with β = 3, the largest moment gradient in the calculations. Overall, Figures 8 and 9 show
that the proposed formula provided better accuracy than Yura’s formula.
COMPARISON BETWEEN PROPOSED FORMULA AND EXPERIMENTAL RESULTS
Since the proposed LTB formula only provided elastic strength, design equations (mapping
functions) were needed to consider the effects of inelastic behavior and initial imperfections.
Generally, such equations are described as functions of plastic moment, elastic buckling
strength and slenderness ratio. In the AISC specification, which is the most widely recognized
formula, the unbraced length Lb was used to represent the slenderness ratio. In the proposed
buckling formula, however, one might notice that the elastic buckling strength M cr* was not
explicitly expressed as a function of Lb . Therefore, the AISC design equations cannot directly
be used for the proposed formula. Based on a Japanese design standard, the following
equations were proposed as functions of non-dimensional slenderness ratio λb = M p / M cr* :
(a) When λb ≤ λ p :
M n* = M p (30)
(b) When λ p < λb ≤ λe :

 λb − λ p 
M n* =  1 − 0.4  Mp (31)
 λe − λ p 

where
*
λb = M p M cr , λe = 1 0.6 , λ p = 0.6 (32), (33), (34)
(c) When λb > λe :
M n* = M cr
*
(35)
In the above equations, M n* is the design buckling strength considering the inelastic behavior of
the beam. λ p is the slenderness ratio where M n* reaches M p , and λe is the ratio of a transition
point between elastic and inelastic buckling. The value of 0.6 for λ p was assumed on the basis
of material and geometrical non-linear buckling analyses separately conducted.
Figure 10 shows a comparison between the experimental data and the design equations given
in Eqns. (30) through (35). The horizontal axis represents the non-dimensional slenderness ratio,
where M cr* is given by the proposed formula of Eqn. (13). The vertical axis represents the
strength normalized by plastic moment, where the strength is given by the experiments or the
above equations. In Figure 10, elastic buckling
strength shown in Eqn. (13) is also indicated as 1.4
*
a reference. In calculating M cr , the fixed support
1.2
condition was applied by using Table 3 and the No. 1
No. 2
moment gradient parameter β was assumed to 1.0 No. 3
be 2, corresponding to a double curvature and
an equal end moment condition. Note that 0.8
M/Mp
setting an appropriate β is relatively difficult

0.6
because of the beam composite action seen in
positive bending. Based on the calculated 0.4 Eqns. (30) to (35)
stresses of the bottom flanges at both ends of Eqn. (13)
the composite beam, β = 2 was approximately 0.2
Experiments
assumed for a comparison with the experiments.
0.0
As seen in the figure, the design equations are 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
on the safe side against the experimental results, λb
providing with relatively good estimates for Fig. 10 – Accuracy of the proposed formula
Specimens No. 1 and No. 2, and with a conservative estimate for Specimen No. 3. This
conservative estimate may be explained possibly by the following two reasons: One is an
additional effect from the rotational restraint at the top of the flange. As seen in Figure 6, the
rotational restraint was conservatively neglected in the model. The other is the moment gradient
definition for a composite beam. As stated previously, β = 2 was assumed in calculating M cr* ,
but the actual gradient subjected to the steel beam might become larger, resulting in some
strength increase. Note that the AISC (Yura’s) formula is not shown together in Figure 10,
because the torsional fixed support condition seemed to be out of scope for the AISC design
equations. Therefore, a comparison of elastic
buckling strength with the same end condition and Table 5 – Accuracy of Yura’s formula
moment gradient as those in Figure 10 was made in Yura's Proposed
Table 5 between Yura’s and the proposed formulae. Specimen no.
M cr /M p M* cr /M p
Yura’s formula provided conservative elastic
No.1 3.26 3.84
strength estimates.
No.2 1.89 2.18
No.3 0.63 0.68
CONCLUSIONS
This study aimed to clarify the behavior and design of LTB of an I-shaped steel beam with the
top flange continuously restrained laterally using a reinforced concrete floor slab and headed
studs. Three sub-assemblage specimens having different flange widths were first tested under
horizontal loading applied to the columns. The loading condition was reverse curvature and
equal end moment, which is typically seen in seismic design. Through the experiments, it was
shown that the floor slab effectively restrained the flexural-torsional buckling for all the cases.
An elastic buckling strength formula was then derived for the restrained steel beam based on an
energy approach. New displacement functions were introduced so that a closed-form formula
was made possible. In comparison with finite element analyses, the accuracy of the proposed
formula together with that of the existing AISC formula (Yura’s formula) was examined. The
comparisons indicated that the proposed formula provided good accuracy while the existing
AISC formula had a relatively large scatter for some cases. Finally, the proposed formula was
compared with experimental results. In doing so, new design equations to consider the inelastic
effect were proposed. The comparison showed that the proposed equations provided
reasonably accurate and conservative strength estimates to the experimental results.
REFERENCES
American Institute of Steel Construction (AISC). (2010). Specification for structural steel
buildings. Chicago, IL: AISC Committee on Specifications.
Bradford M.A. & Gao, Z. (1992). Distortional buckling solutions for continuous composite beams.
Journal of Structural Engineering (ASCE). 118 (1). 73-89.
Kimura Y. & Yoshino Y. (2016). Effect of lateral-torsional restraint of continuous braces on
lateral buckling strength for H-shaped beams with flexural moment gradient. Journal of
Structural and Construction Engineering. 81 (726). 1309-1319 (in Japanese).
Trahair, N. S. (1993). Flexural-torsional buckling of structures. Bosa Roca, FL: CRC Press.
Wakabayashi, M. & Nakamura, T. (1973). Numerical analysis of Lateral buckling strength of H-
shaped steel beams subjected to end moments and uniformly distributed load. Transactions
of the Architectural Institute of Japan. 208. 7-13 (in Japanese).
Yura, J. A. (1995). Bracing for stability - State-of-the-Art. Proceedings of the ASCE Structures
Congress XIII. (pp. 88–103). New York, NY: American Society of Civil Engineers (ASCE).
Ziemian, R. D. (Ed.). (2010). Guide to stability design criteria for metal structures, 6th Edition.
Hoboken, NJ: John Wiley & Sons.
Deformation-based reliability concept for composite beams
Prof. Dr.-Ing. Wolfgang Kurz

Dipl.-Ing. Natali Kostadinova Kostadinova
Institute of Steel Structures, University of Kaiserslautern (TU),
Paul-Ehrlich-Straße 14, 67663 Kaiserslautern, Germany
wolfgang.kurz@bauing.uni-kl.de
natali.kostadinova@bauing.uni-kl.de
Prof. Dr.-Ing. Karsten Geißler

Gregor Korpas M. Sc.
Institute of Design and Steel Construction, University of Berlin (TU),
Gustav-Meyer-Allee 25, 13355 Berlin, Germany
ek-stahlbau@tu-berlin.de
wojciech.g.korpas@tu-berlin.de
Abstract
The safety concept of Eurocodes is based on design values for actions and material or member
resistance. The resistance of materials and members is defined by their strength, not by their
deformation capacity. However deformation capacity of material, members and joints is needed
to realize the performance of structural elements. This is of specific evidence for composite beams
where slip in the joint between steel and concrete is commonly used to create efficient and
economic structures. An analytical model that is based on the differential equation for elastic
composite action was developed to describe slip and strain distribution in composite beams. The
deformability and the distribution of deformations represent an ultimate limit state for the
composite beam. With scattering parameters for the materials and the composite joint, a Monte
Carlo Simulation was used to define the reliability of the whole composite beam as well as the
weighting for the constituent material parameters. This method provides the possibility to evaluate
the behavior of the members during the whole loading process up to the ultimate limit state.
INTRODUCTION AND PUSH-OUT TESTS AT THE UNIVERSITY OF KAISERSLAUTERN

In the 1920s composite beams were tested with regard to their load bearing capacity. The type of
connection played a major role in the bearing and deformation behavior of the composite
structure. Today, the headed studs are the most widely used connectors. In the course of
intensive research activities in the 1960s, international standardization of the test methods for
screening tests was developed.
On the basis of [Eurocode 4 2010], twelve Push-out tests were performed to determine the
deformation capacity parallel to (longitudinal slip) and perpendicular to the force direction of
headed studs in solid concrete slab and to extend the existing test data. The test results show
that the theoretical load-bearing capacity is in good agreement with the shear test value. In
addition, the stiffness of the headed studs in the concrete is taken into account. The maximum
capacities of the Tests are summarized in Table 1 and Figure 1.
Table 1 – Specimen of Push-out (left) and test results of Push-out (right)
Specimen details
Calotte
Speciment at the
connectors
compressive
Kaiserslautern
Variation
average of
average of
Arithmetic
Arithmetic
measured
Number
University of
concrete
of stud
strength
maximal
Longitudinal
Cubic
loads
slip
slip
Transversal
separation
Ø/lengh fck,cube Pu δu(Pu)
Longitudinal [mm] [N/mm²] [kN] [mm]
slip
S1 16/100 3 69,17 126,54 5,99
Transversal
separation S2 22/100 3 69,17 205,97 7,16
Tension rod S3 16/100 3 46,06 115,51 7,46
Mortar bed
S4 22/100 3 46,06 173,31 7,26
Pu
Applied force
90
25 cyclic loading with 1mm/min
[%]
40 failure loading
with 0,25mm/min
5
0 Slip [mm] δu(Pu)
Fig. 1 – Push-out during test realization
Load-deformation curves of all Push-out specimens at the University of

Kaiserslautern (TU)
1800
1600 S1_1KBD16
S1_2KBD16
1400
S1_3KBD16
Cylinder load [KN]
1200 S2_1KBD22
S2_2KBD22
1000 S2_3KBD22
800 S3_1KBD16
S3_2KBD16
600 S3_3KBD16
400 S4_1KBD22
S4_2KBD22
200 S4_3KBD22
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Relative displacement [mm]
Fig. 2 – Load-deformation curves of Push-out test
As a summary, it was confirmed that the headed studs rigidity increases with increasing stud
diameter in case of constant concrete strength. The composite joint uplifted at the top and bottom
with increasing slippage, the uplift in the joint at the upper row of headed studs being
proportionally faster than the uplift at the lower row of headed studs. As a consequence of the
support of the Push-out tests, compressive forces are generated in the upper region of the
specimen perpendicular to longitudinal shear. Because of the eccentric load transmission, a
bending moment appears as well as a tensile force in the lower region of the specimen to obtain
the state of equilibrium. This compressive force leads to relieving friction forces within the joint.
Cracks began to from at the recesses and grew in the direction of compressive struts from the
lower row of headed studs towards the support. These splitting tensile stresses were transferred
by the transverse reinforcement in the test specimen. In addition, cracks grew from the headed
studs to the bearing and, at the same time, a horizontal crack formed below the upper headed
bolt ridges, resulting from bending in the concrete slabs. Figure 3 presents typical crack patterns
and the headed studs shearing-off as failure mode.
a) b) c) d)
Fig. 3 – Push-out specimens after testing: a) S2 – PO-1KBD22 b) S1 – PO-3KBD16

c) Rotation of concrete slaps by S1 – PO-3KBD16 d) Failure by S1 – PO-1KBD16
TESTS ON FRICTION
Friction is a part of the transmission of longitudinal shear force. In addition to the existing test
results, further 31 tests (V1 - V31) were performed in the laboratory for structural engineering at
the University of Kaiserslautern to determine the coefficient of friction between concrete and steel.
For these experiments a direct proportionality between sliding friction and compressive force of
the specimens was assumed. During the test, both sides of the steel plate were pressed with a
perpendicular load. For this reason, the following relation could be used to determine the
coefficient of friction:
𝑃𝑃
µ= (1)
2 · 𝐶𝐶
with: P pulling force and C compressive force of perpendicular hydraulic press systems
Four test variants of different coatings of steel surfaces were tested. Table 2 presents the
summary of results obtained of both coated and uncoated surfaces.
Table 2 – Conclusions of mean values of friction tests (left) and test procedure (right)
contact surface arith. mean
Specimen uncoated of friction
concrete
steel µ [-]
V1 to V10 unoiled unoiled 0,46 P
V11 to V12 unoiled oiled 0,37
V13 to V19 oiled oiled 0,22
steel coated with
concrete C
epoxy C
V20 to V25 unoiled unoiled 0,17
V26 to V31 oiled oiled 0,19
TESTS ON COMPOSITE BEAMS AT THE UNIVERSITY OF KAISERSLAUTERN

In the laboratory for structural engineering at University of Kaiserslautern, six tests on composite
beams with concrete slabs were performed. The specimens were loaded in a 4-point bending test
up to the ultimate limit state. The deflection, the uplift, the slip between concrete slab and steel
beam, the concrete compression and the steel strains were measured. Table 3 presents the
parameters and results of the slab tests. A schematic illustration and the measuring techniques
used can be found in Figure 4.
Table 3 – Specimen details
Headed
Specimens at the University
Maximal cylinder load

Dimensions Stud Materials
Connectors
of Kaiserlslautern
Deflection
Concrete properties on
Thickness of concrete
Length of composite
the day of the Steel

Diameter / Height
Width of concrete
experiment
Section load
Steel beam
Number
beam
plate
slab
Cubic
Young`s
conpressive Strengt
Modul
strenght
fck,cube Ecm fyk

[mm] [mm] [mm] [mm] [mm] - [N/mm²] [N/mm²] [N/mm²] [kN] [mm]
T1 61 62,51 29565 434 683,14 290,6
T2 39 67,88 30473 335 489,62 189,3
16 / 100
IPE 400
T3 89 61,88 27657 335 634,56 311,2

8000
1500
1600
140
T4 59 47,37 31667 349 556,67 222,8

T5 35 45,19 32016 349 444,5 129,3
T6 18 44,87 30812 349 356,34 67,04
Load
front view [mm]
LVDT’s slip
LVDT’s Deflection
150 3050 800 800 3050 150
8000
Fig. 4 – Experimental setup of the composite beam
The composite beams were subjected to a test load which was increased with a constant cylinder
speed of 0.25 mm / min until failure. In several steps, the loading was paused for the crack
formation to be investigated, as well as for the relaxation of materials to be observed. As a result
of the large deformation of the composite beam and large slip of shear connection, the connecting
studs suddenly failed. In the case of 5 tested composite beams, the failure of the thrust joint
occurred over the entire half of the beam. In composite beams with a higher degree of shear
connection rearrangements of the dowel forces were observed.
Fig. 5 – Composite beam during test procedure (left) and shear failure of T5 (right)
ANALYTICAL DERIVATION AND CALCULATION METHOD FOR HYBRID

CONSTRUCTIONS WITH ELASTIC COMPOSITES
In the past, numerous experimental investigations were carried out on composite structures with
regard to the influence of deformations, e.g. slip on the load bearing behavior. Also several
analytic approaches taking account of rigidity of load bearing components (concrete, shear
connectors and steel beams), internal static indeterminacy and cross-section geometry were
developed to describe the elastic load-bearing behavior of composite beams.
The determination of the stress resultants in the different parts of a composite beam leads to the
solution of limit value problem by inhomogeneous differential equations. The derived basic
equation for determining the deformation of a simple supported beam was considered by many
researchers. [Zhou 2014] provided a closed solution for the differential equation of a 4-point
bending test of composite beams with elastic composite action. This solution is illustrated in
Figure 7 and Figure 8. The sum of the partial bending moments in concrete slab and steel beam
results from the difference between total bending moment and a “composite moment” of axial
forces in both partial sections. The relation between curvature w`` and bending moment is shown
in Figure 6.
Strain of a Bending moment Axial force
composite beam component component
- -
EcIc -
+ +
d -
= -
+
EaIa
+ + +
Fig. 6 – Strain distribution of a composite beam
w``(Ec Ic +Ea Ia ) = −(M − N · d) (2)

The deflection is determined by the double integral of the curvature.
P P (a ≤ x ≤ l-a), M = P⋅a (3)
Pɣ sinh(ωa) �sinh(ωx)+sinh�ω(l-x)�� (4)

a l-2a a N=ɣM-
ω sinh(ωl)
1 P�3x 2 +a2 -3lx� ɣdPa (5)
l w=- + 2 -
6 EIeff (1+βeff ) ω EIeff
x Pɣd sinh(ωa) �sinh(ωx)+sinh�ω(l-x)��
ω3 sinh(ωl)EIeff I
Fig. 7 – Structural System of 4 point bending test (left) and theoretical formula (right)
EI=Ec Ic +Ea Ia (6)

EcIc K c KFK a (7)
EcAc EAeff = ( )l
K cKF + K FK a + K cKa
d EAeff d2 (8)
EaAa EIeff = EI(1 + βeff ) whit βeff =
EI
EaIa β K
ɣ = (1+βeff )·d and ω = �EAF · (1 + βeff )
eff eff
Ec A c Ea A a 1
Kc = , Ka = , KF = nk
l l π2
k = Spring stiffness of connectors, n = Number of connectors
Fig. 8 – Model of springs for a composite beam
Stresses and deformations are basic information for the description of material stiffness. The
internal forces can be calculated on the basis of the integrals of the stress distribution. The
deformation is a result of geometry and strains changes. With the equations according to [Zhou
2014], the fundamental equations for the stresses can be calculated over the height of the partial
cross section.
Based on this method a new analytic approach to elastic-plastic behavior was developed.
Therefore the cross-section was divided into lamellae. Material behavior of steel and concrete
was described by bi-linear approximations. In case that stresses in a lamella leave the elastic
sector, Young`s modulus for this lamella is assumed to be zero and stresses remain fixed on yield
strength or compressive strength. In longitudinal direction the beam is divided into sections. By
that method the differential equation established for elastic theory can be used to calculate beams
with non-elastic materials on a segment-by-segment base. Figure 9 presented the cross-section
of beam portioned into lamellae. Fig. 9 shows at the top illustration a Segment were the whole
cross section shows elastic behavior. In the lower illustration the bottom steel flange in the
segment is yielding. Transition conditions between adjoining segments have to be formulated to
use the elastic analytical approach as a finite solution for elastic-plastic beams.
Apportionment of cross-section
Cross-section
Lamella 1
Lamella 2
Elements
……
Stress distribution Effective cross-section

within the steel EcAc
Ieff,a
Aeff,a
fyk
Fig. 9 – Schematic illustration of analytical procedures for the first Step

In the following Figure 10, the results of analytical evaluations are compared to the results of 4-
point bending tests of the composite beam.
Load-deformation curves in a 4 point bending test and analytic

evaluation at the University of Kaiserslautern (TU)
750
700
650 T1
600
550 T2
Cylinder load [KN]
500
450 T3
400
350 T4
300
250 T5
200
150 T6
100
50 Analytic approach
0
0 50 100 150 200 250 300 350
Deflection in the center w [mm]
Fig. 10 – Load-deformation curves for test from T1 to T6 compared to analytical approaches
COMPOSITE BEAM TESTS AT TU BERLIN
The six composite beam tests at the Technical University of Berlin were separated into two series
with three beams in each. They were planned as three point bending tests. Set one included a
span of the composite beam of 6 m with a HEB 400 steel girder (S235) and a concrete plate
(C30/37) with the height of 14 cm and width of 150 cm. At the second set the span of the
composite beams was 10 m and the steel girder was an IPE 400 (S235) with a concrete plate
(C30/37) with the height of 14 cm height and width of 150 cm. In Figure 11 the load deformation
curves (LDC) of the component tests are shown. The dashed line represents the calculations.
The beam with the minimal degree of partial shear-connection of the second set showed a stud
fracture at a load of 270 kN. After that the load was elevated further up to 276 kN with a huge
increase in deformation. The failure of the other beams was characterized through concrete
compressive fracture.
Fig. 11 – Comparison of the load deformation curve at testing and simulation (TU Berlin)
Table 4 – Results of the tests

Set 1
Elastic limit Ultimate limit state
partial degree of
Deformation [cm] Cylinder load [kN] Deformation [cm] Cylinder load [kN]
shear connection
h = 100% S1-100 1,31 750 4,93 1128
h = 70% S1-70 1,29 720 7,45 1152
h = 40% S1-40 1,32 660 9,96 1077
Set 2
h = 100% S2-100 3,25 220 27,30 348
h = 70% S2-70 3,18 190 32,74 315
h = 40% S2-40 3,18 185 16,38 / 22,84 270 / 276
PROBABILISTIC CALCULATIONS
The reliability of buildings and their components is important for its safety and economy. The
reliability could be described by probability of failure pf or reliability index β. Both are directly linked
to each other.
Table 5 – Reliability elements
Reference period
50 years 1 year
Reliability Failure Reliability Failure
Limit state
index β probability 𝑝𝑝𝑓𝑓 index β probability 𝑝𝑝𝑓𝑓
Sustainability 3,85 7,5 ∙ 10−5 4,7 1,3 ∙ 10−6
Serviceability 1,5 6,68 ∙ 10−2 3,0 1,34 ∙ 10−3
If the influence factors αi and the reliability index β are known, it is possible to calculate the design
values for instance for each material property. For lognormal distributed variables, which are
usually used for modeling the resistance properties, the following equation can used.
2
𝑟𝑟𝑑𝑑𝑑𝑑 = 𝑚𝑚𝑅𝑅𝑅𝑅 ∙ exp(−𝛼𝛼�𝑅𝑅 ∙ 𝛼𝛼𝑅𝑅𝑅𝑅 ∙ 𝛽𝛽 ∙ 𝑉𝑉𝑅𝑅𝑅𝑅 − 0,5 ∙ 𝑉𝑉𝑅𝑅𝑅𝑅 ) (9)
The characteristic value is structural independent. In civil engineering it refers to the 5% fractile
(k=1,645).
2
𝑟𝑟𝑘𝑘𝑘𝑘 = 𝑚𝑚𝑅𝑅𝑅𝑅 ∙ exp(𝑘𝑘𝑅𝑅𝑅𝑅 ∙ 𝑉𝑉𝑅𝑅𝑅𝑅 − 0,5 ∙ 𝑉𝑉𝑅𝑅𝑅𝑅 ) (10)
With the calculation algorithm developed at this research project, the range of resistance with
scattering material values is going to be rated. In contrast to previous investigations, not only the
strength of the materials is considered but also the strain is variated and its influence is evaluated.
For this purpose, the significant points at the stress-strain diagram have been varied as shown
principle in Figure 12. The correlations of the material parameters were considered. For the
characteristics of the shear connection (load capacity and description of deflection along the
length) distributions resulting for instance from push out- and friction- tests by TU Kaiserslautern
were used.
The number of resulting random material parameter combinations is huge, including the
characteristics of the composite joint. For each random combination a simulation (with increasing
the load) has been taken through and the load-deformation curve was calculated.
Fig. 12 – Variation of the stress-strain-relationships
In Figure 13 the load deformation relationship of the composite beam test (set 2, test with the
minimal degree of 40 % for shear connection) is superposed with the probabilistic calculation
based on the same beam characteristics. The simulation results were plotted in form of a point
cloud for achievement the ultimate limit state and - the smaller point cloud at the left side of the
diagram – an elastic material limit value.
Every point represents one simulation with a random material combination according to the
distributions for the material properties. The results show a high median and a wide variance with
regard to the possible deflection achieving the limit state of load capacity, is a typical characteristic
for composite beams. The high deflection is caused by the materials (especially the mild steel)
and the flexible composite joint too.
Fig. 13 – Comparison of LDC - test and probabilistic simulation
LIMIT STATE OF LOAD CAPACITY - DESIGN VALUES

Both, stress and resistance, are distributed variables. The calculation of the resistance density
function is performed with random scattering material value combinations. To define the structural
safety, these two variables are assembled into a two dimensional density function. It represents
the event-space with a volume of 1,0 under the function respectively 100 % probability of
occurrence. The space is being separated into failure and survival area by the limit state function,
which encompasses the point in which the resistance and the stress are equal. The structure is
safe when the failure area is smaller than the required failure probability.
The calculations were performed using a limited number of material combinations N = 50.000. To
increase this number and improve the accuracy of the statistical statement, a field around the
design point is created, which extends by 1 % around the design point (figure 15). Afterwards the
material combinations were sorted according the common probability. The combination with the
highest common probability is the design combination.
Fig. 14 – Event-space
Fig. 15 – Two dimensional density
function
LIMIT STATE OF SERVICE LIFE
The serviceability of a structural component could be achieved either by limiting the component
strains with the aim, that no plasticity occurs, through limiting the width of cracks (for concrete
members) or through limiting the component deformation for instance by a criteria depending from
the span like L/300.
Probabilistic simulation were conducted for an exemplary beam that generate a huge number of
load deformation curves (N = 30.000), each for one possible material combination. In figure 16
the 5%, 50% (mean) and 95% fractile curves of the LDC are presented. In addition to that, two
lines were included, which represent the occurrence of plastic strains within the components. For
each load step the design deformation with a probability of failure 6,68 ∙ 10−2 , respectively β= 1,5,
was determined. By a smaller degree of shear connection the plastic deformations within the
composite joint begin at lower load steps, this influences the distribution of the deformation.
Of course it’s possible to analyse the results of the probabilistic calculations in different directions.
For instance in figures 17 and 18 a number of deformation criteria for each load step in the LDC
were examined. One result is the reliability index for a deformation at a certain load. At the figures
is to recognize, that the beam with the full shear connection could develop plastic strains and is
still safe with regard to the deformation criteria. In contrast to that, at the beam with the minimal
degree of shear connection some deformation criteria occur at lower loads (and before plasticity
begin).
Fig. 16 – Probabilistic - LDC Fig. 17 – Reliability index, Fig. 18 – Reliability index,

h=100% h=40%
OUTLOOK
The basic principles of this safety concept were first developed for the well known single span
composite beam. Until now the deformation behavior of the composite joint was idealized and not
exactly considered. New quotas for the description of composite joints are developed.
The main target of the actual research is to develop deformation based criteria to design the
composite beam (in analogy to the steel concrete beam) and to quantify the ductility. Further all
hybrid beams, with a flexible composite joint too, should be calculated with a design concept on
the same basics.
ACKNOWLEDGEMENTS
The presented results were financed by the support of the DFG research project "Probabilistic
method for deflection - based design of composite beams". The project’s numbers are KU 2571/2-
1 / 602657 and GE 2011/4-1 / 10041685. The authors would like to thank Nelson
Bolzenschweißtechnik GmbH & Co.KG and the steel construction company C&P Freiberg for their
support.
REFERENCES
DIN EN 1994-1-1 (2010). Eurocode 4: Design of composite steel and concrete structures – Part
1-1: General rules and rules for buildings. Beuth-Verlag, Berlin, Germany.
Vlascici, Milivoi-R. (2016). Bestimmung der Reibkoeffizienten zwischen Stahl und Beton.
Bachelor thesis, University of Kaiserslautern, Germany.
Kludka, M.; Kurz, W. (2015) Schlussbericht zu dem IGF-Vorhaben Klebstoff als dauerhaftes
Verbundmittel bei Stahlverbundträgern. Final Report IGF 17458 N, Joint Research Center
University of Kaiserslautern, Germany
Korpas, G., Geißler, K. (2016). Zuverlässigkeitstheoretisches Konzept für die
verformungsorientierte Bemessung von Verbundträgern, 20. DASt- Forschungskolloquium
Essen, Germany.
Zhou, D.; Pahn, M.; Kurz, W. (2014). Beitrag zur Berechnung von Verbundträgern mit elastischem
Verbund. Stahlbau 83, Heft 4, Ernst & Sohn Verlag, Berlin, Germany.
EFFECTS OF COMPOSITE SLAB ON SHEAR STRENGTH OF STEEL
PLATE GIRDERS
Peter Y. Wang
Princeton University
Princeton, NJ
pywang@princeton.edu
Maria E. Moreyra Garlock

Princeton University
Princeton, NJ
mgarlock@princeton.edu
ABSTRACT
Steel plate girder bridges are often vulnerable to shear buckling at elevated fire temperatures
due to their slender webs. A composite slab could be effective in increasing the shear capacity
in fire, but a measure of this slab effectiveness has not been done to-date. Using nonlinear
finite element models, this paper evaluates the amount of additional shear strength that a
composite slab can provide to a steel plate girder under fire. The elastic shear buckling load
(Vcr), (ultimate) shear post-buckling load (Vu), and stiffness are evaluated for 16 models with
varying parameters: composite slab (or not), web slenderness ratio, and stiffener spacing. The
steel girders are subject to a temperature of 700°C. Results show that the composite slab
increases Vcr on the order of 10% and Vu on the order of 30% compared to the non-composite
models. The slab is most effective at increasing Vcr and Vu for larger stiffener spacings.
1. INTRODUCTION:
Fire in bridges are not uncommon and have serious economic consequences. Fires underneath
or adjacent to highway bridges can directly expose bridge girders to elevated temperatures. The
effect of fire on bridges has ranged from extensive deformation to complete collapse, often
requiring costly demolition, replacement and detours. Shear buckling is a prominent failure
mode for steel plate girder bridges under fire due to the high slenderness ratios found in plate
girder webs as well as the reduction in mechanical properties of steel at elevated temperatures.
While there have been some studies related to this subject, there have been very few that
discuss how a composite slab affects the shear strength.
Shear buckling was identified as the mode of failure for both the MacArthur Maze bridge fire in
Oakland, California and the bridge fire in Birmingham, Alabama [1] as observed by the typical
diagonal out-of-plane web deformation that characterizes this type of buckling. Following elastic
shear buckling, the thin plates are capable of significantly more strength, referred to as (ultimate)
post-buckling shear strength. While analytical solutions exist for the elastic shear buckling load
of plates [2], the shear post-buckling behavior is more difficult to characterize analytically. This
is true for ambient temperature and elevated temperatures caused by fires. Glassman et al. [3]
have proposed a means of measuring this post-buckling shear strength of a non-composite
steel girder under fire; however, the elastic and post-buckling capacities of plate girders are
more difficult to predict when a composite slab is present.
To the authors’ knowledge, there have been two studies of web shear buckling that consider the
composite slab on a steel girder: Aziz et al. [4] performed an experimental study, while
Glassman et al. [1] performed a comparative analysis using finite element modeling and
experimental results. These two papers reference other publications related to web shear
buckling for non-composite steel girders. Especially relevant are [5], [6], [7] and [8], the last of
which discusses a compression-based shear buckling model and empirical equations for post-
buckling shear strength prediction for plate girders.
Study [1] found that modeling the composite slab added significant shear strength to the steel
girder and matched well with experimental values, while neglecting the slab resulted in post-
buckling shear capacities of about half of the experimental values. Modeling thermal gradients
were not found to be significantly more accurate than modeling a uniform average temperature.
For optimizing efficiency and accuracy, the authors recommended using a uniform temperature
in the steel girder equal to the average temperature of the cross section and to include the slab
in the model if it is a composite section.
The current study builds upon previous studies [1,4]. The objective of this paper is to evaluate
the amount of additional shear strength that a composite slab can provide to a steel plate girder
under fire. The influence of web slenderness and vertical stiffener spacing on the slab
effectiveness for increasing shear strength is evaluated. To this end, 16 finite element models
were analyzed, where each model had a different combination of the following parameters:
composite slab vs. no slab; web plate slenderness (D/tw); and vertical stiffener spacing (a/D).
2. BACKGROUND:
Case study
On May 9, 2013, a tanker truck near Harrisburg, PA crashed on a highway ramp leading from I-
81 Northbound to Route 322 Eastbound. The truck was carrying 7,500 gallons of fuel and
ignited upon collision, directly exposing the bridge girders of Route 322 to fire for a duration of
45 minutes (see Figure 1). While the exposed Span 2 of Route 322 Eastbound did not collapse,
extensive out-of-plane web deformation and waving were observed in the steel plate girder
webs, prompting demolition by PennDOT. This highway bridge girder was used as the prototype
for the current study.
Fig. 1 – Bridge fire near Harrisburg, PA in 2013
The slenderness ratio of a web is defined as the web depth (D) divided by the web thickness (tw).
While typical bridge plate girders from the Federal Highway Administration’s 1982 Standard
Plans have web slenderness ratios ranging from 128 to 270 [9], the Harrisburg bridge girder had
a stockier web, with a slenderness ratio of 80. The dimensions of the original Harrisburg girder
are given in Table 1. Note that this girder did not have vertical stiffeners (as will be discussed
later).
Table 1 - Dimensions of Girder from the Harrisburg Bridge Fire
Cross-Section View
Dimension Description (without slab)
A N/A (no stiffeners) transverse stiffener spacing
D 1.143 m web depth
Tw 0.0143 m web thickness
D/tw 80 web slenderness ratio
a/D N/A (no stiffeners) panel aspect ratio
L 39.2 m span between supports
bf top 0.3048 m top flange width
tf top 0.0159 m top flange thickness
bf bottom 0.4064 m bottom flange width
tf bottom 0.0222 m bottom flange thickness
Elastic Shear Buckling

Equation 1 presents the elastic shear buckling load, Vcr [2]. In this equation, E is the modulus of
elasticity, 𝜐 is the Poisson’s ratio, and D and tw were defined above.
!! !
(1) 𝑉!" = 𝑘 ∗ ! !
∗ 𝐷𝑡!
!"∗ !!!! ∗
!!
The coefficient for elastic shear buckling, k, depends on the a/D ratio, also referred to as the
panel aspect ratio. The parameter a is the transverse (vertical) stiffener spacing. Smaller a/D
ratios result in larger k and therefore greater resistance to elastic shear buckling. Equations 2(a)
and 2(b) describe how to calculate k for a panel that represents flanges as simply supported
and fixed, respectively [2]. Equation 2(c) represents the flange stiffness by considering the
flange tf and web thickness tw.
!.!!
(2a) 𝑘!! = 5.34 + ! !
for a/D ≥ 1
!
!.!" !.!!
(2b) 𝑘!" = 8.98 + ! !
− ! !
for a/D ≥ 1
! !
! ! !! ! !!
(2c) 𝑘 = 𝑘!! + 𝑘!" − 𝑘!! 1− 2− for < <2
! ! !! ! !!
3. FINITE ELEMENT MODELS
Parameters
In the finite element study, three modeling parameters were addressed: composite slab vs. no
slab, web slenderness ratios (D/tw), and panel aspect ratios (a/D). Sixteen different models
were developed based on varying these three parameters for the Harrisburg bridge girder case
study defined earlier. The different models of the parametric study are listed in Table 2. The
naming convention is as follows: C signifies a composite slab and N signifies no slab (thus
essentially assuming a non-composite section); the first number refers to the a/D ratio of the
panel; and the number following the hyphen refers to the web slenderness ratio. For example,
C2-80 signifies a model with a composite slab, an a/D ratio of 2 (panel length a equal to 2*D),
and a web slenderness ratio of 80.
Table 2 – Modeled specimens for finite element parametric study
Model name Slab modeling a/D D/tw tw (m) a (m)
N1-‐80 1 80 0.0143 1.143
N1-‐120 1 120 0.0095 1.143
No slab
N1-‐160 1 160 0.0071 1.143
N1-‐200 1 200 0.0057 1.143
N2-‐80 2 80 0.0143 2.286
N2-‐120 2 120 0.0095 2.286
No slab
N2-‐160 2 160 0.0071 2.286
N2-‐200 2 200 0.0057 2.286
C1-‐80 1 80 0.0143 1.143
C1-‐120 1 120 0.0095 1.143
Composite slab
C1-‐160 1 160 0.0071 1.143
C1-‐200 1 200 0.0057 1.143
C2-‐80 2 80 0.0143 2.286
C2-‐120 2 120 0.0095 2.286
Composite slab
C2-‐160 2 160 0.0071 2.286
C2-‐200 2 200 0.0057 2.286
The dimensions of the modeled slab are based on the tributary width of the Harrisburg case
study: 2.34 m wide and 0.23 m deep. The same concrete slab was modeled for all specimens.
Because the Harrisburg bridge girder was not considered “slender” (D/tw = 80), no transverse
stiffeners were present along the 39.2 m span of the girders. In this study, however, we will
assume a/D equal to 1.0 and 2.0 as shown in Figure 2. The authors realize that for the given D,
the combinations of a/D and D/tw might not be used in design (e.g. D/tw = 80 would not use
stiffeners); however the study is nevertheless effective in giving us insights on the contributions
that a composite slab has on the shear strength of a steel girder in fire.
Fig. 2 - Web panel aspect ratios; left a/D = 1.0, right a/D = 2.0
A range of web slenderness ratios was achieved by varying the thickness of the girder web; web
depth was kept constant at the D value of 1.143 m from the Harrisburg case study (Table 1).
The slenderness ratios tested were D/tw = 80, 120, 160 and 200, where D/tw = 80 corresponds to
the original Harrisburg bridge girder dimensions. Flange dimensions were held constant for all
specimens based on the Harrisburg cross section in Table 1. Note that the thickness of the web
also plays a role in the amount of restraint afforded by the flanges (Equation 2(c)).
Finite Element Properties and Restraints
The steel was modeled as ASTM A572 Gr. 50 (345 MPa). The modulus of elasticity at ambient
temperatures was taken as 200 GPa, while the Poisson’s ratio was taken as 0.3. Stress-strain
curves for the steel at elevated temperatures were given by Eurocode 3 Part 1.2 [10]. Concrete
was modeled with a strength of f’c = 27.6 MPa (4 ksi). The stress-strain curve for concrete was
based on EC2 Part 1.2 [11].
The steel girder is modeled with deformable S4 shell elements with five integration points
through the thickness. The concrete slab is modeled with 3D deformable solid elements
(C3D8R hex elements). The finite element models and mesh are shown in Figure 3. The mesh
size was validated by comparison to the elastic shear buckling, where values lay within 5% of
the theoretical value based on Equation 1. Because the theoretical solution assumed a
symmetric section, and the Harrisburg bridge girder had different top and bottom flanges, the
average thickness of the flanges was used for determining the k (Eqn. 2c) for the theoretical
solution [2, 12]. The concrete slab mesh was chosen to match the mesh size of the steel shell
elements. Although an optimization study could also have been conducted for the concrete
mesh, the size used in this study is reasonable compared to [1], where the model was validated
against experimental results.
A tie constraint was imposed between the bottom surface of the concrete slab and the top
surface of the top flange in order to achieve composite action, where the top flange surface was
selected as the master surface [13]. In addition, the shear studs were not modeled and thus 100%
composite action is assumed in the fire condition. Such an assumption is not unreasonable
given the experimental and numerical conclusions of [1, 4], which indicate that the slab adds
about 50% more shear capacity compared to a member without a slab. Such results imply that
the slab has some composite action in a fire although it cannot be quantified how much. Other
boundary conditions are shown in Figure 3.
(a) Steel Plate Girder Boundary Conditions (b) Slab Boundary Conditions
(c) Boundary conditions defined
Fig. 3 – Finite element model boundary conditions shown for a/D = 1 (a/D = 2 is the same).
Analysis Method and Temperature
An eigenvalue extraction analysis was employed to find the elastic shear buckling load Vcr. The
first positive eigenmode was chosen as the initial imperfection, which was scaled to a factor of
D/10,000. This imperfection magnitude was validated by previous authors as discussed in [1,
12].
A nonlinear solver employing Newton’s method was used to determine the post-buckling shear
strength, Vu, and behavior (load-displacement) of each girder specimen. The post-buckling
strength was taken as the maximum load value on the load displacement curve. The web was
loaded on side “2” of Figure 3. To meet the paper objectives, a pure shear scenario is modeled
using the boundary conditions of Figure 3.
Real fire temperatures change with time and create thermal gradients in the girders. It was
observed by Glassman et al. [1] that modeling the thermal gradients created did not increase
the accuracy of Vu significantly more than modeling the whole girder at a uniform temperature
equal to the average section temperature. Thus, the steel girders were modeled at a steady
state uniform temperature of 700°C, which represents a temperature close to the maximum
average based on experiments [1, 4] and CFD modeling [7]. Aziz et al [4] indicates that when
the average temperature is about 700°C, the average slab temperature is about 100°C.
Therefore a constant slab temperature of 100°C was used in the models.
At 700°C, steel has a modulus of elasticity, proportional limit, and yield stress equal to 0.23,
0.075, 0.13, respectively, times the ambient temperature values. It therefore represents a state
in the fire event where the material has weakened but it is less than the temperature causing a
phase change to austenite [7, 14]. Vcr is proportional to the modulus of elasticity (as seen in
Equation 1), while shear yield, Vy , is proportional to the yield stress of steel. Since the modulus
of elasticity has a greater reduction factor than the yield stress at 700°C, girder sections that
would be governed by shear yield at ambient temperatures may be governed by shear buckling
at fire temperatures.
4. PARAMETRIC STUDY RESULTS

The results of the parametric finite element study are shown in Table 3, which summarizes the
elastic shear buckling load (Vcr) and post-buckling shear capacities (Vu) of each girder specimen
modeled. Vcr / VcrN describes the ratio of the elastic shear buckling load of the specimen to that
of the same specimen with no slab modeled (e.g. C1-80/N1-80, C2-120/N2-120, etc). Similarly,
Vu / VuN describes the ratio of the post-buckling shear strength of the specimen to that of the
same specimen with no slab modeled. Also included in the Table is the value of shear when the
proportional limit is reached, Vp, represented by:
!! !!
(3) 𝑉! = 𝐷𝑡!
!
where σy is the yield stress at ambient temperature (345 MPa) and kp is the proportional limit
reduction factor (0.075 for 700°C) [10]. Note in Table 3 that there are only four values of Vp that
repeat for each set since only tw varies in Eqn. (3).
Table 3 –Finite Element Parametric Study Results
Model name Vcr (kN) Vu (kN) Vp (kN) Vcr/VcrN Vu/VuN

N1-‐80 696 653 245 1 1
N1-‐120 219 396 163 1 1
N1-‐160 93 278 122 1 1
N1-‐200 48 213 98 1 1
N2-‐80 503 550 245 1 1
N2-‐120 164 328 163 1 1
N2-‐160 72 226 122 1 1
N2-‐200 37 169 98 1 1
C1-‐80 764 809 245 1.10 1.24
C1-‐120 237 507 163 1.08 1.28
C1-‐160 102 365 122 1.10 1.31
C1-‐200 53 285 98 1.10 1.34
C2-‐80 592 743 245 1.18 1.35
C2-‐120 186 447 163 1.13 1.36
C2-‐160 81 311 122 1.12 1.37
C2-‐200 42 234 98 1.13 1.38
From Table 3, the following is observed:
 Vp is reached before Vcr for all models with D/tw equal to 80 and 120. Therefore, for stockier
webs, a stiffness change should occur before elastic buckling (stiffness is discussed later).
 The presence of a slab increases Vcr as noticed by the Vcr / VcrN ratios for the C1 and C2
models. On average Vcr increases by 12%.
 The presence of a slab increases Vu as noticed by the Vu / VuN ratios for the C1 and C2
models. On average Vu increases by 33%.
 The Vcr / VcrN ratios are similar for the C1-80 through C1-200 specimens, as well as the C2-
80 through C2-200 models. The slab’s effectiveness in increasing Vcr does not appear to be
affected by plate slenderness for a/D ratios = 1.0 and 2.0.
 The Vu / VuN ratios increase with increasing slenderness ratios for both C1 and C2 models.
The relationship between this ratio and slenderness is plotted in Figure 4, where it is seen that
the relationship is nearly linear.
 The presence of a slab increases Vcr and Vu more for a/D = 2.0 than a/D = 1.0. Such is
observed by the larger Vcr / VcrN and Vu / VuN ratios for C2 models compared to C1 models.
1.40
1.35
Vu/VuN
1.30
1.25 Composite a/D=1

Composite a/D=2
1.20
0 50 100 150 200 250
Web slenderness ratio D/tw
Fig. 4: Ratio of post-buckling strength of composite specimen over that of no-slab specimen vs.
slenderness ratio
Figure 5 plots the shear load versus vertical deformation (of the lower corner) of the plate. Only
the initial stages of loading are shown to highlight the stiffness of each model. These curves peak
at Vu shear values listed in Table 3, which is intended to document strength (whereas in Figure 5
stiffness is documented). Each of the three plots of Figure 5 keeps one variable constant, as
labeled in the figure caption. The following is observed:
 The models with 80 and 120 slenderness show a clear change in stiffness at values close to
the Vp values listed in Table 3. At this point, the paths of the composite vs. no slab models
show clear divergence, where the composite model has more stiffness. Therefore, the slab
appears to be most effective in providing shear stiffness only after reaching the proportional
limit of steel.
 The models with 160 and 200 slenderness change stiffness more gradually; however, a
bifurcation of the composite vs. no slab models is seen at values less than Vp. Since Vcr is
less than Vp in these cases, the slab will begin to contribute more strength (and stiffness) after
Vcr is reached.
 A larger D/tw corresponds to a lower stiffness. This is expected and observed for all
combination of parameters.
 The a/D = 1.0 models show that the slab has a greater effect on stiffness than the a/D = 2.0
models (Figure 5(a) vs.5(b))
 Models with a/D = 1.0 have significantly higher stiffness than a/D = 2.0 (Figure 5(c)).
 The slab is most effective in increasing the girder stiffness when D/tw is large. Note, for
example, in Figure 5(a), how the solid lines (representing composite models) rise further
above the similarly colored dashed lines (representing the non-composite models) for larger
D/tw.
C1-80 N1-80 C2-80 N2-80

C1-120 N1-120
C2-120 N2-120
C1-160 N1-160
600 C1-200 N1-200 600 C2-160 N2-160
C2-200 N2-200
500 500
80
400 400 80
Shear (kN)
Shear (kN)
300 300
120 120
200 160 200
200 160
100 100
200
0 0
0 0.005 0.01 0 0.005 0.01
Disp of bottom of web, far end (m) Disp of bottom of web, far end (m)
(a) constant variable: a/D = 1.0 (b) constant variable: a/D = 2.0
C1-80 C2-80
C1-120 C2-120
C1-160 C2-160 80
600 C1-200 C2-200
500
80
400 120
Shear (kN)
300 160
200
120
200 160
100 200
0
0 0.005 0.01
Disp of bottom of web, far end (m)
(c) constant variable: composite slab
Figure 5 – Load-deflection curves of specimens
5. SUMMARY AND CONCLUSIONS
The objective of this paper was to evaluate the amount of additional shear strength that a
composite slab can provide to a steel plate girder under fire. Using finite element analysis, 16
plate girder specimens were modeled using the following parameters: with and without a
composite slab, two stiffener spacing (a/D) ratios, and four slenderness (D/tw) ratios. The elastic
shear buckling load (Vcr), the (ultimate) shear post-buckling load (Vu), and the stiffness are
evaluated. The following conclusions are based on the study:
• The effect of the composite slab on Vcr and Vu was not negligible. The presence of a
composite slab increased Vcr and Vu by 12 and 33 percent, respectively, on average, for the
parameters studied.
• Regarding a/D, the presence of a composite slab increases Vcr and Vu more for a/D = 2.0
than a/D = 1.0. Conversely, the composite slab has a greater effect on stiffness for a/D =
1.0 compared to a/D = 2.0.
• Regarding D/tw, the composite slab is most effective in increasing the girder stiffness when
D/tw is large. The slab’s effectiveness in increasing Vcr does not appear to be affected by
D/tw.
These conclusions are limited to the parameters and assumptions described throughout. More
research is needed before general conclusions that apply to all steel girders in fire can be made.
For example, larger a/D and D/tw are needed for validation of the proposed relationships.
Experimental studies to verify the predicted post-buckling shear strengths would also benefit the
study. Further, a closer examination of the effectiveness of the shear studs in providing
composite action in areas of high shear and elevated temperature are also needed.
Nevertheless, this paper provides good insight to understand the additional shear strength that
a composite slab can provide to a steel plate girder under fire.
ACKNOWLEDGMENTS
The authors would like to acknowledge Spencer Quiel, José Alos-Moya, and Jonathan
Glassman for their contributions to this work.
REFERENCES
[1] Glassman, J. D., Garlock, M. E. M., Aziz, E. M., and Kodur, V. K. (2016). Modeling
parameters for predicting the postbuckling shear strength of steel plate girders. Journal of
Constructional Steel Research, 121, 136-143.
[2] Timoshenko, S. P. and Gere, J. M. (1961). Theory of Elastic Stability, Second Edition. New
York: McGraw-Hill Book Company, Inc.
[3] Glassman, J., and Garlock, M. (2017). A compression model for ultimate postbuckling shear
strength at elevated temperatures. Journal of Structural Engineering, 143(6).
[4] Aziz, E. M., Kodur, V. K., Glassman, J. D. and Garlock, M. E. M. (2015). Behavior of steel
bridge girders under fire conditions. Journal of Constructional Steel Research, 106, 11-22.
[5] Vimonsatit, V., Tan, K.-H, and Ting, S.-K. (2007). Shear strength of plate girder web panel at
elevated temperature. Journal of Constructional Steel Research, 63, 1442-1451.
[6] Vimonsatit, V., Tan, K.-H., and Qian, Z.-H. (2007). Testing of Plate Girder Web Panel
Loaded in Shear at Elevated Temperature. Journal of Structural Engineering, 133(6), 815-
824.
[7] Alos-Moya, J., Paya-Zaforteza, I., Garlock, M.E.M., Loma-Ossorio, E., Schiffner, D., &
Hospitaler, A. (2014). Analysis of a bridge failure due to fire using computational fluid
dynamics and finite element models. Engineering Structures, 68, 96-110.
[8] Glassman, J., and Garlock, M. E. M. (2016). “A Compression Model for Ultimate
Postbuckling Shear Strength.” Thin-Walled Structures, vol. 102, pp. 258-272.
[9] Standard plans for highway bridges. (1982). Washington, D.C.: U.S. Dept. of Transportation,
Federal Highway Administration (FHWA).
[10] Eurocode 3: Design of steel structures: Part 1.2 General rules - Structural fire design
(together with United Kingdom National Application Document). (2001). London: BSI.
[11] Eurocode 2: Design of concrete structures: Part 1.2 General rules - Structural fire design
(together with United Kingdom National Application Document) / British Standards Institution).
(2000). London: BSI.
[12] Garlock, M. E. M., Glassman, J.D. (2014). Elevated temperature evaluation of an existing
analytical model for steel web shear buckling, J. Constr. Steel Res. 101, 395–406.
[13] Dassault Systemes. (2010). ABAQUS Analysis User's Manual (6.10).
[14] Payá-Zaforteza, I., & Garlock, M. E. M. (2012). A numerical investigation on the fire
response of a steel girder bridge. Journal of Constructional Steel Research,75, 93-103.
STRUCTURAL RELIABILITY OF STEEL-CONCRETE COMPOSITE
COLUMNS AND FRAMES
Mark D. Denavit
University of Tennessee, Knoxville
Knoxville, Tennessee, USA
mdenavit@utk.edu
ABSTRACT
The design provisions for steel-concrete composite columns in the AISC Specification for
Structural Steel Buildings have evolved significantly in recent editions. The changes have made
the provisions more accurate, direct, and broadly applicable. These developments have been
justified through comparisons to prior provisions and analyses that indirectly address reliability.
Thus, a full accounting of the structural safety of the current provisions in the AISC Specification
for the strength of composite members has not been performed. This work represents the first
steps of such an assessment. Reliability analyses are performed using Monte Carlo simulations
and specifically focusing on two aspects. First, the resistance factor for compressive strength of
highly steel dominant members. Second, the transition between the relatively lower resistance
factor for compressive strength to the relatively higher resistance factor for flexural strength for
members subjected to combined load.
INTRODUCTION
Steel-concrete composite columns are a viable alternative to traditional structural steel or
reinforced concrete columns for use within building framing systems. Advantages of composite
columns include high strength, stiffness, and ductility. Advantages can also be realized during
construction. For concrete-filled steel tubes (CFT), whether they be circular (CCFT, Figure 1a)
or rectangular (RCFT, Figure 1b) in shape, the steel tube takes the place and eliminates the
need for temporary formwork. For steel-reinforced concrete (SRC, Figure 1c) columns within
taller buildings, the steel columns can be erected several stories above of the placement of
concrete. In this cost-effective construction process, the steel can support the construction
loads and the different trades can work with minimal interference.
One disadvantage of composite framing systems is that the development of their design
provisions has often lagged behind that of the traditional systems. This has resulted in
provisions that do not consistently reflect the advantages that composite columns and framing
systems can provide. When composite columns were first introduced to the AISC Specification
in the 1986 edition (AISC 1986), the strength of a composite column was based on that of an
equivalent steel column. More mechanistic strength provisions were introduced as part of a
major restructuring of the composite provisions for the 2005 edition (AISC 2005). Further
changes were made for the 2010 edition (AISC 2010) mainly addressing the use of noncompact
or slender tubes in CFT members. In the recently released 2016 edition (AISC 2016), the
provisions for the compressive strength of composite columns have been revised and new
frame stability provisions have been added which were specifically calibrated for composite
systems (Denavit et al. 2016).
Figure 1 – Composite cross sections

Over the same time period, several investigations of the reliability and structural safety of
composite columns were conducted. Kogut and Chou (2004) investigated the possibility of the
use partial resistance factors within the AISC Specification. Mirza and Skrabek (1991, 1992)
investigated the safety of SRC columns designed with ACI Code provisions. Kvedaras and
Kudzys (2006) investigated the safety of CFT columns with an internal void designed with
European provisions. Beck et al. (2009) investigated the safety of CCFT columns designed with
several design codes. Aslani et al. (2016) investigated the reliability of flexural stiffness
provisions, focusing on serviceability limit states. However, a rigorous and complete reliability
analysis of the composite column provisions within the AISC Specification (AISC 2016) has not
been performed since the member strength was based on that of an equivalent steel section
(Lundberg and Galambos 1996). The structural safety of the changes to the design provisions
have been justified based on comparisons to previous provisions or through analyses that
indirectly address reliability.
In this work, the structural safety of steel-concrete composite columns designed by the
provisions of the 2016 AISC Specification (AISC 2016) is evaluated. The focus of this work is on
the compressive strength of composite columns and the strength of short composite columns
subjected to combined axial compression and bending moment.
CURRENT DESIGN PROVISIONS

For practice in the United States, design provisions for composite columns currently exist in
both the AISC Specification (AISC 2016) and the ACI Code (ACI 2014). As noted in the
introduction, the provisions for composite columns within the AISC Specification have been the
subject of significant development over the past several editions. The provisions within the ACI
Code, on the other hand, have been relatively unchanged and have been considered for
removal in future editions.
The axial compressive strength of a composite column per the AISC Specification (AISC 2016)
is given by Equations (1) through (6) noting that CFT members with internal reinforcing or not
classified as compact are excluded in this work.
 Pno  0.658Pno Pe
 when Pno Pe  2.25
Pn    (1)
 0.877 Pe when Pno Pe  2.25
 2 EI eff
Pe  (2)
Lc 2
 Fy As  Fysr Asr  0.85 f cAc for SRC

Pno   (3)
 Fy As  C2 f cAc for CFT
 E I  Es I sr  C1 Ec I c for SRC
EI eff   s s (4)
 Es I s  C3 Ec I c for CFT
 A  Asr 
C1  0.25  3  s   0.7 (5)
 Ag
 
A 
C3  0.45  3  s   0.9 (6)
 Ag
 
where, Pn = axial compressive strength, Pno = axial compressive strength without length effects,
Pe = elastic critical buckling load, EIeff = effective flexural rigidity, Lc = column effective length, Fy
= steel yield stress, As = area of steel, Fysr = reinforcing yield stress, Asr = area of reinforcing, f'c
= concrete compressive strength, Ac = area of concrete, Es = modulus of elasticity of steel, Is =
moment of inertia of steel, Isr = moment of inertia of reinforcing, Ec = modulus of elasticity of
concrete, Ic = moment of inertia of concrete, Ag = gross cross-sectional area, and C2 = 0.85 for
RCFT and 0.95 for CCFT.
Using these equations, the nominal compressive strength of a composite member, Pn, is always
greater than that of the bare steel component, Pn,steel (which is also calculated with Equation (1),
but using Pno = AsFy and EIeff = EsIs). However, since the resistance factors are different (ϕ =
0.75 for composite, ϕ = 0.9 for steel), it is possible for the design compressive strength of a
highly steel dominant composite column, ϕPn, to be less than the design compressive strength
of the bare steel component, ϕPn,steel. Recognizing this possibility, there is a provision within the
AISC Specification (AISC 2016) that ϕPn need not be taken as less than ϕPn,steel. This provision
is logical and ensures that the addition of concrete never results in a reduction of compressive
strength, but it is unclear how this provision could be applied to the determination of strength
under axial compression and bending moment. A potentially more consistent approach could
include a resistance factor for composite columns that increases with the steel ratio for highly
steel dominant members. However, such a recommendation would require strong justification.
There are several methods defined within the AISC Specification (AISC 2016) for determining
the interaction strength of composite cross sections under combined axial compression and
bending moment. The most widely cited and widely used method is the plastic stress distribution
(PSD) method. In this method, the steel components are assumed to have reached the yield
stress, Fy, in either tension or compression and the concrete components are assumed to have
reached a stress of 0.85f′c in compression (a higher stress, 0.95f′c, is permitted for CCFT to
account for confinement). The strength can be computed for a continuum of points in the
interaction diagram based on different locations of the neutral axis. Such an interaction diagram
is shown for the case of a RCFT cross section in Figure 2a. For this cross section, the outside
dimensions of the steel tube are H = B = 304.8 mm and thickness is tn = 7.94 mm, which
corresponds to a steel ratio, ρsn = Asn/Ag = 0.10. Material strengths were selected as Fyn = 344.7
MPa and f'cn = 55.16 MPa. The subscript “n” indicates that these values are nominal (or, for the
case of tn, the nominal design thickness of the steel tube).
Closed-form solutions based on the PSD method are published for five specific anchor points in
the interaction diagram (AISC 2013). These five points, labeled “A”, “E”, “C”, “D”, and “B”, when
connected by straight lines (Figure 2b) form a approximation of the continuous curve. The
interaction diagram for the cross section constructed using the strain compatibility method
defined within the ACI Code (ACI 2014) is shown in Figure 2c.
Figure 2 – Axial compression and bending moment interaction diagrams

The AISC Specification (AISC 2016), does not specifically state how to apply the resistance
factor for strength under combined axial compression and bending moment, only that ϕ = 0.75
for axial compressive strength and ϕ = 0.9 for flexural strength. The commentary to the AISC
Specification provides an approximate method for applying these resistance factors whereby the
ordinate of each calculated point within the interaction diagram is multiplied by ϕ = 0.75 and the
abscissa is multiplied by ϕ = 0.9. This method can lead to potentially unconservative situations
where an insufficient reduction is applied to the “bulge” region near point “D”. Note that in Figure
2b point “D” within the factored interaction diagram is near the line between points “D” and “B” of
the nominal interaction diagram. The potential unconservative nature of this approximate
method of applying the resistance factor is part of the reason it is recommended within the
commentary to use only points “A”, “C”, and “B” when constructing interaction diagrams for the
design of composite columns.
A potentially more consistent approach, and one which would not restrict the number of points
that could be used in an interaction diagram, would be to apply the resistance factor in a similar
manner to the method given within the ACI Code (ACI 2014). Within the ACI Code, the
resistance factor for combined axial load and bending moment is given as a function of the
extreme tensile strain in the steel and computed for each point on the interaction diagram. Thus,
each point has only one resistance factor, effectively moving the point inward to the origin, as
opposed to down and to the left as in the approach defined within the commentary to the AISC
Specification (AISC 2016). This method cannot be directly applied to the AISC Specification
since infinite curvature and thus infinite strain are assumed within the PSD method. However,
this method can be used if the resistance factor is reformulated as a function of the ratio of the
location of the neutral axis to the depth of the section (c/d) as in Equation (7). Such a
formulation would be compatible with the provisions of the AISC Specification and thus an
resistance factor similar to Equation (7) represents a promising alternative to the current
approximate approach given in the commentary to the AISC Specification (AISC 2016).
 0.75 for c d  0.6


  0.75  0.15  d c  5 3 for 0.375  c d  0.6 (7)
 0.9 for c d  0.375

RELIABILITY ANALYSIS
The resistance factor used in the evaluation of a strength limit state can be calculated using
Equation (8) (ASCE 2010).
R
 e VR (8)
Rn
where, α is a linearization coefficient, β is the desired reliability index, VR is the coefficient of
variation (CV) of the resistance, μR is the mean value of the resistance, and Rn is the nominal
resistance.
In this work, the linearization coefficient is taken as α = 0.70 (ASCE 2010) and the desired
reliability index is taken as β = 2.6, a value appropriate for member strength (AISC 2016). When
examining column behavior, the resistance is the axial compressive strength and when
examining short beam-columns it is the radial distance along a line from the origin to a point on
the interaction curve. In both cases, the mean value and CV of the resistance are determined, in
part, through Monte Carlo simulations.
Variations in resistance come from a variety of sources. The sources can generally be
categorized as relating to either material, fabrication, or professional factors (Ravindra and
Galambos 1978). Material variations arise because of the variability in the mechanical
properties of the steel and concrete. Fabrication variations arise because of the variability in
geometric dimensions. Professional variations arise because of the uncertainty in the underlying
theory and design provisions. In this work, material and fabrication variations will be accounted
through Monte Carlo simulations. Each parameter with significant variability will be identified
and assigned a probability distribution. Then for any selected nominal case, a number of
simulations are performed by sampling each probability distribution and using the resulting
parameters to compute the strength. This process is represented by the generic function in
Equation (9) [i.e., f(…)]. The resistance is assumed to be the product of the computed strength
and a random variable representing the uncertainty within the selected model, Xmodel, which
accounts for the professional variations. Following this assumption, the mean value and CV of
the resistance (μR and VR) can be computed per Equations (10) and (11), where μmodel and Vmodel
are the mean value and CV of Xmodel and μf(…) and Vf(…) are the mean value and CV of the
strengths computed from the Monte Carlo simulations.

R  X model f X Fy , X Es , X fc ,  (9)
 R  model  f  (10)
2
VR  Vmodel  V f2 (11)
In this work, the strength of each simulated case is computed using current code provisions
(AISC 2016) (e.g., Equations (1) through (6) for axial compressive strength). The statistical
properties of random variables were identified from the literature and listed in Table 1. For the
steel yield strength, steel modulus of elasticity, and steel tube thickness, the statistical
properties are given for random variables normalized by the nominal value. For the concrete
compressive strength and concrete modulus of elasticity, the statistical properties are not
normalized and are only applicable for concrete with a nominal compressive strength of 55.16
MPa. These values were obtained following the recommendations of Mirza et al. (1979) which
account for variations between in-situ strength and control cylinder strength as well as speed of
loading. Outside dimensions of the steel tube and the length of the columns were taken as
deterministic.
Table 1 – Statistics of random variables
Parameter Mean CV
Steel yield strength, Fy/Fyn (Bartlett et al. 2003) 1.10 0.06
Steel modulus of elasticity, Es/Esn (Bartlett et al. 2003) 0.993 0.034
Steel tube thickness, t/tn [approximates statistical properties
1.00 0.05
recommended for As (e.g., Lundberg and Galambos 1996)]
Concrete compressive strength, f'c (Mirza et al. 1979) 40.71 MPa 0.136
Concrete modulus of elasticity, Ec (Mirza et al. 1979) 27740 MPa 0.105
EXPERIMENTAL DATABASE
A database of experimental results was developed and analyzed to define the random variable
Xmodel. The database is based on similar, previously developed, databases (Goode 2008; Kim
2005) but is limited to columns specimens subjected to concentric axial compression and short
proportionally loaded beam-column specimens with a length to depth ratio, Lc/d ≤ 5. The
database was further specialized by removing CFT specimens not classified as compact and
specimens with outside dimensions less than 76.2 mm.
A total of 410 CCFT, 150 RCFT, and 88 SRC column specimens were included in the database.
For each of these specimens, a test-to-predicted ratio was obtained by taking the ratio of the
experimentally determined strength to the strength computed per Equation (1). Plotting the test-
to-predicted ratio versus length, as in Figure 3 for CCFT, reveals key regions of behavior. For
short columns, there is high variation and high test-to-predicted ratios this is likely due to the
high levels of concrete confinement which can be present in CFT columns but is only partially
accounted for in the design equations. For intermediate length columns, the variation is less and
the test-to-predicted ratios center at about unity. For long columns, the test-to-predicted ratios
begin to rise indicating there again may be bias in the design equations that result in
underprediction of strength. Plotting the test-to-predicted ratio versus other key parameters,
such as steel ratio (Figure 3), does not reveal such definitive trends. Statistical data for the test-
to-predicted ratio for various subsets of the database are presented in Table 2.
Figure 3 – Database results for CCFT column specimens
Table 2 – Test-to-predicted ratio statistics for column specimens
Member Number of Standard
Case Mean CV
Type Specimens Deviation
All 410 1.384 0.348 0.251
Lc/d ≤ 6 225 1.399 0.266 0.190
CCFT
6 < Lc/d ≤ 20 131 1.180 0.170 0.144
20 < Lc/d 54 1.815 0.521 0.287
All 150 1.165 0.219 0.188
Lc/d ≤ 6 104 1.140 0.193 0.169
RCFT
6 < Lc/d ≤ 20 25 1.080 0.088 0.081
20 < Lc/d 21 1.393 0.300 0.215
All 88 1.484 0.849 0.572
SRC Lc/d ≤ 16 60 1.158 0.146 0.126
16 < Lc/d 28 2.184 1.236 0.566
A total of 252 CCFT, 117 RCFT, and 17 SRC short beam-column specimens were included in
the database. A majority of these specimens were subjected to axial load only. Again, for each
of these specimens, a test-to-predicted ratio was obtained. An interaction diagram was
constructed for each specimen using measured material and geometric properties. The test-to-
predicted ratio was taken as the ratio between the distance along a radial line from the origin to
the experimentally determined strength to the distance along the same line from the origin to the
intersection with the interaction diagram. These test-to-predicted ratio results are plotted with
respect to eccentricity and steel ratio in Figure 4 for CCFT beam-columns. The mean value and
variation is significantly higher for the specimens with zero eccentricity and again, no definitive
trends are identified with respect to the steel ratio. Statistical data for the test-to-predicted ratio
for various subsets of the database are presented in Table 3.
A more detailed analysis of these results is warranted to better understand the variations in the
test-to-predicted ratio. For the analyses in the following sections, the statistical data for Xmodel is
taken as μmodel = 1.00 and Vmodel = 0.08 for both RCFT columns and short beam-columns.
Figure 4 – Database results for CCFT short (Lc/d ≤ 5) beam-column specimens
Table 3 – Test-to-predicted ratio statistics for short (Lc/d ≤ 5) beam-column specimens
Member Number of Standard
Case Mean CV
Type Specimens Deviation
All 252 1.328 0.296 0.227
CCFT e=0 206 1.396 0.272 0.195
e>0 46 1.026 0.191 0.186
All 117 1.077 0.138 0.128
RCFT e=0 91 1.094 0.148 0.135
e>0 26 1.019 0.076 0.074
All 17 1.279 0.181 0.142
SRC e=0 9 1.246 0.157 0.126
e>0 8 1.315 0.210 0.159
COLUMNS
The results of a Monte Carlo simulation of a RCFT column are presented in Figure 5. These
results are for the cross section described previously for the interaction diagrams of Figure 2.
Monte Carlo simulations were run, obtaining 500 instances of the cross section following the
statistical data given in Table 1. For each cross section, the compressive strength was
computed using Equation (1) for column lengths ranging from 0 to 50H. At each length, the
minimum, 5th percentile, median, 95th percentile, and maximum strengths were identified and
plotted along with the strength using nominal properties. Histograms of the strength are shown
for Lc/H = 0, 20 and, 40. Note that in this figure, the strength is normalized by the compressive
strength without length effects, Pno, calculated with nominal properties. The variation in the
material and geometric properties is shown in Figure 5. The variation due to uncertainty in the
model (i.e., the professional factor) is not included in Figure 5.
In Figure 5, the nominal strength coincides most closely with the 95th percentile. The primary
reason for this is the mean concrete compressive strength is about 74% of the nominal strength
(Table 1). The value for the mean concrete compressive strength was obtained following the
recommendations of Mirza et al. (1979) which account for variations between in-situ strength
and control cylinder strength as well as speed of loading. Other, more recent, references may
provide different values.
Figure 5 – Column Monte Carlo simulation results

The resistance factor is obtained using Equation (8) and combining the Monte Carlo simulation
results (Figure 5) with the statistical data of the model in accordance with Equations (10) and
(11). The computed resistance factor as a function of length is shown in Figure 6 for the RCFT
cross section as well as variations of the cross section with different steel ratios. The smallest
selected steel ratio is ρs = 0.07, which results in a tube slenderness near the limit for compact
shapes. The largest selected steel ratio is ρs = 0.50, which is an extreme value, but could be
obtained using built up steel shapes. The selected steel ratios are listed in Table 4 including the
ratio of bare steel to composite design strength. Note that for the highest steel ratios, the design
strength of the bare steel exceeds that of the composite member. As seen in Figure 6, there is a
definite trend of the increasing steel ratio resulting in increased computed resistance factor, in
some cases approaching the resistance factor for steel. Nonetheless, the current value (ϕ =
0.75) remains a reasonable average value, especially given that the minimum permitted steel
ratio is ρs = 0.01.
Figure 6 – Column resistance factor results
Table 4 – Steel ratio case data
Tube ϕPn,steel/ϕPn
Steel Ratio, ρsn H/tn
Thickness, tn Lc/H = 0 Lc/H = 20 Lc/H = 40
0.07 5.49 mm 55.5 0.426 0.476 0.658
0.10 7.94 mm 38.4 0.536 0.581 0.737
0.15 12.17 mm 25.1 0.671 0.706 0.821
0.20 16.57 mm 18.4 0.767 0.801 0.912
0.25 21.15 mm 14.4 0.838 0.871 0.978
0.30 25.91 mm 11.8 0.892 0.923 1.029
0.40 36.01 mm 8.5 0.965 0.997 1.097
0.50 46.84 mm 6.5 1.010 1.044 1.137
SHORT BEAM-COLUMNS
The results of a Monte Carlo simulation of a short RCFT beam-column are presented in Figure
7. These results are also for the cross section described previously for the interaction diagrams
of Figure 2. Monte Carlo simulations were run, obtaining 500 instances of the cross section
following the statistical data given in Table 1. For each cross section, the strength interaction
diagram for combined axial compression and bending moment was computed using the PSD
method for enough points to effectively approximate the continuous curve. At various
eccentricities (i.e., angles from the origin), the minimum, 5th percentile, median, 95th percentile,
and maximum strengths were identified and plotted along with the strength using nominal
properties. Histograms of the strength are shown for e/H = 0, 0.2, and infinity (pure bending).
Again, the variability shown in Figure 7 is representative of the variation in the material and
geometric properties only, variation due to uncertainty in the model (i.e., the professional factor)
is additional.
Figure 7 – Short beam-column Monte Carlo simulation results
The resistance factor is obtained using Equation (8) and combining the Monte Carlo simulation
results (Figure 7) with the statistical data of the model in accordance with Equations (10) and
(11). The computed resistance factor, as a function of eccentricity, is shown in Figure 8. Also
shown in Figure 8 are the current resistance factor and one defined by Equation (7). To obtain
the current resistance factor, the nominal interaction diagram was computed and a factored
interaction diagram was obtained following the recommendations in the commentary to the
AISC Specification (AISC 2016), these diagrams are shown in Figure 2a. The resistance factor
as a function of eccentricity was computed as the ratio of the radial distance from the origin to
the factored curve to the radial distance from the origin to the nominal curve. Note the range of
eccentricities where the current resistance factor exceeds 0.9, these are within the “bulge”
region of the interaction diagram where potential unconservative results were noted previously.
To obtain the resistance factor per Equation (7), the location of the neutral axis was recorded
during the computation of each point on the nominal interaction diagram. With Equation (7), a
monotonic transition between the resistance factor for compression (ϕ = 0.75) and the
resistance factor for flexure (ϕ = 0.9) is obtained. This more closely matches the calculated
resistance factor, further indicating it may provide a better solution than the current
recommendations.
Figure 8 – Short beam-column resistance factor results
CONCLUSIONS
In this paper, the reliability of steel-concrete composite columns and their provisions within the
AISC Specification (AISC 2016) was examined. Variations due to uncertainty in the material and
geometric properties were accounted for through Monte Carlo simulations. Variations due to
uncertainty in the prediction of strength was approximated through evaluation of an
experimental database. A RCFT cross section was selected for detailed examination and the
resistance factor was computed for axial compressive strength for a variety of steel ratios and
for short beam-columns. The following conclusions were noted. There remains some
inconsistency in the provisions for the resistance factor for composite columns, notably for
highly steel dominant columns and for the application of the resistance factor to strength under
combined axial compression and bending moment. A sizable portion of the uncertainty
originates in the modeling capability (i.e., the “professional factor”), and further research would
be useful to better understand and reduce the variation. The computed resistance factor is
dependent on cross-sectional properties, including steel ratio, as well as, column length, and
eccentricity. The potential for a more accurate and more consistent method of applying
resistance factors to interaction diagrams based on provisions within the ACI Code (ACI 2014)
was identified, however, further research is required too fully calibrate such an approach.
REFERENCES
ACI. (2014). Building Code Requirements for Structural Concrete and Commentary. American
Concrete Institute, Farmington Hills, MI.
AISC. (1986). Load and Resistance Factor Design Specification for Structural Steel Buildings.
American Institute of Steel Construction, Chicago, Illinois.
AISC. (2005). Specification for Structural Steel Buildings. American Institute of Steel
Construction, Chicago, Illinois.
AISC. (2013). Seismic Design Manual, 2nd Ed. American Institute of Steel Construction,
Chicago, Illinois.
ASCE. (2010). Minimum Design Loads for Buildings and Other Structures. American Society of
Civil Engineers, Reston, Virginia.
Aslani, F., Lloyd, R., Uy, B., Kang, W.-H., and Hicks, S. (2016). “Statistical calibration of safety
factors for flexural stiffness of composite columns.” Steel and Composite Structures,
20(1), 127–145.
Bartlett, F. M., Dexter, R. J., Graeser, M. D., Jelinek, J. J., Schmidt, B. J., and Galambos, T. V.
(2003). “Updating Standard Shape Material Properties Database for Design and
Reliability.” AISC Engineering Journal, (1), 2–14.
Beck, A. T., de Oliveira, W. L. A., De Nardim, S., and ElDebs, A. L. H. C. (2009). “Reliability-
based evaluation of design code provisions for circular concrete-filled steel columns.”
Engineering Structures, 31(10), 2299–2308.
Denavit, M. D., Hajjar, J. F., Perea, T., and Leon, R. T. (2016). “Stability Analysis and Design of
Composite Structures.” Journal of Structural Engineering, ASCE, 142(3), 04015157.
Goode, C. D. (2008). “Composite Columns - 1819 Tests on Concrete-Filled Steel Tube Columns
Compared with Eurocode 4.” The Structural Engineer, 86(16), 33–38.
Kim, D. K. (2005). “A Database for Composite Columns.” M.S. Thesis, School of Civil and
Environmental Engineering, Georgia Institute of Technology, Atlanta, Georgia.
Kogut, G. F., and Chou, K. C. (2004). “Partial resistance factor design on steel–concrete beam-
columns.” Engineering Structures, 26(7), 857–866.
Kvedaras, A. K., and Kudzys, A. (2006). “The structural safety of hollow concrete-filled circular
steel members.” Journal of Constructional Steel Research, 62(11), 1116–1122.
Lundberg, J. E., and Galambos, T. V. (1996). “Load and Resistance Factor Design of
Composite Columns.” Structural Safety, 18(2–3), 169–177.
Mirza, S. A., Hatzinikolas, M., and MacGregor, J. G. (1979). “Statistical Descriptions of Strength
of Concrete.” Journal of the Structural Division, ASCE, 105(6), 1021–1037.
Mirza, S. A., and Skrabek, B. W. (1991). “Reliability of Short Composite Beam-Column Strength
Interaction.” Journal of Structural Engineering, ASCE, 117(8), 2320–2339.
Mirza, S. A., and Skrabek, B. W. (1992). “Statistical Analysis of Slender Composite Beam-
Column Strength.” Journal of Structural Engineering, ASCE, 118(5), 1312–1332.
Ravindra, M. K., and Galambos, T. V. (1978). “Load and Resistance Factor Design for Steel.”
Journal of the Structural Division, ASCE, 104(9), 1337–1353.
CALIBRATION OF THE ELASTIC FLEXURAL RIGIDITY FROM AMBIENT VIBRATION
MEASUREMENTS FOR A BUILDING WITH ENCASED COMPOSITE COLUMNS
Tiziano Perea
Universidad Autónoma Metropolitana, Azcapotzalco
Mexico City, Mexico.
tperea@azc.uam.mx
Miguel A. Garcia
Kiewit Energy de Mexico
miguel.garcia@kiewit.com
Manuel E. Ruiz-Sandoval
Universidad Autónoma Metropolitana, Azcapotzalco
mrh@azc.uam.mx
Roberto T. Leon
Virginia Polytechnic Institute and State University
Blacksburg, Virginia, USA.
rleon@vt.edu
Mark D. Denavit
University of Tennessee, Knoxville
Knoxville, Tennessee, USA.
mdenavit@utk.edu
Jerome F. Hajjar
Northeastern University
Boston, Massachusetts, USA.
JF.Hajjar@northeastern.edu
ABSTRACT
Calibrated results of the elastic flexural rigidity (EIelastic) for the determination of the fundamental
period of a building with encased composite columns, also known as Steel Reinforced Concrete
(SRC) columns, are discussed in this paper. This calibration is carried out from the natural
frequencies measured from an ambient vibration test and from modal analyses using available
semi-empirical equations proposed by researchers and adopted in some codes. The twenty-
five-story building evaluated in this study is an ordinary composite moment frame system (C-
OMF) in the longitudinal direction, with two ordinary X braced frames (C-OBF) in the transversal
direction, and with SRC composite columns integrated by an H built-up steel section encased
with a reinforced concrete section. Vibration periods obtained from the ambient vibration test
show that the use of the effective flexural stiffness for encased composite columns in the
previous AISC Specification were too conservative for the period prediction in modal analysis.
The lowest errors for the evaluated building are those given in Eurocode, ACI, and the one
proposed by Denavit et al. that was recently adopted in the latest AISC Specification.
INTRODUCTION
Structural systems based on steel-concrete composite construction have developed a growing
popularity in past few decades because of their many structural and constructional advantages.
However, the increasing construction of composite structural systems in high seismic risk zones
around the world has shown that existing specifications are not capable of providing the robust
parameters needed to check all critical strength and service limit states. Particularly, the lack of
clear guidelines on the flexural stiffness that must be used in the elastic analysis of structural
systems with composite columns is a significant unresolved issue, making difficult the
determination of the dynamic properties and thus the seismic forces for these systems. The
common practice of only using either the concrete or the steel stiffness generally tends to be
significantly conservative. Other efforts have considered a weighted superposition of each
material stiffness, but some of them have not been fully validated for all the limit states, such as
those regarding the strength, stability, and serviceability, as well as for the determination of the
vibration frequencies or periods, and the lateral interstory drifts. In this paper, unique ambient
vibration tests on a large structure incorporating steel concrete reinforced (SRC or encased)
columns are used to attempt to validate several proposed approaches to determining column
stiffness elastic properties.
AMBIENT VIBRATION TESTS

Description of the building
The evaluated structure in this study is shown in Figure 1. It is a 25-story building with a height
of 98.2 m. intended for commercial and office use, with four basement levels supported on
concrete piles. As shown in Figures 1(a) and 1(c), the structural system is a composite ordinary
moment frames (C-OMF) in the longitudinal direction, and two ordinary composite X braced
frames (C-OBF) in the transversal direction. A typical plan view of the building is shown in
Figure 2. All the columns have encased or SRC composite cross-section as illustrated in figure
1(b), consisting of a steel built-up I-shape encased by a square reinforced concrete section with
105 cm. sides. The steel section is built-up by plates (ASTM A-572 Gr. 50) of 300x19 mm. for
the flanges, and 400x13 mm. for the web. The concrete has a compression strength of fc´ =24.5
MPa (3.6 ksi), reinforced with 12#38 bars (12#12) and 16#32 bars (#10), and confined by
double #13 (#4) ties spaced at 200 mm as shown in Figure 1(b). The floor system consists of a
precast slab supported by the principal steel beams in non-composite action.
(a) North side (b) SRC cross-section (c) West side

Figure 1 – Composite building studied
Ambient vibration measurements
Shortly after most of the structure had been completed, an ambient vibration test was conducted
on the structure. In these types of measurements, the quality of the data depends on the
correct horizontal and vertical distribution of the sensors, as well as the sensitivity and precision
of the instruments.
The equipment used for recording the ambient vibration consisted of a data acquisition (DAQ)
system (Siglab 20-42), with a 20-bit analog-digital converter (A/D), and an anti-alias filter of 90
dB. Also, four seismic sensors (PCB, model 393B31) were used, as well as a laptop to
command the DAQ system and store the vibration data.
Four sensors were used to measure the ambient vibrations for 15 minutes. Figure 4 illustrates
the sensor layout placed on the building roof, with two sensors located at the centroid oriented
in both the longitudinal and transverse directions, and two additional sensors placed at the roof
ends with the aim of evaluating twisting.
It should be noted that, when this test was performed, the building was under construction,
presenting a progress of 100% for the structural members (i.e. beams, columns, braces), but
with no presence of contents or non-structural elements (e.g. facade, partitions, collateral
loads). Thus, the only mass source during the testing was the self-weight of the structural
members and the sources of lateral stiffness could be easily identified.
With the aim to determine dynamic properties of the soil, vibrations were also measured in the
ground using two sensors, one oriented in the north direction and the other in the vertical
direction.
Figure 2 – Plan view and frontal view of the building showing the location of sensors
Data processing
The process to determine the dynamic properties is divided into two stages. The first consists in
recording the vibrations captured by sensors in the central unit. The second is data processing,
including a baseline correction, signal filtering, and finally transforming the vibration data from
time domain to frequency domain. The technique used to make the switch domain is the fast
Fourier transform (Bendat and Piersol, 1989). The Fourier transform can be defined by the
following equation:
T
X!! ( f ) = ò !!
x(t )e -i×2p f ×tdt
0 (1)
which can be written as:
T T
X!! ( f ) = ò !!
x(t ) cos(2p f ×t )dt - i ò !!
x(t ) sen(2p f ×t )dt
0 0 (2)
Fourier amplitudes X!! ( f ) are made up of a real part X!! R ( f ) and an imaginary X!! I ( f ) .
X!! ( f ) = X!! R ( f ) 2 + X!! I ( f ) 2
(3)
The power spectral density of the signals is calculated per the following equation:
S xx ( f ) = X!! ( f ) ×X!! ( f ) (4)
The natural frequencies of the system will produce peaks in the power spectral density,
indicating that they have greater participation in the frequency content of the recorded signals.
These peaks can easily identify the natural frequencies for the principal modes, as shown in
Figures 3(a) and 3(b). The reciprocals of these frequencies correspond to the vibration periods.
(a) Longitudinal or X direction (b) Transversal or Y direction

Figure 3 – Power spectra obtained from the vibration recordings
Table 1 shows the estimated periods for the first three vibration modes corresponding to the
longitudinal, transversal, and torsional directions. Environmental vibrations were also measured
in the field adjacent to the building to determine the site period. After applying the Nakamura´s
technique (1989), the ground period resulted with a value of 0.79 seconds.
Table 1 – Natural vibration periods measured of the building
Period (s)
Mode
Longitudinal (X) Transversal (Y) Torsional
1 2.030 1.890 1.509
2 0.555 0.617 0.443
3 0.302 0.359 0.244
EQUATIONS FOR THE FLEXURAL STIFFNESS OF ENCASED COMPOSITE COLUMNS

One of the main issues related to the design of composite systems regards the determination of
the appropriate stiffness for composite members to be used in the different analysis and
calculations required, including periods, stability and displacements.
Some current design codes include equations for the determination of the effective flexural
stiffness for composite members; however, the values determined with these equations vary
significantly from each other, creating uncertainty about which is the closest to the real
performance value.
Table 2 shows some expressions to calculate the effective flexural stiffness (EIeff) for encased or
SRC composite columns. This table includes both equations contained in design codes and
those proposed by researches. The one proposed by Denavit et al. (2014, 2016) was recently
adopted in the latest AISC Specification (AISC 2016). Eurocode (CEN 2004) has two equations
of the effective flexural stiffness, one for the slenderness calculation in buckling analysis, and a
second one (which is used in this work) for internal forces calculation in second-order analysis.
Table 2 – Available equations of the effective flexural stiffness for encased composite columns
Code / Author Equation X direction Y direction
EI eff = 0.537Ec I g EI eff = 0.514Ec I g
Tikka and Mirza a c Ec ( I g + I s )
EI eff = + 0.8 ( I s + I sr ) EI eff = 14.404Es I s EI eff = 72.304Es I s
(2006) (1 - b d ) EI eff = 0.563Ec I c EI eff = 0.539Ec I c
EI eff = E s I s + E s I sr + C1Ec I c EI eff = 0.72Ec I g EI eff = 0.69Ec I g
Denavit et al. (2014)
æ As + Asr ö EI eff = 19.335Es I s EI eff = 97.224Es I s
C1 = 0.25 + 3 ç ÷ £ 0.70 EI eff = 0.756Ec I c EI eff = 0.724Ec I c
AISC (2016)
è Ag ø
EI eff = E s I s + 0.5E s I sr + C1Ec I c EI eff = 0.328Ec I g EI eff = 0.297Ec I g
AISC (2010) æ A ö EI eff = 8.792Es I s EI eff = 41.893Ec I g
C1 = 0.10 + 2 ç ÷ £ 0.30
s
EI eff = 0.344Ec I c EI eff = 0.312Ec I c
èA +A ø
s c

ACI (2014) EI eff = 0.7 E c I g EI eff = 18.786Es I s EI eff = 98.592Es I s
EI eff = 0.734Ec I c EI eff = 0.734Ec I c
Eurocode (2004) EI eff = 0.9 ( E s I s + Es I sr + 0.5Ec I c ) EI eff = 20.499Es I s EI eff = 107.577Es I s
EI eff = 0.801Ec I c EI eff = 0.764Ec I g
EI gross = 1.325Ec I g EI gross = 1.295Ec I g
Gross flexural EI gross = Es I s + Es I sr + Ec I c EI gross = 35.57 Es I s EI gross = 182.426Es I s
stiffness EI gross = 1.39Ec Ic EI gross = 1.358Ec I g
Table 2 also includes the calculated values for each effective flexural stiffness equation, and for
each direction, using the material and geometric information of the encased composite column
(Figure 1(b)) for the building under study (Figures 1 and 2). These values are normalized with
the concrete Young´s modulus and the gross moment of Inertia (EcIg), the flexural stiffness of
the steel stiffness (EsIs), and the flexural stiffness of the concrete section (EcIc). The gross
flexural stiffness (EcIg ) is defined as the superposition of the steel, reinforcement and concrete
flexural stiffness using a value of Es = Esr = 210 GPa (29,000 ksi) for the steel and Ec = 21.7
GPa (3,150 ksi) for the concrete as determined from the local concrete code.
MODAL ANALYSIS
This section shows the results of the modal analysis for the composite building previously
described, in which the different values of effective stiffness of composite columns that were
presented in Table 2 are used. Figure 6 shows the first three vibration modes of the building.
The structural model of the building includes the concrete walls already built when the vibration
test was performed.
Note that the prediction of the masses in the structure correspond to the best estimate at the
time when the vibrations were measured. Since the building was under construction (and thus
an absence of non-structural members or contents), the main source of masses in the building
correspond to the self-mass of the structural members and the construction loads, which
allowed to reduce the expected dispersion at frequencies of vibration due to uncertainties in the
masses of the system. Therefore, the rigidity of the structure, which mainly depends on the
elastic flexural stiffness (EIelastic) of the composite columns, is the target variable to calibrate for
the accurate determination of the natural vibration periods in composite moment frames.
Figure 4 – Analytical models of the studied building
Table 3 shows the results of the modal analysis for the evaluated building. Although higher
modes and periods were calculated from modal analysis, this table shows only the fundamental
period for each direction. For the beams, the flexural stiffness of the steel was used since they
are in non-composite action with the concrete precast slab. For the encased composite
columns, the effective flexural stiffness enlisted in table 3 were used in the modal analysis.
Table 3 – Fundamental vibration periods obtained from modal analysis

Fundamental vibration period (s)
Code / Author Equation
Longitudinal Transversal Torsional
Tikka and Mirza a c c( g
E I + I s)
EI eff = + 0.8 ( I s + I sr ) 2.282 2.237 1.740
(2006) (1 - b d )
EI eff = Es I s + Es I sr + C1 Ec I c
æ As + Asr ö 2.211 2.075 1.643
C1 = 0.25 + 3 ç ÷ £ 0.70
AISC (2016)
è Ag ø
EI eff = E s I s + 0.5Es I sr + C1 Ec I c
AISC (2010) æ A ö 2.637 2.446 1.966
C1 = 0.10 + 2 ç ÷ £ 0.30
s
èA +A ø
s c
ACI (2014) EI eff = 0.7 E c I g 2.218 2.089 1.591

Eurocode (2004) EI eff = 0.9 ( E s I s + Es I sr + 0.5Ec I c ) 2.198 2.045 1.613
Gross flexural EI gross = Es I s + Es I sr + Ec I c 2.087 1.799 1.431
stiffness
EXPECTED DIFFERENCES WITH FORCED AND SEISMIC VIBRATIONS

Studies reveal that, for different levels of vibrations, such as those generated by an earthquake,
the natural frequencies or periods may change depending on the excitation amplitudes that
affects the structure (Trifunac 1972, Lamarche 2002, Murià 2007, among others). The increase
in the vibration intensity causes a decrease in the fundamental frequency (or an increase in the
period) of the structure relative to the corresponding one that would be obtained with lower
intensities such as those generated by ambient vibrations. This is due to changes in the
structure rigidity and damping, particularly when the system is at the non-linear range.
Therefore, it is expected that the periods obtained from ambient vibrations will be relatively
lower compared to those measured from forced vibrations and seismic vibrations. To illustrate
the possible dispersion, Table 4 shows some ratios reported in the literature of periods obtained
from forced vibrations over periods obtained from ambient vibrations. Although this table does
not report all the available studies on the subject, this reflects that the expected discrepancy
between ambient and forced vibrations is about 1% to 5%. Due to possible changes in the
structure rigidity during strong ground motions, higher values and more dispersion are expected
between periods obtained from seismic vibration and those obtained from ambient vibration
measurements (Rodríguez 2001, Murià 2000, 2001, 2007).
Table 4 – Ratios of forced vibration periods to ambient vibration periods

Ratios
Reference Structural system
Longitudinal Transversal Torsional
Trifunac (1972a) 1.038 1.047 1.035 22-story steel building
Trifunac (1972b) 1.017 1.033 1.011 9-story concrete building
Lamarche (2002) 1.019 1.017 1.017 2-story concrete frame system
Since the vibration periods in a modal analysis are determined from the elastic stiffness of the
system, it is expected that the periods obtained with ambient vibrations are the best estimators
of analytical periods under the actual conditions of mass and stiffness (as assumed in this
study). However, slight errors between analytical and measured periods are possible due to
some uncertainties (e.g. real elastic modulus of the component materials, potential soil-structure
interaction, differences between estimated and actual mass, among others).
Also, note that the reported periods in this paper correspond to the conditions when the building
was measured (which was under construction as described earlier). However, the system
periods will change when the construction is completed and the building is under full operation
(i.e. when the building is under the ultimate conditions of stiffness and mass).
COMPARISON OF RESULTS
This section presents and discusses the obtained periods (for the first vibration mode at each
direction) extracted from the ambient vibration testing (reported in Table 1), and those
calculated from the modal analysis (reported in Table 3) assuming the different effective
stiffness equations of the composite columns (reported in Table 2).
The error values between the periods calculated from the modal analysis and those measured
from ambient vibration testing are listed in Table 5. These errors are also plotted in Figure 7.
Table 5 – Errors between ambient vibrations calculated and measured periods

Code / Author Equation Longitudinal Transversal Torsional
Tikka and Mirza a c Ec ( I g + I s )

EI eff = + 0.8 ( I s + I sr ) 12.4% 18.4% 15.3%
(2006) (1 - b d )
Denavit et al. (2014) EI eff = Es I s + Es I sr + C1 Ec Ic
æ As + Asr ö 8.9% 9.8% 8.9%

C1 = 0.25 + 3 ç ÷ £ 0.70
AISC (2016) è Ag ø
EI eff = E s I s + 0.5Es I sr + C1 Ec I c
AISC (2010) æ A ö 29.9% 29.4% 30.3%

C1 = 0.10 + 2 ç ÷ £ 0.30
s
èA +A ø
s c
ACI (2014) EI eff = 0.7 E c I g 9.3% 10.5% 5.4%
Eurocode (2004) EI eff = 0.9 ( E s I s + Es I sr + 0.5Ec I c ) 8.3% 8.2% 6.9%
Gross flexural EI gross = Es I s + Es I sr + Ec I c 2.8% -4.8% -5.2%

stiffness
Tikka and Mirza (2006)

AISC (2016)
AISC (2010)
ACI (2014)
Logitudinal
Eurocode (2004)
Transversal
Gross
Torsional
-10% -5% 0% 5% 10% 15% 20% 25% 30% 35%

Figure 7 – Errors between periods calculated from the modal analysis
and those measured from ambient vibration testing
As seen by comparing the results and the corresponding errors, the gross flexural stiffness
provides the smaller absolute errors. However, for both the transversal and torsional direction,
these are negative errors, which means that the assumed stiffness is greater than the one
developed by the system (presumed as the one obtained from the ambient vibration test), and
thus resulting in an inconsistency. Consequently, the gross flexural stiffness is likely too high for
general use within an elastic analysis.
Somewhat larger errors (between 5% to 10%) were obtained when the effective flexural
stiffness for the encased composite columns were determined by the equation in Eurocode
(2004), ACI (2014), and Denavit et al. (2014). Note that, for the encased composite columns in
this building, the effective flexural stiffness is near 0.7EcIg with these three equations as shown
in Table 2. For different encased cross-sections, the ACI equation will retain the value of 0.7EcIg,
while the equations by Denavit et al. (2014) and Eurocode (2004) may vary due to the different
participation coefficients within these two equations.
The highest errors were obtained from the equations proposed by Tikka and Mirza (2008), with
an average error about 15%, and from the equation of the AISC Specification (AISC 2010), with
an average error around 30%. It is worth mention that these two equations were originally
calibrated for the slenderness and the axial compressive strength of the available experimental
data of encased composite columns and thus were intended for stability-type calculations.
Thus, these results suggest that the equations proposed by Eurocode (2004), ACI (2014) and
Denavit et al. (2014) provide an adequate estimate of the expected fundamental periods, and so
an adequate estimation of the seismic design forces.
CONCLUSIONS
A calibration study of the effective flexural stiffness for encased SRC composite columns in
moment frames is performed utilizing unique data from an ambient vibration test and modal
analysis. The results of the ambient vibration test demonstrate that this approach is relatively
simple and very efficient to determine some dynamic structural properties such as vibration
frequencies or periods.
Vibration frequencies and periods obtained from the ambient vibration test show that previous
AISC Specification (AISC 2010) was too conservative for use in modal analysis. On the other
hand, the gross flexural stiffness of the composite cross-section overestimate the actual
stiffness, and thus underestimates the periods. The effective stiffness equation that provided
results with the lowest errors for the evaluated building are those in Eurocode (2004), ACI
(2014), and the one proposed by Denavit et al. (2014) that was recently adopted in the latest
AISC Specification (AISC 2016).
The calibration of the elastic flexural stiffness, as presented in this article along with other
methodologies, will allow to understand the evolution and degradation of the effective flexural
stiffness and determine the different responses of structural systems with encased or SRC
composite columns.
NOTATION
EIelastic Elastic flexural stiffness of the composite section.
EIeff Effective flexural stiffness of the composite section.
EIgross Gross flexural stiffness of the composite section.
Es Elasticity modulus of the structural steel shape.
Is Moment of inertia of the structural steel shape.
Esr Elasticity modulus of the reinforcement steel.
Isr Moment of inertia of the reinforcement steel.
Ec Elasticity modulus of the concrete.
Ic Moment of Inertia of the concrete section.
Ig Moment of Inertia of the gross composite section
(superposition of the steel, reinforcement and concrete sections).
C1 Coefficient of the concrete stiffness in the effective stiffness of the composite column.
βd Relationship between the sustained axial force and the maximum required axial force.
αc Ratio of the axial load eccentricity and the element length.
REFERENCES
ACI. (2014). Building Code Requirements for Structural Concrete and Commentary. American
Concrete Institute, Farmington Hills, MI.
Bendat J.S., Piersol A.G. (1986). “Random data, analysis and measurement procedures”. John
Wiley and Sons.
CEN. (2004). Eurocode 4: Design of composite steel and concrete structures – Part 1-1:
General rules and rules for buildings. European Committee for Standardization, Brussels,
Belgium.
Denavit M.D., Hajjar J.F., Leon R.T., Perea. T. (2014). “Analysis and Design of Steel-Concrete
Composite Frame Systems”. Structures Congress 2014, pp. 2605–2616.
Denavit, M. D., Hajjar, J. F., Perea, T., and Leon, R. T. (2016). “Stability Analysis and Design of
Composite Structures.” Journal of Structural Engineering, ASCE, 142(3), 4015157.
Lamarche C.P., Mousseau S., Paultre P., Proulx J. (2002). “A comparison of ambient and
forced-vibration testing of a full scale concrete structure”. Université de Sherbrooke,
Sherbrooke, QC, Canada.
Murià D. (2007). “Experiencia mexicana sobre la respuesta sísmica de edificios
instrumentados”. Instituto de Ingeniería UNAM. México. In Spanish.
Murià D., Fuentes L., González R. (2000). "Incertidumbres en la estimación de las frecuencias
naturales de vibración de edificios". Información Tecnológica, Vol. 11, No. 3, pp. 177-184. Chile.
In Spanish.
Murià D., Rodríguez G. (2001). “Análisis de los registros sísmicos obtenidos de 1993 a 1998 en
el edificio Jalisco”. Instituto de Ingeniería UNAM. Colección: Serie Azul No. 628. México. In
Spanish.
Nakamura Y. (1989). “A method for dynamic characteristics estimation of surface using
microtremors on the ground surface”. QR of RTRI. Vol. 30, No.1, pp. 25-33.
Rodríguez G., Macías M., Murià D., Palacios C. (2001). “Respuesta sísmica de un edificio
instrumentado en un periodo de 10 años”. XIII Congreso Nacional de Ingeniería Sísmica,
Guadalajara, Jalisco, México. In Spanish.
Trifunac M.D. (1972). “Comparisons between ambient and forced vibration experiments”.
Earthquake Engineering and Structural Dynamics, Vol. 1, pp. 133-150. John Wiley & Sons.
USA.
Pushing the Envelope of Composite Column Design Using High-
Strength Materials
Amit H. Varma
Professor, Lyles School of Civil Engineering, Purdue University
West Lafayette, IN, USA
ahvarma@purdue.edu
Zhichao Lai
Postdoctoral Research Associate, Lyles School of Civil Engineering, Purdue University
laizhichao@gmail.com
ABSTRACT
AISC 360-16 prevents the use of high-strength materials (Fy ≥ 525 MPa and f’c ≥70 MPa) for
CFT columns due to lack of adequate research and comprehensive design equations. This
paper addresses this gap using a three-step method. The first step consists of compiling the
experimental database of high-strength rectangular CFT column tests in the literature and
evaluating the possibility of extending the current AISC 360-16 design equations for estimating
the strength of high-strength rectangular CFT short columns. The second step consists of
conducting parametric studies using benchmarked finite element (FEM) models to: (i) address
gaps in the database and (ii) develop effective stress-strain relationships for modeling the high-
strength steel tube and high-strength concrete infill of rectangular CFT members. The third step
consists of proposing new design equations that accurately predict the behavior and strength of
high-strength rectangular CFT short columns.
INTRODUCTION
Concrete-filled steel tube (CFT) members are usually comprised of rectangular or circular steel
tubes filled with concrete. CFT members can have more efficient behavior than reinforced
concrete or structural steel members due to the interaction between the steel tube and concrete
infill. The steel tube provides confinement to the concrete infill, while the concrete infill delays
the local buckling of the steel tube (Lai et al. 2014, Lai and Varma 2015, and Lai et al. 2016). As
an innovative structural component, CFT members have been used widely around the world in
various structures. For example, they have been used as: (i) chords in composite arch bridges
(Huang et al. 2017), (ii) mega columns in high-rise buildings (Lai et al., 2016), (iii) piles in
floodwall structures (Lai and Varma, 2017), and (iv) bridge piers (Chen et al., 2017).
The current AISC Specification (AISC 360-16, 2016) prevents the use of high-strength materials
for CFT columns. According to AISC 360-16 (2016), the steel yield stress (Fy) should not
exceed 525 MPa, and the concrete compressive strength (f’c) should not exceed 70 MPa. CFT
members are classified as high-strength if either Fy or f’c exceeds the specified limits. CFT
members are classified as conventional-strength if both Fy and f’c are less than or equal to the
limits.
Significant experimental studies have been conducted to evaluate the behavior of conventional-
strength CFT members as columns, beams, and beam-columns. For example, experimental
tests on CFT columns have been conducted by researchers such as Furlong (1967), Janss and
Anslijn (1974), Lin (1988), Fujimoto et al. (1995), Schneider (1998), and Han et al. (2001)
among others. Experimental tests on CFT beams have been conducted by researchers such as
Ichinohe et al. (1991), Elchalakani et al. (2001), Wheeler and Bridge (2004), Lennie et al. (2008),
and Jiang et al. (2013) among others. Experimental tests on CFT beam-columns have been
conducted by researchers such as Ichinohe et al. (1991), Prion and Boehme (1994), Nakahara
and Sakino (2000), Uy (2001), and Mursi and Uy (2004) among others.
These studies have been summarized independently by researchers such as Nishiyama et al.
(2002), Kim (2005), Hajjar et al. (2013), Lai et al. (2014), and Lai and Varma (2015) among
others. Additional discussions of these tests and database can be found in Lai (2014). These
experimental tests indicate that the behavior and strength of conventional-strength CFT
members depend on parameters such as the concrete compressive strength (f’c), steel yield
stress (Fy), tube width-to-thickness ratio (b/t) or diameter-to-thickness ratio (D/t), and member
length (L).
Based on the findings from these prior experimental investigations, AISC 360-16 (2016)
provides design provisions for conventional-strength CFT members, which include: (i) steel tube
slenderness limits (i.e., tube width-to-thickness limits) to categorize CFT members into compact,
noncompact, and slender; and (ii) design equations for estimating the compressive, flexural and
beam-column strength of CFT members. The authors have presented the basis (development
and verification) of these design provisions elsewhere in Lai et al. (2014), Lai and Varma (2015),
and Lai et al. (2016).
According to AISC 360-16 (2016), a conventional-strength rectangular CFT column is classified
as compact if the governing tube slenderness ratio (λ) ≤ λp (λp = 2.26 Es Fy ); if λp< λ ≤ λr (λr =
3.0 Es Fy ), the column is classified as noncompact; if λr < λ ≤ λlimit (λlimit = 5.0 Es Fy ), the
column is classified as slender.
AISC 360-16 (2016) also provides the design equations (Eqs. 1-5) for estimating the cross-
sectional strength of conventional-strength rectangular CFT columns (Pno). The cross-sectional
strength (Pno) depends on the classification of the section as compact, noncompact or slender
as per AISC 360-16 (2016), Section I1.4. Eq. (1) is for compact sections; Eqs. (2) and (3) are for
noncompact sections; and Eqs. (4) and (5) are for slender sections.
For compact sections: Pno = Pp = As Fy + 0.85f 'c Ac (1)
Pp − Py
For noncompact sections: Pno = Pnc = Pp − 2
( −  p )2 (2)
( r −  p )
where, Py = As Fy + 0.70f 'c Ac (3)
For slender sections: Pno = Pcr = As Fcr + 0.70f 'c Ac (4)
9Es
where, Fcr = (5)
(b t )
2
Several researchers have experimentally investigated the behavior of high-strength rectangular

CFT columns, including Cederwall et al. (1990), Varma (2000), Uy (2001), Liu et al. (2003),
Mursi and Uy (2004), Sakino et al. (2004), Liu and Gho (2005), Lue et al. (2007), Aslani et al.
(2015), and Xiong et al. (2017) among others. These prior studies have provided valuable
insights into the fundamental behavior of high-strength rectangular CFT columns. However,
there is a lack of comprehensive design equations based on the compilation of these studies.
Consequently, AISC 360-16 (2016) prevents the use of high-strength rectangular CFT columns.
This paper makes a contribution towards addressing this gap, and proposes effective stress-
strain relationships and new design equations for high-strength rectangular CFT members using
a three-step method. The first step consists of compiling an experimental database of high-
strength rectangular CFT column tests and evaluating the possibility of extending the current
AISC 360-16 design equations for estimating the strength of high-strength rectangular CFT
short columns. The second step consists of developing and benchmarking detailed 3D nonlinear
finite element models for predicting the behavior of high-strength CFT columns from the
database. The benchmarked models were then used to perform parametric studies to (i)
address gaps in the database and (ii) develop effective stress-strain relationships for high-
strength steel tube and high-strength concrete infill of rectangular CFT members. The third step
consists of using the effective stress-strain relationships and the enhanced database and to
propose new design equations that more reliably predict the behavior and strength of high-
strength rectangular CFT short columns.
EXPERIMENTAL DATABASE
As discussed in the previous section, several experimental tests have been conducted to
evaluate the behavior and strength of high-strength rectangular CFT columns. This section
compiles an experimental database by reviewing these tests. A total of 124 tests on high-
strength CFT columns were included in the database. The experimental database covers a
broad range of material and geometric parameters. The column length (L) varies from 300 to
3000 mm (length-to-depth ratio varies from 3.0 to 25.0), the width (B) varies from 100.3 to 324.0
mm, the depth (H) varies from 80.0 to 324.0 mm, the flange thickness (tf) and web thickness (tw)
varies from 4.0 to 12.5 mm, the tube slenderness ratio (b/tf and h/tw) varies from 13.0 to 72.0,
the slenderness coefficient (λcoeff) varies from 0.47 to 3.21, the concrete compressive strength
(f’c) varies from 20.0 to 164.1 MPa, and the steel yield stress (Fy) varies from 259.0 to 835.0
MPa. Details of the database have been presented elsewhere by the authors (Lai and Varma,
2017) and are not repeated here for brevity.
Figure 1(a) shows the distribution of test data with respect to the length-to-depth ratio (L/H). As
shown, most tests (111 out of 124) were conducted on short (stub) columns (L/H ≤ 6.0), while
there are only 13 data points for relatively long columns (L/H > 6.0). This paper focuses on stub
(short) columns, i.e., the cross-sectional strength. The corresponding investigations of long
columns are currently being investigated by the authors and will be part of a future paper. Figure
1(b) shows the distribution of the test data with respect to λcoeff. As shown, most tests were
conducted on columns with compact (105 out of 124) and noncompact (18 out of 124) sections,
while there is only one test data for columns with slender sections.
Figure 2 compares the strengths calculated using the AISC design equation (Pno) with those
obtained from the experiments (Pexp) for the 111 short column specimens. The data points
showing these comparisons are identified as “EXP” in the figure. Figure 2 indicates that the
AISC design equations can reasonably estimate the cross-sectional strength of high-strength
rectangular CFT columns. The mean Pexp/Pno ratio is 1.07, and the corresponding coefficient of
variation is 0.09. These comparisons also indicate that the AISC design equations are
conservative for columns with compact sections and slightly unconservative for columns with
noncompact sections. There is only one comparison for columns with slender sections due to
lack of test data. Therefore, no conclusions can be made regarding the conservatism of the
AISC 360-16 equations for designing high-strength rectangular CFT columns with slender
sections, and additional data points are required. This will be addressed later by conducting
finite element analysis.
90 20 Compact Slender
80 18
Noncompact
70 16
Number of tests
Number of tests
60 14
12
50
10
40
8
30 6
20 4
10 2
0 0
0.30
0.60
0.90
1.20
1.50
1.80
2.10
2.40
2.70
3.00
3.30
3.60
3.90
4.20
4.50
4.80
L/H λcoeff
(a) (b)
Fig. 1 - Distributions of test data for high-strength rectangular CFT columns with respect to: (a)
length-to-depth ratio (L/H) and (b) slenderness coefficient (λcoeff)
1.6
1.4
1.2
1.0
Pexp /Pno
0.8
Noncompact
0.6
EXP
0.4 Additional
Compact Slender
FEM
0.2
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
λcoeff = b/t(Fy /Es)0.5
Fig. 2 - Comparisons of the axial compressive strengths calculated using the AISC 360-16
design equations with the corresponding experimental and analysis results
FINITE ELEMENT MODELS

This section develops and benchmarks detailed 3D nonlinear FEM models for high-strength
rectangular CFT short columns. The FEM models were developed using ABAQUS (2016).
Details of the FEM models include: (i) element types, (ii) contact interaction, (iii) steel and
concrete material models, (iv) geometric imperfections, (v) boundary conditions, and (vi)
analysis method. Most of these details are the same as those presented by the authors in a
prior research (Zhichao Lai et al., 2014) for conventional-strength rectangular CFT members.
For example, an idealized bilinear curve was used to specify the steel uniaxial stress-strain
behavior in both compression and tension (see Figure 3a). The only exceptions are the
parameters used to define the concrete material model. This is explained as follows.
This paper used the CDP model to model the material multiaxial behavior of concrete for high-
strength rectangular CFT columns. The default value in ABAQUS was used for eccentricity (ϵ =
0.1). The dilation angle (ψ), concrete biaxial-to-uniaxial compressive strength ratio f’bc/f’c, and
the ratio of compressive to tensile meridians of the yield surface in Π space (Kc) were
determined based on the recommendations by Tao et al. (2013), i.e., ψ = 40o, f’bc/f’c defined by
Eq. (6), and Kc defined by Eq. (7). Also, the Tao et al. (2013) stress-strain relationship (see
Figure 3b) was used to specify the concrete uniaxial behavior in compression. Similar to Lai et
al. (2014), the concrete tensile stress-crack opening displacement behavior was specified using
the empirical model developed by CEB-FIP (2010), as shown in Figure 3(c).
f 'bc
= 1.5(f 'c )−0.075 (6)
f 'c
5.5
Kc = (7)
5 + 2(f 'c )−0.075
1.5 Compression
A B
1
Et=Es/100
0.5
σ/Fy
Stress σ
Es
0
-4 -3 -2 -1 Es 0 1 2 3 4
-0.5 C
Et=Es/100
-1
Tension
-1.5 O Strain ε
ε/εy
(a) Steel uniaxial stress-strain curve (b) Compressive stress-strain relationships
for concrete
4
3.5
3
Stress, MPa
2.5
2
1.5
1
0.5
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Cracking displacement, mm
(c) Tensile stress-crack opening displacement curve for concrete (f’c = 110 MPa)
Fig. 3 - Steel and concrete uniaxial stress-strain behavior used in the FEM analysis
The developed FEM models were benchmarked by using them to predict the behavior and
strengths of the 111 high-strength stub CFT columns in the experimental database. Figure 4
compares the strengths obtained from FEM analyses (PFEM) with those obtained from
corresponding tests (Pexp). As shown, the mean experimental-to-predicted strength ratio
(Pexp/PFEM) ratio is 1.03, and the corresponding coefficient of variation is 0.07. Figure 5 shows
representative comparisons of the axial force-displacement responses. Figures 4 and 5 indicate
that the FEM models can reasonably predict the behavior and strengths of high-strength
rectangular CFT columns. It should be noted that, for some specimens, there are some
discrepancies between the measured and predicted stiffness (especially before the peak load).
However, the reason for this is unclear.
15000
10000
Pexp (kN)
5000
0
0 5000 10000 15000
PFEM (kN)
Fig. 4 - Comparisons of the strengths obtained from the finite element analyses and
corresponding tests
16000
FEM 16000
14000 FEM
EXP 14000
12000 EXP
12000
Axial force, kN
10000
Axial force, kN
10000
8000
8000
6000 6000
Specimen SC-32-80
4000 tested by Varma (2000) Specimen SC-48-80
4000
tested by Varma (2000)
2000 2000
0 0
0 2 4 6 8 10 12 0 5 10 15
Axial displacement, mm Axial displacement, mm
(a) Specimen SC-32-80 (b) Specimen SC-48-80
4000 2000
FEM FEM
3500 1800
EXP 1600 EXP
3000
1400
Axial force, kN
Axial force, kN
2500 1200
2000 1000
1500 800
600
1000 Specimen C3 Specimen C9-2
400 tested by Liu et al. (2013)
tested by Liu et al. (2013)
500 200
0 0
0 2 4 6 8 10 12 0 5 10 15
Axial displacement, mm Axial displacement, mm
(c) Specimen C3 (d) Specimen C9-2
Fig. 5 - Representative comparisons of the axial force-displacement responses
FURTHER EVALUATION OF AISC DESIGN EQUATIONS
To address gaps in the experimental database, additional FEM analyses were conducted using
the FEM models developed and benchmarked in the previous section. Specimen SC-48-80
tested by Varma (2000) was selected as the prototype member. Seven additional FEM analyses
were conducted using the benchmarked FEM model of this specimen by changing the tube wall
thickness while keeping the tube width constant at 400 mm. Figure 2 compares the strengths
obtained from the additional FEM analyses with those calculated using AISC 360-16 (2016).
The data points showing these comparisons are identified as “Additional FEM” in the figure.
These comparisons indicate that the AISC 360-16 design equations are over conservative in
estimating the cross-sectional strength of high-strength rectangular CFT columns with slender
sections. This is because AISC 360-16 (2016) underestimates (i) the critical buckling stress for
steel tube and (ii) the strength contribution from concrete. These two reasons will be explained
in detail later.
EFFECTIVE STRESS-STRAIN RELATIONSHIPS

The FEM models benchmarked previously in this paper were used to conduct 40 analyses on
stub CFT columns with: (i) five steel tube wall slenderness ratios (b/t = 20, 40, 60, 80, and 100),
(ii) three steel yield stress values (Fy = 317 MPa, 525 MPa, and 827 MPa), and (iii) three
concrete compressive strength values (f’c = 21 MPa, 70 MPa and 110 MPa). Results from the
finite element analysis of each stub column model included: (i) complex state of multiaxial
stresses and strains of finite elements modeling the steel tube and the concrete infill, and (ii)
overall axial load-displacement (P- ) response. These results were post-processed to obtain
average stress-strain relationships for the steel tube and concrete infill and develop the effective
stress-strain relationships. In a previous research (Lai and Varma, 2017), the authors have
presented the details regarding the development and validation of these effective stress-strain
relationships; therefore, they are not repeated here for brevity.
Figure 6(a) shows the resulting effective σ-ε relationship developed for the steel tube. As shown,
only two anchor points (εp, σp and ε5, σ2) are required to define this σ-ε relationship: σp is the
peak stress (i.e., steel critical buckling stress), εp is the peak strain, σ5 is the post-buckling stress
at 5εy, and εy is the steel yield strain estimated as Fy divided by Es. The peak strain (εp) can be
calculated using Eq. (8).
p
p = (8)
Es
Eq. (9) defines σp and σ5. This equation was obtained from regression analysis, as shown in
detail by the authors in (Lai and Varma, 2017). In this equation, t is the thickness of the steel
tube wall.
Fcr
= 0.99 − 0.07coeff (9-1)
Fy
 5.0  fc' 
= 0.9 − 0.1 11 + coeff  (9-2)
Fy  F 
 y 
b/t
coeff = (9-3)
Es Fy
Figure 6(b) shows the effective σ-ε relationship developed for the concrete infill. As shown, this
σ-ε relationship consists of the ascending branch (OA), descending branch (AB), and residual
strength branch (BC). The ascending branch (OA) is defined by Eq. (10), where parameters A
and B determine the shape, εc is the strain at peak stress, and the concrete elastic modulus (Ec)
is determined as per ACI 318-14 (2014). The descending branch BC is defined by point B (5εc,
σ5c), where σ5c is the stress at 5εc and can be defined by Eq. (11).
 A X + B  X2
= (10-1)
f 'c 1 + ( A − 2) X + (B + 1) X 2

X= (10-2)
c
Ec  c
A= (10-3)
f 'c
( A − 1)2
B= −1 (10-4)
0.55
 c = 0.00076 + (0.626f 'c − 4.33)  10−7 (10-5)
Ec = 4700 f 'c (10-6)
 5c  f' 
= 1.36 − 0.1 21 c + coeff  (11)
f 'c  
 Fy 
Compression
1 Compression A
Fcr/Fy σp /f'c OA is defined by Eq. (8)
σ5 /Fy 1
0.5  5c  f' 
= 1.36 − 0.1 21 c + coeff 
Es εcr/εy f 'c 
ε5 /εy  Fy 
σ/Fy
0 0.5 B C
σ/f'c
-4 -2 0 2 4 6 8 σ5c/f'c
Es
Fcr
-0.5 = 0.99 − 0.07coeff
Fy O
0
Et=Es/100  f'  -1 0 1 2 3 4 5 6 7 8 9
-1
 5.0
= 0.9 − 0.111 c + coeff 
Fy  F
 y 
Tension -0.5
-1.5 ε/εc
ε/εy Tension
(a) (b)
Fig. 6 - Idealized effective stress-strain relationships for high-strength rectangular CFT columns:
(a) steel tube and (b) concrete infill
DESIGN EQUATIONS
As discussed earlier in the paper, the AISC 360-16 (2016) design equations can reasonably
estimate the cross-sectional strengths of compact high-strength rectangular CFT columns.
However, they are slightly unconservative for those with noncompact sections, and over
conservative for those with slender sections. To explain this, the critical buckling stresses (Fcr)
calculated using AISC 360-16 (Eq. 5) was compared to those calculated by Eq. (9-1). Figure 7
shows representative comparisons for Specimen SC-48-80 tested by Varma (2000). This figure
indicates that Fcr calculated using AISC 360-16 (2016) is conservative when the slenderness
coefficient (λcoeff) is greater than a certain limit (e.g., 3.4 for Specimen SC-48-80). The degree of
conservatism increases as the member cross-section becomes more slender (thinner). For
example, Fcr calculated using AISC 360-16 (2016) is only 56.3% of that calculated by Eq. (9-1)
when λcoeff = 5.0. When λcoeff is less than the limit, Fcr calculated using AISC 360-16 (2016) is
unconservative. AISC 360-16 (2016) assumes that conventional-strength CFT columns with
compact and noncompact sections (λcoeff ≤ 3.0) can develop yield stress in the steel tube.
However, as shown in Figure 7, this is inaccurate for high-strength rectangular CFT columns
because: (i) for noncompact sections, local buckling instead of yielding of the steel tube governs,
and (ii) compact sections develop yielding in the steel tube; however, hoop stresses (resulting
from the interaction with concrete infill) lead to the reducing of the longitudinal stress capacity
required to cause yield of the steel tube in compression (Lai and Varma, 2016). The AISC
design equations also assume that CFT columns with noncompact and slender sections can
develop only 0.70 f’c in the concrete infill. However, as shown in Figure 6(b) (the first quadrant),
this is also inaccurate for high-strength rectangular CFT columns.
1.2
AISC 360-16 (Eq. 5)
1
0.8
Fcr / Fy
0.6
Proposed equation (Eq. 11)
0.4
0.2
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
λcoeff
Fig. 7 - Typical comparisons of the critical buckling stress (Fcr) calculated by AISC 360-16 and
the proposed equation (Eq. 9-1).
In order to more accurately represent the behavior of high-strength rectangular CFT columns
and to better estimate their cross-sectional strengths, the following equation was proposed:
Pc = As Fcr + 0.85f 'c Ac (12)
Eq. (12) assumes that high-strength rectangular CFT columns can develop Fcr in the steel tube.
The concrete compressive stress was limited to 0.85f’c to account for accidental load
eccentricity and difference of field-cured and standard-cured concrete. In this equation, Fcr is
calculated using Eq. (9-1). Eq. (12) implies that, for high-strength rectangular CFT columns, the
AISC 360-16 (2016) steel tube slenderness limits (that are used to categorize CFT section) can
be removed. The cross-section local stability is addressed directly by Eq. (9-1).
Figure 8 shows comparisons of the cross-sectional strengths estimated using Eq. (12) with
those obtained from (i) experimental investigations of the 111 stub columns in the database,
and (ii) seven additional FEM analyses that addressed gaps in the database. In this figure, the
ordinate represents the experimental (or FEM)-to-calculated strength ratio (Pexp/Pc). As shown,
the mean Pexp/Pc ratio is 1.14, and the coefficient of variation is 0.07. These comparisons
indicate that the proposed design equations can reasonably estimate the cross-sectional
strength of high-strength rectangular CFT columns (all three types). It should be noted that Eq.
(12) applies to high-strength rectangular short CFT columns with: (i) steel yield stress Fy ≤ 827
MPa, (ii) concrete compressive strength f’c ≤ 110 MPa, (iii) tube width-to-thickness ratio b/t ≤
100, and (iv) length-to-depth ratio L/H ≤ 6.0. The applicability of Eq. (12) to other cases need to
be further evaluated.
1.6
1.4
1.2
1.0
Pexp /Pc
0.8
Noncompact
0.6
EXP
0.4 Additional
Compact Slender
FEM
0.2
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
λcoeff = b/t(Fy /Es)0.5
Fig. 8 - Comparisons of the axial compressive strengths calculated using the proposed
equations with the corresponding experimental and analysis results.

This paper presented the results of a three-step method that was used to address gaps in AISC
360-16 (2016) for designing high-strength rectangular CFT short columns, and to propose
effective stress-strain relationships and new design equations for these members. In the first
step, an experimental database consisting of 124 high-strength rectangular CFT column tests
was compiled. Test results from the database were used to evaluate the possibility of extending
the AISC 360-16 design equations for estimating the strength of high-strength rectangular CFT
short columns. The evaluations indicated that AISC design equations reasonably estimated the
strength of high-strength rectangular CFT short columns with compact sections. However, they
were slightly unconservative for those with noncompact sections.
In the second step, detailed 3D nonlinear FEM models for high-strength CFT columns were
developed and benchmarked. The benchmarked models were used to perform parametric
studies to (i) address gaps in the database and (ii) develop effective stress-strain relationships
for modeling the high-strength steel tube and high-strength concrete infill of rectangular CFT
members, accounting for the effects of confinement and local buckling. Figure 6 together with
Eqs. (9) - (11) defined these relationships. Further evaluations of the AISC 360-16 design
equations with the enhanced (gaps addressed) database indicated that these equations were
over-conservative for estimating the strengths of high-strength rectangular CFT short columns
with slender sections.
In the third step, the effective stress-strain relationships together with the enhanced database
were used to propose new design equations (Eqs. 9-1 and 12) for estimating the strength of
high-strength rectangular CFT short columns. The conservatism of these proposed design
equations was confirmed by using them to predict the strengths of CFT columns from the
enhanced database.
REFERENCES
ABAQUS. (2016). ABAQUS Version 6.16 Analysis User’s Manuals. Providence, RI, USA:
Dassault Systemes Simulia Corporation.
ACI 318-14. (2014). Building code requirements for structural concrete. Farmington Hills, MI,
USA: ACI.
AISC 360-16. (2016). Specification for structural Steel buildings. Chicago, IL, USA: AISC.
Aslani, F., Uy, B., Tao, Z., & Mashiri, F. (2015). Behaviour and design of composite columns
incorporating compact high-strength steel plates. Journal of Constructional Steel Research,
107, 94–110.
CEB-FIP (Euro-International Committee for Concrete (CEB)-International Federation for
Prestressing (FIP).). (2010). Model Code for concrete structures. (CEB-FIP MC 2010).
London, U.K.: Thomas Telford.
Cederwall, K., Engstrom, B., & Grauers, M. (1990). High-strength concrete used in composite
columns. In T. T. Hester (Ed.), Proceedings of the 2nd International Symposium on
Utilization of High Strength Concrete (pp. 195–214). Berkeley, CA, USA.
Chen, B., Lai, Z., Yan, Q., Varma, A. H., & Yu, X. (2017). Experimental behavior and design of
CFT-RC short columns subjected to concentric axial loading. Journal of Structural
Engineering.
Elchalakani, M., Zhao, X. L., & Grzebieta, R. H. (2001). Concrete-filled circular steel tubes
subjected to pure bending. Journal of Constructional Steel Research, 57, 1141–1168.
Fujimoto, T., Nishiyama, I., Mukai, A., & Baba, T. (1995). Test results of eccentrically loaded
short columns -- square CFT columns. In Proceedings of the Second Joint Technical
Coordinating Committee Meeting, U.S.-Japan Cooperative Research Program, Phase 5:
Composite and Hybrid Structures. Honolulu: National Science Foundation.
Furlong, R. W. (1967). Strength of steel-encased concrete beam Columns. Journal of the
Structural Division, ASCE, 93(ST5), 113–124.
Gonzalez, Lennie, A. R., Kowalsky, Mervyn, J., Nau, J., & Hassan, T. (2008). Reversal cyclic
testing of full scale pipe piles, Technical Report No. IS-08-13, Constructed Facilities
Laboratory, North Carolina State University. North Carolina State University, Raleigh, NC,
USA.
Hajjar, J. F., Gourley, B. C., Tort, C., Denavit, M. D., & Schiller, P. H. (2013). Steel-concrete
composite structural systems, http://www.northeastern.edu/compositesystems. Department
of Civil and Environmental Engineering, Northeastern University. Northeastern University.
Retrieved from http://www.northeastern.edu/compositesystems
Han, L.-H., Zhao, X., & Tao, Z. (2001). Tests and mechanics model of concrete-filled SHS stub
columns, columns and beam–columns. Steel and Composite Structures, 1(1), 51–74.
Huang, W., Lai, Z., Chen, B., Xie, Z., & Varma, A. H. (2017). Concrete-filled steel tube (CFT)
truss girders: experimental tests, analysis, and design. Engineering Structures.
Ichinohe, Y., Matsutani, T., Nakajima, M., Ueda, H., & Takada, K. (1991). Elasto-plastic
behavior of concrete filled steel circular columns. In M. Wakabayashi (Ed.), Proceedings of
3rd International conference on steel-concrete composite structures (pp. 131–136).
Fukuoka, Japan: Association for International Cooperation and Research in Steel-Concrete
Composite Structures.
Janss, J., & Anslijn, R. (1974). Le Calcul des Charges Ultimes des Colonnes Métalliques
Enrobés de Béton. Rapport CRIF, MT, 89, (Bélgica).
Jiang, A., Chen, J., & Jin, W. (2013). Experimental investigation and design of thin-walled
concrete-filled steel tubes subject to bending. Thin-Walled Structures, 63, 44–50.
Kim, D. K. (2005). A database for composite columns [Thesis]. Georgia Institute of Technology.
Lai, Z. (2014). Experimental database, analysis and design of noncompact and slender
concrete-filled steel tube (CFT) members [Dissertation]. Purdue University.
Lai, Z., Huang, Z., & Varma, A. H. (2016). Seismic analysis and performance of high strength
composite special moment frames (C-SMFs). Structures, 9, 165–178.
Lai, Z., & Varma, A. H. (2015). Noncompact and slender circular CFT members: experimental
database, analysis, and design. Journal of Constructional Steel Research, 106(3), 220–233.
Lai, Z., & Varma, A. H. (2016). Effective stress-strain relationships for noncompact and slender
CFT members. Engineering Structures, 124, 457–472.
Lai, Z., & Varma, A. H. (2017). High-strength rectangular CFT members: database, modeling,
and design of short columns. Journal of Structural Engineering.
Lai, Z., & Varma, A. H. (2017). Seismic behavior and modeling of circular concrete partially filled
spirally welded pipes (CPFSWP). Thin-Walled Structures, 113, 240–252.
Lai, Z., Varma, A. H., & Griffis, L. G. (2016). Analysis and design of noncompact and slender
CFT beam-columns. Journal of Structural Engineering, 142(1).
Lai, Z., Varma, A. H., & Zhang, K. (2014). Noncompact and slender rectangular CFT members:
experimental database, analysis, and design. Journal of Constructional Steel Research,
101(10), 455–468.
Lin, C. Y. (1988). Axial capacity of concrete infilled cold-formed steel columns. In W.-W. Yu & H.
S. Joseph (Eds.), Ninth International Specialty Conference on Cold-Formed Steel
Structures (pp. 443–457). St. Louis: University of Missouri - Rolla.
Liu, D., & Gho, W.-M. (2005). Axial load behaviour of high-strength rectangular concrete-filled
steel tubular stub columns. Thin-Walled Structures, 43(8), 1131–1142.
Liu, D., Gho, W. M., & Yuan, J. (2003). Ultimate capacity of high-strength rectangular concrete-
filled steel hollow section stub columns. Journal of Constructional Steel Research, 59(12),
1499–1515.
Lue, D. M., Liu, J.-L., & Yen, T. (2007). Experimental study on rectangular CFT columns with
high-strength concrete. Journal of Constructional Steel Research, 63(1), 37–44.
Mursi, M., & Uy, B. (2004). Strength of slender concrete filled high strength steel box columns.
Journal of Constructional Steel Research, 60(12), 1825–1848.
Nakahara, H., & Sakino, K. (2000). Flexural behavior of concrete filled square steel tubular
beam-columns. In Proceedings of the 12th World Conference on Earthquake Engineering
(pp. 441–448). Upper Hutt, New Zealand: New Zealand Society for Earthquake
Engineering.
Nishiyama, I., Morino, S., Sakino, K., Nakahara, H., Fujimoto, T., Mukai, A., … Hayashi, Y.
(2002). Summary of research on concrete-filled structural steel tube column system carried
out under the U.S.-Japan cooperative research on composite and hybrid structures. Ibaraki
Prefecture, Japan.
Prion, H. G. L., & Boehme, J. (1994). Beam-column behaviour of steel tubes filled with high
strength concrete. Canadian Journal of Civil Engineering, 21, 207–213.
Sakino, K., Nakahara, H., Morino, S., & Nishiyama, A. (2004). Behavior of centrally loaded
concrete-filled steel-tube short columns. Journal of Structural Engineering, 130(2), 180–
188.
Schneider, S. P. (1998). Axially loaded concrete-filled steel tubes. Journal of Structural
Engineering, 124(October), 1125–1138.
Tao, Z., Han, L. H., & Wang, Z. Bin. (2005). Experimental behaviour of stiffened concrete-filled
thin-walled hollow steel structural (HSS) stub columns. Journal of Constructional Steel
Research, 61(7), 962–983.
Uy, B. (2001). Strength of short concrete filled high strength steel box columns. Journal of
Constructional Steel Research, 57(2), 113–134.
Varma, A. H. (2000). Seismic behavior, analysis and design of high strength square concrete
filled steel tube (CFT) columns [Dissertation]. Lehigh University.
Wheeler, A., & Bridge, R. Q. (2004). The behavior of circular concrete-filled thin-walled steel
tubes in flexure. In R. T. Leon & J. Lange (Eds.), Proceedings of the fifth international
conference on Composite Construction in Steel and Concrete (pp. 412–423). Kruger
National Park, Berg-en-Dal, Mpumalanga, South Africa: American Society of Civil
Engineers.
Xiong, M.-X., Xiong, D.-X., & Liew, J. Y. R. (2017). Axial performance of short concrete filled
steel tubes with high- and ultra-high- strength materials. Engineering Structures, 136, 494–
510.
Strength of CFT Columns with Various Material Strengths and Sectional
Slenderness
Ho-Jun Lee
Dept. of Architecture and Architectural Engineering, Seoul National Univ.
Seoul, Korea
hojun1032@gmail.com
Hong-Gun Park
Dept. of Architecture and Architectural Engineering, Seoul National Univ.
Seoul, Korea
parkhg@snu.ac.kr
ABSTRACT
To enhance the structural performance of concrete-filled steel tubular (CFT) columns, there is
an increasing interest in utilizing high-strength materials. Nevertheless, current international
standards have limitations pertaining to material grades in composite constructions. The present
research focuses on assessing the load-carrying capacities of rectangular CFT columns and
beam-columns, incorporating broad ranges of the sectional slenderness as well as material
strengths. Concentric and eccentric axial load tests were conducted to investigate the effects of
steel yield strength and width-to-thickness ratio. Extensive sets of test specimens were further
collected from previous studies to investigate the applicability of current design codes and verify
fiber model analysis. Based on parametric studies, a simplified strength model for the composite
section was developed.
INTRODUCTION
From the viewpoints of enhancing structural performance and saving resources, high-strength
materials are gaining popularity in building constructions. The use of such high-strength
concrete and steel in composite structures is also expected to improve the economy and
constructability. Recently, high-strength materials (concrete compressive strength up to 150
MPa, steel tensile strength up to 780 MPa) have been applied to concrete-filled tube (CFT)
columns mainly for high-rise buildings (Aoki et al., 2012; Endo et al., 2011; Matsumoto et al.,
2012).
However, as observed in Table 1, the available material strengths for composite structures are
limited in current design codes (AIJ, 2008; AISC, 2016; CEN, 2004). Generally, high-strength
concrete is not recommended owing to its brittle behavior. In case of high-strength steel, the
restriction mainly comes from the incompatibility between yield strain of the steel and crushing
strain of the concrete: early crushing of the concrete may occur prior to development of plasticity
in the steel components. Meanwhile, recent studies have been focusing more on adopting high-
strength materials (Aslani et al., 2015; Lee et al., 2016; Liew et al., 2016; Wang et al., 2017).
Test results for high-strength rectangular CFT (RCFT) members are available in latest
literatures. According to the experimental studies, by using high-strength materials, the
structural performance of RCFT columns was definitely improved in terms of strength and
stiffness, while many test results also revealed that the ultimate strengths were generally
overestimated by current specifications. However, as most studies were limited to individual test
results, the effects of material strengths should be further clarified based on comprehensive
databases.
This study focuses on assessing the load-carrying capacities of RCFT columns and beam-
columns, incorporating broad ranges of the sectional slenderness as well as material strengths.
First, concentric and eccentric axial load tests were conducted to examine the effects of steel
yield strength and width-to-thickness ratio. Applicability of current design codes are further
examined based on extensive databases from previous studies. For the use of high-strength
materials and large sectional slenderness, a simplified model to construct P-M interaction curve
was developed.
REVIEW OF CURRENT DESIGN CODES

In ANSI/AISC 360, the sectional class of the steel tube is classified into compact (b/t ≤ λp), non-
compact (λp < b/t ≤ λr), and slender (b/t > λr) sections (Table 1). For non-compact and slender
sections, the theoretical elastic buckling stress Fcr is used in design. In Eurocode 4, only
compact section is allowed. In the recommendations of the Architectural Institute of Japan (AIJ,
2008), a strength degradation factor γs is used to account for the local buckling.
Table 1 – Structural provisions for RCFT columns
AIJ
Provisions Eurocode 4 a
(CFT ANSI/AISC 360-16
recommendations)
Material 235  Fy  460 235  Fy  440 Fy  525
strengths
(MPa) 20  fc  50 18  fc  90 21  fc  69
Compact: b t  2.26 Es Fy  λp
Width-to- Non-compact:
thickness B t  1.78 Es Fy B t  2.44 Es Fy
b t  3.00 Es Fy  λr
ratio
Slender: b t  5.00 Es Fy
Concrete: 0.85fc (compact)

Effective Concrete: 1.0fc Concrete: γ c fc 0.7fc (slender)
material
Steel: Fy Steel: γ s Fy Steel: Fy (compact)
strengths
Fcr (slender)
Stress Rectangle (compact)
Rectangle Rectangle
block Triangle (slender)
a
Intended for use with hollow structural steel sections (i.e., rounded corners)
Effective material strengths and stress distribution assumed in the cross-section design are also
summarized in Table 1. Figure 1 illustrates typical rigid-plastic analysis for cross-sectional
design of compact RCFTs. Although the definition for the compact section is not identical
among the various international standards, all design codes allow the use of yield strength Fy.
Strength factor γ only for infill concrete is different. The AISC code uses γ = 0.85 for ordinary
reinforced concrete columns, while the Eurocode 4 uses γ = 1.0 addressing the lateral
confinement effect. In Eurocode, the slenderness limit for the compact section is stricter than in
AISC code. In the AIJ recommendations, a reduction factor γc addressing the scale effect
(Sakino et al., 2004) is used. These factors are generally consistent in the design of axial
strength and combined axial-flexural strength.
Steel stress Concrete stress

B Fy γfc
t
H N.A.
Fy
Fig. 1 – Typical stress distribution in rigid-plastic analysis
For slender sections in AISC specifications, a strength reduction factor for the infill concrete is
defined as 0.7, addressing the poor lateral confinement. The steel compressive stress is
replaced with Fcr. Also, instead of uniform stress block, triangular stress distribution is assumed
for conservative design. In the case of noncompact sections, the nominal strength is calculated
by interpolating the strengths for compact and slender sections. For construction of P-M
interaction relationships, a bilinear model connecting the nominal axial and flexural strengths is
recommended.
In the axial-flexural design of AIJ recommendations, the effective strength and depth of the
concrete stress block, are reduced when either high-strength concrete ( fc > 60 MPa) or large
width-to-thickness ratio (B/t > 2.4 Es Fy ) is used (Nakahara & Sakino, 2003). Such provision
accounts for the brittle behavior of the high-strength concrete or poor confinement of slender
tube section.
To investigate the structural performance of RCFT columns with high-strength steel thin plates,
concentric and eccentric axial load testing was performed (Lee et al., 2017; Lee et al., 2019).
Major test parameters were eccentricity (eccentricity ratio = e/H), steel yield strength, and the
use of stiffeners (Figure 2 and Table 2). With the same width-to-thickness ratio, a tube section
with mild steel (yield strength = 301 MPa) was classified as compact section while a tube
section with high-strength steel (yield strength = 746 MPa) was classified as slender section
(AISC, 2016). Since a stiffener was located at the center of the tube wall, the width-to-thickness
ratio of the subpanel was decreased from 58.0 to 29.0, satisfying the requirement for compact
section. Higher concrete strength was used in rectangular sections to maintain the similar axial-
load contribution of steel and concrete as in square sections.
H = 300 H = 150 w = 150 150 150
150
150
ts = 5 bs = 60 20 60
B = 300
B = 300
tt = 5
150
150
(a) Section 1, 3 (b) Section 2 (c) Section 4 (d) Section 5
Fig. 2 – Cross sections of test specimens
Table 2 – Test parameters and results of specimens
B H tt Fy,t fc e Ptest

Specimen Note
(mm) (mm) (mm) (MPa) (MPa) (mm) (kN)
C1 300 300 5 746 70.5 - 8,686 -
C2 300 150 5 746 83.6 - 6,152 -
C3 300 300 5 301 70.5 - 6,602 -
C4 300 300 5 746 70.5 - 9,726 Stiffened
C5 300 150 5 746 83.6 - 6,253 Stiffened
E1 300 300 5 746 70.5 40 6,491 -
E2 300 150 5 746 83.6 20 3,873 -
E3 300 300 5 301 70.5 40 4,553 -
E4 300 300 5 746 70.5 40 7,131 Stiffened
E5 300 150 5 746 83.6 20 4,322 Stiffened
Figure 3 shows load-displacement relationships of test specimens. Effects of sectional and

material properties were consistent in two series of testing, regardless of axial-load eccentricity.
In specimens with square sections 1, 3, and 4, the peak strength was the greatest in section 4
(high-strength steel plus stiffeners) and the lowest in section 3 (mild steel). In specimens with
rectangular sections 2 and 5, the load-carrying capacity of section 5 (high-strength steel plus
stiffeners) was greater than that of section 2 (high-strength steel). The test result indicates that
the stiffeners enhanced the strength by restraining the tube local buckling, satisfying the
predictions for compact section. Also, although early local buckling took place in sections 1 and
2 without stiffeners, the load-carrying capacity continued to increase without significant stiffness
degradation.
12000 8000
7000
10000 E4
6000
8000 E1
Axial load (kN)
5000 E3
C3
C1 C4
6000 4000
E5
3000
4000
C5
2000 E2
2000 C2
1000
0 0
0 2 4 6 8 10 12 14 0 5 10 15 20 25 30 35 40
Axial displacement (mm) Mid-height deflection (mm)
(a) Concentric axial loading (b) Eccentric axial loading
Fig. 3 – Load-displacement relationships of test specimens
APPLICABILITY OF CURRENT DESIGN CODES

In the present study, to extend the applicability of current design codes for RCFT columns
covering wide ranges of material strengths [ fc > 70 MPa, Fy > 525 MPa, AISC (2016)] and
sectional slenderness, relevant test results were extensively collected. The constructed
database is categorized into two parts: 1) short columns for axial capacity and 2) beam and
beam-columns for flexural and combined axial-flexural capacity. The properties of existing test
specimens are summarized in Table 3.
Table 3 – Parameter ranges of existing test specimens
Stub-columns Beam-columns
No. of specimens 195 210
B 180 ~ 500 mm 149 ~ 600 mm
H 135 ~ 500 mm 150 ~ 600 mm
b/t 10.4 ~ 131.7 13.6 ~106.7
Fy 211 ~ 835 MPa 211 ~ 913 MPa
b t  Fy Es 0.47 ~ 4.94 0.65 ~ 4.46
fc 20.0 ~ 202.0 MPa 20.0 ~ 183.0 MPa
In selecting test specimens, careful attention was paid to obtain reasonable results: 1) wide
ranges of material strengths and sectional slenderness were included, 2) exceptionally small
specimens were excluded, and 3) the compressive strength of concrete was modified
considering the shape and size of the material test specimens. Small-scale specimens with B ×
H < 1802 mm2 were excluded to minimize the size effect (Nakahara et al., 1998).The concrete
compressive strength fc was defined as the strength of an equivalent concrete cylinder with a
100 mm diameter (Yi et al., 2006). In this study, concrete strength greater than 70 MPa was
regarded as high strength.
Figure 4 shows the comparison between test strengths and predictions for stub-columns.
Generally, the AISC code shows unsafe predictions for noncompact specimens with high-
strength materials. Thus, in the AISC code, the use of high-strength materials is restricted. On
the other hand, the conservative nature of the AISC predictions significantly increases as the
slenderness ratio increases.
While the AIJ predictions accurately estimated the capacity of specimens with high-strength
steel by considering the local buckling, the capacity of specimens with high-strength concrete
was generally overestimated. This is partly because the reduction factor addressing the
concrete size effect does not adequately reflect the early failure of the high-strength concrete.
Eurocode 4 was only applicable for test specimens with specified material strengths (Fy ≤ 460
and fc ≤ 50 MPa) and sectional slenderness (B/t ≤ 1.78 Es Fy ).
1.6 1.6
1.4 1.4
Test result/Prediction
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0 1 2 3 4 5 0 1 2 3 4 5
b Fy b Fy
t Es t Es
(a) ANSI/AISC 360-16 (b) Eurocode 4
1.6
1.4
1.6 Fy≤525MPa, fc≤70MPa
Mean=1.05 Fy>525MPa, fc≤70MPa
1.2
SD=0.096 Fy≤525MPa, fc>70MPa
1.4
1
Fy>525MPa, fc>70MPa
Max=1.31
1.2
0.8
0.6 1
0 1 2 3 4 5
b Fy
0.8
t Es Min=0.80
(c) AIJ Recommendations
0.6
0 1 2 3 4 5
Fig. 4 – Strength evaluation of short columns based on current design codes
Figure 5 shows the comparison between test strengths and predictions for beam-columns. The
test strength was defined as the measured one at the critical section, considering additional
moment due to the axial force (i.e., second-order effect). It is also noted that test results
obtained from experimental set-ups, which may be affected by base confinement, were omitted
in strength evaluation (AIJ, 2008). The results showed similar tendency as in the case of stub-
column specimens.
2.5 2.5
2 2
1.5 1.5
1 1
0.5 0.5
0 1 2 3 4 5 0 1 2 3 4 5
b Fy b Fy
t Es t Es
(a) ANSI/AISC 360-16 (b) Eurocode 4

2.5
2
1.6 Fy≤525MPa, fc≤70MPa
Mean=1.05 Fy>525MPa, fc≤70MPa
1.5 SD=0.096 Fy≤525MPa, fc>70MPa
1.4
Fy>525MPa, fc>70MPa
1 Max=1.31
1.2
0.5 1
0 1 2 3 4 5
b Fy
0.8
t Es Min=0.80
(c) AIJ Recommendations
0.6
Fig. 5 – Strength evaluation of beam-columns
0 1 based on
2 current 3design codes
4 5
STRENGTH MODEL FOR RCFT BEAM-COLUMNS

For the use of high-strength materials and large sectional slenderness, methods to construct
simplified P-M interaction curve were investigated. First, to check strain compatibility between
steel and concrete, section analysis using fiber models was performed and verified with existing
test results. Uniaxial constitutive laws proposed by Sakino et al. (2004) were modified to obtain
better agreement with the database (Lee et al., 2019). Particularly, strength reduction factor was
not applied to the infill concrete. Figure 6 and Figure 7 show comparisons with the test results
and analytical results. Except for high-strength concrete, the predictions agreed with the test
strengths. This indicates that, in the case of high-strength concrete, the ultimate capacity of
composite section is affected by premature failure of the concrete rather than strain compatibility
based on material tests. Therefore, reasonable reduction factors are required for high-strength
concrete (Liew et al., 2016), which is not a scope of this paper.
1.6 1.6
(a) Fy≤525MPa, fc≤70MPa (b) Fy>525MPa, fc≤70MPa
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0 1 2 3 4 5 0 1 2 3 4 5
1.6 1.6
(c) Fy≤525MPa, fc>70MPa (d) Fy>525MPa, fc>70MPa
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0 1 2 3 4 5 0 1 2 3 4 5
b Fy b Fy
t Es t Es
Fig. 6 – Comparison between test results and analytical results for stub-columns
2 2
1.8
(a) Fy≤525MPa, fc≤70MPa 1.8 (b) Fy>525MPa, fc≤70MPa
1.6 1.6
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0 1 2 3 4 5 0 1 2 3 4 5
2 2
(c) Fy≤525MPa, fc>70MPa (d) Fy>525MPa, fc>70MPa
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0 1 2 3 4 5 0 1 2 3 4 5
b Fy b Fy
t Es t Es
Fig. 7 – Comparison between test results and analytical results for beam-columns
For construction of P-M relationships, it is essential to define reasonable performance points for
pure compression and pure flexure (points A and B in Figure 8). Based on parametric study
using fiber model analysis (b/t = 15 ~ 80, fc = 20 ~ 150 MPa, Fy = 200 ~ 900 MPa), simplified
equations for points A and B were suggested.
P/Pn
A Fiber model analysis
Modified rigid-plastic method
M/Mn
B
Fig. 8 – Comparison between fiber model analysis and modified rigid-plastic analysis
The pure compressive capacity Pn of composite section is calculated as follows.
Pn  Ac  αfc   As  βFy  (1)
Where α and β are strength reduction factors for the high-strength concrete and slender tube
section, respectively. In the case of fc ≤ 70 MPa, α = 1.0. For fc > 70 MPa, further test results
are needed to determine appropriate α. The peak stress of steel βFy was calculated based on
an existing post-buckling model (Guo et al., 2007), which was also adopted in the fiber model.
The strain incompatibility of peak stresses between steel and concrete may be significant when
high-strength steel or conventional steel slender section is used. According to the results of
section analysis, in the case of high-strength steel, the peak load of composite section was
generally governed by the failure of steel component. In the case of conventional steel slender
section, the peak load was governed by the failure of infill concrete. In both cases, the effect of
strain incompatibility on the pure compression of composite section was marginal [the difference
between analytical result and Equation (1) was less than 2%]. Consequently, Equation (1) still
gives conservative predictions.
For pure flexure, it is consistent to assess the nominal capacity Mn using the same effective
stresses 1.0 fc and βFy defined for pure compression. For convenience, a conventional rigid-
plastic approach (Figure 1) was modified with the compressive stresses 1.0 fc and βFy. In such
case, depth of the neutral axis is expressed as follows.
2Fy Htw  fcbt f  1  β  Fy bt f

a (2)
2Fy tw  fcb  2βFy tw
According to the parametric study, when normal-strength materials are used, the modified rigid-
plastic method yields Mn comparable or smaller than the value calculated by the fiber model
analysis, which is conservative for practical uses. On the other hand, when high-strength steel
thin plates are used, predictions of the modified approach tend to be greater than those of the
analytical model. Since the overestimation is greater than 5% when steel yield strength is
greater than 800 MPa and concrete compressive strength is less than 30 MPa, the modified
method is not recommended for such combination.
In the range of medium axial-load ratio (Figure 8), the prediction of modified rigid-plastic method
(with effective stresses 1.0 fc and βFy) generally overestimates the analytical result which is
assumed as actual capacity. Thus, the method is not applicable for determining the entire curve.
Instead, it is recommended to use effective concrete stress of 0.85 fc in axial-flexural design of
the composite section with normal-strength materials. In the case of high-strength steel, since
relative strength contribution of the steel component is dominant, it is conservative to use an
equivalent steel-column method (Method 1 of the AISC specifications).
CONCLUSIONS
Concentric and eccentric axial load tests were conducted to examine the effects of steel yield
strength and width-to-thickness ratio on the strength of RCFT columns and beam-columns.
Applicability of current design codes were further examined based on extensive databases from
previous studies, incorporating broad ranges of the sectional slenderness as well as material
strengths. The major findings of this study are summarized as follows.
1) In the specimens with high-strength steel thin plates, despite early local buckling, the load-
carrying capacity continued to increase without significant reduction of the stiffness. In the
stiffened specimens, the load-carrying capacity reached the plastic strength of composite
section owing to restrained local buckling.
2) It was confirmed that the current design codes (ANSI/AISC 360-16, Eurocode 4, AIJ
recommendations) safely predict the test strengths of RCFT columns and beam-columns with
conventional material strengths and width-to-thickness ratios. In particular, the test specimens
with normal-strength materials and large sectional slenderness significantly outperformed the
nominal strengths by the AISC specifications, attaining the plastic capacity of composite section.
3) For application of high-strength materials and large width-to-thickness ratio, strength models
were developed for the composite section subjected to pure compression and flexure. When
constructing the P-M interaction curve for the high-strength steel, it is recommended to use the
method for equivalent steel column.
REFERENCES
AIJ. (2008). Recommendations for design and construction of concrete filled steel tubular
structures. Tokyo: AIJ (Architectural Institute of Japan).
AISC. (2016). Specification for structural steel buildings ANSI/AISC 360-16. Chicago: AISC
(American Institute of Steel Construction).
Aoki, Y., Iwashimizu, T., Yamada, Y., & Nagano, K. (2012). Construction of the ultra-high-
strength concrete CFT pillar of design strength 150 N/mm^2 : 'ABENO HARUKAS' 300
meters super-high-rise compound building. Concrete Journal, 50(8), 683-688.
Aslani, F., Uy, B., Tao, Z., & Mashiri, F. (2015). Predicting the axial load capacity of high-
strength concrete filled steel tubular columns. Steel and Composite Structures, 19(4),
967-993.
CEN. (2004). Design of composite steel and concrete structures, Part 1-1: general rules and
rules for buildings Eurocode 4. Brussels: CEN (European Committee for
Standardization).
Endo, F., Yamanaka, M., Watanabe, T., Kageyama, M., Yoshida, O., Katsumata, H., & Sano, T.
(2011). Advanced technologies applied at the new “Techno Station” building in Tokyo,
Japan. Structural Engineering International, 21(4), 508-513.
Guo, L., Zhang, S., Kim, W.-J., & Ranzi, G. (2007). Behavior of square hollow steel tubes and
steel tubes filled with concrete. Thin-Walled Structures, 45(12), 961-973.
Lee, C.-H., Kang, T. H.-K., Kim, S.-Y., & Kang, K. (2016). Strain compatibility method for the
design of short rectangular concrete-filled tube columns under eccentric axial loads.
Construction and Building Materials, 121, 143-153.
Lee, H.-J., Choi, I.-R., & Park, H.-G. (2017). Eccentric compression strength of rectangular
concrete-filled tubular columns using high-strength steel thin plates. Journal of Structural
Engineering, 143(5), 04016228.
Lee, H.-J., Park, H.-G., & Choi, I.-R. (2019). Compression loading test for concrete-filled tubular
columns with high-strength steel slender section. Journal of Constructional Steel
Research, 159, 507-520.
Liew, J. R., Xiong, M., & Xiong, D. (2016). Design of concrete filled tubular beam-columns with
high strength steel and concrete. Structures, 8, 213-226.
Matsumoto, S., Komuro, T., Narihara, H., Kawamoto, S., Hosozawa, O., & Morita, K. (2012).
Structural design of an ultra high-rise building using concrete filled tubular column with
ultra high strength materials. Paper presented at the 15th World Conference on
Earthquake Engineering, Lisbon, Portugal.
Nakahara, H., Inai, E., & Sakino, K. (1998). Axial compressive load capacities of square steel
tubular columns. Journal of structural engineering. B, 44, 167-174.
Nakahara, H., & Sakino, K. (2003). Flexural behavior of concrete filled square steel tubular
columns using high strength materials. Journal of Structural and Construction
Engineering(567), 181-188.
Sakino, K., Nakahara, H., Morino, S., & Nishiyama, I. (2004). Behavior of centrally loaded
concrete-filled steel-tube short columns. Journal of Structural Engineering, 130(2), 180-
188.
Wang, Z.-B., Tao, Z., Han, L.-H., Uy, B., Lam, D., & Kang, W.-H. (2017). Strength, stiffness and
ductility of concrete-filled steel columns under axial compression. Engineering
Structures, 135, 209-221.
Yi, S.-T., Yang, E.-I., & Choi, J.-C. (2006). Effect of specimen sizes, specimen shapes, and
placement directions on compressive strength of concrete. Nuclear Engineering and
Design, 236(2), 115-127.
BUCKLING RESISTANCE OF HYBRID STEEL-CONCRETE COLUMNS
Pisey KEO
pisey.keo@insa-rennes.fr
Hugues SOMJA
hugues.somja@insa-rennes.fr
Quang Huy NGUYEN

quang-huy.nguyen@insa-rennes.fr
Mohammed HJIAJ
mohammed.hjiaj@insa-rennes.fr
ABSTRACT
This paper presents an assessment of the application of simplified design methods proposed in
EC2 and EC4 for the design of slender hybrid columns with several embedded steel profiles. It
relies on an extensive numerical parametric study on the ultimate load of hybrid columns
subjected to combined axial load and bending moment. To accurately estimate the ultimate load,
a nonlinear beam element which accounts for slip has been developed. The predictions of ultimate
loads by the numerical model are compared against the results given by both EC2 and EC4
simplified methods. This comparison shows that the effective stiffness proposed by EC2 and EC4
are not appropriate for hybrid column design. A new expression of the effective stiffness is then
proposed, which provides an accurate estimation of the ultimate load of hybrid column subjected
to combined axial load and uniaxial bending moment while compared to the predictions obtained
by FE analysis.
INTRODUCTION
Over nearly a century, Steel Reinforced Construction (SRC) consisting of steel structural framing
partially or totally encased in concrete has been adopted by engineers. The advantages of SRC
over reinforced concrete construction are: greater ductility, more compact cross-section, reduced
creep deformation, and faster concrete casting (Morino S. , 1998). Those over steel construction
are: multiple roles of concrete as structural, fireproofing and buckling-restraining elements, higher
stiffness, and greater damping. Modern SRC members (hybrid members) commonly have
extensive transverse and longitudinal reinforcement, and some use shear connectors between
steel section and the surrounding concrete (Roeder, 1998).
In high-rise reinforced concrete buildings, columns or concrete walls containing multiple
embedded steel profiles are often used, because the loads are such that standard reinforcement
with rebars is not sufficient. Those composite elements belong to structures defined as “hybrid”,
which means that they are neither reinforced concrete structures in the sense of EC2 (EC2-1-1,
2005), nor composite steel concrete structures in the sense of EC4 (EC4-1-1, 2005). It is unclear
whether these hybrid elements behave as reinforced concrete elements or as composite ones.
For columns being sensitive to instability, both EC2 and EC4 propose simplified design methods
based on moment magnification approach where the second order bending moment is equal to
the first order bending moment multiplied by a magnification factor k. Different expressions of
factor k have been proposed; the most common formula is k = β/(1−NEd /Ncr ) where NEd is the
applied normal force. The accuracy of moment magnification method strongly depends on two
factors. The first one is the equivalence factor to uniform moment diagram β. The second one is,
as included in Ncr, the effective flexural stiffness (EI) which depends on, among other factors, the
nonlinearity of the concrete stress-strain curve, the creep and the cracking along the column
length. The formulation of the effective flexural stiffness (EI) used to design reinforced concrete
and composite columns has been studied for decades and there is a vast amount of its
formulations in the literature, see among others (Westerberg, 2002) (Tikka & Mirza, 2008) and
(Bonet, Romero, & Miguel, 2011).
This paper aims to present an evaluation of the simplified methods (bending moment
magnification method) proposed in EC2 and EC4 when applied for hybrid columns. To do so, a
FE model has been developed in which the geometrical/material nonlinearities as well as the
partial interaction effect between steel profiles and concrete are taken into account. Since there
is no data available in the literature for buckling experimental tests of hybrid columns, the FE
model is validated by comparing its predictions against experimental results of standard
composite columns. This validation was performed in (Keo, Somja, Nguyen, & Hjiaj, 2015).The
FE model then serves as reference for an extensive parametric study in which the simplified
methods proposed in EC2 and EC4 are evaluated in case of hybrid columns. The comparison
between the results obtained with Eurocode simplified methods and with FEA shows that
simplified method of EC2 and EC4 leads somehow to unsafe results when applied for hybrid
columns. It means that the proposed effective flexural stiffness (EI) of EC2 and EC4 are not
appropriate for slender hybrid column design. As a result, in this paper, a new expression of (EI)
for slender hybrid columns is proposed.
FINITE ELEMENT MODEL

In order to analyze the behavior of slender hybrid columns, a two-dimensional beam-column finite
element formulation was developed based on Euler-Bernoulli kinematics and fiber cross-section
model. The co-rotational approach was adopted. In this context, the element displacements are
separated into rigid-body and deformational degrees of freedom. The element rigid-body motion
is handled separately within the mapping from the co-rotational to global element frames. The
developed FE model is capable to consider the following aspects: a cross-section with more than
one steel section in partial interaction; geometrical and material nonlinearities; initial imperfection;
residual stresses; and confinement of concrete. The detail of FE formulation in co-rotational frame
for hybrid beam-column element can be found in (Keo, Nguyen, Somja, & Hjiaj, 2015). In the
following, we highlight only the degrees of freedom and beam kinematic of hybrid beam-column
with two encased steel profiles in local and global coordinate system. All variables subscripted
with “sk” belong to the embedded steel element and those with “c” belong to the surrounding
concrete component. The quantities in local frame are denoted with the overline variables.
s1j us1 j
g1j
y g1 jvcj xl
yl
us1m ucj ucj
cj
s us1i g2j us 2 j
g1i1i ucm s2j
vci θ g2 j
g1i i
s1i s1j uci
us 2 m
ci
g2i us 2i
ci cj s2i
g 2i
s2i s2j
x
Fig. 1- Degrees of freedom and kinematic of the element
In global co-rotational frame, the element has 10 degrees of freedom: global displacements and
rotation of the node (ci and cj) and slips (gki, gkj) between the steel node (ski, skj) and concrete
node (ci, cj) with k=1, 2, see Figure 1. It is here assumed that the slips (gki, gkj) are perpendicular
to the end of cross-sections. The vector of global nodal displacements is defined by
T
(1) pg  uci v ci i g1i g 2i ucj v cj  j g1 j g 2 j 
yl
Ns1i , us1i N s1j , us1j

Ns1m , us1m
M i , i M j, j
N cm , uc m xl
N c j , uc j
Ns2i , us2i N s2j , us2j
N s2m , us 2 m
Fig. 2- Degree of freedoms of local linear element with two encased steel profiles
In local frame, the local element has 10 degrees of freedom (see Figure 2). The transverse
displacement v is approximated using cubic Hermite interpolations. In order to avoid the
curvature locking, three internal nodes are added in order to use quadratic shape function for axial
displacement interpolation. However, for saving the calculation time, three degrees of freedom
corresponding to the internal nodes will be statically condensed out thereafter to obtain the local
displacement vector containing only the degrees of freedom at the element ends. The material
non-linearity is taken into account by adopting the distributed plasticity model with fiber section
discretization (see Figure 3).
(a) Hybrid column with 3 (b) Mega column with 4

embedded steel profiles embedded steel profiles
Fig. 3- Fiber discretization of sections
EUROCODE DESIGN METHODS FOR SLENDER COLUMNS

For a design of columns against instability, both EC2 and EC4 permit using a geometrically and
materially nonlinear analysis, so-called a general method. However, this method is difficult to
implement. Both Eurocodes propose a simplified method, called “moment magnification method”,
in which the first-order bending moment MEd is modified by a magnification factor k to obtain the
second-order one. The factor k largely depends on the flexural stiffness and equivalent moment
distribution. Hence, two steps are needed in this method. The first one is to compute the effective
stiffness (EI) and the second one is to estimate the first-order moment magnification factor
following the shape of bending moment diagram. In general, not only the factors mentioned
previously influence the flexural stiffness of the columns but also the column slenderness, the
eccentricity, the normal force and the reinforcement ratio. The mathematical formulations of EC2
and EC4 simplified method applied for hybrid columns are summarized in Figure 4.
Second order bending moment: M Ed ,2  k M Ed ,1

Effective flexural stiffness: EI  Kc Ecd Ic  Ks Es I s  Ka Ea I a
EC2 EC4
  2

k  1  1,   ; k  1,
N cr c0 N
1 1  Ed
N Ed N cr ,eff
2 f ck    0, 66  0, 44rm  0, 44
N cr  2
EI ; k1 ; k2  n   0, 2;
L 20 170 2
N cr ,eff  EI ;
N kk L2
n  Ed ; K c  1 2 ; K s  K a  1;
Ac f cd 1  ef K c  0, 45  1, 2 ; K s  K a  0,9;
w0  L / 400 w0 : according to steel shape
Fig. 4- Moment magnification method of EC2 and EC4
ASSEMEMENT OF SIMPLIFIED METHODS OF EC2 AND EC4
The developed FE model which was successfully validated above is used to conduct an extensive
parametric study in order to assess the validity of simplified methods of EC2 and EC4 for the
hybrid columns. To do so, slender hybrid columns with different types of cross-sections (HSSRC1-
5, see Figure 5) are modeled using the present FE model and calculated using Eurocode
simplified methods. The obtained results is compared to highlight the applicability of Eurocode
simplified methods to hybrid columns. For all cases, the limit of elasticity for steel profile is 355
MPa and for reinforcement rebar is 500 MPa. Three classes of concrete strength C35, C60 and
C90 are considered. The parametrical study (totally 1140 data sets) is summarized in Table 2.
(a) Section HSRCC1 (b) Section HSRCC2
(c) Section HSRCC3 (d) Section HSRCC5
(e) Section HSRCC4
Fig. 5-. Cross-sections considered in parametric study
Table 1- Summary of case-studies.
Section HSRCC1–5
Concrete C35/45; C60/75; C90/105
fyk 500 MPa
fa 355 MPa
λ 25–180
e/h 0.0–3.0
δ 0.2–0.62
φef 0; 1.5
In order to evaluate the accuracy of the safety level when applying the simplified design methods
proposed in EC2 and EC4 for hybrid column design, nonlinear material models as well as the
safety format have to be properly described. The comparison of the results provided by simplified
method of EC2 against FE analysis is readily achieved by using the stress-strain relationship
based on the design values of the constitutive model parameters as it is clearly defined in EN
1992-1-1: 5.8.6. Regarding the safety format for nonlinear FE analysis, the Eurocode for
composite structures recommends to use the stress-strain relationships defined in EC2 and EC3
as stated in EN 1994-1-1: 6.7.2(8). Therefore, the material constitutive laws and the partial factors
recommended by EC2 and EC3 are adopted (Figure 6).
sc ss ss
fyd=fy/1,0 0,5fy*
fcd=fck/1,5
fsd=fsk/1,15 -0,5fy* -0,5fy*
Es/300 0,5fy*
h -0,5fy*
0,4fcd
Ecd=Ecm/1,2 Es Ea b 0,5fy*
ec es es h/b 1,2
ec1 ecu1 esd esud ey eyu fy*=235MPa
a) Concrete b) Reinforcement c) Steel profile d) Residual stress
Fig. 6- Material constitutive laws and residual stress distribution used for FE analysis
The numerical model takes into account the residual stresses existing in the steel profiles (Figure
6-d)), as well as a sinusoidal geometric imperfection. The recommendations for the amplitude of
the latter are different in each standard. For steel members, a value of l0/1000 is proposed by
EC3, while EC2 proposes a value of l0/400 for concrete columns. Since the construction of hybrid
columns is being alike to reinforced concrete columns where the uncertainty regarding to the
geometry is significant, we adopt an initial imperfection w0 equal to l0/400 combined with an explicit
representation of the residual stress distribution in order to ensure the best accuracy of the results.
Besides, following the recommendations of Eurocode 4, a complete shear connection is imposed
in FE model.
Before presenting the general conclusion of the comparison of FEA to simplified methods, the
results obtained for different slenderness and loading cases for the HSRCC1 section are shown
in Figure 7 and 8. The concrete used is C60 and the effective creep coefficient is taken equal to
1.5. In the case of pure compression (Figure 7a), it appears that the method is not safe for low
slenderness. In the case of columns subjected to single curvature bending (Figure 7b and c),
Eurocode 2 again gives non-conservative results for low to moderate slenderness. Similar results
are observed in the case of double curvature bending, except for very high load eccentricities
(close to pure bending). (Figure 7d), for which the EC2 method gives good results.
The same comparison is made for the simplified method of Eurocode 4 (Figure 8). Quite
surprisingly, the latter seems to perform less well. Under pure compression (Figure 8a), the
simplified method of EC4 gives safe results regardless of column slenderness. For a small
eccentricity, the ultimate load given by EC4 formulation is safe regardless of column slenderness
(see Fig. 10b to d). For a moderate load eccentricity, EC4 method always overestimates the
ultimate load.
a) Buckling curve b) Symmetrical single curvature bending (rm = 1)
c) Single curvature bending (rm = 0) d) Double symmetrical curvature bending (rm = −1)
Fig. 7- Comparison of simplified method of EC2 against FE analysis results
To generally evaluate the EC2 and EC4 simplified methods, the ratio of the numerical failure load
RFE to the corresponding resistance given by simplified method RSM is selected in our
investigation.
R
(2) R  FE
RSM
2 2 2 2
 N   MFE   N   MSM 
with RFE   FE     and RSM   SM     .
N
 pl ,Rd   Mpl ,Rd   Npl ,Rd   M pl ,Rd 
a) Buckling curve b) Symmetrical single curvature bending (rm = 1)
c) Single curvature bending (rm = 0) d) Double symmetrical curvature bending (rm = −1)
Fig. 8- Comparison of simplified method of EC4 against FE analysis results
The histogram of ratio R obtained with both simplified methods is illustrated in Figure 9. It was
found that the mean value “r” and the standard deviation “s” of the resistance ratio are respectively
equal to 0.996 and 0.104 for EC2 simplified method, and 1.010 and 0.112 for EC4 simplified
method. The percentage of R below 0.97 is 41.84% and 34.86% for EC2 and EC4 simplified
method, respectively. As a general conclusion, it can be pointed out that mean estimations of both
design codes seem to be correct but that their shortcomings lead to a large scatter of the results.
(a) Eurocode 2 (b) Eurocode 4

Fig. 9- Histogram of ratio R for simplified method of Eurocode 2 et 4
PROPOSAL OF A MOMENT MAGNIFICATION DESIGN METHOD FOR HYBRID COLUMN
Before suggesting new expressions for correction factors involved in the moment magnification
method for hybrid column design, some effects are analyzed to get further insight into the physical
behavior of hybrid column. Remarkably, the analysis of the results show that the residual stresses
have a marginal effect on the behavior of hybrid columns and they can be ignored. It is worth to
mention that the steel profiles, being placed eccentrically from the center of gravity of RC section,
are summited to a large normal force. The latter reduces the effect of residual stresses. Therefore,
it can be presumed that the structural steels behave as large rebars in lieu of structural steels
under bending. Considering the above comments, it can be concluded that the new method for
hybrid columns should be based on the EC2 variant rather than from the EC4 version.
On the other hand, the effect of concrete creep is investigated. The analysis of the results show
that, due to compressive creep strains of concrete, longitudinal steel compressive strains can
exceed the yield strain. This implies that the steel modulus that collaborates in the effective
stiffness EI of the hybrid column could not be the elastic modulus but should rather be the secant
modulus which varies with concrete creep. Moreover, for slender columns, plastification in the
compression zone of the steel section may not develop before instability. Hence, the secant
modulus of steel should be a function of the creep coefficient φef and the geometric slenderness
λ. For higher values of the creep coefficient, the value of the secant modulus of the steel section
will be rather lower. However, for higher values of slenderness this modulus will be higher.
Therefore, in addition to the previous cases already analyzed, we need to investigate the effect
of the steel yield stress on the ultimate load of hybrid columns. The parametric study with 1140
data sets presented previously is then extended to 2960 including different yield stress of steel
section (S235 and S460 added).
New expressions for β and the correction factors (Ks, Ka, Kc) involved in the definition of the
effective flexural stiffness EI are proposed. The proposed simplified method based on moment
magnification approach is summarized in the following.
The second-order bending moment, which has to be compared with the plastic bending moment
Mpl,N,Rd corresponding to the applied normal force NEd, is obtained by multiplying the first-order
bending moment by a factor:

(3) k
NEd
1
Ncr
where   0.6rm  0.4  0.4 , and Ncr is the critical buckling load of the column calculated using
the expression of effective flexural stiffness as follow:
(4) EI  Kc Ecd Ic  Ks Es Is  KaEaIa
Two versions of the correction factors has been proposed. The first version was proposed by the
same authors in (Keo, Somja, Nguyen, & Hjiaj, 2015) where the factors used in Eurocode 2 were
adopted for the concrete and the steel rebars expressed as follow:
Kc  k1k2 / (1  ef ); Ks  1
fck  N
k1  ; k2  n  0.2; n  Ed
20 170 N pl ,Rd
The calibration factor for steel profiles in the first version is:
0,0124
f 
0,76  y 
Ka   fck  1
1  105ef exp( 0,078 )
This first version of Ka is calibrated by adopting Kc and Ks of Eurocode 2 to give a good accuracy
in estimating the ultimate load of the hybrid columns. However, it does not give a mechanical
sense when considering the stiffness contribution of steel shape sections. Comparing to rebars
contribution with the factor Ks equal to 1, steel shape sections provide less contribution to the
flexural stiffness since the factor Ka is inferior to 1. For that reason, the second version of
correction factors is proposed in this paper.
Since the behavior of rebars and steel profile used in hybrid columns is similar, they should share
the same contribution. Thus, the same correction factor for steel profile and rebars is proposed in
the second version as follow:
0.450 0.147
Kc  k1k 2 / (1  ef ); K s  K a  K sa  1
1  1.433ef exp  0.027 
fck  N
k1  ; k2  n  0.2; n  Ed
20 170 N pl ,Rd
To give a global view of the accuracy of the proposed second version of formulations, the
frequency histogram shown in Figure 10 was constructed.
Fig. 10- Histogram of frequency of proposed formulations
Mean value of the distribution “r” and standard deviation “s” are 0.9988 and 0.0445,
respectively. The percentage of R below 0.97 is 11.69%. Its overall variability gives a good
estimation of the mean value of the ultimate load with relatively small deviation. The mean value
“r” and the standard deviation “s” provided by the proposed second version of simplified methods
are respectively equal to 0.9988 and 0.0445 which is as accurate as the first version of
formulations.
CONCLUSION
The assessment of the bending moment magnification method proposed in EC2 and EC4 when
applied to hybrid columns subjected to combined axial load and uniaxial bending moment have
been performed. To do so, a FE model has been developed in which the geometrical/material
nonlinearities as well as the partial interaction effect between the steel profiles and the
surrounding concrete are taken into account. The FE model has been validated by comparing its
predictions against experimental results for standard composite columns. To thoroughly analyze
the applicability of EC2 and EC4 variants of the moment magnification method to hybrid columns,
an extensive parametric study with 1140 data sets (cross-sections, column slenderness and creep
coefficient) has been carried out. The comparison between the results obtained with Eurocode
simplified methods and with FE analyzes shows that simplified method of EC2 and EC4 leads to
a wide scatter where half of case-studies are unsafe. It means that the proposed effective flexural
stiffness (EI) of EC2 and EC4, which involves in determining the magnification factor, are not
appropriate for hybrid column design. It was observed that the secant modulus of compressed
part of the steel section varies as a function of the creep coefficient φef and the geometric
slenderness λ. Therefore, in addition to the previous cases already analyzed to evaluate EC2 and
EC4 simplified method, a further investigation of the effect of the steel yield stress on the ultimate
load of hybrid columns has been carried out. This later investigation leaded to an extensive
numerical parametric study with 2960 data sets that serves as a database for developing a
simplified method for hybrid column design. New expressions for the correction factors (for the
determination of effective flexural stiffness (EI)) were proposed in order to take into account the
creep effect and the effect of plastification of the steel profiles. Two versions of the correction
factor Ka for steel shape sections has been proposed. The first version of Ka is calibrated by
adopting Kc and Ks of Eurocode 2 to give a good accuracy in estimating the ultimate load of the
hybrid columns. However, it does not give a mechanical sense when considering the stiffness
contribution of steel shape sections. Comparing to rebars contribution with the factor Ks equal to
1, steel shape sections provide less contribution to the flexural stiffness of the column since the
factor Ka is inferior to 1. For that reason, the second version of correction factors is proposed in
this paper. The same correction factor for steel profile and rebars is proposed in the second
version. The comparisons between the proposed second version of formulations and FE analyzes
show that the developed method provides the ultimate load for typical slender hybrid columns
with adequate accuracy.
ACKNOWLEDGMENTS
The authors gratefully acknowledge financial support by the European Commission (Research
Fund for Coal and Steel) through the project SMARTCOCO (SMART COmposite COmponents:
concrete structures reinforced by steel profiles) under grant agreement RFSRCT-2012-00039.
NOTATIONS
MEd,1 design first-order bending moment
MEd,2 design second-order bending moment
NEd design axial load
Npl,Rd design value of the plastic resistance of the cross-section to compressive normal force
Mpl,N,Rd design value of the plastic resistance moment of the cross-section taking into account
the compressive normal force
Ncr elastic critical normal force
l0 effective column length
β equivalent moment factor
k moment magnification factor
EI flexural stiffness of the compression member
Ec elastic modulus of concrete
Ea elastic modulus of structural steel
Es elastic modulus of reinforcement
Ic moment of inertia of concrete cross section
Ia moment of inertia of structural steel about the axis through the center of hybrid section
Is moment of inertia of longitudinal reinforcing bars about the axis through the center of
hybrid section
Ecd design modulus of elasticity of concrete: Ecd = Ecm/γcE
λ geometric slenderness ratio
ρ geometrical reinforcement ratio
REFERENCES
Al-Shahari, A., Hunaiti, Y., & Ghazaleh, B. (2003). Behavior of lightweight aggregate concrete-
encased composite columns. Steel and Composite Structures, 97–110.
Bonet, J., Romero, M., & Miguel, P. (2011). Effective flexural stiffness of slender reinforced
concrete columns under axial forces and biaxial bending. Engineering Structures, 881-
893.
Chen, C., & Yeh, S. (1996). Ultimate strength of concrete encased steel composite columns.
Proceedings of the third national conference on structural engineering, 2197-2206.
EC2-1-1. (2005). Calcul des structures en béton. Règles générales et règles pour les bâtiments.
EC4-1-1. (2005). Calcul des structures mixtes acier-béton. Règles générales et règles pour les
bâtiments.
Kent, D. C., & Park, R. (1971). Flexural members with confined concrete. Journal of the
Structural Division, 1969-1990.
Keo, P., Nguyen, Q.-H., Somja, H., & Hjiaj, M. (2015). Geometrically nonlinear analysis of hybrid
beam-column with several encased steel profiles in partial interaction. Engineering
Structures, 66-78.
Keo, P., Somja, H., Nguyen, Q.-H., & Hjiaj, M. (2015). Simplified design method for slender
hybrid columns. Journal of Constructional Steel Research, 101-120.
Morino, S. (1998). Recent developments in hybrid structures in Japan-research, design and
construction. Engineering structures, 336-346.
Morino, S., Matsui, C., & Watanabe, H. (1985). Strength of biaxially loaded SRC columns.
Composite and Mixed Construction, Proceedings of the U. S. /Japan Joint Seminar,
ASCE, 185-194.
Roeder, C. W. (1998). Overview of hybrid and composite systems for seismic design in the
United States. Engineering structures, 355-363.
Tikka, T. K., & Mirza, S. A. (2008). Effective flexural stiffness of slender structural concrete
columns. Canadian Journal of Civil Engineering, 384-399.
Westerberg, B. (2002). Slender column with uniaxial bending, International Federation for
Structural Concrete (fib). Technical report, bulletin 16. Design examples for 1996 FIP
recommendations practical design of structural concrete.
The seismic stability and ductility of steel columns
interacting with concrete footings
Hiroyuki Inamasu
Doctoral Assistant, Swiss Federal Institute of Technology (EPFL)
Lausanne, Switzerland
hiroyuki.inamasu@epfl.ch
Amit Kanvinde
Professor, University of California at Davis
Davis, United States of America
kanvinde@ucdavis.edu
Dimitrios Lignos
Associate Professor, Swiss Federal Institute of Technology (EPFL)
Lausanne, Switzerland
dimitrios.lignos@epfl.ch
ABSTRACT
First-story columns in steel moment resisting frames are subjected to various forms of geometric
instabilities during earthquakes. These instabilities have the potential to seriously compromise
structural performance, and expedite structural collapse. In turn, these instabilities are influenced
by the interactions of the column boundary conditions. Of specific concern are base connections
that involve embedding the column into a concrete footing. These connections are nominally
assumed to be fixed, although recent experimental data suggests that they could be significantly
flexible. A finite element parametric study examining the effect of connection flexibility on interior
column seismic performance is presented. Several key variables, including column base flexibility,
cross-section, axial load, and column length are interrogated. Results of the finite element study
are presented in support of developing quantitative relationships between column base conditions
and deformation capacity of the columns. Incorporation of these relationships into modeling
frameworks is discussed.
1. Introduction
A number of recent experimental and numerical studies investigated the hysteretic response of
wide flange steel columns in moment resisting frames (MRFs) under multi-axial cyclic loading
(Newell and Uang 2008; Suzuki and Lignos 2015; Uang et al. 2015; Elkady and Lignos 2015,
2016; Fogarty and El-Tawil 2016). In the studies by Elkady and Lignos (2016), the influence of
the column end boundary conditions on the member hysteretic response was highlighted. Local
and global geometric instabilities resulted into severe flexural strength deterioration and column
axial shortening. In all cases the column was fixed at its base, which is a common assumption for
columns embedded in a concrete footing.
Past experiments on embedded column base connections (Grilli et al. 2017) revealed that they
have an appreciable flexibility due to the elastic deformation of the concrete subject to bearing
stress, the base plate and the column embedment length. Notably, findings from reconnaissance
reports from past earthquakes (Clifton et al. 2011; MacRae et al. 2015) as well as system-level
numerical studies (Lignos et al. 2013) suggest that a flexible column base that may be partially
damaged could enhance the seismic behavior of steel MRFs.
Although the influence of the column base flexibility on the structural behavior has only been
investigated through nonlinear response history analyses (Zareian and Kanvinde 2013), its
influence on the column member hysteretic response has never been examined.
This paper investigates the effect of the embedded column base connection flexibility on the
column hysteretic behavior through rigorous nonlinear finite element (FE) simulations. A
continuum FE model is first developed and validated with recent full-scale column tests. The steel
columns utilize a variety of boundary conditions and cross-sectional characteristics. The FE model
is then extended such that the column base flexibility can be considered. This flexibility is inferred
on the basis of available full-scale tests on steel columns embedded on concrete footings. A
parametric study is then conducted in which several key variables, including the column base
flexibility, cross-section, axial load, and column length are interrogated. The influence of these
variables on the column hysteretic behavior is highlighted through a number of illustrative
examples.
2. Review of experimental data on wide flange steel columns and embedded columns in
concrete footings
2-1. Recent experiments on wide-flange steel columns
Uang et al. (2015) tested twenty-five deep columns subjected to inelastic cyclic loading coupled
with a constant compressive axial load and fixed end boundaries. They found that global and local
slenderness ratios as well as the compressive axial load influence the observed failure modes
that can vary from local to lateral torsional buckling.
Suzuki and Lignos (2015) examined the influence of the loading history on the seismic
performance of cantilever wide-flange columns. Tests on nominally identical columns subjected
to symmetric loading histories were complemented with monotonic and ‘collapse consistent’
loading histories that characterize the ratcheting behavior (Ibarra and Krawinkler 2005) a structure
experiences prior to earthquake-induced collapse. They found that routinely used symmetric
cyclic loading histories provide insufficient information regarding the deterioration characteristics
of steel columns that primarily control structural collapse.
More recently, Elkady and Lignos (2016, 2017a) investigated the influence of boundary conditions
on the seismic performance of interior wide flange steel columns with fixed-flexible boundary
conditions. Referring to Figure 1a, columns with a flexible top boundary exhibit larger plastic
deformation capacity compared to fixed end columns. The response of the latter is controlled by
column axial shortening as shown in Figure 1b.
Fixed top peak
Flexible top peak
Fixed top deform. capacity

Flexible top deform. capacity
(a) (b)
D Axial load, P D Axial load, P
Lateral Lateral
load (V) Di load (V)
Di
Inflection
L Inflection L point
point Li
Li
Chord rotation Chord rotation
=D/L =D/L
Mbase = VLi + PDi Mbase = VLi + PDi

Fixed column top Flexible column top
(c)
Figure 1 – Typical steel column hysteretic response and boundary conditions: (a) Moment-
rotation relation; (b) axial shortening-rotation, and (c) deformed shapes (data from Elkady and
Lignos 2016, 2017a).
2-2. Recent experiments on embedded column base connections

Grilli et al. (2017) recently investigated the seismic performance of embedded column base
connections subjected to axial load, P, and lateral drift demands. Columns were designed to be
elastic such that the observed failures were concentrated at the column base connection. Table
1 summarizes the main test parameters including the embedment depth, dembed, and the applied
axial load. The same table summarizes the measured maximum flexural strength, Mmax and
stiffness, b of the embedded column base connection (after subtracting the elastic contribution of
the column). All the columns were tested in a cantilever fashion. Figure 2 illustrates the hysteretic
response of a typical embedded column base connection. Table 1 as well as Figure 2 suggest
that an embedded column base connection has an appreciable column base flexibility. The
reported b values correspond to about 0.4 - 0.8% rotation under the design base moment.
It should also be noted that regardless of the column base flexibility, the specimens exhibited a
considerable plastic deformation capacity without any strength loss.
Table 1 – Summary of embedded column base connection test data (Grilli et al. 2017)
Column Column
Specimen P dembed Mmax b
cross- length
ID (kN) (mm) (kNm) (105kNm/rad)
section (m)
#1 W14x370 2.84 445 508 2613 3.23
#2 W18x311 2.84 445 508 2324 3.84
#3 W14x370 3.10 0 762 3741 3.07
#4 W14x370 3.10 445 762 4124 3.38
#5 W14x370 3.10 -667* 762 3800 3.25
* minus sign indicates tension
Axial
d
load (P)
Lateral
load (V)
Column drift
=d/z
M= V×z
Figure 2 – Representative hysteretic response of embedded column base connection (Grilli et

al. 2017)
3. Finite element model development and verification

In order to investigate the hysteretic response of steel columns embedded on column footings a
FE model should be developed. This is done in two steps. In the first step, the FE model is
validated with column specimens tested by Elkady and Lignos (2016, 2017a) that their base
boundary was nominally fixed. Table 2 summarizes key features of the employed test specimens
including the applied constant compressive axial load ratios (P is the applied axial load; Py = fyA;
fy is the expected yield stress; A is the column cross-section area) and the corresponding member
slenderness, Lb/ry (Lb and ry are the unbraced length and the radius of gyration about weak-axis
of the cross-section, respectively). Both fixed end and fix-flexible column specimens are
considered. The FE model is developed in the commercial finite element analysis software
ABAQUS (version 6.14-1) (ABAQUS 2014). Wide flange columns are modelled with shell
elements based on the modeling approach discussed in Elkady and Lignos (2015, 2017a, 2017b).
The material nonlinearity is considered with a combined isotropic/kinematic material hardening
law (Lemaitre and Chaboche 1994) and residual stresses due to hot-rolling. Local and member
geometric instabilities are triggered by assigning local and out-of-straightness geometric
imperfections within the ASTM (2003) and AISC (2010b) limits.
Figures 3 illustrates comparisons of the column base moment, Mbase, – chord rotation, q, relations
obtained from FE simulations and test results. The column base moment, Mbase, is the reaction
moment at the column bottom surface. The second order moment demand has been subtracted
in all cases. The chord rotation, q, is defined as the horizontal displacement at the top divided by
the actual column height, L by considering the corresponding column axial shortening if any. The
comparisons suggest that the developed FE model simulates relatively well the test results
throughout the entire loading history regardless of the employed cross-section and member end
boundary conditions. Therefore, the FE model is deemed to be reliable and is further extended to
assess the influence of the column base flexibility on the column hysteretic response.
Table 2 – Selected column specimens for the FE model development and verification (Elkady
and Lignos 2016, 2017a)
Specimen Cross Slenderness Axial load Boundary conditions EI/L*1
5
ID -section ratio, Lb/ry ratio, P/Py (Top-Bottom) (10 kNm/rad)
C1 W24x146 79 20% Fixed – Fixed 0.923
C2 W24x146 79 50% Fixed - Fixed 0.923
C3 W24x146 79 20% Flexible - Fixed 0.923
C7 W24x84 51 20% Flexible - Fixed 0.488
*1E: Young’s modulus of steel, and I: the second moment of inertia about strong axis
(a) (b)
(c) (d)
Figure 3 – Comparison between tests and FE simulations: (a) C1; (b) C2; (c) C3; and (d) C7.
(data from Elkady and Lignos 2016, 2017a)
4. Refinement of finite element model boundary conditions
Figure 4 shows schematically the boundary conditions of a typical interior steel MRF column. In
order to simulate the corresponding rotational stiffness attributed to the embedded column base
connection, a rotational spring is assigned at the column base. The elastic stiffness, b of this
spring is inferred based on the corresponding values of Table 1 (Grilli et al. 2017) by considering
notional column sizes corresponding to the assumed column base connection strength.
Figure 4 – Schematic representation of column end boundary conditions
Referring to Figure 4, three rotational springs are assigned at the column top to represent its
member top end boundary conditions. The spring assigned with respect to the y-axis
characterizes the in-plane rotational stiffness of the girders intersecting the column. A bilinear
hysteretic response is assigned to this spring to consider the inelastic behavior of the steel girders.
The rotational spring assigned with respect to the x-axis, represents the out-of-plane flexibility
due to the shear tab beam-to-column connection (Liu and Astaneh-Asl 2000, 2004) and the
torsional resistance of the girders. This spring is characterized by a tri-linear hysteretic behavior.
A third spring is assigned at the column top end with respect to the z-axis that represents the
torsional resistance due to bending of the shear tab plate, the torsional resistance of the 2nd story
column and the weak-axis bending resistance of the girders. The contributions of these three
components are lumped in a drift-dependent linear behavior that is idealized based on
experimental findings summarized in Zhang and Ricles (2006). The authors plan to conduct
further physical testing to characterize this relationship.
Prior experiments on beam-to-column connections that utilized deep columns and typical beams
with reduced beam sections (RBS) (Chi and Uang 2002; Zhang and Ricles 2006) indicated that
deep columns experience additional torque demands after the onset of severe local buckling
within the RBS region. These demands are considered in the present study. In particular, the
corresponding torque force – lateral drift relation is established on the basis of mechanics-based
equations and experimental observations (Chi and Uang 2002; Zhang and Ricles 2006; Lignos et
al. 2010; Lignos and Krawinkler 2012).
5. Parametric study to assess the effect of column base flexibility
A parametric study is conducted to evaluate the influence of the column base flexibility on the
column performance based on the refined FE model discussed in Section 4. In all cases it is
assumed that the concrete footing strength is sufficiently large such that the structural damage
always occurs in the steel column. Table 3 and Figure 5 provides a summary of the investigated
cases. Most of the employed cross-sections are highly ductile as per AISC-341-10 (AISC 2010a).
Note that the rotational stiffness, b corresponding to the one deduced from experiments is referred
as ‘real’ (Grilli et al. 2017). The AISC symmetric cyclic lateral loading protocol is employed in the
context of this study.
Table 3 – Variables in the parametric study

Column base Length Axial load ratio
Cross-section
rotational stiffness, b L P/Py
5 cases 3 cases 4 cases 18 cases
(Fixed, Real, 75%Real, 50%Real, Pin) (3.0, 4.5, 6.0m) (5, 20, 35, 50%) (see Figure 5)
Figure 5 – Selected steel column cross-sections for the parametric study
6. Results and Discussion

6-1. Qualitative assessment of representative cases
Figure 6 shows the FE analysis results obtained from representative cases examined (i.e.,
W24x146-L4.5m-P/Py=20%) with different column base flexibilities. Figure 6(a) shows a typical
column base moment - chord rotation relation. The emphasis on this paper is on the maximum
base moment, Mmax*, corresponding to the capping rotation, qmax*, and the post-capping plastic
rotation qpc* that characterizes the post-buckling behavior of a member. These parameters are
commonly used in component deterioration models employed in nonlinear frame analysis
(PEER/ATC 2010).
Figure 6(b) shows the influence of the assumed column base flexibility on the first-cycle envelope
curve of a steel column. From this figure, while the column base flexibility increases, the onset of
local buckling is delayed. Therefore, qmax* increases while the column base stiffness decreases.
Referring to Figure 6b, the rate of flexural strength deterioration, which is related to the local
buckling progression seems to be more or less the same regardless of the employed column base
stiffness. This is to be expected because the concrete footing is assumed to remain elastic in the
context of the present study. The maximum column base moment was found to be almost the
same regardless of the column base flexibility. Note that the torque force demand applied at the
column top did not practically influence the hysteretic response of the column for the considered
lateral loading protocol. This is attributed to the fact that the column strength mostly deteriorated
due to local buckling rather than lateral torsional buckling. FE simulations with asymmetric lateral
loading histories indicate that lateral torsional buckling is strongly influenced by the torque force
demand.
Mmax*
dv
q
qlast *
qmax * qpc *
Mbase
(a)
(b) (c)
Figure 6 – FE analysis results of representative specimens (W24x146-L4.5m-P/Py=20%): (a)
Typical moment-rotation relation, (b) first-cycle envelope, and (c) vertical displacement of the
column top
Figure 6(c) shows the influence of the column base flexibility on the vertical displacement of the
column top, dv (see Figure 6(a)), which is strongly related to the observed axial shortening once
local buckling occurs. dv is distinguished from axial shortening since the value of dv depends on
the column dimensions. The figure suggests that the vertical displacement in a ‘real’ case is
smaller than the nominally “fixed” column base case. For instance, at a reference chord rotation
of 2% radians representative of design basis seismic events, the column axial shortening
reduction is more than 50% compared to the “fixed” case. This is attributed to the fact that the
cumulative plastic rotation in flexible column bases is smaller than that of a fixed base case at a
given drift.
6-2. Capping rotation, qmax*
Figure 6(b) suggests that the capping rotation, qmax* increases while the elastic concrete footing
stiffness decreases. Prior research (Elkady and Lignos 2015) suggests that the pre-peak plastic
rotation, qp* (i.e., qp* = qmax* - qy) increases exponentially while the local and/or global slenderness
and/or the column axial load ratio decrease. Same observations hold true in the present study.
Figure 7 shows the qmax* versus h/tw relation for 4.5m long steel columns subjected to P/Py = 20%
with four different column base flexibilities. The FE simulations suggest that if the concrete footing
stiffness decreases then qmax* increases in most cases. This finding hold true regardless of the
member length and the applied axial load ratio. For all practical purposes, the qmax* increase
compared to the ideally “fixed” condition may be approximated as 0.5% radians for the ‘real’ case
and 1.0%rad for the ‘50% of real’ case.
Figure 7 – Capping plastic rotation, qmax* – web local slenderness ratio relation (L4.5m-
P/Py=20%)
6-3. Post-capping plastic rotation, qpc*

Figure 6(b) indicates that the post-capping plastic rotation, qpc* is practically not influenced by the
column base flexibility given that the column base strength is assumed to be infinite (i.e., elastic
concrete footing). Therefore, regardless of the employed column base flexibility, the predictive
equations proposed by Hartloper and Lignos (2017) may be adopted. Note that the deduced qpc*
values are in the range of 5-10% radians if P/Py =20% and 20 < h/tw < 45. At higher axial load
ratios and/or higher h/tw ratios, the qpc* values range between 2-5%radians.
6-4. Column axial shortening, daxial

In order to assess the column axial shortening among models with different length, the vertical
displacement at the column top of each model, dv, is subtracted by the vertical displacement of
same model with pinned column base, dv,pin. Therefore axial shortening, daxial* = dv - dv,pin.
Figure 8 shows the daxial* - h/tw relation for different column lengths and column base flexibilities.
In general, the observed axial shortening at a given rotation is slightly smaller for longer columns
because the yield rotation is larger due to the smaller column flexural stiffness. However, column
axial shortening is almost the same regardless of the column length. The reason is that column
axial shortening mainly depends on the onset of local buckling, which in turn depends on the
employed cross-sectional properties.
Referring to Figure 8(b) the column axial shortening is shown at characteristic drift demands.
Notably, at a drift of 2% the column axial shortening is minor if the column base flexibility is
considered. In that respect, the column-base ‘fixed’ assumption is fairly conservative. From the
same figure, if a 50% reduction of the real column base flexibility is assumed, the corresponding
axial shortening at a lateral drift of 2% is almost zero in most of the cases. Similarly, if a 50%
reduction of the real column base flexibility is assumed, the corresponding axial shortening
reduction is, 80% and 30% at a lateral drift of 2% and 4%, respectively.
(a) (b)
Figure 8 – Column axial shortening – web local slenderness ratio relation:(a) P/Py=20% and b
= Fixed and (b) P/Py=20% and L = 4.5m
7. Summary and Conclusions

This paper investigated the influence of the column base flexibility on the hysteretic behavior of
first story interior wide flange steel columns typically seen in moment-resisting frames (MRFs).
This was achieved through a parametric study that was conducted with rigorous finite element
(FE) simulations. The parameters that were investigated include the column base flexibility, the
applied axial load ratio and the geometric properties of the employed cross-section. The
developed FE model was validated with representative test results from full-scale steel column
experiments. The boundary conditions of the FE model were further refined to reflect
representative column base connection flexibilities inferred from recent full-scale embedded
column base tests. The main findings are summarized as follows,
1. The corresponding capping rotation, qmax* increases while the elastic stiffness of the
column base foundation decreases. This increase is on average, 0.5% radians when the
‘real’ column base flexibility is considered relative to the nominally fixed-base case. The
corresponding, qmax* increases by 1.0%rad if the ‘real’ column base flexibility is reduced
by 50%. It was also found that the influence of the column base flexibility on the post-
capping rotation, qpc* was practically negligible.
2. The column axial shortening is overestimated if the embedded column base flexibility is
neglected. In particular, the FE simulation results suggest that the column axial shortening
reduction relative to the theoretically ‘fixed’ base can be more than 30% at a reference
lateral drift demand of 4%.
References
ABAQUS. (2014). ABAQUS analysis user’s manual version 6.14-1. Dassault Systems Simulia
Corp., RI, USA.
AISC. (2010a). Seismic Provisions for Structural Steel Buildings. ANSI/AISC-341-10. American
Institute of Steel Construction.
AISC. (2010b). Specification for Structural Steel Buildings. ANSI/AISC 360-10. American Institute
of Steel Construction.
ASTM. (2003). Standard Specification for General Requirements for Rolled Structural Steel Bars,
Plates, Shapes, and Sheet Piling. American Society for Testing and Materials.
Chi, B., and Uang, C. M. (2002). Cyclic Response and Design Recommendations of Reduced
Beam Section Moment Connections with Deep Columns. Journal of Structural
Engineering, 128(4), 464–473.
Clifton, G. C., Bruneau, M., MacRae, G. A., Leon, R., and Fussell, A. (2011). Steel Structures
Damage from the Christchurch Earthquake Series of 2010 and 2011. Bulletin of the New
Zealand Society for Earthqauke Engineering, 44(4), 297–318.
Elkady, A., and Lignos, D. G. (2015). Analytical investigation of the cyclic behavior and plastic
hinge formation in deep wide-flange steel beam-columns. Bulletin of Earthquake
Engineering, 13(4), 1097–1118.
Elkady, A., and Lignos, D. G. (2016). Dynamic Stability of Deep and Slender Wide-Flange Steel
Columns - Full Scale Experiments. Proceedings of the Annual Stability Conference
Structural Stability Research Council.
Elkady, A., and Lignos, D. G. (2017a). Full-Scale Cyclic Testing of Deep Slender Wide-Flange
Steel Beam-Columns under Unidirectional and Bidirectional Lateral Drift Demands. 16th
World Conference on Earthquake Engineering (16WCEE), Santiago, Chile, num. 944.
Elkady, A., and Lignos, D. G. (2017b). Stability Requirements of Deep Steel Wide-Flange
Columns under Cyclic Loading. Proceedings of the Annual Stability Conference Structural
Stability Research Council, San Antonio, Texas.
Fogarty, J., and El-Tawil, S. (2016). Collapse Resistance of Steel Columns under Combined Axial
and Lateral Loading. Journal of Structural Engineering, 142(1), 4015091.
Grilli, D. A., Jones, R., and Kanvinde, A. M. (2017). Seismic Performance of Embedded Column
Base Connections Subjected to Axial and Lateral Loads. Journal of Structural
Engineering.
Hartloper, A., and Lignos, D. G. (2017). Updates to the ASCE-41-13 Provisions for the Nonlinear
Modeling of Steel Wide-Flange Columns for Performance-Based Earthquake Engineering.
EUROSTEEL 2017, Copenhagen, Denmark.
Ibarra, L. F., and Krawinkler, H. (2005). Global Collapse of Frame Structures under Seismic
Excitations. John A. Blume Earthquake Engineering Center Technical Report 152.
Stanford Digital Repository. Available at http://purl.stanford.edu/dj885ym2486
Lemaitre, J., and Chaboche, J.-L. (1994). Mechanics of solid materials. Cambridge university
press.
Lignos, D. G., Hikino, T., Matsuoka, Y., and Nakashima, M. (2013). Collapse Assessment of Steel
Moment Frames Based on E-Defense Full-Scale Shake Table Collapse Tests. Journal of
Structural Engineering, 139(1), 120–132.
Lignos, D. G., Kolios, D., and Miranda, E. (2010). Fragility assessment of reduced beam section
moment connections. Journal of Structural Engineering, 136(9), 1140–1150.
Lignos, D. G., and Krawinkler, H. (2012). Development and utilization of structural component
databases for performance-based earthquake engineering. Journal of Structural
Engineering, 139(8), 1382–1394.
Liu, J., and Astaneh-Asl, A. (2000). Cyclic Testing of Simple Connections Including Effects of
Slab. Journal of Structural Engineering, 126(1), 32–39.
Liu, J., and Astaneh-Asl, A. (2004). Moment–Rotation Parameters for Composite Shear Tab
Connections. Journal of Structural Engineering, 130(9), 1371–1380.
MacRae, G. A., Clifton, G. C., Bruneau, M., Kanvinde, A. M., and Gardiner, S. (2015). Lessons
from Steel Structures in Christchurch Earthquakes. Proceedings of the 8th International
Conference on Behavior of Steel Structures in Seismic Areas (STESSA), Shanghai,
China.
Newell, J. D., and Uang, C.-M. (2008). Cyclic Behavior of Steel Wide-Flange Columns Subjected
to Large Drift. Journal of Structural Engineering, 134(8), 1334–1342.
PEER/ATC. (2010). Modeling and acceptance criteria for seismic design and analysis of tall
buildings, PEER/ATC 72–1. Prepared for Pacific Earthquake Engineering Research
Center (PEER) by Applied Technology Council (ATC), Redwood City, San Diego.
Suzuki, Y., and Lignos, D. G. (2015). Large Scale Collapse Experiments of Wide Flange Steel
Beam-Columns. Proceedings of the 8th International Conference on Behavior of Steel
Structures in Seismic Areas (STESSA), Shanghai, China.
Uang, C. M., Ozkula, G., and Harris, J. (2015). Observations from cyclic tests on deep, slender
wide-flange structural. Proceedings of the Annual Stability Conference Structural Stability
Research Council, Tennessee, Nashville., 247–263.
Zareian, F., and Kanvinde, A. (2013). Effect of Column-Base Flexibility on the Seismic Response
and Safety of Steel Moment-Resisting Frames. Earthquake Spectra, 29(4), 1537–1559.
Zhang, X., and Ricles, J. M. (2006). Experimental Evaluation of Reduced Beam Section
Connections to Deep Columns. Journal of Structural Engineering, 132(3), 346–357.
Performance based evaluation of a multiple CFT bridge pier with
buckling restrained braces
Chi-Ho Jeon
Chung-Ang University, Dept. of Civil and Env. Engineering
Seoul, Korea
chihobeer@cau.ac.kr
Dong-Wook Kim
Seoul, Korea
chihobeer@cau.ac.kr
Chang-Su Shim
Seoul, Korea
csshim@cau.ac.kr
ABSTRACT
In recent construction industry, accelerated bridge construction method has been risen as a
main issue especially in urban construction. As an alternative of typical columns for
accelerated bridge construction method, a pier of four concrete filled steel tube (CFT)
columns with braces has been introduced. Experiments of the multiple CFT columns with
buckling restrained braces (BRB) have been conducted using static and quasi-static load to
investigate structural behavior. In this paper, fiber element analysis has been simulated to
evaluate the performance of the CFT columns. The OpenSEES software is used for detailed
modeling of piers and both of geometrical and material nonlinearities and behavior of
connections are considered at them. A spring element in OpenSEES was used to express
buckling restrained brace details. According to required performance criteria, design
parameters of the CFT bridge pier with BRBs were determined.
1. INTRODUCTION
The CFT bridge pier was developed from the accelerated bridge concept. Unlike the typical
RC substructures, this bridge pier consists of four CFT columns and braces, and it could be
modularized and is designed with standardized commercial products as shown in Figure 1.
In the previous research, Shim et at. (2014) found that low cycle fatigue failure was occurred
between brace and CFT column at the connection of small-scale test. The author, therefore,
performed tests on four different brace connections, and proposed enhanced connection
detail. Kim et al. (2016a) conducted full-scale test of CFT bridge pier with the enhanced
connection detail and Buckling Restrained Brace (BRB). The author verified that the
enhanced connection detail and BRB improved structural performance of the CFT bridge
pier. Also, he suggested plastic-hinge analysis concept to the structure. Kim et al. (2016b)
also conducted analytical study on the CFT bridge pier according to directivity of load. It was
found that stress concentration at the bracing joint could be prevented by BRBs, and
displacement ductility of the structure was satisfied.
In this paper, the CFT bridge pier was evaluated by performance based concept according
to the brace details.
Fig. 1 Schematic of the CFT bridge pier
2. Experimental program
2.1 Test specimen

In the previous research (Kim et al, 2016a), a bridge pier with four CFT columns and braces
was designed to investigate the performance under static and cyclic loading conditions. The
height of specimen was 7,950 mm, and the four CFT columns had the dimension of Φ508 x
16 x 5,800 with distance of 2,500 mm and 2,000mm between columns in strong and weak
axes, respectively. The dimension of pier cap was 5,000 mm x 3,000 mm x 1,900 mm. In the
pier cap, 2-I407x428x20x35 beams were embedded on the top of columns in strong axis.
The footing had the dimension of 4,800 mm x 4,200 mm x 1,200 mm and embedded length
of columns in the footing was 1.5 times(770 mm) of diameter of the CFT section. The design
compressive strength of concrete was 40 MPa, and measured strength on the day of
experiment was 42 MPa. The yield and tensile strength of CFT columns and braces were
315 MPa and 490 MPa, respectively.
The experiments were planned to consider two types of braces (Figure 2), and both were
used without CFT columns change in sequence; after the first test was conducted with
circular hollow steel tube (CHST) braces under cyclic loading in elastic range, the second
test was conducted with buckling restrained brace (BRB) up to drift ratio of 2.76. The
experiments followed loading program in the previous research (Kim et al, 2016a).
1.992
500 992 500
Splice plate 16t
800
Φ216.3x7t
Gusset plate 16t
(a) Circular hollow steel tube brace for first test
1.992
446 1.100 446
180 1.632 180

Splice plate 16t
C Φ267.4x6t
660
180
C Bracing 16t
Gusset plate 16t
(b) Buckling restrained brace for second test
Fig. 2 Brace details
3. Buckling restrained brace
3.1 Introduction
In the extreme event limit state design, it is important to provide enough energy dissipation
against to earthquake for the safety of structures. The buckling restrained braces (BRBs)
were introduced for those objectives, initially used in frame structures (Roeder an Popov,
1977). The design concept that BRBs have a role of sacrificial members to dissipate the
seismic energy, while other main components maintain their integrity is known as the
structural fuse concept.
The typical BRBs consist of steel core-plate and surrounding steel tube casing filled with
grout or concrete as shown in Figure 3(a) (Ei-Bahey and Bruneau, 2011). Other steel core
sections such as single plate or multiple plates could be used. Figure 3(b) shows the
comparison of behavior in compression between typical buckling brace and BRB. Due to the
buckling behavior, the core area of typical buckling brace is designed for seismic load based
on the critical buckling stress of the section, while BRB section is sized by yield stress. This
advantage is reflected in the larger response modification coefficient, R, assigned to BRBFs
(R=8) by ASCE 2010. In contrast, R values for Ordinary Concentrically Braced Frame
(OCBF) and Special Concentrically Braced Frames (SCBFs) are 3 1/4 and 6, respectively.
(a) (b)
Fig. 3 (a) Typical BRB assembly; (b) Buckling versus buckling restrained brace behavior
( Ei-Bahey and Bruneau, 2011)
3.2 Research on bridge bents with BRBs

As the recent seismic design standard provides more strict provisions for important bridges,
several existing bridges which followed old design standard can no longer guarantee seismic
safety. Therefore, many researchers have tried to use BRBs to bridge bents, especially, in a
retrofitting manner, because of its structural advantage and maintenance benefit. Bazaez
and Dusicka (2015) introduced idealized response when BRBs are used to RC bent as
shown in Figure 4. The BRB needs to be designed in such way to reduce the demands on
the bare bent, therefore, the bare bent would respond elastically under the 500-year and
1000-year event. However, the increased forces in the bent system due to BRBs, sometimes,
have the demands in the concrete elements exceed their capacity. For that reason, minor
inelastic behavior is allowed to occur for the 1000-year event. Xiaone and Bruneau (2016)
conducted case study on application of BRB to bridge bents. The author intended full elastic
column response for conservative design objective by limiting demands in the columns to
their yield flexural strength. This study showed at least 50% of drift reductions with increase
in base shear demands no greater than 20%. Wang et al. (2016) conducted also a case
study using a three-span RC box girder bridge. In the study, it was found that BRBs reduces
column bent drift ratio up to 75% and prevents deck unseating. Also, the BRBs redistribute
the stiffness of bridge bents and abutments, which results in dissipation of shear demand
from 6 to 15%. The authors conclude that BRBs prevent or delay the occurrence of column
plastic hinge from an evidence of the concrete and reinforcement strains in the columns
were reduced.
Fig. 4 Idealized pushover response (Bazaez and Dusicka 2015)
4. Performance based evaluation using analysis
4.1 Definition of performance level of CFT columns.

The researchers introduced above referred to the stress or strain condition of columns when
BRBs were installed. However, they did not suggest the performance level for the columns
with the BRBs. It is necessary to consider performance base design so that the columns give
proper capacity to their signifiance. The performance based design has four performance
levels; fully operational, delayed operational (which can be classified into immediate
occupancy with short-term restoration and life safety and long-term restoration), and
collapse prevention. The structural fuse concept is valid up to delayed operational level,
however, the performance of the columns according to previous design process (AISC 2010)
could be less safe level because section of BRBs are decided from the load acting on the
BRBs and limited deformation.
In order to evaluate the CFT bridge pier according to the performance based design, the
performance level of CFT columns should be defined. The definition of performance level of
CFT columns is shown in Table 1. For the fully operational level, column should be in elastic
range after seismic excitation, which can be expressed by ε < ε y (ε y : yield strain of steel
tube). For delayed operational level, it could be defined as the CFT columns are in status
from yield to buckling of steel tube. The collapse prevention level, therefore, from strain at
buckling to fracture strain. The behavior of CFT columns varies with not only aspect ratio
(L/D) and material properties, which are similar to typical RC columns, but also axial load
ratio (N/N u ) and diameter to thickness ratio (D/t). Thus, buckling behavior could not be
defined by only strain of steel tube. SAISHO et al. (2004) conducted a research on local
buckling behavior of CFT columns considering the factors above. Figure 5 plots one of the
buckling behaviors of CFT column in the research under the condition of L/D = 10.0,
concrete compressive strength σ c = 30 & 120 MPa, and steel tube tensile strength σ u = 440
MPa. The considered value in this study was N/N u = 0.3, σ c = 30 MPa, which cause the
value of Φ lb /Φ u = 9. In which, N u = σ c A c + σ u A s , Φ lb = deformation of CFT column when the
steel tube buckled locally, Φ u = upper bound of elastic deformation (M e /K 0 , M e : elastic
bending moment right before yielding, K 0 : elastic bending stiffness)
Table 1 Qualitative and quantitative performance levels

CFT column Qualitative Quantitative
performance level performance description Performance description
No cracks
Fully operational 0 < ε < εy
No inelastic deformation
Initiation of inelastic deformation

Delayed operational ε y < ε < ε lb
No buckling
Occurrence of buckling
Collapse prevention ε lb < ε < ε u
No failure
Fig. 5 Plastic deformation until CFT column buckles locally with aspect ratio L/D=10.0)
(SAISHO et al., 2004)
4.2 Verification of analysis model

In the previous research (Jeon et al. 2015), the single CFT model was verified using
OpenSEES analysis program as shown in Figure 6(a). The analysis model was compared
with pure bending test specimen, and the simulated strength, initial stiffness, and reduction
of stiffness were well corresponded with test result. Figure 6(b) shows comparison of push
over test result from CFT bridge pier and cyclic analysis result modeled with verified CFT
column model and BRB model. The difference was around 6 % in elastic range and an
average of 12 % in inelastic range. The simulation of analysis model showed similar
behavior to the full-scale test such as plastic hinge location and stress distribution
3000
Opensees
analysis
Experim 2500
ent
Moment (kN-m)
2000
Load (kN)
1500
1000
500 Experiment
Envelope curve (analysis)
0
0 20 40 60 80 100 120 140
θ(Deg.) Displacement(mm)
(a) (b)
Fig. 6 Verification of analysis model for (a) single CFT column and (b) CFT bridge pier
4.3 Analysis result

In order to clearly identify the behavior of the braces, analysis model was revised from test
specimen as shown in Figure 7. The braces were inclined to behave effectively in lateral
load, and two types of braces were compared; one is steel tube brace which has dimension
of Φ216.3x4t, and another one is BRB with same dimension of the test.
CFT
Φ508×16t Steel tube
or BRB
Fig. 7 The revised analysis model
The buckling behavior of steel tube braces was simulated using strength deterioration
constitutive model (Yongtao BAI et al. 2012). According to the model, the buckling strain of
steel tube (Φ216.3x4t) is ε lb = 0.022 and CFT (Φ508x16t) is 0.021, respectively. Figure 8(a)
shows comparison of load-displacement curves according to the brace types. At the drift
ratio of 1.4%, buckling of steel tube brace was occurred. The steel tube reached buckling
strain. As shown in Figure 8(b), both analysis models could not show the buckling strain of
CFT column because section of CFT was class 1, therefore, two models were judged to
have fully operational performance level.
4000
3000
2000
Steel tube 1000

Load (kN)
Buckling
0
-150 -100 -50 0 50 100 150
-1000
-2000
-3000 Steel tube

BRB
-4000
Displacement(mm)
(a)
0.012
Steel tube
0.01 BRB
0.008
0.006
Strain
0.004
0.002
0
-150 -100 -50 0 50 100 150
-0.002
Displacement(mm)
(b)
Fig. 8 Analysis result of (a) Load-displacement curve and (b) strain development of CFT
at plastic hinge
For the further research, it is necessary to simulate different size of BRB and CFT columns
up to higher drift ratio for application of performance based design because the analysis
model above showed the behavior up to 2.1 due to convergence problem. Also, it is
expected that delayed operational performance level of CFT columns will be occurred in
higher drift ratio according to the BRB section.
5. Conclusion
1) In the previous research, a multiple CFT bridge pier was tested under cyclic loading,
and evaluated according to two brace types; steel tube and BRB
2) An analysis model was built in a previous research for the multiple bridge pier, and
verification work have been done.
3) The performance based evaluation for the multiple CFT bridge pier was conducted
using the analysis model based on performance level definition.
4) The analysis showed fully operational performance level of the multiple CFT bridge pier
when BRB and steel tube braces were used.
5) For the performance based design of the multiple bridge pier, it is required to conduct
up to higher drift ratio and to confirm strain level of CFT column whether it reaches
buckling strain
6. References
AISC (2010). Seismic provisions for structural steel buildings, (ANSI/AISC 341-10),
American Institute of Steel Construction, Chicago, IL.
Bazaez R, Dusicka P. (2015). Design implementation of buckling restrained braces for
seismic retrofitting of reinforced concrete multi-column bridge bents. In: Structures congress.
Portland, OR.
C. H. Jeon, D. W. Kim, and C. S. Shim (2015). Cyclic behavior of multiple modular bridge
piers in terms of diagonal load analyzed by OpenSEES, Proceedings of 11th International
Conference on Advances in Steel and Concrete Composite Structures (pp.775-781),
Tsinghua University, Beijing, China.
Dongwook Kim, Chiho Jeon, and Changsu Shim (2016a), Cyclic and static behaviors of CFT
modular bridge pier with enhanced bracings, Steel and Composite Structures, 20(6), 1221-
1236, DOI: 10.12989/scs.2016.20.6.1221
Dongwook Kim, Chiho Jeon, and Changsu Shim (2016b), Analytical evaluation of a modular
CFT bridge pier according to directivity, Steel and Composite Structures, 20(6), 1193-1203,
DIO: 10.12989/scs.2016.20.6.1193
Kersting RA, Fahnestock LA, López WA. Seismic Design of Steel Buckling-Restrained
Braced Frames. NIST GCR 15–917-342015.
Motoo SAISHO, Masahiro KATO, and Shuyun GAO (2004). LOCAL BUCKLING OF CFT-
COLUMN UNDER SEISMIC LOAD, Proceedings of 11th World Conference on Earthquake
Engineering (pp. 2614), B.C., Canada
S. El-Bahey, M. Bruneau. (2011) Buckling restrained braces as structural fuses for the
seismic retrofit of reinforced concrete bridge bents. Engineering Structures, 33, 1052–1061
W. Xiaone, M. Bruneau. (2016) Case Study on Applications of Structural Fuses in Bridge
Bents. Journal of Bridge Engineering, 21(7),
Yongtao BAI, Akihiko KAWANO, Keita ODAWARA, and Shintaro MATSUO (2012),
CONSTITUTIVE MODELS FOR HOLLOW STEEL TUBES AND CONCRETE FILLED
STEEL TUBES CONSIDERING THE STRENGTH DETERIIORATION, J. Struct. Constr. Eng
AIJ., 77(677), 1141-1150
Yuandong Wang, Luis Ibarra, Chris Pantelides. (2016). Seismic Retrofit of a Three-Span RC
Bridge with Buckling-Restrained Braces. Journal of Bridge Engineering, 21(11).
Seismic Performance of an Innovative Composite Column with
Replaceable Steel Slit Dampers: FEM Analysis and Design Method

Xiamen, China
Yifan Liao
Postgraduate of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
Zixiong Guo
Xiamen, China
Yingting Lu
Ph.D. Candidate of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
Xiaojuan Liu
Lecturer of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: liuxjty@hqu.edu.cn
Bahram M. Shahrooz
a
b
Xiamen, China
E-mail: shahrobm@ucmail.uc.edu
Abstract
Over the last few decades, industrialization of building construction and rapid restoration of
functionality of structures after earthquakes have emerged as important issues facing the
structural engineering profession. An innovative composite column with replaceable steel slit
dampers has been developed for prefabricated reinforced concrete (RC) frame structures and
reinforced concrete-steel (RCS) hybrid frame structures. This new composite column, which can
be prefabricated, consists of several steel slit dampers, connection plates, and high strength
bolts. The column was designed based on “strong column-weak damper” philosophy. The
influences of different parameters, such as axial compression ratio, height/depth ratio, and
replaceable details, on the cyclic behavior of this new structural system were studied using FEM
method. The results suggested very good cyclic behavior. Plastic deformation was found to be
concentrated only in the steel slit dampers under large drift angles, while the other major
components stayed nearly elastic. Hence, the columns could easily be repaired. Based on the
numerical simulations, the seismic behavior of this new composite column was further
investigated, and a method for computing its flexural capacity was proposed.
INTRODUCTION
The post-earthquake repair cost of a building and the interruption in the use of the building
cause major economic losses. In order to minimize economic losses, more attention has been
paid to rapid post-earthquake repair techniques, e.g., by concentrating all the inelastic
deformations in replaceable components while keeping the other components in the elastic
range. After earthquakes, damaged components can be replaced with an equivalent one in
order to restore the building function. At present, a variety of replaceable components, such as
replaceable coupling beams and replaceable beam ends, have been developed. An innovative
“fuse” steel coupling beam developed at the University of Cincinnati [Fortney, et al. 2007] allows
post-damage repair/replacement and minimizes the repairing expense. A bolted end-plate
connection was recommended to permit post-damage repair without costly repair of the wall
pier-coupling beam connections. The viability of this system was demonstrated experimentally,
and was subsequently validated through additional tests [Farsi, et al. 2016]. An integrated
analytical and large-scale experimental study was conducted at the University of Toronto to
extend and validate the concept of replaceable links in steel braced frames [Shen, et al. 2011]. It
was found that end-plate links exhibited greater energy dissipation capacity than bolted web
links, although latter achieved significantly higher rotational capacity. Steel slit dampers were
regarded as effective “fuse” elements. Research showed that steel slit dampers had a good
energy dissipation capacity, and its mechanical properties could be predicted well [Saffari, et al.
2013. Climent, et al. 1998]. Recently, steel slit dampers have also been used in beam-column
connections. A steel slit damper system was developed to prevent formation of damage in
conventional columns and beams by dissipating lateral loads at the beam-column connections
[Köken and Köroğlu 2015]. Experimental and analytical results showed that plastic deformations
were concentrated in the damper, and damage to the beam and column was prevented.
During a high intensity earthquake, columns are expected to sustain significant damage at their
base. Due to gravity axial force, post-earthquake repair of the column base is very difficult,
costly, and could be impractical. A post-tensioned (PT) column base connection was designed
to eliminate structural damage due to seismic loads at column bases in self-centering moment
resisting frames (SC-MRFs) [Chi and Liu 2012]. The softening behavior at the connection was
provided by gap opening and elongation of PT bars rather than yielding in the column. Both
experimental and analytical studies were carried out to investigate the cyclic response of PT
column base connections.
Because of its low yield point, good energy dissipation, and stable performance, mild steel has
been used in the development of effective replaceable components. In line with these research
trends, an innovative composite column base with replaceable steel slit dampers was proposed
and evaluated by the authors.
OVERVIEW OF THE REPLACEABLE COMPOSITE COLUMN BASE
A schematic configuration of the innovative composite column base is shown in Figure 1. On

each side of the column base, two fuses made of mild steel plate with slits are placed between
the rocking column base and a fixed T-shape steel plate (or a T stub) by connecting plate I and
connecting plate II, respectively. Over a certain height from the base, the steel tube column is
filled with concrete to prevent local buckling. In conventional construction, the column base is
rigidly connected to the foundation. Therefore, the earthquake energy is dissipated mainly
through plastic deformations near the column base. By contrast, the new composite column
base is designed as a rocking element in which the flexural capacity is provided by steel slit
dampers, designed according to the “strong column-weak damper” philosophy. Steel slit
dampers introduced at locations are expected to experience inelastic deformations. The other
components are then designed to remain elastic under the force corresponding to yielding of the
damper. Because the inelastic deformation is concentrated within the damper, damaged links
can be quickly inspected and replaced following a major earthquake.
Fig. 1 - Construction details of the proposed composite column
In an effort to evaluate the behavior of the proposed replaceable composite column base,
numerical analyses were conducted by using the finite element analysis program
ABAQUS(Hibbit, K. 2014). The objectives of these analyses were to investigate (1) the
damage distribution, (2) the hysteretic behavior, and (c) the lateral loading capacity of the
replaceable column base.
FE MODELS
The cross section of the column was a 400 mm square tubular with a wall thickness of 10 mm.
The following parameters were considered in the analyses: (1) installation location of steel slit
dampers: only parallel to the loading plane (represented with “S”), or both parallel and
perpendicular to the loading plane (represented with “D”); (2) thickness of steel slit dampers (t =
15, 20, 30 mm); (3) column height-to-depth ratio (= 3, 5.5); and (4) ratio of the applied axial
compressive load to design axial load capacity (n = 0.15, 0.4). For example, S20-3-0.15
represents a specimen with dampers installed only parallel to the loading plane, steel slit
damper thickness of 20 mm, height/depth = 3, and applied axial compressive load = 0.15 times
the design axial load capacity. All the replaceable steel slit dampers had identical dimensions
and details (shown in Figure 2) but had different thicknesses (15, 20, 30 mm).
Fig. 2 - Dimension details of the replaceable link
The finite element model of the composite column was shown in Figure 3. Steel slit dampers
were modeled using 20-node quadratic brick, reduced integration elements (C3D20R). The
stiffeners welded to the T-shape plate were modeled with 10-node modified quadratic
tetrahedron elements (C3D10M). The other components (the steel tube column, the infilled
concrete, connecting plates, T-shape plate and the foundation) were modeled by 8-node linear
brick, reduced integration, hourglass controlled elements (C3D8R).
Tie constraints simulating complete bond were applied to the interface between different
components, ignoring the relative slip between two interfacial surfaces. The interaction between
the bottom face of the steel tube column and the top surface of the reinforced concrete
foundation was modeled with a “hard” contact, using a friction coefficient of 0.6 [Rabbat and
Russell 1985]. The bottom face of the foundation was fixed. A constant axial compressive load
and cyclic lateral displacement with increasing amplitudes were applied at the top of the column.
One cycle was performed at the following drift amplitudes: 1/500, 1/250, 1/150, 1/100, 1/75,
1/50, 1/25, and 1/20.
P Mo
de
ling
by the
“Ti bol
e” t
co con
ns ne
tra cti
in on
“Hard” contact on s
the column base
Element type
C3D20R
Z Element type
C3D10M
Y X Fixed end
Fig. 3 - Finite element model
A concrete damage plasticity model with compressive constitutive relationship [Kent and Park
1971] was adopted for infilled concrete. Linear tensile stress-strain relationship was assumed
for concrete [Nguyen and Kim 2009]. After reaching the peak tensile strength, concrete tensile
stress declined linearly to 0 at the ultimate tensile strain 0.005. The compressive strength and
peak tensile strength were taken as 26.8 MPa and 2.39 MPa, respectively. A value of 32,500
MPa was used as the concrete modulus of elasticity. The steel was assumed to exhibit elastic-
perfectly plastic behavior. The elastic modulus of the steel was taken as 200000 MPa. The yield
strengths of the steel slit damper and other steel components were 235 MPa and 345 MPa,
respectively. The foundation was assumed to remain elastic with an elastic modulus of 32500
MPa.
VERIFICATION OF NUMERICAL MODEL
The results from a number of relevant past experimental studies were utilized to verify the
aforementioned FE modeling technique. The studies examined steel slit dampers similar to that
used in the reported research. One research program (Chan and Albermani, 2008) focused on
the dampers only, and another study (Köken and Köroğlu, 2015) evaluated beam-column
connections with dampers attached to the beam. The test results and numerical results were
compared in Figure 4. Figure 4 (a) showed some differences between the calculated and
measured curves. As seen from Figure 4 (b), the numerical result of specimen N15 agrees well
with the experimental result, while the numerical curve of specimen N12, which had a thinner
damper than specimen N15, shows a larger initial stiffness and a smaller maximum strength
than the experimental result. Overall, the numerical simulations can reasonably replicate the
test results.
(a) Steel slit damper (b) Beam-to-column connection
Fig. 4 - Comparison between test and numerical results
SIMULATION OF COMPOSITE COLUMNS: RESULTS AND DISCUSSION
The plastic strain contour of a typical specimen S15-3-0.15 at 1/20 drift ratio was shown in
Figure 5. The plastic deformation was concentrated in the slit dampers while the other
components remained nearly elastic.
Fig. 5 - Plastic strain contour
Hysteretic curves and yield points of the replaceable composite columns were provided in
Figure 6, which showed a similar trend for all the cases. The composite column base was
characterized by gap opening and closing at the column-foundation interface. In the initial stage,
the specimen remained elastic with a linear relationship between the horizontal load and
horizontal displacement. When the moment capacity of the steel slit dampers and the
counteracting moment from the applied axial load (which had an eccentricity with respect to the
point at which the column rotated when it was uplifted) were exceeded, the column base
decompressed and gap opening began to occur. Subsequently, struts in the slit dampers (see
Figure 2) began to yield. As the displacement was increased, more struts yielded before
developing the maximum strength. In spite of significant pinching, the hysteretic loop was stable.
Additional dampers (the “D” series) installed perpendicular to the loading plane improved the
performance of the composite column but did not significantly affect the shape of hysteresis
loop; appreciable pinching still occurred.
(a) (b) (c)
(d) (e) (f)

Fig. 6 - Hysteresis curves
The maximum strength versus damper thickness was plotted in Figure 7. Obviously, the
maximum strength almost linearly increased as the damper thickness increased. It should be
noted that the thickness needs to be limited to ensure the replaceable dampers are the energy
dissipation mechanism. Larger axial compressive load suppressed the level of column uplift,
which enhanced the lateral strength of the composite column. For specimens with identical
column height/depth, the axial compressive load ratio had little influence on the slope of curves.
As shown later (Eq. 6) the damper thickness and level of axial compressive load independently
contributed towards the strength. As the aspect ratio increased, the lateral strength decreased
significantly.
Skeleton curves were shown in Figure 8. Before gap opening, the horizontal displacement was
mainly composed of the elastic deformation of the column. Therefore, the initial stiffness did not
noticeably change as the damper thickness and axial compressive load were increased, but
decreased as the column height/depth ratio became lager.
Fig. 7 - Maximum strength versus damper thickness relationship
(a) (b) (c)

Fig. 8 - Skeleton curves
Steel slit dampers installed in the column base were subjected to combined cyclic tension and
shear force, which is more severe than conventional steel plate dampers under pure shear.
Numerical simulation of these two different loading conditions were conducted to a series of
FEM analyses. The results indicated that the ultimate load carrying capacity was not influenced
by the differences in the loading. However, Figure 9 clearly indicated that strength degradation
of slit dampers in specimen S15-3-0.15 (solid line) was appreciably more severe than a
comparable damper subjected to pure shear only (dashed line).
The shear yielding strength of steel slit dampers in the composite column base was computed
based on Eqs. (1) and (2) [Oh, Kim, Ryu 2009]. The first term in Eq. (1) implies yielding is
governed by flexure, whereas the second term is based on yielding in shear.
 σ y tB 2 2σ y tB 
Fq, s  min k ,k  (1)
 2L ' 3 3 
L '  L  2r 2 / LT (2)
Where k is the number of struts; t is thickness of damper; y is yield stress of steel; B is width of
struts; L′ is the equivalent height indicated by Eq. (2).
As evident from Figure 9, the capacity computed from Eqs. (1) and (2) is a reasonable
conservative estimate of the values obtained from the FEM analysis. The FEM capacity is at
most 1.24 times larger than the predicted capacity from Eqs. (1) and (2).
Fig. 9 - Vertical force versus shear angle relationship
DESIGN METHOD
With reference to Figure 10, a design methodology was developed. The nominal yield flexural
strength My of specimen can be calculated based on Eq. (3).
M y  Md, v  Md, h (3)
Where
Md,v is the flexural strength provided by vertical shear forces of the steel slit dampers.
Md,h is the flexural strength provided by horizontal tension and compression force of the struts of
the dampers due to the rotational deformation of the column foot.
Md,v is calculated from Eq. (4).
Md, v   Fq, s xi (4)
Where
xi is the horizontal distance from the centroid of struts of the damper to the rotation point
indicated in Figure 10.
Md,h can be calculated based on Eq. (5).

Md, h    nσ
j 1,2
h, j tydy (5)
Where
n is the number of dampers at one side of column. σh,j is the normal stress of the steel slit
dampers. The subscript j of σh,j indicates each side of the column has a set of steel slit dampers.
The integral variable y and the coordinate system were both shown in Figure 10. Determination
of σh,j needs to account for the elastic lateral stiffness of the T-shape plates, but it will complicate
the calculation process. For design purposes, Md,h was ignored, leading to a conservative
estimation of the flexural strength. Based on the rotational equilibrium of forces shown in Figure
10 with respect to the rotation point, we can figure out the design horizontal force of the
composite column as follows:
N(h  Δ)   Fq, s xi
Pdesign  2 (6)
H
Where
Δis the hypothetical horizontal yielding displacement which is approximately assumed to be
0.01H, based on the results of FEM analyses.
H is the height of contra flexural point of the composite column.
Fig. 10 – Design model
The horizontal yielding load calculated from Eq. (6) was compared against that from FEM
analysis in Figure 11. The average value and coefficient of variation of Pdesign/PFEM was 0.70 and
0.01, respectively. Therefore, Eq. (6) can conservative be used. Further studies are needed on
the contribution of Md,v.
CONCLUSIONS
This paper proposed an innovative composite column, which could be replaced in-situ after a
major earthquake. The finite element model of the column was established, verified and used to
investigate the load transferring mechanism and seismic performance of the column. The
following conclusions can be drawn based on this research:
Fig. 11 - Comparison of calculate result with FEM
1) Designed based on “strong column-weak damper” philosophy, the innovative composite

column showed stable hysteresis behavior.
2) Steel slit dampers were the main source of energy dissipation. Plastic deformation was
found to be concentrated in the dampers; hence, the damaged components may easily be
replaced after an earthquake.
3) The flexural strength of the composite column depends on the initial axial force and damper
thickness. For the same height/depth ratio, the yield moment My increased as the initial
axial force and damper thickness became larger.
4) A design method was developed to calculate the flexural strength of columns with
replaceable steel slit dampers. Based on nonlinear FEM simulations, the method was found
to produce a reasonable and conservative estimate of the strength.
ACKNOWLEDGMENTS
Ph.D. Candidate H. Chen from Huaqiao University gave a great help on the modification of the
pictures in this paper. The author sincerely appreciates his help. The research was funded by
the National Natural Science Foundation of China (Grant No. 51578254) and the National
Natural Science Foundation of China (Grant No. 51378228), which is greatly acknowledged.
REFERENCE
Fortney, P. J., Shahrooz, B. M., Rassati, G. A. (2007). Large-scale testing of a replaceable “fuse”
steel coupling beam. Journal of Structural Engineering, 133(12): 1801-1807.
Farsi, A., Keshavarzi, F., Pouladi, P., et al. (2016). Experimental study of a replaceable steel
coupling beam with an end-plate connection. Journal of Constructional Steel Research, 122:
138-150.
Shen, Y., Christopoulos, C., Mansour, N., et al. (2011). Seismic design and performance of steel
moment-resisting frames with nonlinear replaceable links. Journal of Structural Engineering,
137(10): 1107-1117.
Saffari, H., Hedayat, A. A., Nejad, M. P. (2013). Post-Northridge connections with slit dampers to
enhance strength and ductility. Journal of Constructional Steel Research, 80(1): 138-152.
Climent, A. B., Oh, S. H., Akiyama, H. (1998). Ultimate energy absorption capacity of slit-type
steel plates subjected to shear deformation. Bull. math. statist, 28: 115-126.
Köken, A., & Köroğlu, M. A. (2015). An experimental study on beam to column connections of
steel frame structures with steel slit dampers. Journal of Performance of Constructed Facilities,
29.
Chi, H., & Liu, J. (2012). Seismic behavior of post-tensioned column base for steel self-
centering moment resisting frame. Journal of Constructional Steel Research, 78(11): 117–130.
Hibbit, K. (2014). ABAQUS documentation, Version 6.14.
Guo, Z., Zhu, Q., Liu, Y., et al. (2012). Experimental study on seismic behavior of a new type of
prefabricated RC column-steel beam frame connections. Journal of Building Structures, 33(7):
98-105. (in Chinese)
Liu, Y., Guo, Z., Dai, J., et al. (2013). Experimental study on seismic behavior of prefabricated
RCS frame joints with different failure mechanisms. China Civil Engineering Journal, 46(3): 18-
28. (in Chinese)
Rabbat, B. G., & Russell, H. G. (1985). Friction coefficient of steel on concrete or grout. Journal
of Structural Engineering, 111(3): 505-515.
Kent, D. C., & Park, R. (1971). Flexural members with confined concrete. Journal of the
Structural Division, 97: 1969-1990.
Nguyen, H. T., & Kim, S. E. (2009). Finite element modeling of push-out tests for large stud
shear connectors. Journal of Constructional Steel Research, 65(10):1909-1920.
Chan, R. W. K., & Albermani, F. (2008). Experimental study of steel slit damper for passive
energy dissipation. Engineering Structures, 30(4): 1058-1066.
Oh, S. H., Kim, Y. J., Ryu, H. S. (2009). Seismic performance of steel structures with steel
dampers. Engineer Structures, 31(9): 1997-2008.
LIFE-CYCLE PERFORMANCE OF CONCRETE-FILLED STEEL
TUBULAR COLUMN SUBJECTED TO CORROSION AND IMPACT
Lin-Hai Han
Department of Civil Engineering, Tsinghua University, Beijing, 100084 China
lhhan@tsinghua.edu.cn
Chuan-Chuan Hou
Department of Civil Engineering, Tsinghua University, Beijing, 100084 China
houcc@tsinghua.edu.cn
ABSTRACT
A finite element analysis (FEA) model was established to simulate the life-cycle performance of
a concrete-filled steel tubular (CFST) column under the coupling effects of progressive and
shock degradations. The imposed effects on the composite column covered a wide range of
factors that may deteriorate its mechanical performance during the entire life-cycle, including
construction issues (preload, residual stress, and imperfection), sustained load (axial load from
upper structure), progressive degradation (chloride corrosion) and shock degradation (impact
load). The numerical model was stage-by-stage verified with experimental results and showed
good accuracy. It was then employed to analyze the performance of CFST column under the
coupled degradations. Particularly, the effect of the coupled effects on the remaining capacity of
the column was evaluated, which highlights the importance of the life-cycle design and
assessment for the CFST columns.
INTRODUCTION
It is well known that life-cycle analysis (LCA) of structures, including design, performance
assessment, management and maintenance, has been widely addressed in the past several
decades (Biondini & Frangopol, 2016; Frangopol, 2011; Frangopol, Saydam, & Kim, 2012). One
key aspect of the LCA is to assess the deterioration of structural capacity under various types of
damage mechanisms, such as environmental aggressiveness, operating actions, and extreme
hazards. These damage actions can be generally subdivided into two categories based on the
relative length of their durations, namely progressive (continuous) degradation and shock
(sudden extreme) degradation (Sanchez-Silva, Klutke, & Rosowsky, 2011).
The modeling of the structural degradation process is to provide the information of the time-
variant remaining capacity of the structures. Lots of efforts have been devoted to the
establishment of such models, some of which were reviewed in Frangopol, Saydam, & Kim
(2012) and Sánchez-Silva & Klutke (2016). For instance, several researchers conducted LCA
analysis of the seismic resilience of corroded RC structures. The effects of corrosion on the
material properties of reinforcement and concrete were defined and the prediction of seismic
resilience of corroded RC structures was carried out by some simplified single degree of
freedom system (Akiyama, Frangopol, & Matsuzaki, 2011), beam element model with corrosion
effect incorporated (Biondini, Camnasio, & Titi, 2015), or fiber-based model with OpenSees
software package (Yuan, Guo, & Li, 2017).
In China, concrete-filled steel tubular (CFST) has been widely used as major structural
components in large-scale structures in the past several decades, such as columns in high-rise
buildings, subway stations, electricity pylons, and piers, arches and deck systems in bridges
(Han, Li, & Bjorhovde, 2014c). Many of the constructions are located at relatively harsh
environments with high level of humidity or chloride, such as the bridges across rivers and the
Zhoushan electricity pylon near the East China Sea. These structures inevitably suffer from
progressive degradation during their service life; on the other hand, the risk of extreme hazards
like earthquake or impact from moving vehicles, vessels or floating ice cannot be ignored. The
performance and design of CFST structures under these progressive and shock degradations
have been separately addressed in the design codes as well as research works in the past
(Han, Hou, & Wang, 2012; Hou, Han, & Zhao, 2013; Han, Hou, & Wang, 2014a; Hou, Han,
Wang, & Hou, 2016; Bambach, Jama, Zhao, & Grzebieta, 2008; Deng, Tuan, & Xiao, 2012;
Wang, Han, & Hou, 2013; Han, Hou, Zhao, & Rasmussen, 2014b). However, no significant
attention has been paid to the life-cycle performance of CFST under the combination of such
progressive and shock degradations.
This paper is thus an attempt to study the life-cycle performance of CFST subjected to
progressive and shock degradations. A finite element analysis (FEA) model is established to
simulate the structural behaviour with various types of actions incorporated. The FEA model
was verified with experimental data and used to predict the capacity deterioration of the
structure during its life-cycle. It was then employed to analyze the coupling effects of the
progressive and shock degradations on the mechanical performance of the CFST structures,
particularly the remaining capacity.
ESTABLISHMENT OF THE FEA MODEL

Example structure and loading path
A typical CFST column located in the corrosive environment is employed as the example
structure for the life-cycle analysis, as shown in Figure 1(a), where N0 and Nu standard for the
axial sustained load and ultimate axial strength of the column. Due to the harsh environment,
progressive corrosion occurs at the outside surface of the steel tube. At some time point during
its life-cycle, the column might be impacted by a moving object, bringing permanent lateral
deformation to the column. A detailed explanation of the loading path is presented below.
Axial
compression (N)
D (0, t3, Nu)
N0 Nu
N0
N0
Nu ④
A ② B (0, t1, N0) C (0, t3, N0)
(0, 0, N0) N0
③
Impact ①
O Time (t)
t1 t2 t3
Corroded P
(Fp, t2, N0)
Fp
① ② ③ ④ Impact force (F)
(a) Actions on the column (b) Loading path

Fig. 1 – A schematic view of the loading path for a CFST column
As illustrated in Figure 1(b), the lateral impact force (F), time (t) and axial compression (N) are
used as (x, y, z) coordinates for the loading path. Figure 1(b) treats the corrosion procedure and
impact procedure as time dependent parameters, while the short-term axial loading is seen as
independent of time.
Stage① (O-A): the column is axially compressed to a certain level (N0) to simulate the initial
operation loading state;
Stage② (A-B): the column is then under sustained compression and corroded. During a period
of t1 (several years to decades), the outer steel tube of the column will be gradually corroded
while the axial compression is kept constant;
Stage③ (B-P-C): the column is then impacted by a moving body, which can be described with
an impact force curve. In Figure 1(b), the impact starts from point B and there is usually a peak
point P in the curve. It then ends at point C and the impact duration is t3-t1, which is generally a
very short period (of order 102 milliseconds);
Stage④ (C-D): it is assumed that the column does not collapse after the impact. To evaluate
the residual strength of the column, it is axially compressed to failure in the final stage and the
obtained axially compressive strength of the column is Nu.
Degradation condition assumptions
In the current analysis, only uniform corrosion of the steel tube is considered and it is assumed
that uniform corrosion occurs over the full outside surface of the tube. Several models have
been developed to predict the corrosion rate of steel (Biondini et al. 2016). In the current
analysis, as uniform corrosion is assumed, the steel tube thickness loss ts instead of the
corrosion rate is directly used as the indicator of the corrosion severity. As a result, the
corrosion rate and the time duration of the corrosion are not discussed in the analysis.
In real life collision incidents, the moving object that impacts the column generally undergoes
plastic deformation and consumes part of the impact energy. For the simplification of the current
analysis, the object is assumed to be a rigid mass body with a mass as m0 and velocity as V0.
General description
In the past studies, FEA models for CFST members under various types of static or dynamic
loads have been developed. For example, Han et al. (2004) modeled the behavior of CFST
columns under long-term sustained load; Wang et al. (2013) and Han et al. (2014b) modeled
the behavior of CFST beams and columns under lateral impact; Han et al. (2014a) and Hou et
al. (2013) developed FEA models for CFST stub columns and beams under sustained load and
corrosion. In the above studies, the FEA models have been verified with experimental results
and proven to be reasonable. The FEA model with the combined actions, as a result, can be
established by assembling the above models with the different aforementioned actions
considered. As long as the modeling in each loading stage is reliable, the FEA model is robust.
Material properties
The outer steel tube is modeled with a five-stage elastic-plastic model, as has been used by
Han, Yao, & Tao (2007). During the impact process, the strain rate of the materials is relatively
high and the strain rate effect of steel needs to be accounted. The Cowper-Symonds model
(Abramowicz & Jones, 1984) is able to predict the dynamic yield strength of steel under
dynamic loading. It has been proved to be suitable in the simulation of CFST under drop
hammer impact by Wang et al. (2013) and Han et al. (2014b).
The core concrete is modeled with the uniaxial stress-strain model of core concrete presented in
Han et al. (2007), which is able to consider the confinement effect of the outer steel tube on the
core concrete. For circular CFST, the model is shown as follows:
 2 x  x2 x 1
 x
y x 1 (1)
 β0 x  1  x
 2
where, x=/0, y=/0, 0=fc′, 0=c+8000.2×10-6, c=(1300+12.5fc′)×10-6; fc′ is the cylinder

strength of concrete (unit: MPa) and =(Asfy)/(Acfck) is the confinement factor; As and Ac are the
sectional area of the circular steel tube and core concrete, respectively; fy and fck are the yield
strength of steel and characteristic strength of concrete, respectively. Coefficient 0 can be
calculated as:
7
β0  (2.36 105 )[0.25(ξ 0.5) ] ( f c)0.5  0.5  0.12 (2)
The long-term sustained load will influence the strain development of the core concrete due to
shrinkage and creep, while the corrosion will change the confinement of steel tube on the core
concrete. Both effects should be considered in the material model. The long-term effect of
concrete is considered by modifying the strain development of the uniaxial stress-strain curve
and the influence of corrosion is considered by using different confinement factors for concrete
in the initial state and corroded state. More details can be found in Han et al. (2004). The tensile
behavior of concrete is described with the model presented in Shen, Wang & Jiang (1993).
The strain rate effect of concrete should also be considered. After reviewing the various models
for strain rate effects on concrete strength, Wang et al. (2003) and Han et al. (2014b) chose the
model presented in Comite Euro-International du Beton (1993) for concrete under compression
and the model developed by Malvar & Ross (1998) for concrete under tension, as follows:
(d / s )1.026 d  30s1  (d / s ) d  1s1
f cd / f c   , f td / f t   (3)
  (d / s ) d  30s1  (d / s ) d  1s 1
1/ 3 1/ 3
where, fc′ is the static cylinder strength of concrete while fcd′ is the dynamic cylinder strength of
concrete under strain rate d . s is the strain rate of the static loading. The values for s , and
 can be found in Comite Euro-International du Beton (1993); ft and ftd are the tensile strength of
concrete under static loading and dynamic loading with strain rate as s and d , respectively.
The values for s ,  and  can be found in Malvar & Ross (1998).
Pre-load and initial imperfections

One merit of CFST construction is that the steel tube could be first installed and used as
formwork for the core concrete. So it is common that the steel tube bears pre-load from the
upper structure before the concrete part gains strength and works together with the steel tube.
The pre-load on the steel tube should be properly incorporated in the numerical model and can
be achieved by applying pre-load on the steel tube and then establishing the contact and bond
between the steel tube and core concrete (Li, Han, & Zhao, 2015).
Cold-forming and welding are common procedures for the fabrication of steel tubes. Residual
stresses would be induced due to the weldment and cold-forming processes while the former
might bring increase of yield strength to the corner area of the section. The distributions of
residual stresses of the typical square and circular steel tubes can be defined with the models
presented in Moen, Igusa, & Schafer (2008), Masubuchi (1980), and Gao, Usami, & Ge (1998).
Local geometrical imperfection of the steel tube is set to be the lowest buckling mode while bow
imperfection with amplitude as L0/1000 can be used to represent the global imperfection model
(Gardner & Nethercot, 2004).
Modeling strategies
Different numerical algorithms should be used for different loading types. Implicit algorithms,
such as the Newton method in Abaqus (Hibbitt, Karlsson & Sorensen Inc., 2005), are generally
used for solving static problems. So for the simulation of the initial loading stage, corrosion
stage, and the ultimate loading stage, the implicit algorithm is used, while for the simulation of
impact process with large kinetic energy, short period and complex contact behavior, the explicit
algorithm is a good choice.
As two different types of algorithms are used in the simulation and they are not compatible in
the same model, the results need to be transferred between them. Abaqus provides the
commands ‘*Restart’ and ‘*Import’ to transfer results between Standard and Explicit package in
the analysis. Regarding the loading path in Figure 1(b), the initial loading stage and corrosion
stage are simulated with Abaqus/Standard package. The analysis results are written to the
restart file with ‘*Restart’ command. Abaqus/Explicit package is then used to simulate the
impact loading stage. The stress, strain and deformation information of the initial loading stage
are imported into the current model with the command ‘*Import’ prior to the dynamic simulation.
When the impact analysis is over, the results are again written to a restart file, which is imported
into the ultimate loading model to predict the residual axially compressive strength of the CFST
column.
Contacts
The contact behavior between different geometrical parts in the FEA model should be well
considered. For the interaction between the steel tube and the core concrete, hard contact is
applied in the normal direction of the contact surfaces, which does not allow penetration through
the contact surfaces. Coulomb friction model is used in the tangential direction and the friction
coefficient is set to be 0.6. The bond strength between the steel tube and core concrete is
considered with the model developed in Han et al. (2007).
Another contact is the collision between the column and impact body during the impact
simulation. ‘Hard contact’ is applied in the normal direction of the contact surfaces and Coulomb
friction model is used in the tangential direction. The friction coefficient is set to be 0. ‘Surface to
surface contact’ algorithm is used in the implicit solution while ‘general contact’ algorithm is
used in the explicit solution.
The corrosion of the outer steel tube is modeled with the ‘*Model change’ command in the
software. The corroded part of the steel tube is deactivated in the corrosion step with the ‘Model
change’ command.
Elements and boundaries
8-node 3D solid elements with reduced integration are used to model both the outer steel tube
and core concrete. The impact rigid body is modeled with 4-node 3D rigid element, as shown in
Figure 2. Structural meshing technique is used to achieve uniformly distributed mesh style.
Higher mesh density is used in the impact area of the column as there is very complex contact
behavior and large deformation. Boundaries of the column are set to be simply supported.
Impact rigid body
CFST column Core concrete
Outer steel tube
Fig. 2 – FEA model for impact simulation
VERIFICATION OF THE FEA MODEL

The FEA model should be verified against experimentally measured data to test its validity.
Currently there are no experimental results with CFST columns that fully cover all the load
stages in the aforementioned FEA model. Alternatively, test data for CFST columns under those
separate loads or actions, such as long-term load, corrosion, and impact, are available. The
verification of the FEA model, as a result, is carried out stage-by-stage with the available
experimental results.
Hou et al. (2016) employed accelerated corrosion apparatus to carry out corrosion tests on
circular CFST beams and recorded the failure modes and bending moment (M)-lateral
deflection (u) curves for the specimens. Han et al. (2014b) carried out impact tests on CFST
beams with drop hammer apparatus and obtained failure modes and impact force (F) history of
the specimens. The experimental results in the aforementioned tests were used to stage-by-
stage verify the accuracy of the FEA model and representative comparisons of the measured
and predicted failure modes and force curves were presented in Figures 3 and 4, respectively.
(a) CFST beam under corrosion (b) CFST beam under impact
Fig. 3 – Comparison between the observed and predicted failure modes
60 800
600 Measured
40 Predicted
M/kN·m
F/kN
400
20 Measured 200
Predicted
0 0
0 5 10 15 20 25 30 0 0.01 0.02 0.03 0.04 0.05
u/mm t/s
(a) Bending moment (M) versus lateral displacement (u) (b) Impact force (F) curve
Fig. 4 – Comparison between experimental and numerical results
It shows that the FEA model is capable of accurately predicting the failure modes and force
curves of the tested specimens. Evaluation of some other parameters, such as the ultimate
strength of the corroded beams and the permanent deflection of the impacted beams also show
good agreement between measurements and predictions. It shows that the established FEA
model is sufficiently accurate for the prediction of the behaviour of CFST under the separate
actions and can be reasonably assumed to be capable of predicting the effects of the coupled
actions on CFST structures.
DESIGN OF NUMERICAL EXAMPLES AND ANALYSIS CASES

The established FEA model is used to do full-range analysis of circular CFST column under
sustained load, corrosion and impact. First, numerical examples are designed to be used as
representative cases for the analysis.
Basic parameters of the example
A CFST column example is used to shown the analytical procedure. The basic parameters of
the FEA model for the example are set as follow: D0×ts×L0=400×9.3×4000 mm, where D0 and ts
are the sectional diameter and thickness of the steel tube, and L0 is the length of the column;
the yield strength of steel (fy) is set to be 345 MPa and the cubic strength of concrete is set as
60 MPa. Elastic modulus of steel (Es) and concrete (Ec) are set as 201 GPa and 36.5 GPa,
respectively.
The sustained axial compression N0 is determined through the parameter axial loading ratio (n),
which is defined as:
N0
n (4)
N u0
where N0 is the sustained axial compression, as shown in Figure 1(a) and Nu0 is the ultimate
axially compressive strength of the column under short-term static loading. The axial loading
ratio depends on the design load level in the engineering structures. In the analysis, n is set as
0.4. For the simplification of analysis, it is assumed that the sustained load is applied on the
whole CFST section, so the pre-load effect of steel tube needs not to be considered in the
current example.
Han et al. (2004) showed that the long-term deformation of CFST will stabilize after about 100
days. As a result, the corrosion duration (t1) is set to be 120 days. To better observe the
influence of corrosion, a relatively large corrosion depth (ts) as 0.25ts is used in the
simulations.
The impact energy is another important parameter to be designed. As has been studied in
Wang et al. (2013) and Han et al. (2014b), impact will bring lateral deflection to the column, the
value of which will increase with increasing impact energy. It is reasonable to conclude that the
column will collapse due to the impact if the impact energy is large enough and there should be
a critical impact energy for the axially compressed column, beyond which the column will be
damaged under the impact. The calculation of the critical impact energy can be a future topic for
the analysis. In the current research scope it is intended to analyze the influence of corrosion
and impact on the full-range behavior as well as residual axially compressive strength of the
CFST column. It is assumed that the composite column doesn’t fail during the impact stage.
As a result, several trial-and-error calculations were taken out to find an impact energy which is
under the critical energy for the current column. Finally, the mass of the impact rigid body (m0) is
set as 1000kg and the initial velocity is set as 10m/s2, resulting in an impact energy of 50kJ. The
impact location is set at the mid-span of the column. It was found that the column still stands
after the corrosion and impact stages.
Simulation cases
The above parameters are used in the modeling of the CFST column. To analyze the effect of
different actions on the structure behaviour, several different simulation cases are considered,
each with a different loading path:
Case 1: no sustained load, corrosion or impact load are considered. The column is statically
loaded to failure;
Case 2: the effect of sustained load and impact are considered. The column is first loaded to N0,
and then impacted at mid-span;
Case 3: the effect of sustained load and corrosion are considered. The column is first loaded to
N0, and then is corroded;
Case 4: the coupling effect of sustained load, corrosion and impact are considered. The column
is first loaded to N0, and then is corroded. After the corrosion stage, the column is impacted by
the rigid body at mid-span.
Failure modes
The failure mode of the CFST column in Case 4 is analyzed. The deformed CFST column in
each loading stage is presented in Figure 5, in which S33 stands for the longitudinal stress in
the steel tube (in Pa). The corresponding loading stages can be found in Figure 1.
Stage① Stage② Stage③ Stage④
Fig. 5 – Deformed CFST at the end of each loading stage
It shows that the steel tube is under tension and compression in the initial loading stage①,
which is due to the existence of tensile residual stress in the steel section. The compressive
longitudinal stress then has a 10%-20% increment while the tensile stress has a decrease of the
similar degree in the corrosion stage②. The reason is that 25% of the tube thickness is
corroded and the load carried by the corroded part was gradually transferred to the remaining
section, resulting in higher longitudinal stress.
The column is then impacted at mid-span from the left direction in stage③. It can be seen that
there is obvious lateral deflection (u0) in the column and the stress level is much higher at this
stage, especially around the impact area. Another effect of impact is that very clear bending
appears around the mid-span area. The left side of the section is under compression while the
right side is under tension.
The bending moment obviously comes from the second order effect of the axial compression:
as there is relatively large lateral deflection in the column, the axial compression will bring an
extra bending moment at the mid-span of the column, which can be estimated as M0=N0×u0,
where u0 is the lateral deflection of the column brought by impact. This is the coupling effect
between the axial compression and the lateral impact. The coupling of axial compression and
bending brings much higher axial stress level in the column, as can be seen in the figure. It can
be expected that if the axial compression or the impact energy is large enough so the moment
brought by the second order effect (M0) is high enough, the column might be damaged at this
stage of loading. This second order effect will be further presented in the next section.
The column is axially loaded to failure in stage④ of the modeling. It shows that the lateral
deflection of the beam kept increasing as it was axially compressed. The column is under
compression and bending in the final state and the failure mode is global buckling.
Axial compression (N) versus displacement () curves
The axial compression (N) versus displacement () curves of the four simulation cases are
illustrated in Figure 6. Note that the fluctuation of the axial compression in the impact stage due
to the vibration of the column has been eliminated in the curves for clearer presentation of the
results. The curves are calculated and analyzed stage-by-stage.
10000 1
D1
D3
7500 0.75
D2 D4
Nu/ Nu0
N/kN
5000 0.5
A Case1
B C′ C Case2
2500 0.25
Case3
Case4
0 0
0 5 10 15 20 25 Case1 Case2 Case3 Case4
/mm
Fig. 6 – Axial compression (N) versus axial displacement () Fig. 7 – Ultimate strength ratios
The four curves shown in Figure 6 coincide with each other in the initial loading stage① (O-A),
in which the columns are generally under elastic deformation. The curves then diverge
dramatically in the following stages.
For the column under short-term loading (Case 1), it was kept loaded till failure. The curve goes
from elastic stage to plastic and hardening stages, and then begins to descend after the peak
point. The axially compressive strength of the column is obtained from point D1 as 9484kN.
In the other three cases, the axial compression (N0) was kept constant while corrosion and/or
impact developed. It shows that the axial displacement values of the columns keep increasing
while the axial compression is constant. For the column under corrosion (Case 3, range A-B),
the development of  at this stage is 0.4mm, while for the column under impact (Case 2, range
A-C′), the development is 3.6mm. It shows that increment of  under impact is much larger than
that under corrosion. The reason is that the axial displacement of the columns is caused by
different mechanisms. For the corroded column (Case 3), the axial displacement is due to the
axial strain development, which is the result of the axial stress increment, as shown in Figure 5.
While for the impacted column (Case 2), the axial displacement increment is mainly attributed to
the lateral deflection of the column brought by the impact, as can be understood from Figure 5.
For the column under corrosion and impact (Case 4, range A-B-C), the axial displacement
increment is dramatically larger than that of the other cases. It is the result of the coupling effect
of the sustained load, corrosion and impact. Han et al. (2014b) showed that the impact
resistance of CFST will decrease with decreasing steel ratio (=As/Ac). As corrosion reduced the
steel ratio and thus reduced the impact resistance of the CFST column, the lateral deflection of
the column will increase compared with the no-corrosion case. Simulation results showed that
the lateral deflection brought by impact is 46.5mm for Case 2 while it is 64.0mm for Case 4. As
the axial displacement of the column increases with increasing lateral deflection, a larger axial
displacement is observed in the column of Case 4.
The ratios of the ultimate strength of the columns (Nu) to the ultimate strength of the column
under short-term loading (Nu0) are presented in Figure 7. It shows that corrosion and impact
cause different ratios of strength deterioration to the columns. Corrosion causes 9% reduction of
the strength while impact causes 25% reduction. For the column under combined corrosion and
impact, the strength reduction ratio is 44%, which is even larger than the summation of the other
two cases (36%). The coupling of the two degradations causes higher deterioration to the
column strength compared with the simple summation of deterioration ratios brought by the
individual actions.
This phenomenon can be explained by the second order effect of the axial compression: the
corrosion reduces the axial loading capacity as well as impact resistance of the column; the
lateral deflection of the corroded column brought by impact will be larger under the same impact
energy; larger lateral deflection means larger mid-span bending moment due to the second
order effect. FEA results show that the impact on the corroded column (Case 4, Point C) brings
a much larger bending moment than the impact on the no-corrosion column (Case 2, Point C).
And the axially compressive strength of the column decreases with increasing bending moment
level.
The above analysis shows that the coupling of sustained load, corrosion and lateral impact
brings reduction to the ultimate axially compressive strength of the CFST column. It can be
summarized as below:
(1) Corrosion will reduce the static as well as lateral impact resistance of the CFST column.
(2) As the impact resistance of the column is reduced, a larger lateral deflection occurs in the
column compared with the no-corrosion situation under the same amount of impact energy.
(3) A higher bending moment will be generated at the impact location of the column due to the
second order effect of the axial compression.
(4) The remaining axial strength of the column will be largely reduced due to the reduction of
section steel ratio as well as the second order effect of the axial compression. The analysis
shows that the reduction percentage of ultimate strength is higher than the summation of
reduction percentages brought by the individual actions.
CONCLUDING REMARKS
This paper is an attempt to analyze the life-cycle performance of CFST columns under the
coupled degradations of chloride corrosion and impact. FEA model was established to simulate
the full-range behavior of circular CFST columns under the coupling of sustained load, corrosion
and impact loading. In the model, the pre-load on steel tube, residual stresses and initial
imperfection as well as the effects of several different types of actions were taken into account.
Experimental data were employed to stage-by-stage verify the FEA model and reasonably good
results were achieved. Subsequently, the models were used to do full-range analysis of circular
CFST columns under the coupled actions. It shows that the coupling of sustained load,
corrosion and impact brings notable deterioration to the remaining capacity of the circular CFST
column. Meanwhile, the second order effect of the axial compression may intensify this
deterioration. It demonstrates the significant influence of load history on the remaining capacity
of deteriorated structures.
FEA modeling has been used in the present study to analyze the life-cycle performance CFST
columns. In the next stage of the project, experimental study and further mechanical analysis,
such as the quantitate relationship between the corrosion depth and the critical impact energy of
corroded CFST columns, as well as simplified methods for the remaining life prediction of CFST
structures, will be the future research interests.
ACKNOWLEDGEMENTS
The research reported in the paper is part of the Project 51378290 and supported by the
National Natural Science Foundation of China (NSFC) and Suzhou-Tsinghua Initiative Program
(2016SZ0212). The financial support is highly appreciated.
REFERENCES
Abramowicz, W., & Jones, N. (1984). Dynamic axial crushing of square tubes. International Journal of
Impact Engineering, 2(2), 179-208. doi:http://dx.doi.org/10.1016/0734-743X(84)90005-8
Akiyama, M., Frangopol, D. M., & Matsuzaki, H. (2011). Life-cycle reliability of RC bridge piers under
seismic and airborne chloride hazards. Earthquake Engineering & Structural Dynamics, 40(15),
1671-1687. doi:10.1002/eqe.1108
Bambach, M. R., Jama, H., Zhao, X. L., & Grzebieta, R. H. (2008). Hollow and concrete filled steel hollow
sections under transverse impact loads. Engineering Structures, 30(10), 2859-2870.
doi:https://doi.org/10.1016/j.engstruct.2008.04.003
Biondini, F., Camnasio, E., & Titi, A. (2015). Seismic resilience of concrete structures under corrosion.
Earthquake Engineering & Structural Dynamics, 44(14), 2445-2466. doi:10.1002/eqe.2591
Biondini, F., & Frangopol, D. M. (2016). Life-cycle performance of deteriorating structural systems under
uncertainty: review. Journal of Structural Engineering-ASCE, 142(9). doi:10.1061/(asce)st.1943-
541x.0001544
Comite Euro-International du Beton. (1993). CEB-FIP Model Code 1990: Design Code: T. Telford.
Deng, Y., Tuan, C. Y., & Xiao, Y. (2012). Flexural behavior of concrete-filled circular steel tubes under
high-strain rate impact loading. Journal of Structural Engineering-ASCE, 138(3), 449-456.
doi:10.1061/(asce)st.1943-541x.0000464
Frangopol, D. M. (2011). Life-cycle performance, management, and optimisation of structural systems
under uncertainty: accomplishments and challenges. Structure and Infrastructure Engineering,
7(6), 389-413. doi:10.1080/15732471003594427
Frangopol, D. M., Saydam, D., & Kim, S. (2012). Maintenance, management, life-cycle design and
performance of structures and infrastructures: a brief review. Structure and Infrastructure
Engineering, 8(1), 1-25. doi:10.1080/15732479.2011.628962
Gao, S. B., Usami, T., & Ge, H. B. (1998). Ductility of steel short cylinders in compression and bending.
Journal of Engineering Mechanics-ASCE, 124(2), 176-183. doi:10.1061/(asce)0733-
9399(1998)124:2(176)
Gardner, L., & Nethercot, D. A. (2004). Numerical modeling of stainless steel structural components - a
consistent approach. Journal of Structural Engineering-ASCE, 130(10), 1586-1601.
doi:10.1061/(asce)0733-9445(2004)130:10(1586)
Han, L. H., Hou, C., & Wang, Q. L. (2012). Square concrete filled steel tubular (CFST) members under
loading and chloride corrosion: experiments. Journal of Constructional Steel Research, 71, 11-25.
doi:10.1016/j.jcsr.2011.11.012
Han, L. H., Hou, C. C., & Wang, Q. L. (2014a). Behavior of circular CFST stub columns under sustained
load and chloride corrosion. Journal of Constructional Steel Research, 103, 23-36.
doi:10.1016/j.jcsr2014.07.021
Han, L. H., Hou, C. C., Zhao, X. L., & Rasmussen, K. J. R. (2014b). Behaviour of high-strength concrete
filled steel tubes under transverse impact loading. Journal of Constructional Steel Research, 92,
25-39. doi:10.1016/j.jcsr.2013.09.003
Han, L. H., Li, W., & Bjorhovde, R. (2014c). Developments and advanced applications of concrete-filled
steel tubular (CFST) structures: members. Journal of Constructional Steel Research, 100, 211-
228. doi:10.1016/j.jcsr.2014.04.016
Han, L. H., Tao, Z., & Liu, W. (2004). Effects of sustained load on concrete-filled hollow structural steel
columns. Journal of Structural Engineering-ASCE, 130(9), 1392-1404. doi:10.1061/(asce)0733-
9445(2004)130:9(1392)
Han, L. H., Yao, G. H., & Tao, Z. (2007). Performance of concrete-filled thin-walled steel tubes under pure
torsion. Thin-Walled Structures, 45(1), 24-36. doi:10.1016/j.tws.2007.01.008
Hibbitt, Karlsson & Sorensen Inc. (2005). ABAQUS/Standard User’s Manual (6.5.1 ed.): New York: Rhode
Island.
Hou, C., Han, L. H., & Zhao, X. L. (2013). Full-range analysis on square CFST stub columns and beams
under loading and chloride corrosion. Thin-Walled Structures, 68, 50-64.
doi:10.1016/j.tws.2013.03.003
Hou, C. C., Han, L. H., Wang, Q. L., & Hou, C. (2016). Flexural behavior of circular concrete filled steel
tubes (CFST) under sustained load and chloride corrosion. Thin-Walled Structures, 107, 182-196.
Karren, K. W. (1967). Corner properties of cold-formed shapes. Journal of the Structural Division, 93(1),
401-433.
Li, W., Han, L. H., & Zhao, X. L. (2015). Behavior of CFDST stub columns under preload, sustained load
and chloride corrosion. Journal of Constructional Steel Research, 107, 12-23.
doi:10.1016/j.jcsr.2014.12.023
Malvar, L. J., & Ross, C. A. (1998). Review of strain rate effects for concrete in tension. ACI Materials
Journal, 95(6), 735-739.
MaSubuchi, K. (1980). Analysis of Welded Structures: Residual Stresses, Distortions and Their
Consequences: Pergamon, New York.
Moen, C. D., Igusa, T., & Schafer, B. W. (2008). Prediction of residual stresses and strains in cold-formed
steel members. Thin-Walled Structures, 46(11), 1274-1289.
doi:http://dx.doi.org/10.1016/j.tws.2008.02.002
Sánchez-Silva, M., & Klutke, G.-A. (2016). Reliability and Life-Cycle Analysis of Deteriorating Systems:
Springer.
Sanchez-Silva, M., Klutke, G. A., & Rosowsky, D. V. (2011). Life-cycle performance of structures subject
to multiple deterioration mechanisms. Structural Safety, 33(3), 206-217.
doi:10.1016/j.strusafe.2011.03.003
Shen, J. M., Wang, C. Z., & Jiang, J. J. (1993). Finite Element Analysis of Reinforced Concrete & Ultimate
Analysis of Plate and Shell Structures: Beijing, China (in Chinese).
Wang, R., Han, L. H., & Hou, C. C. (2013). Behavior of concrete filled steel tubular (CFST) members
under lateral impact: Experiment and FEA model. Journal of Constructional Steel Research, 80,
188-201. doi:10.1016/j.jcsr.2012.09.003
Yuan, W., Guo, A., & Li, H. (2017). Seismic failure mode of coastal bridge piers considering the effects of
corrosion-induced damage. Soil Dynamics and Earthquake Engineering, 93, 135-146.
doi:10.1016/j.soildyn.2016.12.002
Research on deformation limits of SRC columns based on
experiment data
Jing Ji
State Key Laboratory of Subtropical Building Science, South China University of Technology
School of Civil Engineering and Transportation, South China University of Technology
Guangzhou, China
cvjingji@scut.edu.cn
Lei Zhang
City, Guangzhou, China
zhanglei.lk@qq.com
Xiaolei Han
State Key Laboratory of Subtropical Building Science, South China University of Technology
xlhan@scut.edu.cn
Tao You
asdf_7283@qq.com
Yuan Cao
137868396@qq.com
ABSTRACT
Based on the experimental results of 206 steel reinforced concrete (SRC) columns collected at
home and abroad, the influence of five principal parameters that are shear span ratio, flexure
shear ratio, axial load coefficient, transverse reinforcement ratio and shaped steel ratio on the
failure modes of SRC columns was analyzed and subsequently a method was put forward using
shear span ratio and flexure shear ratio to classify the failure modes. According to the current
seismic design codes and the summary of phenomena related to different performance levels,
the performances of SRC columns were divided into five levels, and limits of plastic rotations
angle of these levels were obtained based on the feature points of the skeleton curves. It was
found that axial load coefficient was the key parameter, and criteria of deformation limits
corresponding to five performance levels were put forward via statistical analysis and verified
via reliability analysis of 206 test specimens. Results show that the proposed criteria of failure
modes can preferably predict the failure modes of SRC columns; the reliability of deformation
limits of SRC columns derived from axial load coefficient stays within a reasonable range.
KEYWORDS: SRC column; experimental data collection; performance levels; deformation
limits; regressive analysis
0. INTRODUCTION
The performance based design method (PBSD) has been known worldwide as an advanced
theory of structural seismic design, in which the control of structural design switches from force
to deformation. In the performance based design method, structures can be designed with the
required performance targets and corresponding seismic details to satisfy the preference of
owners and functions of structures. Recently, plenty of experiments have been conducted on
the seismic performance of reinforced concrete (RC) members and steel reinforced concrete
(SRC) members (Qi Yongle, 2012; Ji Jing, Xiao Qiyan, Huang Chao, Han Xiaolei, and Li
Shoufang, 2010; Han Xiaolei, Cui Jidong, and Gong Huanjun, 2015; ASCE/SE 41-13, 2014).
Meanwhile, performance criteria of RC beams, columns and walls that agree with the Chinese
codes were proposed, which can be applied to estimate the seismic performance of building
structures (Han Xiaolei, Xie Candong, and Cui Jidong, 2016). However, unlike RC members,
performance criteria of SRC columns which are widely used in super high rise buildings has not
been thoroughly studied.
In this study, experimental results of 206 SRC columns were collected at home and abroad.
Based on the classification by failure modes and performance levels of SRC columns, critical
parameters involved with the deformation limits of members have been analyzed, deformation
limits corresponding to different performance levels for various failure modes were then
obtained. Lastly, accuracy of the proposed deformation limits has been verified via reliability
analysis.
1. CLASSIFICATIONS OF FAILURE MODES
1.1 COLLECTION OF EXPERIMENTAL RESULTS

Based on the analysis of failure modes of 206 collected SRC columns, it is found that shear
span ratio λ, flexure shear ratio m, axial load coefficient n, transverse reinforcement ratio ρv and
shape steel ratio ρs are the five critical parameters that determine the failure mode of SRC
columns. As the difference in calculating these parameters of the collected data, calculations of
the five parameters have been normalized according to the design codes JGJ138-2012,
GB50010-2010 and GB50011-2010, which eliminated the deviation of experiment results.
(1) The experimental axial load coefficient n is expressed as follows:
=n N / (fc Ac + fa Aa ) (1)
where N is the axial load force, fc is the average value of axial compressive strength of
concrete, fa is the average yield strength of steel, and Ac and Aa are the section area of
concrete and steel, respectively.
(2) The flexure shear ratio m is expressed as follows:
m = Mu / (Vu H ) (2)
where Mu and Vu are the ultimate flexural capacity and ultimate shear capacity calculated by
section 6.2 in the code (JGJ138-2012), respectively. H is the height of inflection point.
Moreover, shape steel ratio ρs is defined by the design code JGJ138-2012. Shear span ratio λ
and transverse reinforcement ratio ρv are calculated according to the code GB50010-2010.
Table 1 gives the summary of normalized experiment results.
Table 1 – Experimental data of SRC column
Refere Number
Column
nce of λ n m ρv/% ρs/%
height/mm
index specimen
[10] 3 3 0.3, 0.5 0.55~0.88 0, 1.49 5.16, 8.88 600
[11] 8 4.75， 0.12~0.25 0.4~0.58 0.34~1 0.9~2.8 1930, 2489
6.13
[12] 6 5.6 0.12 0.22~0.45 0.24 1.54~4.53 2075
[13] 20 1~2.5 0.2~0.33 0.73~1.37 0.8~1.6 6.11 220, 330, 440,
550
[14] 2 4 0.32, 0.48 0.38, 0.41 1.08 5.26, 8.85 1000
[15] 17 1.4~3.2 0.18~0.53 0.37~0.93 1.02~2.04 4.98 336, 444, 564,
5 780
[16] 12 3 0.12~0.35 0.73~0.87 0.8~1.4 4.6~6.8 630
[17] 13 1~3 0.22~0.66 0.47~1.06 0.82 5.6 240, 360, 480,
600, 720
[18] 10 1~3 0.22~0.66 0.47~1.06 0.82 5.6 240, 360, 480,
600, 720
[19] 8 3 0.4 0.44~0.51 1.27~2.41 7.22 900
[20] 2 1 0.34, 0.57 0.77, 1.03 1.05 1.38 333
[21] 2 3 0.4, 0.45 0.34, 0.35 2.71 2.88 750
[22] 6 2.75 0.15~0.45 0.39~0.7 0.78~1.19 4.53 825
[23] 5 3 0.31~0.44 0.42~0.57 1.01 1.23~4.88 750
[24] 9 4.23 0.31~0.48 0.32~0.51 0.95~2.05 6.15 1100
[25] 3 5 0.29 0.35~0.41 1.12~1.92 6.08 1000
[26] 4 3 0.29 0.63~0.81 0.89 3.98~8.42 600
[27] 4 5 0.32~0.34 0.45~0.57 0.89 3.98~7.26 1000
[28] 4 3 0.29 0.6~0.77 0.89~1.93 6.08 600
[29] 3 3 0.3 0.71~0.8 0.89~1.5 6.08 600
[30] 4 3 0.24~0.38 0.72~0.8 0.89 6.08 600
[31] 3 5 0.24~0.33 0.48~0.53 0.89 5.03 1000
[32] 6 2.2 0.43, 0.48 0.42~0.53 0.8, 1.6 5.6 440
[33] 12 1.75, 3 0.25~0.73 0.23~0.89 0.35, 0.7 3.01, 3.68 350, 600
[34] 32 1.5~2.5 0.32~0.56 0.47~1.12 0.8~2.2 4.52 300, 400, 500
[35] 5 1.88 0.32~0.57 0.51~1.07 0.6, 1.2 6.89 300
[36] 3 1.5 0.24~0.59 0.63~1.15 0.63 4.45 300
1.2 CLASSIFICATION OF FAILURE MODES

General failure modes of SRC columns contains four types of failure: shear-diagonal-
compression failure, shear-bond failure, flexural-shear failure and flexural failure. To obtain the
critical parameters that influence the failure mode, shear span ratio λ, flexure shear ratio m,
axial load coefficient n, transverse reinforcement ratio ρv and shape steel ratio ρs are analyzed
as illustrated in Figure 1~Figure 4.
It can be found from Figure 1~Figure 4 that axial load coefficient n, transverse reinforcement
ratio ρv and shape steel ratio ρs have little effect on the failure modes. Thus, failure modes of
SRC columns cannot be classified by these three parameters. On the other hand, shear span
ratio λ and flexure shear ratio m have a significant relevance with the failure modes. By
considering above two factors, criteria of failure mode classification are obtained in Table 2.
1.6 3.0
Transverse reinforce ratio (%)

Shear-diagonal-compression Shear-diagonal-compression
1.4 Shear-bond
2.5
Shear-bond
Moment-shear ratio
Flexure-shear Flexure-shear
1.2
Flexure 2.0 Flexure
1.0
0.8 1.5
0.6 1.0
0.4
0.2 0.5
0.0 0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
Shear-span ratio Shear-span ratio
Fig. 1 –Failure modes influenced by shear-span ratio Fig. 2 –Failure modes influenced by shear-span ratio
and flexure-shear ratio and transverse reinforce ratio
0.8 Shear-diagonal-compression 10.0 Shear-diagonal-compression
0.7 Shear-bond Shear-bond
Axial load coefficient
Flexure-shear 8.0 Flexure-shear

Shape steel ratio
0.6
Flexure Flexure
0.5 6.0
0.4
0.3 4.0
0.2
2.0
0.1
0.0 0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
Shear-span ratio Shear-span ratio
Fig. 3 –Failure modes influenced by shear-span ratio Fig. 4 –Failure modes influenced by shear-span ratio
and axial load coefficient and steel ratio
Table 2 – Failure mode classification of SRC column

Failure mode Criteria of classification
Shear-diagonal-compression failure λ≤1.5, m＞0.6
Shear-bond failure (1.5<λ<2.5, m＞0.5) or (λ≤1.5, m≤0.6)
Flexural-shear failure (1.5<λ<2.5, m≤0.5) or (λ≥2.5, m＞0.7)
Flexural failure λ≥2.5, m≤0.7
Based on the classification criteria of failure modes shown in Table 2, all the collected
experimental results were re-classified and analyzed as presented in Table 3.
Table 3 – Accuracy of the classification of SRC column
Total Number Number of accurate
Failure mode Accuracy(％)
of specimens classification
Shear-diagonal-
47 39 83
compression failure
Shear-bond failure 39 29 74
Flexural -shear failure 15 11 73
Flexural failure 103 85 83
Total 204 164 80
As can be seen from Table 3, the accuracy of failure mode classification is 80% when using the
two parameters, shear span ratio λ and flexure shear ratio m.
2. DEFORMATION LIMITS OF SRC COLUMNS
2.1 DIVISION OF PERFORMANCE LEVELS

Performance level of a structural member under earthquake with a given intensity refers to the
maximum degree of damage that the structural member can resist. In this study, performance
levels of a structural member have been divided into five phases, namely, intactness, slight
damage, moderate damage, relatively severe damage and severe damage respectively. Failure
phenomena of the five phases are given in Table 4.
Former study (Qi Yongle, 2012) shows that plastic rotation angle can preferably reflect the
damage degree of structural members, and the plastic rotation angle was selected as the
deformation index for structural members. Symbols θy, Δ1, Δ2, Δ3 and Δ4 were used to represent
the total rotation angle limits corresponding to intactness, slight damage, moderate damage,
relatively severe damage and severe damage level, respectively. Besides, symbols θ1, θ2, θ3
and θ4 represent the plastic rotation angle (with yield rotation angle θy subtracted from total
rotation angle) corresponding to slight damage, moderate damage, relatively severe damage
and severe damage level, respectively. On the other hand, rotation angle limits of each
performance level were obtained from the skeleton curves of SRC columns. It was concluded
from failure modes of 206 SRC columns that once the peak capacity is reached, the load-
rotation angle curve decreases almost linearly and slowly. For this reason, rotation angle limits
of moderate damage and relatively severe damage level were equally interpolated in the range
between slight damage and severe damage level, which have been given in Table 4. Figure 5
illustrates the relationship between rotation angle limits corresponding to the five performance
levels and loads on the skeleton curves, where Py and Pmax represent the yield load and peak
capacity of SRC columns respectively.
According to the method given in Table 4, lateral displacements of SRC columns corresponding
to intactness, slight damage and severe damage were extracted from the skeleton curve. The
plastic rotation angle limit can be obtained by subtracting the yield rotation angle from the total
rotation angle, which is the ratio of lateral displacement and the height of column, as given in
Table 5.
Pmax
Py θ1
0.7Pmax θ2
θ3
θ4
θy Δ1 Δ2 Δ3 Δ4
Δ/H
Fig. 5 –Rotation angles corresponding to 5 performance levels
Table 4 – Failure phenomena and rotation angle limits under 5 performance levels of
SRC column
Performance
Failure phenomena Rotation angle limits
level
No significant cracks occur in concrete, R-park method, equivalent energy
Intactness longitudinal bars have not been yield, no method, and general yield
residual deformation after unload moment method
Fissure width<2mm, longitudinal bars are
Rotation angle corresponding to
Slight yield in tension, steel flange is yield in
the peak capacity, as the rotation
damage compression, concrete coverage spalling,
angle of point C in Figure 5
load is close to the peak capacity
Fissure width≥2mm, stirrups are partially yield,
Moderate steel flange is yield in tension, concrete
θ2=θ1+1/3（θ4-θ1）
damage coverage spalling further, decrease of
capacity<10%
Most concrete coverage spalling, concrete
Relatively starts to crashing, most stirrups are yield or
severe slip out, the bottom of SRC column has a θ3=θ1+2/3（θ4-θ1）
damage relatively severe damage, decrease of
capacity>10%
All the concrete coverage have spalled, most
concrete have been crushed, fissures θ4 is the rotation angle when the
Severe penetrate through the bottom of SRC capacity decrease to 70% of the
damage columns, the bottom of SRC column has a peak value, as the rotation angle
severe damage, longitudinal bars and stirrups of point D in Figure 5
all bloat out, decrease of capacity>20%
Table 5 – Rotation angle limits of SRC column
Yield Displacement Ultimate

Reference Number of
displacement at peak load displacement θy θ1 θ4
index specimens
δy /mm δ1/mm δ4/mm
[10] 3 6.25~7.32 11.6~17.41 18.15~84.76 0.01~0.012 0.009~0.017 0.020~0.13
[11] 8 25~46 39.37~72.47 201.91~245.3 0.013~0.018 0.006~0.025 0.063~0.114
[12] 6 28.49~48.41 56.95~70.08 148.27~312.11 0.014~0.023 0.009~0.019 0.057~0.127
[13] 20 1.36~3.74 2.41~7.99 4.76~21.46 0.005~0.009 0.003~0.01 0.015~0.041
[14] 2 4.01, 5.77 10.97, 11.19 26.36, 43.15 0.004, 0.006 0.005, 0.007 0.021, 0.039
[15] 17 3.69~5.97 6.98~15.18 10.39~41.83 0.005~0.014 0.004~0.016 0.013~0.051
[16] 12 3.12~3.93 7.78~10.57 14.65~29.08 0.005~0.006 0.007~0.011 0.018~0.04
[17] 13 3~6.15 4.65~12.21 12.49~35.18 0.006~0.013 0.003~0.008 0.011~0.07
[18] 10 2.71~4.68 3.88~5.58 9.82~22.62 0.006~0.012 0.001~0.004 0.021~0.039
[19] 8 5.18~6.49 11.21~17.23 29.32~38.84 0.006~0.007 0.006~0.012 0.026~0.036
[20] 2 0.94, 1.15 2.32, 2.87 7.19, 9.39 0.003 0.004, 0.006 0.019, 0.025
[21] 2 5.5, 5.9 9.9, 10.04 26.27, 32.82 0.007, 0.008 0.005, 0.006 0.027, 0.036
[22] 6 5.28~6.07 12.51~19.53 42.9~64.45 0.006~0.007 0.008~0.017 0.045~0.071
[23] 5 3.71~6.69 9.18~20.41 16.21~40.31 0.005~0.009 0.005~0.02 0.013~0.045
[24] 9 5.87~8.95 10.22~15.6 25.67~32.5 0.005~0.008 0.004~0.008 0.016~0.023
[25] 3 5.1~6.8 11.12~12.11 19.9~22.54 0.005~0.007 0.005~0.007 0.015~0.017
[26] 4 2.5~3 5.46~7.2 9.76~12.6 0.004~0.005 0.005~0.007 0.012~0.016
[27] 4 4.02~7.76 9.04~13.64 25.02~45.23 0.004~0.008 0.004~0.007 0.021~0.038
[28] 4 2.2~3.52 5.11~6.91 8.42~11.28 0.004~0.006 0.004~0.007 0.009~0.015
[29] 3 2.74~2.85 5.18~6.78 13.4~15.32 0.005 0.004~0.007 0.018~0.021
[30] 4 2.52~3.33 5.38~7.64 7.82~15.67 0.004~0.006 0.004~0.008 0.009~0.021
[31] 3 5.36~7.96 11.43~13.24 29.67~34.74 0.005~0.008 0.005~0.006 0.022~0.028
[32] 6 8.52~12.4 15.12~22.19 35.89~49.76 0.019~0.028 0.013~0.031 0.058~0.092
[33] 12 0.36~9.33 0.92~10.42 3.12~32.78 0.001~0.016 0.001~0.011 0.007~0.058
[34] 32 1.01~1.64 2.13~4.93 4.29~9.07 0.003~0.005 0.002~0.01 0.008~0.027
[35] 5 4.2~5.31 8.74~14.71 18.28~29.06 0.014~0.018 0.014~0.033 0.045~0.081
[36] 3 — — — — — —
·2.2 VERTIFICATION OF DEFORMATION LIMITS FOR SRC COLUMNS

According to section 2.1, the way to obtain rotation angle limits is clear for the performance
levels of intactness, slight damage and severe damage. However, for the performance levels of
moderate damage and relatively severe damage, rotation angle limits were set up by the Equal
Division Method. That is to say, rotation angle limits for performance levels of moderate
damage and relatively severe damage are assumed to be at the points of trisection of rotation
angle limits for performance levels of slight damage and severe damage, and be calculated as
formulas presented in Table 4. In order to verify the accuracy of this method, 9 SRC columns
with detail experiment records were collected and the analysis results are shown in Table 6.
Meanings of the parameters in Table 6 are as follows:
δt: Displacement corresponding to the experiment phenomena in Table 4 during the loading
process.
H: Height of the inflection point;
Experimental rotation angle θt: θt = δ t / H (3)
Mean divided value of rotation angle θd: rotation angle limits obtained from Table 4;
Errors: ( θt − θd ) / θt (4)
Table 6 – Error analysis of rotation angle by Equal Division Method
Moderate damage Relatively severe damage

H
Specimens δt θt θd Errors δt θt θd Errors
(mm)
(mm) (rad) (rad) (%) (mm) (rad) (rad) (%)
SRHC-1 600 4.5 0.008 0.007 12.5 7.5 0.013 0.010 23
SRHC-2 600 5.5 0.009 0.009 0.0 8.3 0.014 0.011 21.4
SRHC-3 600 6 0.010 0.009 10.0 8.2 0.014 0.010 28.6
SRHC-4 600 6 0.010 0.010 0.0 8 0.013 0.013 0.0
SRC-1 825 24.75 0.030 0.035 16.7 41.25 0.050 0.053 6.0
SRC-2 825 24.75 0.030 0.025 16.7 33 0.040 0.041 2.5
SRC-3 825 16.5 0.020 0.022 10.0 24.75 0.030 0.035 16.7
SRC-4 825 24.75 0.030 0.025 16.7 33 0.040 0.040 0.0
SRC-5 825 24.75 0.030 0.030 0.0 41.25 0.050 0.048 4.0
According to Table 6, the errors of rotation angle limits for moderate damage level is less than
20% by the Equal Division Method. For relatively severe damage level, the errors is less than
30%, which is still in a reasonable range. It can be assumed that the rotation angle limits for
moderate damage level and relatively severe damage level of SRC columns obtained by the
Equal Division Method were reasonable.
3. ROTATION ANGLE LIMITS OF SRC COLUMNS
3.1 ROTATION ANGLE LIMITS OF DIFFERENT PERFORMANCE LEVEL

Correlation analysis has been carried out to study the relevance between rotation angle limits
and 5 parameters in the study (Liu Han, 2016). It is found that the axial load coefficient that in a
certain range is the most relevant parameter for rotation angle limits of different performance
levels of different failure modes. Accordingly, the boundary values of axial load coefficient for
different failure modes were presented in Table 7. Within the axial load boundary, the rotation
angle limits were little changed, while the limits changed linearly when exceeding the boundary.
Table 7 – Boundary values of axial load coefficient for different failure modes
Failure mode Axial load coefficient n

Shear-diagonal-compression failure n≤0.3,n≥0.5
Shear-bond failure n≤0.3,n≥0.5
Flexure-shear failure n≤0.2,n≥0.4
Flexure failure n≤0.2,n≥0.4
Table 8 – Rotation angle limits of Shear-diagonal-compression members with n≤0.3
Rotation angle standard Rotation

min max mean
limits deviation angle limits
θy 0.006 0.013 0.009 0.002 0.007
θ1 0.003 0.016 0.008 0.004 0.004
θ2 0.014 0.035 0.022 0.007 0.015
θ3 0.024 0.061 0.036 0.013 0.023
θ4 0.033 0.088 0.050 0.018 0.032
Table 9 –Rotation angle limits of SRC columns (rad)
Performance level
Failure mode Light Moderate Relatively severe Severe
Intactness
damage damage damage damage
Shear-diagonal-
λ≤1.5，m>0.6
compression failure
n≤0.3 0.007 0.004 0.015 0.023 0.032
n≥0.5 0.003 0.004 0.008 0.012 0.016
Shear-bond failure 1.5<λ<2.5，m>0.5
n≤0.3 0.007 0.006 0.014 0.021 0.029
n≥0.5 0.002 0.003 0.006 0.009 0.012
Flexure-shear failure 1.5<λ<2.5，m≤0.5 or λ≥2.5，m>0.7
n≤0.2 0.005 0.008 0.015 0.023 0.030
n≥0.4 0.005 0.006 0.012 0.018 0.024
Flexure failure λ≥2.5，m≤0.7
n≤0.2 0.009 0.008 0.026 0.042 0.056
n≥0.4 0.005 0.003 0.012 0.020 0.027
The value of limits was calculated as mean minus one time standard deviation using plastic
rotation angles (total rotation angle minus elastic rotation angle).
The value of limits for shear-crush failure mode was calculated as shown in Table 8, and other
cases were calculated the same way. Finally the limits criteria were summarized in Table 9
(Interpolation is permitted between these values).
3.2 GUARANTEED RATE OF ROTATION ANGLE LIMITS

Experimental data of different failure modes was collected. On this basis, rotation angle limits
was verified by comparing the limits between Table 9 and experiments. This paper only
validated the rotation angle limits for intactness, light damage and severe damage performance
level, rotation angle limits for other performance levels were set up by means of Equal Division
Method. The calculation method of guaranteed rate was based on ASCE/SEI 41.
Take shear-diagonal-compression failure as an example. 47 specimens with shear-diagonal-
compression failure were collected, and the guaranteed rate of rotation angle in Table 9 is
shown in Figure 6(a), (b), (c).
Intactness Light damage Severe damage

0.6 0.8 0.8
0.5 0.7 0.7
0.6 0.6
0.4 0.5 0.5
log (θtest/θ1)
log (θtest/θ4)
log (θtest/θy)
0.3 0.4
0.4
0.2 0.3
0.3
0.1 0.2
0.1 0.2
0.0 0.1
0.0
-0.1 0.0
-0.1
-0.2 -0.2 -0.1
-0.3 -0.3 -0.2
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Axial load coefficient n Axial load coefficient n Axial load coefficient n
(a) (b) (c)
Fig. 6 – Success ratio of rotation angles correspongding to intactness, light

damage and severe damage
Figure 6(a), (b), (c) shows the logarithmic distribution of the ratio between the experimental
rotation angle and the limits in Table 9 for intactness, light damage and severe damage
performance level. θtest means the rotation angle limits obtained from experiments and θi
(i=y,1,4) is obtained from Table 9. It can be assumed that θtest/θi follows the lognormal
distribution, the rotation angle limits in Table 9 is reasonable when log(θtest / θ i ) ≥ 0 , and the
guaranteed rates of rotation angle limits for intactness, light damage and severe damage
performance level are 80.6%, 81.6% and 71.1%, respectively.
Using the same method, the guaranteed rates of rotation angle limits of shear-bond failure,
flexure-shear failure and flexure failure were calculated, and the rates for intactness, light
damage and severe damage performance levels were shown in Table 10.
According to Table 10, the rotation angle limits for severe damage level have a low guaranteed
rate of 71.1%. Mainly because there are many judgment conditions for severe damage failure
and the data are of high discreteness. Overall the guaranteed rates of proposed rotation angle
limits range from 70% to 80%, which can meet the practical engineering need.
Table 10 – Success ratio of rotation angle under different performance levels(%)
Performance level
Failure mode
Intactness Light damage Severe damage
Shear-diagonal-compression
80.6 81.6 71.1
failure
Shear-bond failure 81.7 83 72
Flexure-shear failure 79.8 79.9 69.6
Flexure failure 73 74.6 65
4. CONCLUSIONS
(1) Categorization criteria of failure modes of SRC columns based on shear-span ratio and
moment-shear ratio were proposed and the accuracy reaches 80%.
(2) Five performance levels of SRC columns were put forward: intactness, light damage,
moderate damage, relatively severe damage and severe damage. The standards of
determining these states are also presented.
(3) Considering the influence of five parameters including shear-span ratio, axial load
coefficient, moment-shear ratio, stirrup ratio and steel ratio on the rotation angle limits for
different failure modes, the boundary value of axial load coefficient was determined, based on
which the deformation limits criteria of SRC columns was summarized. Verifying the
deformation limits in comparison with experiments, the guaranteed rate can be up to 70%,
which can meet the practical demands in engineering.
REFERENCE
Qi Yongl (2015). Research on Deformation Limits of RC Beams, Columns and Shear walls
based on material strain. Guangzhou, South China University of Technology.
Ji Jing, Xiao Qiyan, Huang Chao, Han Xiaolei, Li Shoufang (2010). Research on deformation
limits of performance-based RC shear walls controlled by flexure. Journal of Building
Structures, 31(9), 35-41.
Han Xiaolei, Cui Jidong, Gong Huanjun, et al (2015). Experimental study on seismic behavior of
RC beams with different seismic grades. Journal of Building Structures, 36(S2), 299-308
ASCE/SE 41-13 Seismic evaluation and retrofit of existing buildings (2014). Reston, VA,
American Society of Civil Engineers.
Han Xiaolei, Xie Candong, Cui Jidong, et al (2016). Seismic performance evaluation of frame-
supported shear wall structures based on component deformation. J. Huazhong Univ. of
Sci. & Tech.(Natural Science Edition), 44(07), 105-109+115.
Speicher M S, Harris J L (2016). Collapse Prevention seismic performance assessment of new
eccentrically braced frames using ASCE 41. Engineering Structures, 117, 344-357
JGJ138-2012, Code for design of composite structure (2010). Beijing, China Architecture &
Building Press.
GB 50010-2010, Code for design of concrete structures (2010). Beijing: China Architecture &
Building Press, 2010
GB 50010-2010，Code for seismic design of buildings (2010). Beijing: China Architecture &
Building Press, 2010
Gan D, et al (2011). Seismic behavior and moment strength of tubed steel reinforced-concrete
(SRC) beam-columns. Journal of Constructional Steel Research. 67(10), 1516-1524
James M, Ricles S D P (1994). Seismic performance of steel-encased composite columns.
Journal of Structural Engineering, 2474-2494
Hsu H L, Jan F J, Juang J L (2009). Performance of composite members subjected to axial
load and bi-axial bending. Journal of Constructional Steel Research, 65(4), 869-878
Li Junhua (2005). Study on the performance of steel reinforced high-strength concrete columns
under low cyclic reversed loading. Xi’an, Xi’an University of Architecture and Technology.
Wang Qiuwei (2013). Experimental study on seismic behavior of steel reinforced concrete
columns with new-type cross sections. Journal of Building Structures, 11, 123-129.
Ma Hui (2013). Research on seismic performance and design calculation methods of steel
reinforced recycled concrete columns. Xi’an, Xi’an University of Architecture and
Technology.
Zheng Shansuo, Wang Bin, Hou Piji, et al (2011). Experimental study of the damage of
SRHSHPC frame columns under low cycle reversed loading. China Civil Engineering
Journal, 09, 1-10.
Ping Zhendong (2009). Experimental Research on ductility of SRHSHPC frame columns. Xi’an,
Xi’an University of Architecture and Technology.
Peng Chunhua (2008). Experimental Study and Theory Analysis on Limit Value of Axial
Compression for SRHSHPC columns. Xi’an, Xi’an University of Architecture and
Technology.
Chen Xiaogang, Mou Zaigen, Zhang Jubing, et al (2009). Experimental study on the seismic
behavior of steel reinforced concrete columns. Journal of University of Science and
Technology Beijing, 31(12), 1516-1524.
Du Yongshan (2007). An Experimental Study on Hysteretic Behavior of Reinforced Concrete
Transnomal Short Columns and Steel Reinforced Concrete Transnomal Short Columns.
Harbin, Harbin Institute of Technology.
Xie Xiadi (2008). Experiment Study on Seismic Design Details and Axial Compressive
Performance of CSRC Columns. Xiamen, HuaQiao University.
Zhang Zhiwei (2007). Experiment Study on deformation performance and hysteretic
characteristic of SRC columns. Xiamen, Huaqiao University.
Liu Yang (2006). Experiment Study on Seismic Behavior of Core Steel Reinforced Concrete
Columns. Xiamen, Huaqiao University.
Chen Caihua (2007). Experiment Study and Numerical Analysis on Steel Reinforced Concrete
Compression-bending Members. Beijing, China Academy of Building Research.
Jiang Chuanxing (2007). Study on Effect of Stirrup Ratio on Steel Reinforced High-strength
Concrete when Shear-span Ratio is 5. Chongqing, Chongqing University.
Chen Zijing (2007). Experimental Study on Proper Ratio of Steel Skeleton of Steel Reinforced
High-Strength Concrete Columns. Chongqing, Chongqing University.
Yu Lei (2006). Study on Reasonable Steel Contents of Steel High-Strength Reinforced
Concrete Columns. Chongqing, Chongqing University.
Yang Dingfeng (2007). Experimental Research on the Effects of Loading Performance about
SRHC Column with different stirrup ratio and λ=3 on the repeatedly horizontal load.
Chongqing: Chongqing University.
Ma Xiang (2009). Experimental research on the effects of stirrup ratios on bearing performance
of SRHC columns with studs. Chongqing: Chongqing University.
Liu Wei (2007). Study on Limited Value of Axial Compression Ratio of Steel Reinforced High
Strength Concrete Column. Chongqing, Chongqing University.
Shen Qinchuan. Study on Limit Value of Axial Compression Ratio of Steel Reinforced High
Strength Concrete Column. Chongqing, Chongqing University
Sun Cangbai (2005). The Study of seismic behavior on steel high-strength reinforced concrete
columns. Xi’an, Xi’an University of Architecture and Technology.
Jiang Donghong (2001). Study on mechanical performance and seismic performance of steel
reinforced high-strength concrete frame column. Shenyang, Northeastern university.
Jia Jingqing (2000). The research of SRHC short columns and high-strength concrete short
columns on mechanics performance. Dalian: Dalian University of Technology.
Li Xiangmin, Zhang Yu, Chen Zongliang (1999). The Limited Values of Steel Reinforced High
Strength Concrete Column. Journal of Building Structures, 07, 10-13.
Zhang Sufang (1999). The Ductility of SRC Frame Short Column under Low Reversed Cyclic
load. Journal of Southwest JiaoTong University, 02: 112-118.
Liu Han (2016). Research on deformation limits of SRC beam and column based on experiment
data. Guangzhou: South China University of Technology.
Update to ASCE/SEI 41 Concrete Provisions (2007). US: Federal Emergency Management
Agency.
Update of Design Provisions for Moment-Resisting Joints between
Steel Beams and Reinforced Concrete Columns
Luis B. Fargier-Gabaldón
Director, Diseño y Construcciones Integrados,
Mérida, Venezuela
lfargier@dcintegrados.com
Paul Cordova
Senior Project Manager, Simpson Gumpertz & Heger
ppcordova@sgh.com
Gustavo J. Parra-Montesinos
C.K. Wang Professor of Structural Engineering. University of Wisconsin-Madison
gustavo.parra@wisc.edu
Gregory Deierlein
John A. Blume Professor. Stanford University
ggd@stanford.edu
Abstract
Composite moment frames consisting of steel beams continuous through reinforced concrete or
concrete-encased steel composite columns offer an attractive alternative to concrete or steel
special moment frames for buildings constructed in regions of high seismicity. The current state-
of-practice to design connections between steel beams and concrete columns is based on the
document “Guidelines for design of joints between steel beams and reinforced concrete columns”,
which was published in 1994 and does not incorporate subsequent research on the seismic
performance of the connections. This paper summarizes an effort to update the 1994 Guidelines.
Relevant test data on connection behavior are reviewed and summarized, including joint response
under reversed cyclic loading of interior, exterior and roof-T connections, the use of high strength
concrete, and connection details not included in the 1994 Guidelines, such as joints with steel
band plates and transverse beams. Revised (updated) equations are presented to determine joint
shear strength for various joint configurations, including validation against test data that
demonstrates applicability to regions of high seismicity.
Introduction
Composite frames offer advantages over traditional reinforced concrete (RC) or steel structures
with cost-savings and improved efficiency of construction, while exhibiting adequate performance
under large displacement reversals. Continuous steel beams are more efficient than RC beams
for long beam spans and the through-joint detail eliminates the risk of weld fractures at the beam-
to-column flanges, as observed in steel-framed buildings that were damaged in the Kobe and
Northridge earthquakes (Tremblay et al. 1996, FEMA-351). On the other hand, the RC columns
can be significantly more cost-effective to resist axial forces than the heavy steel counterparts
(Griffis 1996). Composite frame construction can be faster than that of traditional RC or steel
structures because the steel frame construction can advance as in a typical steel frame
construction, but without the need of time consuming full penetration welds at beam-column joints.
Light steel erection columns support the continuous steel beams and steel deck, which can be
erected a few stories above the concrete operation related to assemblage of reinforcement cages
and casting of the columns and slab. Alternatively, the composite frame can be constructed using
precast concrete column – steel beam assemblies.
The design of moment connections between RC or concrete-encased columns and steel beams
continuous through the joint was first addressed in the “Guidelines for design of joints between
steel beams and reinforced concrete columns”, published in 1994 by the ASCE Task Committee
on Composite Structures. The document was based on the experimental work conducted at the
University of Texas at Austin by Sheikh et al. (1989) and Deierlein el al. (1988). Since its
publication, several experimental programs have been conducted in the US, including the work
of Kanno (1993), Bugeja et al. (2000), Parra-Montesinos and Wight (2000), Liang and Parra-
Montesinos (2004), Fargier-Gabaldón and Parra-Montesinos (2005); in Japan, Kuramoto and
Noguchi, (1997), Nishiyama et al. (1998, 2000), Noguchi and Kim (1997, 1998), Baba and
Nishimura (2000); and in Taiwan, Cordova et al. (2004), Chen et al. (2004). This paper presents
a summary of the most relevant experimental research programs conducted over the past 30
years and an update of the design provisions for moment-resisting joints between through-joint
steel beams and RC columns.
Research at the University of Texas at Austin
The work at the University Texas at Austin included the testing of seven specimens under
monotonic loading and eight specimens under reversed cyclic loading (Sheikh et al. 1989 and
Deierlein el al. 1989). All specimens featured a 200 mm deep built-up steel beam and a 510 mm
square RC column. Joint details included the use of vertical stiffeners welded to the beam at the
column faces, commonly referred to as face bearing plates, steel erection columns, and extended
face bearing plates welded to the flanges of the steel beam to mobilize the concrete regions
located outside the width of the beam.
Test results indicate that the web panel of the steel beam carries part of the joint shear, as in
beam-column joints of steel frames (Figure 1a). Face bearing plates (Figure 1b) increased the
connection strength by approximately 70% when compared to the steel web panel alone. The
addition of extended face bearing plates increased the connection strength by 180%. Joints that
featured face bearing plates in combination with steel columns experienced a nearly 130%
increase in strength. Based on these observations, the joint shear strength was divided into three
mechanisms: (a) the strength of the steel web panel (Figure 1a), (b) the contribution of an inner
diagonal concrete strut that forms between flanges of the steel beam and the face bearing plate
(Figure 1b), and (c) the contribution of a compression field or truss mechanism outside the width
of the steel beam that is activated with the use of shear keys welded to the beam flanges and
horizontal tie reinforcement above and below the beam (Figure 1c and 1d). Connections failing
in shear exhibited ductile behavior and sustained its shear strength through large deformations
(Figure 2a).
Test also showed localized concrete crushing above and below the top and bottom beam flanges,
respectively, which led to rigid body rotations of the steel beam inside the joint (Figure 2b),
pinched hysteresis, and reduced energy dissipation of the specimens. Vertical joint reinforcement
welded to the top and bottom flanges of the steel beam near the column faces was found to be
effective in increasing joint bearing strength and preventing or postponing the occurrence of a
bearing failure.
The results from the research conducted at the University of Texas at Austin led to the
development of the “Guidelines for design of joints between steel beams and reinforced concrete
columns” published by an ASCE Task Group on Composite Connection in 1994. Joint details
covered in the Guidelines are shown in Figure 4, except for the steel band plates (Figure 3c) and
cover plates (Figure 3f) which are addressed in the updated provisions.
Research at the Cornell University
Kanno and Deierlein (1993) conducted an extensive experimental program at Cornell University,
which included the testing of nineteen specimens under reversed cyclic loading. One specimen
was designed for beam yielding, two for a simultaneous joint shear failure and beam yielding, and
the remaining ten and six to fail in shear and bearing, respectively. Three specimens featured 100
MPa strength concrete and the remaining used 40 MPa strength concrete. All specimens featured
a 406 mm square RC column and face bearing plates. Other joint details included steel band
plates wrapping around the column regions just above and below the steel beam, cover plates
around the column faces within the beam depth, transverse steel beams, wide bearing plates,
headed studs welded on the beam flanges, horizontal joint reinforcement in volumetric ratios
ranging from 0.4 to 1.2%, and vertical joint reinforcement. An axial load corresponding to an
average axial stress based on the gross concrete area of approximately 0.3𝑓𝑐′ was applied to five
of the specimens.
Test results suggest that: (a) the axial load helps to mobilize the concrete regions located outside
the flanges of the steel beam, (b) the use of high strength concrete increases the connection
strength compared to the use of regular strength concrete, (c) vertical joint reinforcement is very
effective in increasing the bearing strength of the joint, and (d) transverse beams enhance the
participation of the concrete regions located outside the beam flanges to resist shear.
Kanno and Deierlein (1993) also studied the influence of confinement on the bearing strength of
the concrete regions located just above and below the steel beam through tests of fifty small
concrete blocks. It was found that heavily confined specimens failed at a stress of approximately
3.3 times larger than that corresponding to the 28-day concrete compressive strength. Given
these findings, they concluded that the use of steel band plates above and below the steel beam
significantly increases the bearing strength of the composite joint.
Research at the Texas A & M University
Bugeja et al. (1999) tested five interior and one exterior composite connections under reversed
cyclic loading in two orthogonal directions. In all specimens, one steel beam ran continuous
through the joint and the transverse steel beams were attached with a bolted moment connection
to the continuous beam. Specimens were first tested in the continuous direction until a drop in the
lateral load occurred, then they were rotated 90 degrees and tested to failure in the perpendicular
direction. In one specimen, a reversed order was used and the specimen was tested to failure in
the continuous beam direction.
The steel beams in all specimens were designed for composite action with a concrete slab over
a metal deck in both directions. Joint details included face bearing plates, steel columns welded
to the flanges of the steel beam, cover plates wrapping around the column faces within the beam
depth, and cover plates placed at 45 degrees with respect to the longitudinal axis of the beams,
forming an octagon in a plan view, which encased the RC column region within the beam depth.
The specimens featured a 380 mm square RC column and a W310 x 32.7 (W12 x 22) hot-rolled
steel section as a beam in both orthogonal directions. All specimens showed a stable load versus
displacement response and large energy dissipation capacity, significant beam yielding, and
limited joint damage in the continuous direction. In the discontinuous direction, three of the six
specimens experienced beam yielding with significant plastic rotation capacity. The failure of two
specimens in the discontinuous direction was characterized by a relative rotation between beam
segments located at either side of the joint, while the exterior joint exhibited a bearing failure. The
test results showed that the effective flange width of the concrete slab correlated well to the
recommendations given in the AISC Manual (1994), except for the cases of edge beams in
positive bending for which the effective flange width can be taken as the column width. Steel
beams exhibited full composite action throughout testing.
Research at the University of Michigan
Several researchers studied composite beam-column joint behavior at the University of Michigan
between 1994 and 2005. Parra-Montesinos and Wight (2000) tested nine exterior connections
under reversed cyclic loading. Joint details included face bearing plates, steel columns, steel band
plates, transverse steel beams, and horizontal joint reinforcement in volumetric ratios ranging
from 0.5 to 0.9%. Two specimens had the joints reinforced with discontinuous fiber reinforcement.
All specimens featured a 400 mm square RC column. The beam-to-column depth ratio
approached 0.6 for five of the specimens and 1.0 for the remaining four specimens. Results from
this investigation showed: (a) all specimens exhibited a stable load versus displacement response
with significant energy dissipation, (b) a hoop volumetric ratio of 0.9% was sufficient to confine
the concrete in the joint region, (c) yielding of the steel web panel occurred over a width of
approximately 0.8 times the column depth, and (d) steel band plates effectively activate an outer
concrete strut through direct bearing on the concrete. A minimum beam depth-to-column
longitudinal bar diameter ratio of 20 was recommended for use in design, except for the case of
joints with steel band plates, in which a ratio of 16 was deemed appropriate.
Liang and Parra-Montesinos (2004) tested two interior and two exterior RC column-to-steel beam
connections with a composite concrete slab over a metal deck under reversed cyclic loading. Joint
details included face bearing plates and steel columns in all specimens in combination with either
steel band plates and transverse steel beams or joint confinement reinforcement in a volumetric
ratio of 1.1%. The column-to-beam flexural strength ratio was 2.2 and 1.3 for the exterior and
interior connections, respectively. The specimens were designed for beam hinging, with the
connections proportioned for a target joint shear deformation of 0.5% using a joint deformation
model proposed by Parra-Montesinos and Wight (2001). All specimens were constructed with a
380 mm square RC column and W310x38.7 (W12x26) steel beam with a 90/50 mm steel deck.
All specimens exhibited significant energy dissipation with full hysteresis loops and a response
dominated by beam hinging and limited joint damage, as intended in the design. The load versus
drift hysteretic response of Specimen 2 (interior connection) with steel band plates, transverse
steel beams and without joint transverse reinforcement is shown in Figure 5. Concrete crushing
of the slab at the column face occurred in all specimens during positive bending of the composite
beam. Shear studs in the hinging zone showed no signs of damage at the end of the test.
Maximum joint shear distortions of approximately 0.005 and 0.003 rad. and bearing deformations
of approximately, 0.007 and 0.008 rad. were measured in the interior and exterior connections,
respectively. Results from this investigation suggest that properly detailed composite connections
between RC columns and a composite continuous steel beam can be designed for beam hinging
and limited joint damage during a major seismic event.
Fargier-Gabaldon and Parra-Montesinos (2006) tested two interior T joints under reversed cyclic
loading to investigate joint details, including column longitudinal bar anchorage, that would allow
adequate mobilization of joint concrete to resist shear. Specimens 1 and 2 featured a joint depth-
to-bar diameter ratio of 12 and 20, respectively, and a 400 mm square RC column. Joint details
used in Specimen 1 included face bearing plates, steel band plates and a steel column. No
transverse reinforcement within the joint region was used in this specimen and the column
longitudinal reinforcement was terminated above the top flange of the steel beam with heads to
eliminate hooks and reduce reinforcement congestion. Joint horizontal reinforcement (0.5%
volumetric ratio) was added to the joint region in Specimen 2 in combination with a hot-rolled steel
channel running transversely above the top flange of the steel beam. The column longitudinal
reinforcement ran through the channel and was anchored with headed bars that bear against the
web of the steel section. Both specimens featured vertical joint reinforcement in the form of
inverted U-shaped stirrups placed over the top flange of the steel beam. Specimen 1 was
designed for joint failure and Specimen 2 was designed for column hinging with limited joint
damage. Specimen 1 showed good energy dissipation capacity up to 3% drift. Beyond 3% drift,
concrete crushing occurred underneath the heads of the longitudinal bars, which led to large
beam rigid body rotations that accounted for nearly the totality of the applied drift at the end of the
test. Specimen 2 exhibited large column plastic hinge rotations and limited joint damage, as
intended in design, with beam rigid body rotations accounting for almost 30% of the applied drift
throughout the test. Test results indicated that a joint depth-to-column bar diameter ratio of 20 is
reasonable for composite T-connections, provided there is a sufficient bearing area at the end of
the bars.
Fargier-Gabaldón and Parra-Montesinos (2005) tested two exterior column-to-steel beam

connections to investigate effective outer panel width in connections where the concrete column
is significantly wider than the steel beam. The two specimens featured a column-to-beam width
ratio of 3.6, in which a 400 x 600 mm rectangular RC column was connected to a W360x57.8
(W14x38) hot-rolled steel beam. Both specimens featured face bearing plates and an embedded
steel erection column. Additionally, Specimen 1 featured horizontal joint transverse reinforcement
in a volumetric ratio of 0.7%, while Specimen 2 featured transverse steel beams and steel band
plates. Except for the column width, Specimens 1 and 2 were nominally identical to Specimens 6
and 9, tested by Parra-Montesinos and Wight (2000), in which a 400 mm square RC column was
used with a column-to-beam width ratio of 2.4. The hysteresis loops indicated that both specimens
exhibited excellent energy dissipation and stiffness retention capacity. Extensive joint diagonal
cracking was observed on the lateral faces of the RC column, which indicated that the entire
column core was mobilized to transfer shear. Specimen 1 experienced a bearing failure,
characterized by crushing of the concrete regions located just above and below the top and
bottom flanges of the steel beam, with beam rigid body rotations accounting for most of the applied
drift at the end of the test. Specimen 2, with steel band plates and transverse beams, showed an
improved bearing response when compared to that observed in Specimen 1. The band plates
effectively confined the concrete regions located above and below the connection, forcing a shear
failure of the joint region while beam rigid body rotations were kept below 1.5% for drift cycles up
to 4%. Shear distortions measured on the column lateral faces exceeded 1.5% in both beam-to-
column connections. Specimens 1 and 2 experienced a 50% increase in the joint strength when
compared to that measured in Specimens 6 and 9, respectively, tested by Parra-Montesinos and
Wight (2000). Test results indicated that the use of steel columns and band plates was effective
to mobilize the full core of the concrete column even in connections between wide concrete
columns and steel beams.
Collaborative research program between the National Center for Earthquake Engineering,
Taipei, Taiwan and Stanford University
A full-scale three-bay, three-story composite frame was tested at the National Center for
Earthquake Engineering in Taipei, Taiwan (Deierlein et al. 2005, Cordova et al. 2004, Chen et al.
2004). The test frame included 650 mm square precast RC columns. The steel beams were 600,
500 and 396 mm deep at the first, second and third story, respectively. Welded shear studs over
the steel beam length ensured composite action with the concrete slab over a metal deck. Joint
details included face bearing plates, top and bottom steel band plates and transverse steel
beams. Column longitudinal bars in the top level were tag-welded to a steel cap plate located on
top end of the RC column. The frame was designed to meet the code requirements of a structure
built in California and Taiwan and was subjected to four pseudo-dynamic tests scaled to represent
seismic hazard levels with 50%, 10%, and 2% chance of exceedance in 50 years. Test results
showed that these events correlated well with limited structural damage, repairable damage, and
collapse prevention, respectively, and that the structure exhibited stable behavior through large
deformations. A final push-over test was conducted up to an inter-story drift of approximately 10%.
At this drift level, the lateral mechanism consisted of hinging concentrated at the base of the first
story columns, top of the second story columns, and beam ends of the first story. Beyond 7% drift,
bolted beam splices located at the third points of the beam span fractured, which led to a decrease
in the lateral strength of the frame. Tests results showed that the behavior of the frame exceeded
the performance expectations implied by modern design codes and was consistent, if not better,
than that expected of conventional steel or RC moment frame buildings.
Update of design provisions for moment-resisting joints between steel beams and
reinforced concrete columns
Based on the 1994 Guidelines and the results from experimental research conducted over the
past 30 years, new design provisions for moment-resisting joints between steel beams and RC
columns are currently under development. The design provisions broaden the scope of the 1994
Guidelines to consider the following: (a) explicit consideration of joint deformations, (b) high
strength concrete within the joint region, (c) designs with steel band plates and transverse beams,
and (d) design for high seismic regions (e.g., seismic design category D in ASCE 7). A comparison
between the equations given in the Guidelines (1994) and those to appear in the updated design
provisions can be found in Table 1. The detailing requirements given in Sections 4.3, 4.4 and 4.5
of the 1994 Guidelines, related to the thickness of face bearing plates, flanges of the steel beam,
and extended face bearing plates, respectively, remain unchanged in the updated provisions.
The updated design provisions have significant advantages over the 1994 Guidelines in terms of
computational effort: (a) the joint strength can be estimated without the need for iteration to
determine the value of jh in equations 7 and 7a shown in Table 1, and (b) the calculation of the
bearing strength is done in terms of the bearing forces acting on the inner panel (Fig. 4) and
expressed in terms of horizontal shear (Equation 5’ in Table 1). This allows for the use of a single
equation to estimate the joint strength (Equation 7’ in Table 1), as opposed to the procedure
described in the Guidelines, in which the bearing strength and shear strength of the joint are
checked independently (Equations 5 and 7 in Table 1, respectively). Note that a typical joint
designed with the updated provisions requires the application of 10 to 12 compact equations,
while the application of the 1994 Guidelines requires the use of 18 to 20 equations, some of which
are lengthy, and in some cases requiring an iterative process.
The calculated-to-measured strength ratios of the updated design provisions for 29 composite
connections are shown in Table 2. The average ratio is 0.88, with a standard deviation of 0.09
and a maximum and minimum of 1.09 and 0.70, respectively. The updated design provisions are
expected to be released in 2017 with perhaps minor modifications to the equations presented in
Table 1, after the feedback received during the review process before its publication.
Conclusions
Results from tests on composite joints conducted over the past 30 years between RC columns
and continuous steel beams, as well as from the test of a full-scale three-story, three-bay
composite frame under pseudo dynamic loading, validate the use of the composite frames in
zones of high seismicity and provide substantial information to update the design procedure given
in the 1994 Guidelines. The updated design provisions are calibrated using a database that
contains reversed cyclic loading test results of interior and exterior connections with a variety of
joint details, many of which are not included in the original guidelines. Details include face bearing
plates, steel columns, steel cover plates, steel band plates, extended face bearing plates, and
vertical joint reinforcement. The updated provisions allow the use of high strength concrete in the
connection region, explicitly account for connection deformations, and require less computational
effort than the 1994 Guidelines. The updated design provisions for composite connections
presented herein can be used to proportion composite frame joints of buildings located in high
seismic zones with a reasonable balance between the required computational effort and accuracy
in the strength estimation, as confirmed by a calculated-to-measured strength ratio of 0.88 with a
standard deviation of 0.09 for 29 subassembly connections tested to failure.
References:
ACI-ASCE Committee 352. (2010). Recommendations for design of beam column joints in
monolithic reinforced concrete structures. Report No. ACI 352R-10. Farmington Hills, MI:
American Concrete Institute.
AISC. (1994). Manual of Steel Construction - Load and resistance factor design (LRFD). Chicago:
American Institute of Steel Construction.
ASCE Task Committee on Design Criteria for Composite Structures in Steel and Concrete.
(1994). Guidelines for Design of Joints Between Steel Beams and Reinforced Concrete
Columns. Journal of Structural Engineering, 120(8), 2330-2357.
ASCE 7 (2010). Minimum Design Loads for Buildings and Other Structures, ASCE/SEI 7-10,
American Society of Civil Engineers.
Baba, N., & Nishimura, Y. (2000). Stress transfer on through beam type steel beam-reinforced
concrete column joints. Proceedings 6th ASCCS International Conference on Steel–
Concrete Composite Structures, 753-760.
Bugeja, M., Bracci, J. M., & Moore, W. P. (2000). Seismic behavior of composite RCS frame
systems. Journal of Structural Engineering, 126(4), 429-436.
Chen, C. H., Cordova, P., Lai, W. C., Deierlein, G. G., & Tsai, K. C. (2004). Pseudo-dynamic test
of full-scale RCS frame: Part 1 – Design, construction and testing. Paper No. 2178. 13th
World Conference on Earthquake Engineering. Vancouver, CA.
Cordova, P., Lai, W.C., Chen, C.H., & Tsai, K.C. (2004). Pseudo-dynamic test of full-scale RCS
frame: Part 2 – Analyses and design implications. Paper No. 674. 13th World Conference
on Earthquake Engineering. Vancouver, CA.
Deierlein, G. G., Sheikh, T. M., Yura, J. A., & Jirsa, J. O. (1989). Beam–column moment
connections for composite frames: Part 2. Journal of Structural Engineering, 115(11),
2877-2896
Deierlein, G. G., Cordova, P., Tsai, K. C., Chen, C. H., & Lai, W. C. (2005). Composite moment
frames. Concrete International, 27(05), 39-44.
Fargier-Gabaldon, L. B., & Parra-Montesinos, G. J. (August 2005). Seismic Behavior of RCS
Connections with Special Configurations. Report No. UMCEE 05-11. Ann Arbor, Michigan:
Department of Civil and Environmental Engineering, University of Michigan.
Fargier-Gabaldon, L. B., & Parra-Montesinos, G. J. (2006). Behavior of Reinforced Concrete
Column–Steel Beam Roof Level T-Connections under Displacement Reversals. Journal
Structural Engineering, ASCE, 132(7), 1041-1051.
FEMA. (2000). Recommended Seismic Evaluation and Upgrade Criteria for Existing Welded Steel
Moment-Frame Buildings, prepared by the SAC Joint Venture for the Federal Emergency
Management Agency-351. Washington, DC: Federal Emergency Management Agency.
Griffis, L. (1986). Some design considerations for composite-frame structures. AISC Engineering
Journal, 23(2), 59-64.
Kanno, R., & Deierlein, G. G. (1993). Strength, deformation, seismic resistance of joints between
steel beams and reinforced concrete columns. Report No. 93-6. Ithaca, N.Y.z: Cornell
University.
Kuramoto, H., & Noguchi, H. (1997). An overview of Japanese research on RCS systems.
Proceedings of ASCE Structures Congress XV (pp. 716-720). Reston, Virginia: ASCE.
Liang, X., & Parra-Montesinos, G. J. (2004). Seismic behavior of reinforced concrete column-steel
beam subassemblies and frame systems. Journal Structural Engineering, 130(2), 310-
319.
Nishiyama, I., Itadani, H., & Sugihiro, K. (1998). Bi-Directional Seismic Response of Reinforced
Concrete Column and Structural Steel Beam Subassemblies. Paper Ref. T177-2.
Proceedings of Structural Engineers World Congress (p. 8). Reston, Virginia: ASCE.
Nishiyama, I., Kuramoto, H., Itadani, H., & Sugihiro, K. (2000). Bi-Directional Behavior of Interior-
, Exterior, and Corner-Joints of RCS System. Paper No. 1911/6/A. Proceedings 12th
World Conference on Earthquake Engineering, (p. 8).
Noguchi, H., & Kim, K. (1997). Analysis of beam–column joints in hybrid structures. Proceedings
ASCE Structures Congress XV (pp. 726-730). Reston, Virginia: ASCE.
Noguchi, H., & Kim, K. (1998). Shear strength of beam-to-column connections in RCS system.
Paper Ref. T177-3. Proceedings Structural Engineers World Congress (p. 7). Reston,
Virginia: ASCE.
Parra-Montesinos, G., & Wight, J. K. (2000). Seismic behavior, strength, and retrofit of exterior
RC column-to-steel beam connections. UMCEE 00-09. Michigan: Department of Civil and
Environmental Engineering, University of Michigan.
Parra-Montesinos, G. J., & Wight, J. K. (2001). Modeling shear behavior of hybrid RCS beam-
column connections. Journal Structural Engineering, 127(1), 3-11.
Sheikh, T. M., Deierlein, G. G., Yura, J. A., & Jirsa, J. O. (1989). Beam–column moment
connections for composite frames: Part 1. Journal Structural Engineering, 115(11), 2858-
2876.
Tremblay, R., Bruneau, M., Nakashima, M., Prion, H. G., Filiatrault, M., & DeVall, R. (1996).
Seismic Design of Steel Buildings: Lessons from the 1995 Hyogo-ken Nanbu Earthquake.
Canadian Journal of Civil Engineering, 23(3), 757-770.
Table - 1. Summary of design equations
ASCE Guidelines (1994) Updated provisions

Width of the 𝑏𝑖 = max⁡(𝑏𝑓 , 𝑏𝑝 ) 𝑏𝑖 = 𝑏𝑓
inner panel
Connections with steel columns, or extended face Connections with steel columns, or extended face
bearing plates: bearing plates:
𝑏𝑜 = 𝐶(𝑏𝑚 + 𝑏𝑖 ) < 2𝑑𝑜 ⁡⁡⁡⁡⁡⁡[1] 2
Width of the 𝑏𝑜 = 𝑦 + 𝑥 − 𝑏𝑓 ⁡ ≤ 𝑏 − 𝑏𝑓 ⁡⁡⁡⁡[1′ ]
3
outer panel
(𝑏𝑓 + 𝑏) 𝑥𝑦 Connections with steel band plates
𝑏𝑚 = ≤ 𝑏𝑓 + ℎ ≤ 1.75𝑏𝑓 ⁡… 𝐶 = ⁡⁡[1𝑎]
2 ℎ 𝑏𝑓 ⁡⁡⁡⁡⁡⁡⁡𝑏𝑜 = 12𝑡𝑏𝑝 ≤ 𝑏 − 𝑏𝑓 ⁡⁡[1′ ]
Nominal 𝑉𝑠𝑛 = 0.6⁡𝐹𝑦𝑠𝑝 𝑡𝑠𝑝 ⁡𝑗ℎ⁡⁡⁡⁡[2] 𝑉𝑠𝑝𝑛 = 0.6⁡𝐹𝑦𝑠𝑝 ⁡𝑡𝑠𝑝 ⁡𝛼𝑠𝑝 ⁡ℎ⁡⁡⁡⁡[2′]⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡
horizontal
shear
An iterative process is required to find 𝑗ℎ (see 𝛼𝑠𝑝 = 0.9⁡ interior joints, 𝛼𝑠𝑝 = 0.8 exterior
strength of
Equation 7a) No iteration is required
the steel web
(inner panel)
Nominal 𝑉𝑐𝑠𝑛 = 1.7√𝑓𝑐′ 𝑏𝑝 ℎ ≤ 0.5⁡𝑓𝑐′ ⁡𝑏𝑝 𝑑𝑤 ⁡⁡⁡[3] 𝑉𝑖𝑐𝑛 = 𝛼𝑐𝑠 ⁡1.7⁡√𝑓𝑐′ ⁡𝑏𝑖 ⁡ℎ⁡ ≤ 0.5⁡𝑓𝑐′ ⁡𝑏𝑓 ⁡𝑑𝑗 ⁡⁡[3′]
horizontal
shear
strength of Does not diferentiate between interior and exterior 𝛼𝑐𝑠 = 1.0⁡ interior joints, 𝛼𝑐𝑠 = 0.6 exterior. Similar to
the concrete connections. the ACI 352 (2010) provision.
strut (inner
panel)
Strength given by a compression field mechanism: Strength given by the contribution of a diagonal
concrete strut (𝑉𝑜𝑛 )
𝑉𝑐𝑓𝑛 = V𝑐′ + V𝑠′ ≤ 1.7√𝑓𝑐′ 𝑏𝑜 ℎ⁡⁡⁡[4]
Nominal
𝑉𝑜𝑛 = 𝛼𝑐𝑠 1.25⁡√𝑓𝑐′ 𝑏𝑜 ℎ⁡⁡[4′ ]
horizontal
shear V𝑐′ = 0.4√𝑓𝑐′ ⁡𝑏𝑜 ℎ⁡⁡⁡⁡[4𝑎]
strength of 𝛼𝑐𝑠 = 1.0⁡ interior joints, 𝛼𝑐𝑠 = 0.6 exterior
the outer
𝐴𝑠ℎ 𝐹𝑦𝑠𝑝 0.9ℎ 𝐴𝑠ℎ
concrete V𝑠′ = …⁡[4𝑏] … ≥ 0.004⁡𝑏⁡⁡[4𝑐] Joints with steel band plates and transverse beams:
panel 𝑠ℎ 𝑠ℎ ⁡⁡
no horizontal ties are required within the beam depth.
Other cases: provide horizontal ties in a stirrup
Does not differentiate between interior and volume-to-joint volume of 0.01.
exterior connections
Vertical bearing strength of the joint is considered The vertical bearing strength of the inner panel in
appropriate if [5] is satisfied: terms of horizontal shear is given by (Figure 4):
∑𝑀𝑐 ≤ 𝜙[0.7ℎ𝐶𝑐𝑛 + ℎ𝑣𝑟 (𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 )] (𝑀𝑣 − 𝑉𝑏 ⁡ℎ)

− 0.35ℎ∆𝑉𝑏 ⁡⁡⁡⁡⁡[5] 𝑉𝑖𝑏 = ⁡⁡⁡[5′ ]
𝑑𝑗
𝐶𝑐𝑛 = 2𝑓𝑐′ 𝑏𝑗 𝑎𝑐 ⁡⁡⁡[5𝑎] … 𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 ≤ 𝐶𝑐𝑛 ⁄2 ⁡⁡[5𝑏] ⁡⁡⁡𝐶𝑐𝑛 = 𝛼𝑐𝑛 ⁡𝑓𝑐′ ⁡𝑏𝑓 ⁡(0.3⁡ℎ)⁡⁡⁡[5𝑎′]
Bearing ℎ ℎ2 𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 ≤ 𝐶𝑐𝑛 ⁄2 ⁡⁡⁡[5𝑏 ′ ]

strength 𝑏𝑗 = 𝑏𝑖 + 𝑏𝑜 ⁡⁡⁡[5𝑐] … 𝑎𝑐 = − √ − 𝐾 ≤ 0.3ℎ⁡⁡⁡⁡[5𝑑]
2 4
𝑀𝑣 = 𝐶𝑐𝑛 ⁡(0.7⁡ℎ) + ℎ𝑣𝑟 (𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 )⁡⁡⁡
1
𝐾= [∑𝑀𝑐 + ∆𝑉𝑏 ℎ⁄2⁡ 𝛼𝑐𝑛 = 2.5 band plates, 𝛼𝑐𝑛 = 2.0 other
𝜙2⁡𝑓𝑐′ 𝑏𝑗
− 𝜙(𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 )ℎ𝑣𝑟 ]⁡⁡⁡[5𝑒]
Joint bearing strength is checked with Equation [6′]
𝜙 = 0.7
Equation 5a’ takes into account the beneficial effect of
steel band plates in the joint bearing strength
ASCE Guidelines (1994) Updated provisions
𝜙⁡𝑉𝑖 = 𝜙[𝑉𝑠𝑛 + 𝑉𝑐𝑠𝑛 ]⁡⁡⁡[6] 𝜙⁡𝑉𝑖 = min[⁡𝜙𝑠 (𝑉𝑠𝑝 + 𝑉𝑖𝑐 )⁡⁡⁡;⁡⁡𝜙𝑏 ⁡𝑉𝑖𝑏 ]⁡⁡⁡⁡[6′]
Nominal
strength of The strength of the inner panel given in [6] is 𝜙𝑠 = 0.85, 𝜙𝑏 = 0.75
the inner independent of the connection bearing strength
panel in which is checked with Eq. (5)
The strength of the inner panel is taken as the lesser
terms of of the horizontal shear strength of the inner panel
horizontal 𝜙 = 0.7 𝜙𝑠 ⁡(𝑉𝑠𝑝 + 𝑉𝑖𝑐 ) and the vertical bearing strength of the
shear
inner panel in terms of horizontal shear (𝜙𝑏 ⁡𝑉𝑖𝑏 ) (Eq.
5’)
The shear strength of the joint is checked with [7]: The shear strength of the joint is checked with [7′ ]:
∑𝑀𝑐 − 𝑉𝑏 𝑗ℎ ≤ 𝜙[𝑉𝑠𝑛 𝑑𝑓 + 0.75𝑉𝑐𝑠𝑛 𝑑𝑤 𝛴⁡𝑀𝑏

− 𝑉𝑐 ≤ 𝑘⁡[𝜙𝑉𝑖𝑛 + 𝜙𝑠 𝑉𝑜𝑛 ]⁡⁡⁡⁡[7′ ]
+ 𝑉𝑐𝑓𝑛 (𝑑 + 𝑑𝑜 )]⁡⁡⁡⁡[7] 𝑑𝑗
Strength of
the
connection ∑𝑀𝑐 k = 1.0 where large joint deformations are allowed,
𝑗ℎ = ≥ 0.7ℎ⁡⁡⁡⁡[7𝑎] and k = 0.85 where the intent is to limit joint
∆𝑉
𝜙(𝑇𝑣𝑟𝑛 + 𝐶𝑣𝑟𝑛 + 𝐶𝑐 ) − 𝑏⁄2 deformations
An iterative process is required to find 𝑗ℎ

Ties required
above and 𝑉𝑐𝑓𝑛 𝑉𝑜𝑛
𝐴𝑡 ≥ ⁡⁡⁡⁡[9] 𝐴𝑡 ≥ ⁡⁡⁡⁡[9′ ]
below the 𝐹𝑦𝑡 𝐹𝑦𝑡
steel beam
Joint depth-
to-bar (𝑑 + 2𝑑𝑜 )⁡ 𝑑𝑗 ⁡420
⁡ ≥ 20⁡⁡⁡⁡⁡[10] ≥ 20⁡⁡⁡⁡⁡⁡[10′ ]
diameter 𝑑𝑏 𝑑𝑏 ⁡𝐹𝑦𝑟
ratio
 Interior, exterior, top T and top corner  Interior, exterior, top T and top corner (knee)
(knee) joints. joints.
 Joint aspect ratio 3/4 ≤ h/d ≤ 2  Joint aspect ratio 2/3 ≤ h/d ≤ 3/2
 𝑓𝑐′ ≤ 40⁡𝑀𝑃𝑎  𝑓𝑐′ ≤ 100⁡𝑀𝑃𝑎
Scope of the  Reinforcing steel F𝑦𝑟 ≤ 410⁡𝑀𝑃𝑎  Reinforcing steel F𝑦𝑟 ≤ 410⁡𝑀𝑃𝑎
provisions
 Structural steel F𝑦 ≤ 345⁡𝑀𝑃𝑎  Structural steel F𝑦 ≤ 345⁡𝑀𝑃𝑎
 Zones of low-to-moderate seismicity  Zones of high seismicity
 Applicable to control joint damage
Table - 2. Calculated-to-measured strength ratios
(updated design provisions)
Calculated-to-
measured Failure
Research Program Specimen Joint Details
strength mode
ratio
4 FBP 1.06 Shear
5 Thick FBP 1.03 Shear
Sheikh et al. (1989) 6 Doubler plates 0.75 Shear
7 Wide FBP (1.5 bf) 0.80 Shear
8 Extended FBP 0.77 Shear
10 FBP 1.08 Shear
Deierlein et al. 11 FBP 0.93 Bearing

(1989) 15 FBP+SC 0.91 Shear
17 FBP+SC+VJR 1.09 (max) Bearing
OB1-1 FBP+EFBP 0.92 Shear
OBJS1-1 FBP+EFBP 0.90 Shear
OBJS2-0 FBP 0.83 Bearing
OJS2-0 FBP+SC 1.02 Bearing
OJS3-0 FBP+EFBP 0.88 Bearing
OJS5-0 FBP+EFBP+BP 0.82 Bearing
Kanno (1993) OJS6-0 FBP+EFBP 1.00 Bearing
HJS1-0+ FBP+EFBP 0.92 Bearing
HJS2-1+ FBP+EFBP 0.89 Bearing

OJB-1 FBP+SC 0.85 Shear
OJB4-0 FBP+SC 0.76 Shear
1 FBP+SC 1.03 Bearing

Parra-Montesinos
6 FBP+SC 1.01 Shear
and Wight (2001)
9 FBP+SC+SBP 0.88 Bearing
Fargier-Gabaldón
and Parra- 1 FBP+SC+SBP 0.70 (min) Shear
Montesinos (2005)
2 FBP+SC+SBP 0.76 Bearing
FBP: Face bearing plates
EFBP: Extended FBP
Average () 0.88
SC: Steel column
Std () 0.09
SBP: Steel band plate
+ Featured 100 MPa concrete.
h b
Vspn
tsp
Steel panel
sp h
(a) Steel web panel
Inner concrete strut
Vicn
bi
Face bearing
plate
(b) Inner diagonal concrete

strut
Steel column or
extended face
bearing plates
Von
Outer concrete bo/2

strut
bj
Tie reinforcement
below the beam
(c) Outer diagonal concrete strut (case of steel
column or extended face bearing plates)
Steel band
plate
Von
Outer bo/2
concrete strut
(e) Outer diagonal concrete strut (case of steel band plates)

Fig. - 1 Joint shear resisting mechanisms
Concrete
crushing
Gap
(a) (b)
Fig. - 2 Joint failure modes: (a) Panel shear and (b) Vertical bearing
Extended face Stiffener Embedded
bearing plates Steel column
Face bearing plates

(a) (b)
Stiffener
Steel band plate
Face bearing
Transverse plate
(c) (d)
beam
Cover plate
Vertical joint
reinforcement Note: Transverse steel
(f) beam shall be used in
(e)
combination with steel
band for joint confinement
Fig. - 3 Joints details
Cvr Tvr
hvr
Web panel
cn 𝑓𝑐′
0.3 h Face bearing plates
dj
Vertical joint Concrete strut

reinforcement 0.3 h
cn 𝑓𝑐′
Tvr Cvr
hvr
Fig. - 4 Bearing forces acting on the inner panel
Fig. - 5 Hysteresis of Specimen 2 reported in Liang and Parra-Montesinos (2004)
Notation
𝐴𝑡 = minimum cross sectional area of ties or band plates located within a distance 0.4d above and below the beam;
𝐴𝑠ℎ = cross-sectional area of reinforcement parallel to beam (including cross ties) with spacing sh;
𝑎𝑐 = length of concrete bearing zone;
𝑏 = width the concrete column measured perpendicular to the beam;
b𝑓 = width of steel beam flanges;
𝑏𝑖 = width of inner concrete panel;
𝑏𝑗 = effective width of joint panel;
𝑏𝑚 = maximum effective width of joint region
𝑏𝑜 = width of outer concrete panel;
𝑏𝑝 = width of FBP;
𝑏𝑝 ′= width of steel column or extended FBP;
𝐶 = mobilization coefficient;
𝐶𝑐 = compression force in concrete bearing zone;
C𝑐𝑛 = nominal compression strength of bearing zone;
C𝑣𝑟 = compression force in vertical reinforcement;
C𝑣𝑟𝑛 = nominal compression strength of vertical reinforcement;
𝑑⁡ = depth of steel beam measure parallel to de column axis;
𝑑𝑏 = diameter of column longitudinal bars;
d𝑏𝑝 = depth of steel band plate measured parallel to column;
𝑑𝑐 = steel column depth measured parallel to beam;
𝑑𝑓 = distance between beam flange centerlines;
𝑑𝑗 = effective joint depth = d − t w ;
𝑑𝑤 = distance between beam flanges (height of web);
𝐹𝑦 ⁡= specified yield strength of the structural steel;
𝐹𝑦𝑟 = specified yield strength of the column longitudinal bars;
F𝑦𝑠𝑝 = specified yield strength of steel web panel in the joint region;
𝑓𝑐′ = specified compression strength of concrete within the joint region;
ℎ = depth of concrete column measured parallel to the beam axis;
h𝑣𝑟 =horizontal distance between vertical joint reinforcement;
𝑗ℎ = horizontal distance between bearing force resultant;
𝑘 = deformation-based strength reduction factor;
𝑀𝑣𝑏 ⁡= bearing moment strength of joint;
𝑠ℎ = center-to-center spacing of column ties;
𝑇𝑣𝑟 = tension force in vertical joint reinforcement;
𝑇𝑣𝑟𝑛 = nominal tension strength of vertical joint reinforcement;
𝑡𝑓 = thickness of beam flanges;
t 𝑠𝑝 = thickness of steel web panel;
V𝑏 = beam shear;
𝑉𝑐𝑓𝑛 = 𝑉𝑐′ + 𝑉𝑠′ =nominal horizontal contribution of the compression field strength;
V𝑐 = column shear;
𝑉𝑖𝑏 = vertical bearing strength of the inner panel in terms of horizontal shear;
V𝑖𝑛 = nominal horizontal shear strength of inner panel;
V𝑖𝑐𝑛 = horizontal shear strength of inner diagonal concrete strut;
𝑉𝑠′ = reinforcing steel contribution to nominal compression field strength;
𝑉𝑐′ = concrete contribution to nominal compression field strength;
V𝑛 = nominal horizontal shear strength of the connection;
V𝑜𝑛 = nominal horizontal shear strength of outer panel (same as the outer diagonal concrete strut);
V𝑗 ⁡= horizontal joint shear;
V𝑠𝑝𝑛 = nominal horizontal shear strength of steel web panel;
V𝑢𝑗 = required connection strength in terms of horizontal joint shear;
𝑥 = h for the case of extended face bearing plates, h/2 + d c/2 for the case of steel columns;
𝑦 = width of the steel column flange or extended face bearing plate;
𝜙 = resistance (capacity reduction) factor;
∆𝑉𝑏 = net vertical beam shear transferred into column;
∆𝑉𝑐 = net horizontal column shear transferred into beam;
𝜌𝑠 = tie-to-joint volume ratio (volume of ties within the beam depth divided by the joint volume, (dj h b);
∑𝑀𝑐 = net column moments transferred through joint; ∑𝑀𝑏 = net beam moments transferred through joint.
Innovative Slip-Critical Blind Bolts and Connections
for Concrete-filled RHS Columns
Wei Wang1,2, Mingxiao Li1,2, Yiyi Chen1,2 and

1 State Key Laboratory of Disaster Reduction in Civil Engineering,
Tongji University, Shanghai 200092, China;
2 Department of Structural Engineering, Tongji University, Shanghai 200092, China
weiwang@tongji.edu.cn
Xiaogang Jian3
3 School of Mechanical Engineering, Tongji University, Shanghai 200092, China
jianxgg@tongji.edu.cn
ABSTRACT. This paper presents an experimental investigation on beam-to-column blind bolted

endplate connections using innovative slip-critical blind bolts (SCBB). Eight full-scale beam-to-
column cruciform extended endplate connections using the SCBB were subsequently tested
under cyclic loading. Three types of stiffening strategies were adopted for the RHS columns, i.e.
inner diaphragms, thickened column wall, and filled concrete. For all the cases, the SCBB
performed very well and exhibited almost identical behaviour to the traditional high strength slip-
critical bolts. Two failure modes of the connections, namely, beam bucking (mode 1) and column
damage (mode 2), were generally observed and were found to be influenced by the column
stiffening strategies. The stiffening measures were also found to greatly influence the stiffness
and strength of the connections. The test result generally showed that a properly design SCBB
connection could exhibit very satisfactory cyclic performance, demonstrating a great potential of
such concrete filled RHS column-to-beam connection for seismic applications.
1 INTRODUCTION
Concrete filled columns have become an attractive choice for steel and composite building
construction, owing to several advantages over open sections. A square or rectangular hollow
section (SHS or RHS) column is a typical kind of tubular column which presents the advantage of
good bending and torsional resistance, due to its much larger flexural and torsional modulus than
open sections. Besides, the column wall can provide constraints for filled concrete, a case which
increases the compressive capacity of the concrete. Moreover, the surface painting becomes
much easier for concrete filled tubular structures and the appearance is more elegant from an
architectural point of view. Meanwhile, concrete filled RHS/SHS columns can be connected with
H-beam more easily than circular hollow section (CHS) columns.
Despite the advantages described above, traditional bolting is not readily applicable to the
connections between H-beams and concrete filled RHS/SHS columns due to the inaccessibility
of the internal space of the tube. Nowadays, welding is still the main method employed for such
connections, where using interior diaphragms, through diaphragms, and external diaphragms are
typical solutions. These kinds of connections usually have large initial stiffness, and thus they can
be considered as rigid connections according to Eurocode 3 [1]. However, compared with the
bolting solution, welding is more demanding in terms of construction effort and may cause quality
issues, recalling that a large number of welded connections were severely damaged in the 1994
Northridge and 1995 Hyogoken-Nanbu earthquake. Alternatively, blind bolts, which can be
installed and fastened from the outer side of the column, offer a convenient way for connecting
H-beams to concrete filled tubular columns. Several types of blind bolts have been developed
over the past several decades, including BOM, HSBB and Ultra-Twist from the US, Flowdrill
technology from Netherland, Hollo-bolt, RMH and EHB from the UK and Ajax Oneside from
Australia. However, it is found that almost none of them can achieve slip-critical bolted
connections except Ajax-Oneside from Australia.
Mourad et al. [2] conducted experiments on connections using HSBB (High Strength Blind Bolt)
under cyclic loading, but the sleeves were peeled off, leading to decreased preloading level of the
bolts. Mourad et al. [3] proposed a design method for extended endplate connections using Ultra-
Twist (a new blind bolt in the USA), and some constructional requirements were also given.
Harada et al. [4] found that TCBB (Torque-control high-strength Blind Bolt, the same as Ultra-
Twist) experienced unexpected pulling out failure of the blind bolt. Yeomans et al. [5] tested
several endplate connections with hollo-bolts (published by the British Steel Tubes & Pipes [6])
and also found that the bolts were pulled out, which indicated low pre-tightening forces of the blind
bolts. Elghazouli et al. [7] studied the monotonic and hysteretic behavior of hollo-bolt connections,
and good ductility was exhibited. A theoretical model for the initial stiffness and bearing capacity
of the hollo-bolt connections was proposed by Málaga-Chuquitaype [8]. Tizani et al. [9-10]
reported an investigation into the cyclic behaviour of endplate connections to concrete filled
tubular columns using the EHB. The results showed that the performance of the blind bolted
connections was mainly influenced by the failure modes, and connections with relatively weak
columns could exhibit higher energy dissipation and ductility. Lee et al. [11-12] reported an
experimental program investigation on blind bolted connections to unfilled hollow section columns
using Ajax Oneside [13] from Australia.
To further explore the potential of blind bolts for H-beam to concrete filled RHS/SHS columns, an
innovative Slip-Critical Blind Bolt (SCBB) is proposed at Tongji University. The component parts,
pre-tightening mechanism, installing procedures and the performances of SCBB are totally
different with other blind bolts. A novel split-type spacer, allowing almost the same pre-tightening
mechanism as traditional high strength bolt, was introduced into this blind bolt. This special
detailing provides possibility for SCBB to possess sufficient pre-tightening force recommended by
major codes and makes the blind bolt slip-critical. Experiments were conducted to investigate the
connecting performances of the bolts and the hysteretic behaviour of the connections using SCBB
with various types of stiffening measure, and the failure modes, strength capacity, ductility, initial
stiffness, and energy dissipation are fully examined.
2 EXPERIMENTAL PROGRAM
2.1 Test specimens
Eight full-scale beam-to-column cruciform extended endplate connections using the SCBB were
tested under cyclic loading. The specimens are summarized in Table 1. For ease of identification,
each specimen was assigned with a short name, which starts with the thickness of column wall,
followed by the strengthening strategies, bolt types or other special characteristics. For example,
C6DA5 represents the connection with 6mm column wall, inner diaphragms and axial load ratio
0.5 for the column. The nomenclature used for other specimens follows the same format, in which
CF stands for the concrete filled in the column, HB stands for hollo-bolt, and SB stands for strong
beam. Three types of stiffening strategies, namely, inner diaphragms, thickened column wall and
concrete filled columns, were adopted to strengthen the SHS columns, and the connection
detailing is shown in Figure 1. The SHS columns used in the tests were all cold-formed, with the
thickness (tc) varying between 6 and 12 mm. The beams for specimens S1 through S7 were H-
shaped (250×125×3.2×4.5) built-up sections and were fabricated by welding. A hot rolled H-
shaped steel beam with a larger section (300×150×6×8 mm) was employed for specimen C6SB.
Four rows of M24 SCBB (except C6HB) were used to connect the endplate (550×200×24 mm) to
the column, and stiffeners were used between the beams and the extended parts of the endplates.
For the parameter matrix among the eight specimens, specimen C6 did not have any stiffening
measures. Inner diaphragms were added into the SHS columns of specimens C6D and C6DA5,
with a column axial load ratio of 0.2 and 0.5, respectively. The local wall thickness (at the
connection zone only) of the column of C12 was twice the thickness of C6, such that the stiffness
of the column panel zone was increased. C6CF and C12CF were designed to study the influence
of concrete filled columns, in comparison to C6 and C12, respectively. Hollo-bolt (designed by
Lindapter, UK) was used in C6HB, designed to check the differences between the two types of
blind bolts. Beams of larger sectional dimensions were adopted in C6SB, thus making C6SB a
typical specimen of column failure.
Table 1 – Overview of the eight specimens
Column Beam Axial
No Specimen Notes
/mm /mm load ratio
S1 C6 250×6 No stiffening
0.2
S2 C6D 250×6 Diaphragm & axial load ratio 0.2
S3 C6DA5 250×6 0.5 Diaphragm & axial load ratio 0.5
S4 C12 250×6~12 250×125 Thickened column wall
S5 C6CF 250×6 ×3.2×4.5 Concrete filled
Thickened column wall & concrete
S6 C12CF 250×6~12
0.2 filled
S7 C6HB 250×6 With hollo-bolt
300×150
S8 C6SB 250×6 Strong beam-to-weak column
×6×8
The material properties of the specimens are shown in Table 2. Tensile coupon tests conforming
to [14] were carried out to obtain the yield strength, ultimate strength and elongation at fracture of
the varying parts of the steel specimens. The compressive strength of the 150 mm concrete cubes
were tested soon after the experiment, with an average fc,cube value of 35MPa.
(a) C6, C6CF, C6SB (b) C12, C12CF
(c) C6D, C6DA5 (d) C6HB
Figure 1. – Configuration of the connections.
Table 2. – Material properties.
Steel type fy (MPa) fu (MPa) Elongation at fracture
Thin column 313 461 35%
Thick column 411 568 23%
Beam (except C6SB) 350 466 33%
Beam (C6SB) 418 584 29%
Endplate 351 527 30%
2.2 Test set-up

The general test set-up is shown in Figure 2. The effective length of the column is 3180 mm, and
the beam span is 1800 mm. Both far ends of the beams were pinned in plane, and spherical
hinges were set at the top and bottom of the columns, respectively. To resist the horizontal
reaction load, a bracing with a pipe section was placed between the top of the column and the
reaction frame. Two actuators were vertically positioned at the east and west ends of the beams.
Both beams were laterally supported by universal ball joints, avoiding out-of-plane stability
problems. Besides, to study the influence of the axial load ratio, a constant axial load was applied
to the column through a jack at the top.
Jack
Spherical hinge
+3.525
+3.330
Top horizontal bracing
Reaction frame
SHS column
Out-of-plane bracing omitted for clarity
+1.920
H-beam H-beam
West East
Actuator Actuator
Spherical hinge
±0.000 Ground girder
Figure 2. – Schematic of test set-up.
2.3 Loading procedure
Quasi-static cyclic loading procedure based on AISC 341-10 [15] was adopted, as illustrated in
Figure 3, in which the whole procedure is displacement-controlled. The two actuators were
connected to the same pump, and the loads at both beam ends were applied synchronously.
150 0.08
0.06
100
0.04
Displacement  (mm)
50
Drift ratio  (rad)

0.02
0 0
-0.02
-50
-0.04
-100
-0.06
-150 -0.08
0 2 4 6 8 10 12 14 16 18 20 22
Cycle number
Figure 3. – Illustration of AISC cyclic loading procedure.
3 FAILURE MODES AND OBSERVATIONS

The end-plate was designed according to Chinese code CECS102-2002 [16], thus it is relatively
strong in comparison to the H-beams and concrete filled SHS columns. Therefore, the failure
modes were mainly influenced by the beams and columns, and the typical failure modes of the
connections are shown in Figure 4. Two types of failure modes were generally identified which
are described as follows:
Failure mode I: severe inelastic buckling of beam sections. The beams of this failure mode were
relatively weaker than the column and the connection zone, thus buckling occurred to the beam
sections after its initiation of section yielding. C12CF (shown in Figure 4(a)) is a typical specimen
exhibiting failure mode I, where the column and the connection zone generally stayed elastic.
Failure mode II: excessive plastic deformation of column walls. The strength of the column wall
was relatively small, and the column acted as the primary source of deformation. Due to the
pulling-out tendency of the bolts, gaps were developed between the end-plates and the column
walls, and were kept broadening with the increasing of the inter-story drift angle. Test C6SB is
taken as representative of failure mode II, as shown in Figure 4.
(a) Connection with failure mode I (b) Connection with failure mode II
Figure 4. – Representative failure of the connections.
3.1 Specimens C6 & C6CF
For specimen C6, at the beginning of the loading process, the tensile action of the blind bolts
caused local bending deformation of the column front walls accompanied by out-of-plane
deformation of the column side walls. Soon afterwards, the column entered into the inelastic
deformation stage. Due to the relatively thin column wall, a gap, with a maximum value of 22 mm,
between the endplate and the column front wall was evidently formed. The steel beams
experienced evident local buckling when the inter-story drift ( ) approached 7% and the load
carrying capacity began to fall. Fractures occurred to the welds connecting the web and flange of
the east beam (=9%), leading to a sharp drop (more than 50%) of the load carrying capacity.
The filled concrete in specimen C6CF restricted the concave deformation of the column wall, but
the gap between the end-plate and the column front wall was still visible, although narrower than
that of C6. The white wash applied over the panel showed that initial yielding started at 0.375%
drift. Yielding of the beam was first recorded at 1.5% drift. The beams started to buckle at 3% drift
and fractured at 6% drift, and afterwards the load carrying capacity started to decrease. For these
two connections, failure mode II appeared first, but they finally failed by mode I (inelastic buckling
of the beams).
3.2 Specimens C6D & C6DA5 with inner diaphragms
For specimen C6D, the diaphragm restricted the deformation of the column wall to some extent,
thus reducing the aforementioned gap. At 1.5% drift, a gap developed between the endplate and
the column at the height where the diaphragms located. The load reached its peak value at 4%
drift, and then dropped very quickly upon buckling of the beam flanges and webs. It is recalled
that the ductility of specimen C6D was lower than that of specimen C6. However, much higher
initial stiffness was obtained in specimen C6D, and this phenomenon will be discussed in detail
in section 4.2.
The observed failure mode of specimen C6DA5 (axial load ratio r=0.5) was similar to that of C6D.
Due to the larger axial load ratio, the column in C6DA5 yielded earlier, and its initial stiffness was
therefore slightly lower than C6D.
3.3 Specimens C12 & C12CF with thickened column wall
Due to the increased strength and stiffness of the panel zone, severe buckling occurred to the
beams very early and led to a rapid fall of the load carrying capacity of specimens C12 and C12CF.
Only a very small gap could be seen between the endplate and the column wall at 1.5% while the
beam section experienced full yielding. The beams started to buckle at =2%, a drift level which
is much smaller than that observed in the previously discussed specimens. The column of
specimen C12 almost remained elastic and the gap at the endplate was very small. The initial
stiffness was greatly improved due to the thickened column wall, but the ductility of the connection
was lower than specimen C6 and this was attributed to the early buckling of the H-shaped beams.
The column of specimen C12CF was further strengthened by the presence of the filled concrete,
and it almost stayed elastic during the entire test procedure. The load reached its peak value at
only 1.5% drift and the plastic hinges were observed at the sections where the stiffeners and
beams intersected. Beams buckled earlier compared with specimen C12.The loading capacity fell
by 75% at 5% drift due to the fracture of the welding. White wash peeled off from the beam surface
in a large area of the plastic hinge. The above test observations clearly indicate that the two
specimens (C12 and C12CF) fail by failure mode I.
3.4 Specimen C6HB with hollo-bolts
Hollo-bolts were used in specimen C6HB to examine the influence of bolt types on the connection
behavior. In a previous tensile test on hollo-bolts conducted by the authors, it was found that the
hollo-bolt is susceptible to shear failure of the sleeves, leading to decreased tensile capacity when
compared with the traditional high strength bolt. This explains the large difference between
specimen C6HB (hollo-bolt) and C6 (SCBB) which was that the column wall and the endplate in
specimen C6HB became separate where the bolts located when encountering tensile forces,
indicating low pre-tightening forces of the hollo-bolts. Some bolts even slacked at 8% drift with a
tendency of being pulled out, after which the hollo-bolts almost quitted working. Beams remained
elastic while the columns experienced excessive plasticity. This connection exhibited failure mode
II accompanied with bolt failure.
3.5 Specimen C6SB with stronger beams
The beams in specimen C6SB were relatively strong, leading to a ‘strong beam, weak column’
connection. The column demonstrated very large out-of-plane deformation and huge gaps were
observed between the endplate and column wall. The load carrying capacity was higher than that
of the other specimens because of the larger beam sections. Some of the bolt holes began to
crack at 8% drift, resulting in relaxation of the SCBB. Severe concave deformation occurred to
the column wall and the connection finally failed by mode II coupled with cracking of bolt holes.
4 MOMENT-ROTATION RESPONSES AND FURTHER DISCUSSIONS

4.1 Moment-rotation curves
The moment-rotation curves of the specimens are shown in Figure 5.
80 80
Mbp Mbp
60 60
Mby Mby
40 40
Moment M(kNm)
Moment M(kNm)
20 20
0 0
-20 -20
C6 C6D
-40 -40
Mby Mby
-60 -60
Mbp Mbp
-80 -80
-100 -80 -60 -40 -20 0 20 40 60 80 100 -80 -60 -40 -20 0 20 40 60 80
Drift ratio  (mrad) Drift ratio  (mrad)
(a) (b)
80 80
Mbp Mbp
60 60
Mby Mby
40 40
Moment M(kNm)
Moment M(kNm)
20 20
0 0
-20 -20
C6DA5 C12
-40 -40
Mby Mby
-60 -60
Mbp Mbp
-80 -80
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
(c) (d)
80 80
Mbp Mbp
60 60
Mby Mby
40 40
Moment M(kNm)
Moment M(kNm)
20 20
0 0
-20 -20
C6CF C12CF
-40 -40
Mby Mby
-60 -60
Mbp Mbp
-80 -80
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
(e) (f)
80 90
Mbp
60
Mby 60
40
Moment M(kNm)
Moment M(kNm)
30
20
0 0
-20
C6HB -30
-40 C6SB
Mby -60
-60
Mbp
-80 -90
-150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
(g) (h)
Figure 5. – Moment-rotation curves of each specimen.
Figure 5a and e show the hysteretic curves of specimens C6 and C6CF, both undergoing mode
II but finally failed in mode I. The curves exhibited pinching effect at the beginning due to the
excessive out-of-plane deformation of the column side walls and the gap between the column wall
and the endplate. After the column experienced yielding, the stiffness of the column became much
smaller, which led to the pinching hysteretic curves of the specimens. The curves experienced
sharp drops when severe welding fracture occurred in the last two or three cycles, and some
minor decrease of load took place due to beam buckling and micro-cracks of welds. The hysteretic
curve of specimen C6CF demonstrated lower ultimate displacement due to the filled concrete
which restricted the deformation of column and hence causing earlier buckling of the beams.
Figure 5b,c,d and f show the hysteretic curves of specimens C6D, C6DA5, C12 and C12CF that
failed by mode I (inelastic buckling of beams). The hysteretic curves dropped very early compared
with specimens C6 and C6CF due to the existence of inner diaphragms or thickened column walls,
leading to early buckling of the beams.
Figure 5g and h show the hysteretic curves of specimens C6HB and C6SB that failed by mode II
(column failure). These curves exhibited significant pinching effect and the beams did not buckle
during the whole test. The load carrying capacity of C6SB fell slowly with the cracking of the bolt
holes. The beams in specimen C6SB were stronger, and the ultimate load is significantly below
the one causing yielding of the beam section.
4.2 Initial stiffness and connection classification
The connections can be classified by their initial stiffness (Sini) and strength according to Eurocode
3 [1]. A connection may be classified as rigid when Siniis no less than kbEIb/Lb (kb equals 8 and 25
for non-sway and sway frames, respectively; E and Ib are the elastic modulus and moment of
inertia of the beam; Lb is the span of the beam, taken as 4 m in this paper), or nominally pinned
when Sini is no greater than 0.5EIb/Lb. A semi-rigid connection is that with Sini between kbEIb/Lb
and 0.5EIb/Lb. From a strength point of view, a connection may be classified as full-strength when
its design moment resistance (MRd) is no less than either the design plastic moment of the beam
(Mb,pl,Rd) or the column (Mc,pl,Rd), or nominally pinned when MRd is no greater than a quarter of the
minimum value of Mb,pl,Rd and Mc,pl,Rd, and again, a partial-strength connection is that with the
moment resistance in between.
It should be noted that when calculating the connection stiffness, some sources of deformation,
including beam elastic bending, column elastic bending and the rigid-body rotation of the column,
should be excluded. The connection classification result is shown in Figure 6 and Table 3.
Connection C6 without any stiffening measures is classified as semi-rigid with relatively low initial
stiffness. Specimens S2 through S5 all have very large stiffness and can be classified as rigid
(non-sway), which indicates that the inner diaphragms, the thickened column wall, or the filled
concrete can greatly increase the initial stiffness of the connections. Specimen C12CF has a
further increased stiffness due to the existences of both the thickened column wall and the filled
concrete, and thus it can be classified as rigid in sway frames. The result above also reveals that
the initial stiffness of the considered specimens is mainly influenced by the column panel zone.
Larger stiffness can be obtained in the connections with stronger columns. Specimens S1 through
S6 all failed in mode I finally, so they exhibited full-strength behavior. The remaining two
connections which failed in column or blind bolt are partial-strength connections.
90
Rigid (sway) Rigid (non-sway)
80
S8 S1-C6
70 S2-C6D
Full-strength
Moment M (kNm)
60 (Except S8) S3-C6DA5

S7 S4-C12
50
S5-C6CF
40 S2 S1
S6-C12CF
S5
30 S7-C6HB
S3
20 S4 S8-C6SB
S6
10 Nominally pinned
0
0 0.03 0.06 0.09 0.12 0.15 0.18 0.21
Drift ratio  (rad)
Figure 6. – Comparison of M- envelope curves.
Table 3. – Initial stiffness and connection classification.
Sini By
Specimen Features By stiffness
(109Nmm/rad) strength
C6 Not stiffened 3.894 Semi-rigid
Inner diaphragms
C6D 9.895
(r=0.2)
Inner diaphragms
C6DA5 8.993 Rigid (non- Full-
(r=0.5)
sway) strength
Double thickened
C12 13.974
column
C6CF With concrete filled 12.528
C12CF With concrete filled 33.414 Rigid(sway)
C6HB Hollo-bolt 1.650 Partial-
Semi-rigid
C6SB Stronger beams 4.649 strength
5 SUMMARY AND CONCLUSIONS

Eight full-scale beam-to-column cruciform extended endplate connections using innovative slip-
critical blind bolts have been tested under cyclic loading. Three types of stiffening strategies,
namely, inner diaphragms, thickened column wall, and filled concrete, were adopted for the SHS
columns. The load carrying capacity, ductility, initial stiffness, connection classification, stiffness
degradation and energy dissipation of the specimens were discussed in this paper. Confidence
was gained from the experimental study towards future application of the considered beam-to-
concrete filled RHS column blind bolted connections against seismic action. The main conclusions
can be summarized as follows:
(1) The SCBB performed very well and exhibited almost identical behavior to the traditional
high strength slip-critical bolts. On the contrary, some hollo-bolts used in specimen C6HB
experienced early relaxation prior to bolt hole damage, a phenomenon which indicated
possible shear failure of the sleeves and its low pre-tightening forces.
(2) Two types of failure modes were observed, i.e. plastic beam buckling failure (mode I) and
column failure (mode II). The connections with failure mode II exhibited pinching effect in
their moment-rotation curves, and led to higher ductility but lower damping ratio compared
with the specimens with failure mode I.
(3) The inner diaphragms, thickened column wall, or the concrete filled column can greatly
improve the initial stiffness of the connection, allowing a rigid connection behavior to be
exhibited according to Eurocode 3. The strength of the connections is influenced by the
failure mode, and is further affected by the relative resistance of each element.
(4) The SCBB and the corresponding connections have been proved to achieve the basic
standard of field weld-free construction of steel and composite buildings in high seismic
regions. The potential application of such connections is envisaged in the future.
ACKNOWLEDGEMENTS
The research presented in this paper was sponsored by the National “Twelfth Five-Year” Plan for
Science & Technology Support Program through Grant No. 2015BAL03B01-2.
REFERENCES
[1] BS EN 1993-1-8:2005. Eurocode 3: Design of Steel Structures - Part 1-8: Design Of Joints.
British Standards, 2005.
[2] Mourad S, Ghobarah A, Korol R M. “Dynamic Response of Hollow Section Frames with Bolted
Moment Connections”. Engineering Structures, 1995, 17(10):737-748.
[3] Mourad S, Korol R M, Ghobarah A. Design of extended end-plate connections for hollow
section columns. Canadian Journal of Civil Engineering, 1996,23(1):277-286.
[4] Harada Y, Arakaki T, Morita K. Structural behaviour of RHS column-to-H beam connection
with high strength bolts. Steel Structures, 2002,2:111-121.
[5] Yeomans N F. Rectangular hollow section column connections using the Lindapter Hollo-
Bolt. Tubular Structures VIII, 1998.
[6] SHS Jointing: Flowdrill & Hollo-Bolt. British Steel Tubes & Pipes,Corby, 1997.
[7] Elghazouli A Y, Málaga-Chuquitaype C, Castro J M, et al. Experimental monotonic and cyclic
behaviour of blind-bolted angle connections. Engineering Structures, 2009,31(11):2540-2553.
[8] Málaga-Chuquitaype C, Elghazouli A Y. Component-based mechanical models for blind-
bolted angle connections. Engineering Structures, 2010,32(10):3048-3067.
[9] Tizani W, Al-Mughairi A, Owen J S, et al. Rotational stiffness of a blind-bolted connection to
concrete-filled tubes using modified Hollo-Bolt. Journal of Constructional Steel Research,
2013,80:317-331.
[10] Tizani W, Wang Z Y, Hajirasouliha I. Hysteretic Performance of a New Blind Bolted
Connection to Concrete Filled Columns under Cyclic Loading: An Experimental Investigation.
Engineering Structures, 2013, 46(1):535-546.
[11] Lee J, Goldsworthy H M, Gad E F. Blind bolted T-stub connections to unfilled hollow section
columns in low rise structures. Journal of Constructional Steel Research, 2010, 66(8):981-
992.
[12] Lee J, Goldsworthy H M, Gad E F. Blind bolted moment connection to unfilled hollow section
columns using extended T-stub with back face support. Engineering Structures, 2011,
33(5):1710-1722.
[13] ONESIDE brochure. Ajax Engineered Fasteners - BN012 data sheet, 2002.
www.ajaxfast.com.au
[14] GB/T228.1-2010. Metallic materials – Tensile testing – Part 1: Method of test at room
temperature. Chinese National Code, 2010.
[15] ANSI/AISC 341-10. Seismic Provisions for Structural Steel Buildings. American National
Standard, 2010
[16] CECS102:2002. Technical Specification for Steel Structure of Light-weight Buildings with
Gabled Frames. China Association for Engineering Construction Standardization, 2002.
Braced Frame Connection with Square CFT Columns Penetrated by
Gusset Plates
Peng Wang#1
South China University of Technology
Guangzhou, China
pwang1987@126.com
Jianrong Pan#2*
Guangzhou, China
ctjrpan@scut.edu.cn
Jingmin Liang#3
Guangzhou, China
897864197@qq.com
Huashen Xie#4
Guangzhou, China
1450093791@qq.com
Zhan Wang#5
Guangzhou, China
wangzhan@scut.edu.cn
ABSTRACT
This paper describes an investigation into the behavior of braced frame connections with
square CFT columns, comprised of a penetrating gusset plate. Four simplified connections
with different construction details of the gusset plates were subjected to compressive load.
The force transferring mechanism between steel and concrete and the composite action in
the column were investigated. The experimental results revealed that the forces between
steel and concrete with penetrating brace frame connections were delivered through the
bearing caused by the edge of gusset plate, of the ribs or a hole added to the plate into the
concrete infill. The construction details including with ribs or a single hole significantly
enhanced the efficiency of force transmission and the composite action level in the column.
The penetration braced frame connections with square CFT columns assure that square CFT
column and composite connection are adequate to develop the required resistance for a wide
range of braced frame systems.
INTRODUCTION
Braced frames with CFT columns are stiff, strong and economical structural systems for the
reason that concrete-filled tube columns provide large axial stiffness and bracing system
provides substantial lateral strength and stiffness. Therefore, a wide range of
brace-beam-column (BBC) connection types have been used to connect braces and beams
to composite columns in braced frames with CFT columns, such as penetration connection
and internal diaphragm connection.
Braced frames with CFT columns usually require large diameter tubes, and the design bond
stress of this column is usually too small to satisfy the shear transfer demands. Penetration
connection is one kind of the viable BBC connections to improve the shear transfer capacity
of most CFT columns. Penetrating gusset plate connections aid in the force transferring by
adding direct bearing of steel on concrete, and this bearing mechanism could transfer forces
between the steel and concrete readily. There are several ways to achieve the penetrations of
connection, and the economical constructions of gusset plate penetrations are illustrated in
Figure 1. The connection gusset plate fits within slots in the steel tube during fabrication and
then welded to the tube. This type of steel-to-steel connection type is made for the beam and
brace attachments, and it permits quick and economical fabrication.
Shear connectors
Concrete Concrete
filled beam filled beam
tube tube
Brace Brace
(a)Gusset plate (b)Gusset plate with shear connection

Holes cut into Ribs welded to
gusset plate gusset plate
Concrete Concrete
filled beam filled beam
tube tube
Brace Brace
(C)Gusset plate with holes (d)Gusset plate with ribs

Fig.1- Typical braced frame connections with CFT
columns penetrated by gusset plates
Both circular and square CFT columns are widely used in recent years. Nevertheless, more
research studies have been conducted on circular CFT columns and connections. A series of
non-linear analyses (MacRea, Kimura and Roeder 2003) and experimental studies (MacRae,
Roeder, Gunderson and Kimura 2003) were completed to evaluate the behavior of
penetration connections of circular CFT columns. The results showed that penetration
connections of circular CFT columns in braced frames have enhanced the composite action
effect between steel and concrete and is revealed the law of load transfer and distribution
between steel plate and concrete. Design guidance is obtained from these studies results.
Braced frames with square CFT columns have more regular shape and better adaptability to
construction. However, little research has been carried out on the penetration connections of
braced frames with square CFT columns. This limits the generalization of braced frame with
square CFT columns. In this paper, experiments were conducted on the basis of simplified
braced frame connection with square CFT columns penetrated by gusset plates. The
connection behavior was discussed and the composite action of each specimen was
evaluated.
CONNECTION SIMPLIFIED
The common actual braced frame connections with square CFT columns penetrated by
gusset plates as show in Figure 2(a), would be difficult and expensive to carried out
experiment with realistic boundary conditions. Therefore, a simplified connection could be
used instead of the actual braced frame connection to be tested. Figure 2(b) shows how the
simplified connection relates to the prototype one. The simplified one represents the bottom
half of a gusset plate penetrated through a square CFT column, which is used to transfer the
downward brace force shown in Figure2 (a) into the concrete infill. This force is represented
in the experimental tests by a uniform vertical load applied at the top of the gusset plate. Due
to symmetry in the connection, the upper half of the connection prototype would have the
same behavior for an upward brace force. This simplified method has been proved to be
viable (Gunderson 2003).
Applied Gravity Load Applied

Brace Force Applied Load
Work Point
Vertical Component
of Brace Force Applied Load
Applied Gravity Load Applied Planform

Plus Brace Force Brace Force
(a)Brace-Beam-Column connection (b)Simplified CFT connection

idealization test setup
Fig.2 – Simple frame connection
Test specimens
The experimental study of simplified connection was carried out to determine the
effectiveness of each connection detail in transferring forces from the gusset plate to the
concrete infill and the tube, as well as the level of composite action of each connection type.
Four specimens were designed and built to simulate practice employed in common prototype
square CFT-BBC connections. For the standard specimen with plain gusset plate, as shown
in Figure 3, the simplified column adopted in test was 1000mm in height and 400mm in width.
The gusset plate fits within slots in the tube, which were cut to a depth of 400mm and width of
2mm plus the gusset plate thickness. The projection of the gusset plate above the tube was
approximately 50mm, which was used for loading. The gusset plate was welded on the
outside of each side of the tube all the way around by using 8mm fillet welds. The
construction details of each specimen are showed in Table 1.
The elastic modulus, yield strength and ultimate strength of the steel used were determined
by the material test before the connection test, providing relevant basis for test loading
analysis. Tension tests were carried on steel coupons cut from the tube and plate in the
connection tests. These included 9 steel coupons (three for each steel tube and plates) were
tested quasi-statically under monotonic tension to determine the stress-strain properties, the
mean values of three same specimens were summerised in Table 2, besides, three standard
concrete cylinders for the conrete slab were air cured beyond 28 days, the average
compression strength of concrete cylinders was 30.8Mpa.
500
50 200 200 50 400
D1 D2
25 25
D3 D4
50
90 100 100 100 100 50
Loading D5 D6
Head
400
8mm Fillet
Weld
Gusset
Plate
10
1000
300
100 100 100
Strain
Gauge
100
50 150
200
(a)Front view (b)Lateral view

Fig.3 – Typical test specimen dimensions and strain gauge instrumentation
Tab.1 – Details of each specimen
Specimen Steel Tube Size Plate Thickness

Plate Detail
Number (mm) (mm)
SJ-1 400×10×400 12 Plain
SJ-2 400×10×400 18 Plain
SJ-3 400×10×400 12 Single Hole:
SJ-4 400×10×400 12 6 Solid square ribs
Tab.2 –Mechanical properties for the structural steels
Material Thickness Yield Stress fy Ultimate Stress fu Elasticity Modulus

Sample (mm) (MPa) (MPa) Es(105MPa)
Plate-18mm 17.92 400 550 2.07
Plate-12mm 11.59 440 580 1.88
Tube 9.49 440 580 1.89
Test set-up and instrumentation

As shown in Figure 3, the displacement instruments were set on the loading device and both
the top of the steel tube and the concrete infill, used to measure the loading displacement and
the slippage between the steel and concrete. The strain gauges were attached to the main
part of the gusset plate and the steel tube, as shown in Figure 3 and Figure 4. The main
contents of the test include: vertical load of the gusset plate, vertical displacement of the
loading head, slippage between the steel and the concrete infill, strain of the tube and gusset
plate.
1 1
100 100 100 100 50
100 100 100 100 50
Strain Strain
Gauge Gauge
450
450
500 500
12 18
1-1 1-1
1 1
(a)Plain:12mm (b)Plain:18mm
1 1
67 67 100 50
100 50
Strain Strain
Gauge Gauge
18 18 18
15°
450
450
R 10
200
0
110
100
18 18
500 500
12 12
1-1 1-1
1 1
(c)Single hole (d)6 Solid square ribs

Fig. 4 – Gusset plate dimensions and strain gauge instrumentation
Loading procedure
As the monotonic loading test progressed, the axial load was applied to the gusset plate by
100kN each step until the yield of the gusset plate was achieved. Then, the load was applied
down to 50kN each step and stopped when out-of-plane bucking occurred at the gusset plate.
EXPERIMENTAL BEHAVIOR
General Behavior
The load applied to the plate was transferred to the square steel tube through the welds and
to the concrete by means of bearing and friction. Strain in the tube and concrete increased
with greater applied load. The final state of the specimen showed that there was obvious
plastic deformation in the top 50mm of the gusset plate, and the out-of-plane buckling
emerged at the gusset plate. The contact surface between the square steel tube and the
concrete was stripped.
Due to varying gusset plate thicknesses and eccentricities between the loading head and the
gusset plate, the yield load and the ultimate load of each specimen differed. The yield load is
defined as the load when the top 50mm of the gusset plate yielded and the ultimate load only
indicates the maximum applied load that the gusset plate could carry during the loading
process, and this value does not represent the failure in the connection. At the same load
level, the test results between the individual specimens can be compared. The main
parameters of the test result are summarized in Table3.
Tab.3 –Main parameters of test results
Yield Ultimate Max Max Composite

Specimen
Description Load Fy Load Fu Displace Slippage Action
Number
(kN) (kN) (mm) (mm) (%)
SJ1 Plain:12mm 2500 2950 7.10 0.342 47.5
SJ2 Plain:18mm 3500 4550 12.55 0.635 54.7
SJ3 Single hole 2100 2850 9.70 0.292 69.4
SJ4 6 Solid square ribs 2600 3200 10.50 0.185 85.4
Composite Behavior
The amount of composite action of specimen was described by the percentage of composite
action, Gpc, which was computed according to Eq.1(a). The εgauge in Eq.1(a) is equal to the
strain measured in the square steel tube during the experiment, εsteel is the expected strain
in the steel tube if there is no concrete and it is computed as P/(AtubeEsteel), where Atube is the
cross-section area of the tube, and Esteel is the elastic modulus of the steel of 189GPa, and P
is the axial load on the test specimen. The strain expected in the specimen if it is to remain
fully composite, εcomp , is computed according to Eq.1(b) where Aconc is the area of concrete in
the tube and Econc is the elastic modulus of concrete.
Respectively, when εgauge is completely equal to εsteel , the steel is carrying all of the force
and the present composite Gpc is equal to 0%, and if εgauge is equal to εcomp , then the
present composite Gpc is equal to 100%. The percent composite values were initially greater
than 100% in some cases because initial bond between the steel and concrete could have
contributed to the amplified.
The precent composite Gpc was calculated from strain gauges at 200mm of the steel, from
those 300mm of the steel, from the concrete strain 300mm, from the base and from
displacements in the bottom 600mm of the specimen as shown in Figure3. The square steel
tubes never reached their yield point within the lower 400mm, so the use of the maximum
change in strain can be used to establish an average stress. The Gpc from the strain gauges
at 200mm and 300mm of the steel were similar indicating little bond transfer in between these
gauges. The Gpc from the concrete stress gauges and from the displacements at the bottom
of the test specimen also indicated similar trends to those obtained from the steel strain
gauges. The final composite action values for each specimen represent a weighted average
of the four methods described above. As mentioned before, the composite action was high at
the initial loading phase, and the design process mainly considers the elastic phase before
the yield. Therefore, the average of percent composite action was calculated from the results
during the load degree of 1000kN to the yield.
εsteel -εgauge
Gpc = ×100% (1a)
εsteel -εcomp
P
εcomp = (1b)
Atube Esteel +Aconc Econc
120
Percent composite ( %)
100
80
60
40
SJ1
20 SJ2
SJ3
SJ4
0
0 1000 2000 3000 4000 5000
Load (kN)
Fig.5 – Composite action
The average of percent composite action for each specimen was computed according to
Eq.1(a), as shown in Table 3. The standard specimen SJ1 with 12mm plain gusset plate
showed the lowest percent composite action with result of 47.5%. Specimen SJ2, with the
thicker (18mm) gusset plate and greater bearing area, has a better composite action result of
55.03%. Specimen SJ3 had composite action of 69.4%, indicating that the gusset plate with a
single hole was more effective than the simple plain gusset plate. When the gusset plate was
welded six transverse ribs, the specimen SJ 4 possessed the best composite action of 85.4%,
which turned out that the gusset plate with ribs made a remarkable contribution on the
amount of composite action.
The percentage of composite action for each specimen at different levels of loading was
illustrated in Figure5. The result curve generally decreases as chemical bond, friction and
binding of the square CFT column were lost, but in some cases it slightly increased again
possibly maybe due to the fact that the bearing became more significant.
Relative slippage between concrete and steel

The relative slippage between the concrete infill and the square steel tube was another index
to indicate the composite performance of each specimen. These results demonstrate the
effectiveness of each construction detail of the gusset plate to minimize slippage and
increase the level of composite action. The slippage was measured at both the top of the
steel tube and the concrete infill and represent slippage deformation in the tube. The results
of the slippage are shown in Figure 6.
6000
5000
4000
Load (kN)
3000
2000
SJ1
SJ2
1000
SJ3
SJ4
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Slippage (mm)
Fig.6 – Slippage between steel tube and concrete core
The slippage between the concrete and steel at the top of the square steel tube raised with
increasing force and the relationship was approximately linear. The slippage results indicate
that the level of composite action and point at which the bond connection between the steel
and concrete breaks down. The graph reveals that all specimens had a slippage that was
approximately equal to zero under 500kN. The transition point of the slope at 500kN suggests
that the chemical bond between the concrete and steel has broken down. Friction controlled
the slippage during the loading degree of 500kN to 1000kN. And then the mechanical transfer
mechanisms such as bearing began to take over after approximately 1000kN. Since both the
concrete and square steel tube were restricted at the base, the amount of slippage was
limited.
The specimen SJ4 with ribs on the gusset plate exhibited the best result with the lowest
slippages of 0.185mm, which was reduced by 46% compared with SJ1. Low levels of
slippage was seen with ribs because they provide an increase area 6 times that of a plain
plate and acted as a key to lock the concrete and steel together. The slippage of specimen
SJ3 with a single hole cut into gusset plate was 0.292mm, which was reduced by 14%
compared with SJ1. The hole cut into gusset plate increased the bearing area.
Strain Along the Plate

The effectiveness of the gusset plate details were analyzed by placing strain gauges along
the mid-line location of the gusset plate, as showed in Figure4.The envelopes of results for
1000kN through the ultimate load of each specimen are showed in Figure7. The difference
between the gauge readings along the plate provides an indication of the transfer of forces
from the plate into the concrete.
400 400
Location along plate (mm)

Location along plate (mm) 300 300
200 200
1000kN 1000kN
100 2000kN 100 2000kN
2500kN 3500kN
2950kN 4550kN
0 0
0 1000 2000 3000 0 1000 2000 3000
Strain (με) Strain (με)
(a)SJ1 (b)SJ2
400 400

300 300
200 200
100 1000kN 100 1000kN

2000kN 2000kN
2100kN 2600kN
2850kN 3200kN
0 0
0 1000 2000 3000 0 1000 2000 3000
(c)SJ3 (d)SJ4
Fig.7 –Strain distribution along the plate
For the specimens with a plain gusset plate, SJ1 and SJ2, the force mainly transferred from
the plate into concrete through chemical bond and friction before reach to the yield load. The
curves are flat with small difference of strain between the top and the bottom, which means a
little amount of force transferred into the concrete during this stage. When exceeding the yield
load, the strain firstly decreased at the height of 400mm to 100mm and then increased at the
bottom 100mm. It turns out that some force transferred into the concrete infill through directly
bearing between the plate subface and the concrete infill.
The stain distribution along the plate for specimen SJ3, as seen in Figure7(c), firstly
increased with the maximum value at the height of 300mm, the location of the top of the hole.
Afterwards, the strain began to decrease obviously from the top to the end of the hole (at the
height of 900mm to 700mm). The distribution of the strain represents that large amount of
force had transferred into the concrete through the bearing area caused by the hole cut into
the plate. Furthermore, the strain at the bottom 100mm of the gusset plate was not increased
as described above for specimens SJ1 and SJ2, but showed a decreasing rule. It means the
force at bottom of the gusset plate was significantly reduced.
For specimen SJ4 with 6 ribs welded on the plate, the strains above and below the ribs were
slightly different at the loading degree of 1000kN. It was resulted of the small amount of initial
slippage between the steel tube and concrete. Nevertheless, the strain values above and
below the ribs showed a shape decrease at the maximum loading degree, especially the
difference of the top-most rib. The higher demands activated the top rib and allowed the
forces to be transferred from the plate into the concrete.
Strain Along the tube

The strain distributions along the tube of the specimens, as seen in Figure8, have the same
general shape, but it still shows some difference that caused by different gusset plate
construction details.
At the height of 900mm close to the loading head, the strain of the steel tube had a large
value and then decreased along the height of 900mm to 620mm, indicating that some force
was transferred from the tube to the concrete during this height range. The strain at the height
of 580mm, just under the gusset plate, was significantly increased because of the directly
bearing between the plate subface and the steel tube. And then the strain distribution in the
bottom 580mm height of the specimen showed a continuous decrease all the way down to
the base of the tube. This was caused by a bond mechanism between the steel tube and the
concrete infill that allowed the force to be transferred. Chemical bond, friction, and binding
between the steel and concrete were all possible mechanisms that would allow the force
transfer from the tube into the concrete infill.
At the loading degree of 1000kN and 2000kN, the strain distribution along the tube was
similar for all specimens. At the yield loading degree and the ultimate loading degree of
specimen SJ2, the sharp increase of strain along the tube as seen in Fig8(b) were due to
localized yielding in the tube along the welds. The stains distribution of the tube of specimen
SJ3 at the ultimate loading degree was significantly reduced at the height of 900mm-700mm.
With more transfer bearing area on the plate caused by the ribs, the demand on the tube of
SJ4 was much lower than SJ1. Moreover, the strain just under the gusset plate was also
lower for SJ4 than for SJ1 with a plain gusset plate. This reflects a better composite action of
specimen SJ4.
1000 1000
900 900
Location Along Tube (mm)
800 800
700 700
600 600
500 500
400 400
300 300
200 1000kN 200 1000kN
2000kN 2000kN
100 2500kN 100 3500kN
2950kN 4550kN
0 0
0 1000 2000 3000 0 1000 2000 3000 4000
(a)SJ1 (b)SJ2
1000 1000
900 900
800 800

700 700
600 600
500 500
400 400
300 300
1000kN 1000kN
200 2000kN 200 2000kN
100 2100kN 100 2600kN
2850kN 3200kN
0 0
0 1000 2000 3000 0 1000 2000 3000
(c)SJ3 (d)SJ4
Fig.8. –Strain distribution along the tube
CONCLUSTIONS
According to the experimental study on the simplified connection of brace frames with square
CFT columns penetrated by gusset plates, the efficiency of different construction details of
the connection gusset plate in force transferring and composite action could be estimated
through stain distribution, slippage, and the percentage of composite action. In general, the
following conclusions can be drawn:
1. The penetrating gusset plain plate can effectively transfer the applied load to the square
steel tube and the concrete infill resulted in a percent composite action value of 47.5%.
The applied load transfers to the concrete infill through friction and bearing, while it
transfers to the steel tube through the connection by the weld.
2. Increasing the thickness of the gusset plate was unable to improve the efficiency of the
concrete infill at the height range of the gusset plate. However, it could offer a greater
bearing area at the bottom between the gusset plate and the concrete, so as to improve
the efficiency of the concrete in the bottom part of the specimen, and resulted in a better
composite action value of 54.7% compared to the standard specimen SJ1.
3. The construction with a hole cut into the gusset plate could increase the bearing area
between the gusset plate and the concrete infill at the height range of the gusset plate,
and consequently increased the composite action from the value of the comparison
specimen SJ1 47.5% to 69.4%.
4. Comparisons of the various connection details showed connection with transverse ribs
welded on the gusset plate was the most effective in transferring force to the concrete
infill of the column due to the ribs detail provided the largest bearing area close to the face
of the gusset plate. The large transfer of force that occurred by the ribbed construction
details resulted in the highest percent composite action values of 85.4%.
ACKNOWLEDGMENTS
This study was supported by the National Nature Science Foundation of China (51378009;
51378219; 51638009), the State Key Laboratory of Subtropical Building Science, South
China University of Technology (2017ZB28).
REFERENCE
AISC. (1993). Load and resistance factor design specification for structural steel buildings,
American Institute of Steel Construction, Chicago, USA.
Gunderson, Chad. (2002). “Braced frame connections with concrete-filled tube (CFT)
columns.” MSCE thesis, Univ. of Washington, Seattle, USA.
Hajjar, J.F. (2008). “Composite steel and concrete structural systems for seismic engineering.”
Steel Construction, 58(5–8):703-723.
Jingmin L. (2017). “Braced fame connections with square CFT Columns penetrated by gusset
Plates.” MSCE thesis, South China Univ. of Technology, Guangzhou, China.
Macrae, G., Roeder, C.W., Gunderson, C., & Kimura, Y. (2004). “Brace-beam-column
connections for concentrically braced frames with concrete filled tube columns.” Journal of
Structural Engineering, 130(2), 233-243.
Macrae, G.A., Kimura, Y., & Roeder, C. (2004). “Effect of column stiffness on braced frame
seismic behavior.” Journal of Structural Engineering, 130(130), 381-391.
Mckenry M.(2002) “Behavior of concrete filled steel tubes in concentrically braced frame.”
MSCE thesis, Univ. of Washington, Seattle, USA.
Roeder, C.W. (1998). “Overview of hybrid and composite systems for seismic design in the
United States.” Engineering Structures, 20(97), 355–363.
Roeder C.W., MacRae G.A., &Waters C.(2000). “Seismic behavior of steel braced frame
connections to composite columns”. Steel Connections IV Workshop.
Roeder, C.W., Cameron, B., & Brown, C.B. (1999). “Composite action in concrete filled tubes.”
Journal of Structural Engineering, 125(5), 477-484.
INFLUENCE OF COMPOSITE SLAB ON THE NONLINEAR RESPONSE OF EXTENDED END-
PLATE BEAM-TO-COLUMN STEEL JOINTS
Roberto Tartaglia1,
University of Naples ‘Federico II’
Department of Structures for Engineering and Architecture
Naples, Italy
roberto.tartaglia@unina.it
Mario D’Aniello2,
Naples, Italy
mdaniel@unina.it
Gian Andrea Rassati3,

University of Cincinnati
Department of Civil and Architectural Engineering and Construction Management
Cincinnati OH, United States
gian.rassati@uc.edu
James Swanson4,
james.swanson@uc.edu
ABSTRACT
Extended end-plate connections are widely used in seismic resisting structures. Nowadays, this
type of joints are prequalified as all-steel assemblies, namely disconnecting the slab from any
protruding part of the connection in order to avoid composite behavior and to control the
hierarchy of resistance between beam and column. However, the most of the building frames
are designed considering steel-concrete composite slabs. The presence of the slab can strongly
influence both the local and the global behavior of structure and the joint. Indeed, the presence
of the composite slab increases both the stiffness and the strength of the joint thus affecting the
beam-to-column hierarchy. The aim of this work is to investigate the influence of the composite
slab on the joint response by means of finite element analyses. In particular, both the pure
flexural capacity and the robustness behavior were investigated for two joint configurations,
namely one fully steel, and the other considering also the composite slab with the
corresponding reinforcements and shear connectors.
Keywords: composite, finite element analysis, beam-to-column joints, extended end-plate

connection, prequalification.
INTRODUCTION
Nowadays, both AISC358-16 and EN1998-1-1 allow disregarding the interaction between the
slab and the steel joint provided that the concrete slab is disconnected from any protruding part
of the joint. Indeed, the current pre-qualification rules for steel beam-to-column joints (i.e.
AISC358-16 in United States of America and those recently developed within the framework of
Equaljoints project in Europe) have been developed for beam-to-column assemblies with
disconnected slab in the zone of the connection. However, the presence of composite slab can
influence both the response of qualified beam-to-column joints and the hierarchy of resistance
between the beam and the column in moment resisting frames (MRF) behavior. Indeed, even
though the concrete slab is locally disconnected both the continuity of the rebars and the
presence of transverse beams connected to the joints can interact together with the steel joint.
Therefore, several analytical and experimental studies have been carried out to evaluate the
contribution of concrete slab to the behavior of steel joints [Aribert (2006), Leon (1998),
Thermou (2004), Ciutina (2013), Lee (1989), and Simones (2001)]. The numerical response of
composite joints has been also investigated by several Authors [Amadio (2011, 2017), Braconi
(2008), Danku (2013), Salvatore (2005) Vasdravellis et al. (2009) Tartaglia et al. (2017b)], which
demonstrated the effectiveness of Finite element (FE) numerical models in predicting the
stiffness, strength, ductility, and energy dissipation capacity of composite joint assemblies. An
example of the accurate interpretation of the experimental results by FEM was shown by
Vasdravellis et al. (2009).
In light of these considerations, this paper describes and discusses the results of a preliminary
FE study devoted to investigate the influence of composite slab on the inelastic response of
steel joints subjected to both pure bending actions and catenary action induced by column loss
scenario. Among the possible types of beam-to-column joint, this study focuses on extended
stiffened end-plate bolted joints designed to behave as full strength according to the European
prequalification procedure (i.e. according to EQUALJOINTS). In order to highlight the role of the
slab, two different configurations were modeled, namely (i) all-steel joint without the presence of
the slab, (ii) the joint with the presence of the slab, with code-compliant details to avoid
theoretically the composite action between the joint and the slab. In this second configuration,
the FE model takes into account the steel sheeting, the concrete deck, the shear connectors
and the steel rebars.
The paper is organized into three main parts. The first part describes the investigated
parameters and the design assumptions considered for the examined joint assemblies. The
second part focuses on the modeling assumptions. The last part discusses the results of the
finite element simulations.
Description of the investigated joints

The investigated beam-to-column joints are extracted from a set of reference buildings properly
designed according to EN1998-1-1 and EN1993-1-1 (see Fig. 1a).
Secondary Beam
Column
(IPE330)
Primary Beam
Column
(HE340B)
(IPE450) Rib
Concrete slab
Secondary
Ly beam
Secondary Beam
(IPE330) Primiry beam
Continuity Rib
y
plate
(CP) Supplementary web plate
Lx (SWP)
x
a) b)
Figure 1: The reference structure and the features of the investigated joints.
The investigated assembly (see Fig. 1) consist in three beams connected to an external column
(HE340B); the two beams in the Y direction (IPE330, hereinafter called secondary beam) were
designed as composite with the slab according to EN1994-1-1. Contrariwise, the beam in the X
direction (IPE450, hereinafter called primary beam) was designed as all-steel. The secondary
beams were designed to resist solely gravity loads as well as their end-connections, which are
made of bolted double cleats (see Fig. 2b) that were designed according to EN1993-1-8 to
transfer all the shear force coming from the slab to the column.
Extended end-plate joint connects the primary beam to the column (see Fig. 2c). This joint was
designed to resist seismic actions according to component method (EN1993-1-8), namely the
joint was designed to guarantee the formation of plastic hinge in the beam without plastic
engagement of the connection and column web panel. The geometrical features of the
examined joints are reported in Tables 1 and 2.
The composite slab was properly designed according to EN1994-1-1 and EN1998-1-1 in order
to carry the vertical loads and to behave as diaphragm under seismic event. The thickness of
the concrete deck is equal to 75mm and A55-P600 G5 profiles sheeting with a depth equal to
55mm was used (see Fig. 3).
Table 1: Geometry of the end-plate connection.

Diameter End-Plate Rib
Bolts hole hEP bEP tEP p0 p1 p2 tRib Slope
[mm] [mm] [mm] [mm] [mm] [mm] [mm] [mm] [mm] [-]
30 33 870 280 25 50 75 180 20 40°
Table 2: Geometry of the double cleats connection.

Diameter double cleat
Bolts Hole hAP bAP tAP lAP e0 e1 e2
[mm] [mm] [mm] [mm] [mm] [mm] [mm] [mm] [mm]
16 18 240 90 5 180 55 65 65
Secondary
Double cleat beam
End-plate Column
End-plate (tEP)
Rib SWP
p0
p1
Primary beam
p2
Double cleat
Secondary
beam
Primary beam
hEP
a)
Secondary beam
Double cleat (tAP) Bolts M30
e0 Rib (tRib)
e1
hAP
e2
bEP
Bolts M16
bAP
b) c)
Figure 2: Connection geometry: a) planar view, b) angular connection and c) extended stiffens
connection.
200mm 200mm
55mm 75mm
130mm
a) b) c)
Figure 3: Details of the composite slab.
According to EN1994-1-1 pr. 7.4 in order to limit the crack opening on the concrete slab, it was
assumed a mesh 200mmx200mm of φ10 steel rebars with a concrete cover of 30mm.
Shear connectors were used to guarantee the composite action between the slab and the
beams. The distribution of the studs on the secondary beam is uniform with a spacing of 150mm
to ensure a perfect connection.
For that concerns the primary beams, it should be highlighted that the MRF was designed with
steel-only dissipative zones (concept c EN1998-1-1 pr. 7.1.2). Therefore, in order to prevent the
involvement of the slab in the resistance of the dissipative zones, a circular area around the
column with a radius equal to 2 times beff (where beff is the larger of the primary beams effective
widths) was maintained in accordance with EC8. In line with this hypothesis, also the connection
between the slab and the column should be avoided; in particular a gap equal to 50 mm
between the concrete and any protruding part of the connection was adopted.
THE FE MODEL
The influence of the concrete slab on the joint global behavior was investigated by means of
numerical analyses performed in Abaqus 6.14.
Two finite element models (FEMs) were developed. The first model (ES-F-AS) (see Fig. 4a) was
built disregarding the presence of the slab, but considering its contribution to the lateral torsional
buckling with the introduction of additional constrains on the beams (as described by Tartaglia
et al. (2018)). The second model (ES-F-WS) is more refined and it accounts for the concrete
deck, the steel sheeting, the steel rebars and the shear connectors (see Fig. 4b and 4c).
a) b)
Concrete slab
Steel Rebars
Nelson studs
Nelson studs
c)
Figure 4: The two investigated models: a) ES-F-AS, b) and c) ES-F-WS.
Most of the model parts (beams, bolts, plates, stiffeners and concrete slab) were modelled using
C3D8R (i.e. An 8-node linear brick, reduced integration, hourglass control) solid finite elements.
Only the reinforcement and the shear connectors (in ES-F-WS model) were modelled by means
of beam elements (B31 a 2-node linear beam in space) with their respective cross section in
order to reduce the computational effort. The steel profiled sheeting was modelled by means of
continuum shell elements (SC8R an 8-node quadrilateral in-plane general-purpose continuum
shell, reduced integration with hourglass control, finite membrane strains) with the thickness of
the real steel sheet (1 mm). The material properties were defined for each element individually
in function of the adopted finite element (beam, shell or solid). Therefore, the material assumed
for the beams, the column and the stiffeners is an European S355 steel with average yield
stress equal to 1.25Fy, as recommended by EN1998-1-1, a steel density equal to 7850 kg/m3
and a Young’s modulus and Poisson ratio respectively equal to 210 GPa and 0.3. The plastic
hardening was simulated using both nonlinear kinematic and isotropic hardening law on the
basis of a set of coupon tests performed at University of Naples “Federico II”. Shear connectors
and steel deck have the same elastic properties but a simplified linear branch was considered
for the post-yielding behavior. The mechanical behavior of the bolts was simulated as reported
in Tartaglia et al. (2018). The material properties of the concrete were simulated using the
concrete damage plasticity model proposed by Hillerborg et al. (1976). This model takes into
account the reduction of the elastic stiffness caused by plastic straining both in tension and
compression and it accounts as well for the stiffness recovery effects under dynamic loading
conditions (Abaqus User’s Manual, 2014). The characteristics of C25/30 concrete were defined
in accordance with EN1992-1-1, i.e. density of 2500 kg/m3, Young’s modulus of 31GPa, and
Poisson’s ratio of 0.2. Table 3 summarizes the adopted material properties of all the introduced
elements.
Explicit analyses were performed. A General Contact interaction was defined in order to
simulate the model interactions (i.e. between the bolts and the column and end-plate).The
Normal and Tangential Behavior were defined respectively introducing ‘Hard Contact’ and
‘Penalty’ friction formulation (with a coefficient equal to 0.3). In the ES-F-WS model the
connection between the shear studs and the beams was modelled using Multi-point constraints
(MPCs). Both the reinforcement and the shear studs were modelled with wire elements
(respectively wire-truss and wire-beam) and embedded in the concrete, while the steel sheet
was tied to the concrete slab.
Table 3: Material properties

Yielding Strenght Ultimate strength
Elements Material
[N/mm2] [N/mm2]
Beams S355 443.75 752.5
Column S355 443.75 753.5
Stiffners S355 443.75 754.5
Reinforcements B450C 450 540
Shear connectors SMD19105 235 360
Steel deck A55-P600 G5 260 450
Bolts Seismic conn. (M16) 10.9 802.5 916.3
Bolts Angular conn. (M30) 10.9 802.5 916.3
Both ES-F-AS and ES-F-WS models have the same boundary conditions, with the only
remarkable difference in the introduction of additional torsional constraints in the ES-F-AS
model to prevent lateral torsional buckling in the beam. The mesh was adequately defined in
function of both element type (solid, continuum shell and wire) and the expected stress
concentration patterns. Moreover, the computational time/result quality ratio was analyzed in
order to optimize the FEAs.
The bolt preloading was modelled as proposed by Pavlović et al. (2015), namely by applying a
temperature variation, to simulate an artificial bolt shrinkage. Both the pure monotonic actions
and the column loss scenario were applied at the primary beam extremity; in case of the seismic
loads the vertical displacement was monotonically increase up to 6% joint rotation. For the
analyses simulating the column loss scenario, the vertical displacement was increased up to a
rotational level of 20% on the connection, while preventing the horizontal displacement.
RESULTS
In this paragraph, the results of the FEAs performed on both ES-F-AS and ES-F-WS are
described in terms of moment rotation curve, yielding lines distribution and failure mode. As
anticipated in the previous section, the influence of the composite slab, on the joint behavior,
was investigated considering two possible scenarios: (i) monotonic bending and (ii) column
removal.
Monotonic bending condition
The monotonic bending condition was examined in order to investigate the joint response when
lateral forces (i.e. wind and earthquake) act on building frames. Moreover, due to the
asymmetry of the ES-F-WS model, two monotonic analyses, considering both the hogging (ES-
F-WS-M-) and the sagging (ES-F-WS-M+) moment, were analyzed.
As expected (see Fig. 5), despite the fact that the steel joint is disconnected from the slab, the
moment rotation curves show a slight difference between the responses of the two models. With
respect to the ES-F-AS, ES-F-WS presents an increase of capacity; indeed both the hogging
and the sagging moment-rotation curves show an increase of resistance almost equal to 5%.
These results confirm that, in spite of all the EN1994-1-1 and EN1998-1-1 rules being respected
there is an interaction between the steel joint and the slab.
It should be noted that the increase of flexural capacity does not depend on the moment
direction. Therefore, the resistant mechanism is not due to the direct interaction between the
slab and the steel resistant joint. Instead, it depends on the torsional resistance of the
secondary beam that offers an additional restraint to the slab.
1200
1000
Moment [kNm]
800
600
E2-F-AS
400
ES-F-WS-M+
200 ES-F-WS-M-
0
0.00 0.02 0.04 0.06
Chord rotatin [rad]
a) b)
c) d)
Figure 5: Seismic results in terms of Moment rotation curve (a) and PEEQ distribution (b, c and
d).
This observation is schematized in Fig. 6, where, for both hogging and sagging moment, it is
reported the relation between the rotation and vertical displacement of the secondary beam with
the slab induced displacements. Larger differences between the two models can be observed
comparing their elastic stiffness. Indeed, as reported in Fig. 5a, both hogging and sagging
moment show an increase of 15% in the case with the slab.
However, both the FE models show that the primary joint behave as full strength and the
hierarchy of resistance between the beam and the column is not compromised.
a) b) c)
Figure 6: Torsional effect on the secondary beam in case of hogging (b) and sagging (c)
moment
It is also interesting to note that both FE models show a similar value of the maximum plastic
deformation (0.068 and 0.071 for ES-F-AS and ES-F-WS respectively) in the primary beam.
Moreover, both models show a small contribution of the column and the bolts to the global
dissipated energy equal to 3.5% and 2% respectively.
a) b)
c) d)
Fig. 7: The development of the tension damage on the ES-F-WS subjected to hogging moment
at: 2% (a), 3% (b), 4% (c) and 6% (d) of chord rotation.
The tension damage (DAMAGET in Abaqus) on the concrete slab is reported in Figure 7. The
first crack opening can be observed at 2% of rotation (see Figure 7a) where the damage
distribution is mainly perpendicular to the primary beam axis around the shear studs. Increasing
the rotation up to 3%, it can be observed (see Figure 7b) a larger distribution of the tension
damage on the slab developing form the shear studs parallel to the primary beam. Finally, at 6%
of chord rotation (see Fig. 7d) the distribution of the tension damage is almost uniform on the
entire slab.
On the other hand, when the joint is subject to a sagging moment, the upper part of the slab will
be in compression and no significant damage can be observed (see Fig. 8a). All the tension
damage is in the lower part of the slab (see Fig. 8b).
a) b)
Figure 8: The ultimate tensional damage on the ES-F-WS subject to sagging moment.
Column loss scenario

The column removal scenario implies the loss of a column adjacent to the investigated joint. In
this case, the rotation demand is larger than that imposed in the pure monotonic scenario. In
addition, there is the development of the catenary action in the joint. Indeed, the presence of
axial restraints and the large deformation regime substantially modify the local demand in the
joint components as showed by Tartaglia et al. (2017a). Figure 9 show the moment-rotation
response, where the moments are computed as the algebraic sum of the first order moment (i.e.
the shear force at the tip of the cantilever beam by the corresponding shear length) and the
second order contribution due to the catenary effects (i.e. the catenary action N multiplied by the
deflection at the tip of the cantilever beam), which counterbalances the first order contribution
(see Fig. 9b). Moreover, the evolution of normal action was plotted, in order to underline the
great differences between the two FE models.
Indeed, the introduction of the composite slab strongly influences both the elastic stiffness
(around 38%) and the bending capacity (around 33%) of the joint. A large part of these
increments are due to the presence of the secondary beam (see Fig.10). Moreover, the
presence of the slab strongly increases the beam resistance to the normal actions.
ES-F-AS ES-F-WS
1600 2500
1400 2000
Normal Action [kN]

Moment [kNm]
1200
1000 1500
800 1000
600 500 E2-F-AS
E2-F-AS
400 ES-F-WS
200 ES-F-WS 0
0 -500
0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20
Chord rotatin [rad] Chord rotatin [rad]
a) b)
c) d)
Figure 9: Column loss scenario results: Moment rotation curve (a), evolution of catenary action
(b) and plastic deformation (PEEQ) (c and d).
Figure 10: The influence of secondary beams on the concrete slab deformation.
CONCLUSIONS
In light of the results of the finite element simulations presented and discussed in the previous
paragraphs, the following conclusions are drawn:
• A more refined model considering, not only the steel joint but also the presence of the
concrete slab, the rebar, the shear connectors and the secondary beams, allows
evaluating more accurately the joint behavior. In particular, the model with the slab (ES-
F-WS) shows an increase of both elastic stiffness and bending capacity.
• The increase of elastic stiffness is mainly due to the presence of the concrete slab and
its interaction with the steel joint. However, the increase of bending capacity is not due to
the direct transfer of shear force between the primary beam and the slab, but from the
torsional contribution of the secondary beams. Therefore, the modelling of secondary
beams ensures a more accurate characterization of the joint ultimate capacity; indeed
the underestimation of the joint capacity can modify the local hierarchy between the
column and the beam and compromise the structural ductility.
• The FE models show that the examined details allows the formation of plastic hinge in
the beam without interference with the slab.
• The influence of the slab is more evident when the column loss scenario is investigated.
Indeed, despite the joints failure mode are similar, with the involvement of the column
and the connection in the failure mechanism, both the bending capacity and the elastic
stiffness differ due to the interaction between the secondary beam and the slab.
REFERENCES:
Amadio C, Bella M, Bertoni V, Macorini L. (2011). Numerical modeling and seismic assessment
of steel and steel-concrete composite frames, the line 5 of the ReLUIS-DPC 2005–2008
Project, 409–448. Napoli, Italy: Doppiavoce.
Amadio C., Bedon C., Fasan M., Pecce M. R. (2017). Refined numerical modelling for the
structural assessment of steel-concrete composite beam-to-column joints under seismic
loads. Engineering Structures 138, 394–409.
http://dx.doi.org/10.1016/j.engstruct.2017.02.037
Aribert J, Ciutina A, Dubina D. (2006). Seismic response of composite structures including
actual behaviour of beam-to-column joints. Composite Construction Steel Concrete. V:708–
17.
Braconi A, Bursi OS, Fabbrocino G, Salvatore W, Taucer F, Tremblay R. (2008). Seismic
performance of a 3D full-scale high-ductility steel–concrete composite moment-resisting
structure— Part II: test results and analytical validation. Earthquake Engineering and
Structural Dynamics. 37:1635–55.
Cassiano D., D’Aniello M., Rebelo C., (2017) Parametric finite element analyses on flush end-
plate joints under column removal. Journal of Constructional Steel Research, 137, 77–92.
Ciutina A., Dubina D. and Danku, G. (2013) Influence of steel-concrete interaction in dissipative
zones of frames: I – Experimental study. Steel and Composite Structures. 15(3), 281-000.
http://dx.doi.org/10.12989/scs.2013.15.3.281.
Danku G., Dubina D., and Ciutina A. (2013). Influence of steel-concrete interaction in dissipative
zones of frames: II - Numerical study. Steel and Composite Structures. 15(3), 305-000.
http://dx.doi.org/10.12989/scs.2013.15.3.305
Dassault (2014), Abaqus 6.14 - Abaqus Analysis User's Manual, Dassault Systèmes Simulia
Corp.
EN 1993:1–1, Design of Steel Structures - Part 1–1: General rules and rules for buildings. CEN,
2005.
EN 1993:1–8, Design of Steel Structures - Part 1–8: Design of Joints. CEN, 2005.
EN 1994:1–1, Design of composite steel and concrete structures - Part 1–1: General rules and
rules for buildings. CEN, 2005.
EN 1998-1, Design of Structures for Earthquake Resistance - Part 1: General Rules, Seismic
Actions and Rules for Buildings. CEN, 2005.
EN 1992:1–1, Design of Concrete Structures - Part 1–1: General rules and rules for buildings.
CEN, 2005.
EQUALJOINTS – European pre-QUALified steel JOINTS: RFSR-CT-2013-00021. Research
Fund for Coal and Steel (RFCS) research programme.
Hillerborg A., Modeer M., and Petersson P-E. (1976). Analysis of crack formation and crack
growth in concrete by means of fracture mechanics and finite elements. Cement and
Concrete Research. 6, 773-782. Pergamon Press, Inc. Printed in the United States.
Lee S-J, Lu L-W. (1989) Cyclic tests of full-scale composite joint subassemblages. Journal of
Structural Engineering. 115(8):1977–98
Lee J., and Fenves G.L. (1998). Plastic-Damage Model for cyclic loading of concrete structures.
Journal of Engineering Mechanics 124(8) 892-900.
Leon, RT. (1998). Analysis and design problems for PR composite frames subjected to seismic
loads. Engineering and structures 20(4-6) 364-71.
Pavlović, M., Heistermann, C., Veljković, M., Pak, D., Feldmann, M., Rebelo, C., Simões da
Silva, L. (2015). Connections in towers for wind converters, part I: Evaluation of down-scaled
experiments. Journal of Constructional Steel Research.
Salvatore W, Bursi OS, Lucchesi D. (2005). Design, testing and analysis of high ductile partial-
strength steel–concrete composite beam-to-column joints. Computers and Structures. 83(28–
30):2334–52.
Simões R, Simões da Silva L. (2001). Cyclic behaviour of end-plate beam-to-column composite
joints. Steel and Composite Structures. 1(3):355–76.
Tartaglia R., D’Aniello M., (2017a) Nonlinear performance of extended stiffened end plate bolted
beam-to-column joints subjected to column removal. The Open Civil Engineering Journal, 11,
3-15.
Tartaglia R., D’Aniello M., Rassati G.A., Swanson J., Landolfo R., (2017b). Influence of
composite slab on the nonlinear response of extended end-plate beam-to-column joints. Key
Engineering Materials, 763:818-825.
Tartaglia R., D’Aniello M., Zimbru M., Landolfo R., (2018) Finite element simulations on the
ultimate response of extended stiffened end-plate joints. Steel and Composite Structures.
27(6):727-745.
Thermou GE, Elnashai AS, Plumier A, Doneux C. (2004). Seismic design and performance of
composite frames. Journal of Constructional of Steel Research. 60:31–57.
Vasdravellis G, Valente M, Castiglioni CA. (2009). Behavior of exterior partial-strength
composite beam-to-column connections: experimental study and numerical simulations.
Journal of Constructional Steel Research. 65:23–35.
NUMERICAL INVESTIGATION OF MOMENT-RESISTING SLIM-FLOOR
BEAM-TO-COLUMN CONNECTIONS
Cristian Vulcu, Rafaela Don, Adrian Ciutina, Dan Dubina

Politehnica University of Timisoara, Department of Steel Structures and Structural Mechanics
Timisoara, Romania
cristian.vulcu@upt.ro; rafaela.don@student.upt.ro; adrian.ciutina@upt.ro; dan.dubina@upt.ro
ABSTRACT
Slim-floor systems present various advantages: (i) additional floor-to-floor space; (ii) natural fire
protection for steel beams; (iii) reduction of material consumption. Slim-floor solutions are
currently used in non-seismic regions, and there are few studies that consider continuous or semi-
continuous fixing of slim-floor beams. Consequently, a study was performed with the aim to
develop reliable connections for slim-floor beams, which will ultimately be applicable to buildings
located in areas with seismic hazard. The current paper presents the undertaken numerical
procedures, and the outcome of the study. In a first stage, a finite element numerical model was
calibrated based on a four point bending test of a slim-floor beam. Furthermore, a case study was
performed in order to investigate continuous slim-floor beam-to-column connections. Two
different types of connections were considered: (i) bolted end-plate, (ii) welded cover plate. The
response was investigated considering both sagging and hogging bending moments.
INTRODUCTION
The structural solutions provided by the usage of composite elements are regarded as an effective
method of enhancing structural performance. A series of advantages emerge as concrete, steel
and additional components are integrated into a more resistant and ductile member (Arcelor-
Mittal: “Slim Floor an innovative concept for floors”). In particular, the slim-floor building system is
attractive to constructors and architects due to the integration of steel beams in the overall height
of the floor, which leads to additional floor-to-floor space, used mostly for acquiring additional
stories. The concrete slab offers natural fire protection to the steel beams, while the use of novel
corrugated steel sheeting reduces the concrete volume, and replaces the secondary beams (for
usual spans of steel structures).
Slim-floor solutions are currently used mostly in non-seismic regions – Hauf (2010) & Braun et al.
(2014), and there are few studies that consider continuous or semi-continuous fixing of slim-floor
beams. Malaska (2000) showed that the semi-continuous joining of slim-floor beams improves
the flexural stiffness of slim-floor beams and allows for the use of shallower beam and floor
sections, and for a better performance of beams under service conditions, by reducing cracking,
deflections and vibrational problems. Wang et al. (2009) and Bernuzzi (1995) showed that in case
of increasing gravitational loads, the continuous fixing of the slim-floor beams may lead to ductile
plastic hinges, both at beam-ends and middle spans. In contrast, the usual seismic behavior rely
on increased frame lateral stiffness and failure mechanisms by dissipation of seismic input
energy, through the plasticization of dissipative elements or connections. Consequently, in the
case of Moment-Resisting-Frames (MRF) or dual frame configurations considering MRF
contribution, the beams or the beam-to-column joints of MRF will dissipate energy through plastic
hinges. Therefore, the application of slim-floor beam systems in seismic zones should consider
moment-resisting connection with columns, thus developing hogging bending, too. However,
certain aspects characteristic for slim-floor systems should be considered (see Fig. 1):
 the concrete slab encases the top steel flange and needs one layer of hogging reinforcement;
 the natural bonding and/or the concrete dowels contribute significantly to the steel-to-
concrete connection. In many cases there is no need for additional connectors;
 the bottom part of steel profile is generally larger than the top flange, in order to accommodate
the concrete supporting system: shallow decking or precast concrete slabs;
Fig. 1 Slim-floor system.
The current paper presents a study performed with the aim to develop reliable connections for
slim-floor beams, with a view to the use in buildings located in areas with seismic hazard. The
paper presents the numerical investigations undertaken and the outcome of the study. In a first
stage, a finite element numerical model was calibrated based on a four point bending test of a
simply supported slim-floor beam. Furthermore, a case study was performed in order to
investigate continuous slim-floor beam-to-column connections. Two different types of connections
were considered: (i) bolted end-plate, (ii) welded cover plate. The response was investigated
considering both sagging and hogging bending moments.
FEM CALIBRATION OF A SLIM-FLOOR BEAM

Research background
The numerical analysis of slim-floor beam-to-column joints is based on an initial calibration of a
finite element (FE) numerical model based on the experimental investigation conducted by Hauf
(2010) on a four-point bending test of a slim-floor beam. This phase was needed due to the lack
of data regarding continuous connections for slim-floor beams. This also represented a
comparison basis which appropriately checked the accuracy of the beam-to-column connection
models. This information became crucial with regard to the behavior of a composite element, the
steel-to-concrete friction coefficient, the modeling procedure, the importance of “concrete dowels”
and reinforcement, the meshing techniques and interactions.
Calibration of the numerical model
The calibration of the numerical model was performed based on the data available in Hauf (2010).
This study presents the experimental testing of an 8 m long slim-floor assembly, represented by
a 4-point-load arrangement, as shown in Fig. 2. The various components of the slim-floor beam
are presented in Fig. 3. The complete technical information (e.g. geometry, boundary conditions,
materials, etc.) is available in: Hauf (2010), Braun et al. (2014) and Braun et al. (2015). The
numerical investigation was performed by means of the Abaqus v6.13 finite element modelling
software. It is to be noted that the measuring units adopted for the definition of the geometry,
material properties and loading conditions were millimeters (mm) and Newtons (N).
The material characteristics are defined within the numerical model for the following: concrete
(C30/37), structural steel (S355) and reinforcement bars (S450), based on the real mechanical
characteristics provided by the authors. The elastic behavior of the steel elements is described
by: Young’s modulus E=210000 N/mm2 and Poisson’s ratio ν=0.3, whereas the plastic behavior
is illustrated in Fig. 4a. The input characteristics for the concrete elastic domain are: Young’s
modulus E=32500 N/mm2 and Poisson’s ratio ν=0.2. The plastic domain behavior is defined by
the stress-strain curve shown in Fig. 4b, without considering the tensile resistance.
Fig. 2 Slim-floor beam – 3D view: static scheme; reinforcing bars arrangement.
Fig. 3 Slim-floor beam components: reinforcing bars, concrete deck, steel profile, and steel plate.
1000 40
35
800 C30/37
30
Stress [N/mm2]
Stress [N/mm2]
600 25
20
400 15
10
200 S450
5
S355
0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.000 0.005 0.010 0.015 0.020 0.025 0.030
(a) Strain [mm/mm] (b) Strain [mm/mm]
Fig. 4 Material model: (a) true stress - true strain curve for structural steel (S355) and reinforcing
bars (S450); (b) stress-strain relationship for concrete.
Three types of interactions were used in this numerical model: embedded, tie and rigid body. In
order to replicate the interface behavior of steel and concrete, an interaction law (and contact)
was defined with a normal and tangential behavior. The normal behavior was assigned by “hard
contact”, which allows for surface separation. The tangential behavior was characterized by
“penalty“ with a friction coefficient in amount of ν=0.6. The reinforcement was connected to the
concrete part using the embedded constraint.
The “Dynamic Explicit” type of analysis was considered, due to the large amount of contact
surfaces (e.g. between the steel profile and the surrounding concrete). Contact problems are
more easily solved by using the “Dynamic Explicit” analysis, in contrast to the “Static General”
analysis. In terms of the finite element type, the following have been used: (i) B31 for reinforcing
bars; (ii) C3D8R for steel profile, steel plates, concrete dowel and concrete. The approximate
global size of finite elements / mesh was considered as follows: reinforcing bars (30 mm), concrete
slab (23 mm), concrete dowel (12 mm), steel profile (13 mm), bottom steel plate (18 mm), lateral
steel plate (13 mm). An overall view of the meshed assembly is shown in Fig. 5.
Fig. 5 Overall assembly view: mesh discretization.
Calibration results
The results of the numerical investigation are presented in terms of “bending moment – vertical
displacement” curve (M-δ) in Fig. 6a. The deformation of the slim-floor beam model is illustrated
in Fig. 6b. As can be observed, the numerical model is able to accurately reproduce the response
of the tested specimen, in terms of both initial stiffness and composite assembly’s capacity. Being
justified by the symmetrical loading and boundary conditions, the response is illustrated only for
the left side. The stress distribution and the plastic strain are also presented for the following
components: concrete (Fig. 7a), steel profile (Fig. 7b), reinforcing bars (Fig. 7c).
2000
Bending moment [kNm]
1600
1200
800
400 Test
FEM
0
0 50 100 150 200 250
(a) Vertical displacement [mm] (b)
Fig. 6 Bending moment–vertical displacement curve (M-δ), Vertical displacement.
It can be easily observed in Fig. 7b that large plastic deformations (strain) appear in the steel
profile’s bottom flange, including the welded steel plate, similarly to the real failure mode.
Fig. 7a shows the stress distribution and the plastic strain that correspond to the concrete slab.
Following the maximum bending formation, maximum stresses are logically concentrated in the
middle of the span. However, in addition to the mid-length stress concentration, longitudinal
yielding appears nearby the supports, justified by the presence of the shear force, absent between
the two loading points.
Fig. 7c shows the stress distribution and the plastic strain in reinforcement bars. The maximum
values are found in the middle of the span. An important aim of the numerical simulation was to
involve the reinforcing bars and steel meshes in the load transfer mechanism. The inclined bars
contribute to a large extent to the load transfer mechanism by connecting the concrete slab to the
concrete located between the steel profile’s flanges, thus preventing the separation of the two.
Moreover, the reinforcing bars that pass through the web perforations, together with the effect of
concrete dowels, contribute to the composite action of the assembly.
(a)
(b)
(c)
Fig. 7 Mises stress distribution and plastic strain in: (a) concrete, (b) steel profile, (c) reinforcing
bars and meshes.
Concluding remarks
The calibration of the slim-floor beam has allowed setting various parameters related to the FE
modelling. Moreover, the numerical investigation has also allowed for the assessment of the load
transfer and the failure mechanism. In particular, the assembly failure mechanism resided in the
formation of a plastic hinge in the mid span, justified by the large plastic deformations of the steel
profile and the steel plate.
The FE calibration of the behavior of the slim-floor beam model leads to some important modeling
features:
 the finite element investigation of the slim-floor assembly implies modeling both the
transverse and the longitudinal reinforcing bars, as well as the inclined reinforcing bars. The
adding of reinforcement bars to the model may improve the overall behavior of the assembly
by enhancing its load bearing capacity. The overall involvement of the reinforcing bars and
steel meshes in the load transfer mechanism is evident in this case;
 the inclined bars emphasize great significance in the load transfer mechanism by connecting
the concrete slab to the concrete located between the steel profile flanges, thus preventing
the separation of the two. The reinforcing bars that go through the web perforations, together
with the concrete dowels, contribute to the composite action of the assembly.
NUMERICAL INVESTIGATION OF SLIM-FLOOR BEAM-TO-COLUMN CONNECTIONS
Configuration of the joint models
Based on previous calibration, six beam-to-column joint models were developed. Table 1 offers
an overview of the joint models, and shows for each model the following information: (i)
abbreviation, (ii) steel grade, (iii) concrete slab, (iv) transversal and longitudinal reinforcement, (v)
inclined reinforcement, (vi) connection type.
Table 1 Investigated beam-to-column joint assemblies with slim-floor beams.

Concrete T+L* Inclined
Nr. Abbrev. Steel Connection
slab reinforcement Reinforcement
1 M1-S S355 NO NO NO Bolted end-plate
2 M2-SC S355 YES Ø6/100 NO Bolted end-plate
3 M3-SC S355 YES Ø6/100 Ø6/100 Bolted end-plate
6 M6-SC S355 YES Ø10/100 Ø6/20 Welded cover plate
*T+L presence of the transversal and longitudinal reinforcement bars.
The numerical models of the joints were developed based on an iterative process. E.g., in order
to assess the importance of adding concrete and reinforcement bars to an assembly, the first
model (M1-S) only includes steel elements. The models developed later included the concrete
slab and different types of reinforcement.
Fig. 8 shows the configuration of the investigated beam-to-column joint model assemblies. In
particular, the T-shape joint assembly (see Fig. 8ab) is composed of a slim-floor beam and a steel
column. With a view to a better visualization of the joint components, the joint assembly in Fig. 8a
has been disassembled into components in Fig. 8b to Fig. 8g.
The concrete slab integrating the steel beam and the reinforcement (transversal, longitudinal and
inclined) has an effective width of 1200 mm and a height of 70 mm (computed according to the
current EN 1994-1 (2004) conditions) and surrounds the column over a width of 450 mm. The
concrete in the bottom troughs has been ignored in the analysis and consequently not modeled.
It is to be noted that the reinforcement present in M3-SC to M6-SC models satisfy the
reinforcement connection requirements of the Annex C of EN 1998-1-1 (2004).
The steel column profile is an HEB500, while the steel beam is composed by a bottom steel plate
(Pl-20x390x1300 mm) welded to a half of an IPE450 profile. The column length is of 3200 mm,
while the beam length is 1300 mm. An additional web plate and continuity plates have been used
in the column web panel. Transversal, longitudinal and inclined reinforcement has been used for
the concrete slab (see Fig. 8 c-d-e). The continuity of longitudinal reinforcing bars is assured
around the column (Fig. 8d). It is to be noted that the usual diameter of reinforcement bars for
slabs is Ø=6 mm, which justifies the choice of the first diameter used in the study (see Table 1).
Moreover, the longitudinal reinforcing bars have been included in order to contribute to the
negative bending moment capacity within the connection zone.
The connection between the slim-floor beam and the column is realized as: (i) bolted extended
end-plate connection (Fig. 8d), (ii) welded cover plate connection (Fig. 8g). The bolted connection
is realized using an extended end-plate and five bolt rows of M22-Gr.10.9 bolts (see Fig. 8d),
while the welded connection is realized using cover plates (see Fig. 8g).
(a) (b) (c) (f)
(d) (e) (g)

Fig. 8 Configuration of beam-to-column joint models: (a) overall joint, (b) slab reinforcement,
(c)/(d)/(e) steel beam and reinforcement (longitudinal and transversal), (f)/(g) bolted end-plate /
welded cover plate connection and inclined reinforcement.
Modelling procedure
The numerical investigations of the slim-floor beam-to-column joint assemblies (see Fig. 8) were
performed by means of the Abaqus v6.13 software. Finite beam elements have been used for the
reinforcement, and solid elements for the other components (bolts, plates, concrete, etc.). For the
definition of the geometry and mechanical characteristics, millimeters and Newton’s have been
used as measuring units. The material characteristics have been defined for the following nominal
characteristics: concrete (C30/37), structural steel (S355), bolts (H.R.10.9), and reinforcement
bars (S400), by considering both elastic and plastic properties. Fig. 9 illustrates the true stress –
true strain curves (except for the elastic deformation) for bolts (Gr.10.9), reinforcement (S400)
and structural steel (S355). The material model for bolts has been defined based on a previous
calibration of a T-stub characterized by failure mode 3 (i.e. bolt failure) – see Dubina et al. (2015).
For all steel elements, the elastic modulus for steel has been taken as 210000 N/mm2, and the
Poisson coefficient was 0.3. The material model behavior used for concrete is shown in Fig. 4b.
A “Dynamic Explicit” type of analysis has been adopted, justified by the complex contacts between
the various parts of the joint assembly (i.e. concrete, steel beam, bolts, end-plate, and column).
For the contact between the various parts of the joint assembly, a contact law was defined with
normal and tangential properties. The normal behavior was assigned through “hard contact”,
which allows for surface separation. The tangential behavior was characterized by “penalty“, with
a friction coefficient in amount of ν=0.6. The connection between the reinforcement and the
concrete slab was defined by using the “embedded region” constraint type. The discretization of
the components was performed by using: (i) B31 finite element types for the reinforcing bars, (ii)
C3D8R finite elements for the rest of the components. The global mesh size was adapted to
different FE: reinforcing bars – 20 mm; concrete – 17 mm; steel profile – 12 mm; bottom steel
plate – 14 mm; column – 16 mm; end plate – 12 mm; column web plate and stiffeners – 14 mm;
bolts – 7 mm. The discretization of the beam-to-column joint assembly / components is illustrated
in Fig. 10. Boundary conditions were specified for the column and the beam: (i) at the top and the
bottom ends of the column, a simple and respectively a fixed support were defined; (ii) at the tip
of the beam, the load was applied in displacement control, which resulted in a positive or negative
bending moment within the connection zone.
1400.0
Gr. 10.9
1200.0
S400
1000.0
Stress [N/mm2]
S355
800.0
600.0
400.0
200.0
0.0
0 0.05 0.1 0.15 0.2 0.25
Strain [mm/mm]
Fig. 9 True stress - true strain curve for steel (S355), reinforcement bars (S400) and bolts (Gr.10.9).
(a) (b) (c)

Fig. 10 Discretization of the beam-to-column joint assembly / connection.
Numerical result
The numerical models of the six beam-to-column joint assemblies were subjected to negative and
positive bending moments. The numerical program allowed assessing the response in terms of
moment-rotation curve, stress distribution and plastic strain. The failure mechanism was obtained,
and from the comparison of moment-rotation curves, the influence of each parameter was
assessed. In order to reach a uniformity of the results, the bending moments and the rotations
were computed at the column face.
The response of the M1-S joint assembly is illustrated in Fig. 11. Under negative bending moment
(Fig. 11a), the upper two bolt rows showed large plastic deformation (Fig. 11b), while under
positive bending moment (Fig. 11c), the lower two bolt rows failed (Fig. 11d). Consequently, under
both loading conditions, the failure mode was similar. The corresponding moment-rotation curves
are presented in Fig. 11e. The comparison showed identical stiffness but with slightly higher
capacity under negative bending, justified by the slightly un-symmetrical positioning of the bolt
rows.
500
400

300
200
100
0
-100
-200
M1-S [M-]
-300 M1-S [M+]
-400 M1-S [M-]
-500
(a) (b) -0.1 -0.075-0.05-0.025 0 0.025 0.05 0.075 0.1
Rotation [rad]
(e)
(a) von Mises stress distribution and (b) plastic

strain corresponding to negative bending
moment; (c) von Mises stress distribution and
(d) plastic strain corresponding to positive
bending moment; (e) moment-rotation curves
corresponding to positive and negative
bending moment.
(c) (d)
Fig. 11 Response of M1-S joint assembly.
Fig. 12 shows the response of the M2-SC and M3-SC joint configurations in terms of moment
rotation-curve (positive and negative). From the comparison between M1-S and M2-SC (see Fig.
12a), it can be observed that the presence of the reinforced concrete slab led to increased
capacity and stiffness. Under a positive bending moment, the lever arm increased due to the
presence of the concrete slab (70 mm thick above the upper beam flange). Under a negative
bending moment, although there was an obvious increase of the initial stiffness, the longitudinal
reinforcement within the slab (subjected to tension) slightly increased the capacity. The M2-SC
model showed a separation between the concrete deck and the concrete within the beam flanges.
This confirmed the importance of the inclined reinforcement that connected the two regions.
From the comparison between M2-SC and M3-SC (see Fig. 12b), it can be observed that the
presence of the inclined reinforcement (Ø6/100 mm) led to a slight increase of capacity under a
negative bending moment. Moreover, it was observed that the separation effect between the two
concrete regions (i.e. the concrete slab and the concrete within steel flanges) was reduced.
500 500
400 400
300 300
200 200
100 100
0 0
-100 -100
M2-SC [M+] M3-SC [M+]
-200 -200
M1-S [M+] M2-SC [M+]
-300 -300
M1-S [M-] M2-SC [M-]
-400 M2-SC [M-] -400
-500 M3-SC [M-]
-500
-0.1 -0.075 -0.05 -0.025 0 0.025 0.05 0.075 0.1 -0.1 -0.075 -0.05 -0.025 0 0.025 0.05 0.075 0.1
(a) Rotation [rad]
(b) Rotation [rad]
Fig. 12 Moment-rotation curves under positive and negative bending moment: (a) comparison
between M2-SC and M1-S assembly; (b) comparison between M3-SC and M2-SC assembly.
In the case of the M4-SC joint configuration, the longitudinal and the transversal reinforcement
was increased to Ø10/100 mm (see Table 1). The response of the M4-SC joint assembly is
illustrated in Fig. 13. Under both loading conditions, the failure modes were similar, i.e. failure of
bolts under tension. The corresponding moment-rotation curves are shown in Fig. 13e. The
comparison showed slightly higher stiffness and capacity under a positive bending moment. A
higher ductility corresponds to the negative bending moment, due to the presence of the
reinforced concrete slab.
(a) (b)
(c) (d)
500
400
300 (a) von Mises stress distribution and
200 (b) plastic strain corresponding to

100 negative bending moment;
0 (c) von Mises stress distribution and
-100 (d) plastic strain corresponding to
-200 M4-SC [M+]
M3-SC [M+]
positive bending moment;
-300 (e) moment-rotation curves
M3-SC [M-]
-400
M4-SC [M-] corresponding to positive and
-500
-0.1 -0.075 -0.05 -0.025 0 0.025 0.05 0.075 0.1
negative bending moments
(e) Rotation [rad]
Fig. 13 Response of the M4-SC joint assembly.
From the comparison between M4-SC joint model and M3-SC, respectively M5-SC – it was
observed that a higher amount of reinforcement led to a slightly higher capacity. For the same
amount of longitudinal and transversal reinforcement, the increase of the inclined reinforcing bars
did not improve the response significantly.
Fig. 14 shows the response of the M6-SC joint assembly (with welded cover plate connection).
Under negative and positive bending moments (see Fig. 14a / Fig. 14c), large plastic deformations
developed in the steel profile (see Fig. 14b and Fig. 14d), i.e. the failure mechanism consisted in
the formation of a plastic hinge in the slim-floor beam. The presence of the concrete slab led to a
slightly higher capacity under positive bending moment (Fig. 14e).
(a) (b)
(c) (d)
800
600
(a) von Mises stress distribution and (b)
400
200
plastic strain corresponding to negative
0
bending moment; (c) von Mises stress
distribution and (d) plastic strain
-200
corresponding to positive bending
-400
M6-SC [M+] moment; (e) moment-rotation curves
-600 M6-SC [M-] corresponding to positive and negative
-800 M6-SC [M-]
bending moments
-0.1 -0.075 -0.05 -0.025 0 0.025 0.05 0.075 0.1
(e) Rotation [rad]
Fig. 14 Response of the M6-SC joint assembly.
Fig. 15a shows an overview with regard to the response of three investigated joint assemblies.
The lowest stiffness and capacity corresponds to the bare steel joint (M1-S), and the highest
corresponds to the slim-floor beam-to-column joint with welded cover plate connection (M6-SC).
800 Rigid 0 M1-S [M-]

600 zone -100 M4-SC [M-]
MRd,beam M7-SC [M-]

400 -200 M6-SC [M-]
200
-300
0
M6-SC [M+] -400
-200 M4-SC [M+]
-400 M1-S [M+] -500
MRd,beam M1-S [M-]
-600 -600
Rigid M4-SC [M-]
-800 zone M6-SC [M-] -700
-0,1 -0,05 0 0,05 0,1 -0,1 -0,075 -0,05 -0,025 0
(a) Rotation [rad] (b) Rotation [rad]
Fig. 15 (a) Overview of slim-floor beam-to-column joint response; (b) Influence of the increased
amount of longitudinal reinforcement (in an additional joint model M7-SC) compared to M4-SC.
CONCLUSIONS
The current study has been performed with the aim to develop reliable connections between slim-
floor beams and columns – with a view to the application in structures located in seismic zones.
In a first stage, a numerical model was calibrated based on a four point bending test of a slim-
floor beam. Furthermore, a case study was performed in order to investigate two types of
continuous slim-floor beam-to-column connections: (i) bolted end-plate, (ii) welded cover plate.
The numerical calibration of the slim-floor beam allowed assessing the load transfer and the
failure mechanism (plastic hinge at mid span). It proved the feasibility and the performance of the
steel-concrete connection, i.e. reinforcing bars passing through the profile web and thus forming
concrete dowels. The FEM investigation on slim-floor beam-to-column joints have shown that:
 The influence of the concrete slab is effective in sagging bending when it contributes to the
global increase of both the stiffness and the bending capacity. In hogging, its influence is
negligible and the connection is practically based on the steel components;
 For the modeled cases, the bolted end-plate connections can provide only limited resistance
and stiffness due to the limited height, leading to semi-continuous joints (semi-rigid and partial
resistant). Further studies are intended in order to observe their applicability in the case of dual
frames in different seismic zones. In order to achieve continuous joints, welded cover plate
connections (M6-SC type) may be a solution;
 The failure mode of the end-plate steel joint assembly was represented by the failure of bolt
rows in tension (brittle failure). The presence of the reinforced concrete slab led to a limited
increase in capacity and stiffness. The inclined reinforcement contributed to the load transfer
mechanism by connecting the concrete slab to the concrete within flanges. A significant
increase in longitudinal reinforcement will lead to a higher capacity under a negative bending
moment (see M7-SC in Fig. 15b). In contrast, the failure mode of M6-SC joint assembly was
represented by the formation of a plastic hinge in the slim-floor steel beam.
Based on existing studies and on the current study, it is proven that the slim-floor beams may be
adapted to Seismic-Resistant Structures and the key-aspect is related to the behavior of slim-
floor beam-to-column joints. Future research activities will involve experimental investigations as
well as structural numerical investigations, in order to assess the applicability of such systems.
REFERENCES
Abaqus v6.13 [Computer software]. Dassault Systèmes, Waltham, MA.
Arcelor Mittal, “Slim-floor - innovative concept for floors”. Long Carbon Europe Sections & Merchant Bars.
Bernuzzi C., Gadotti F., Zandonini R., (1995). Semi-continuity in slim floor steel–concrete composite
systems, 1st European Conference on Steel Structures. Eurosteel 1995.
Braun M., Obiala R., Odenbreit C., Hechler O., (2014). „Design and Application of a new Generation of
Slim-Floor Construction”, 7th European Conf. on Steel & Composite Structures, Naples, Italy.
Braun M., Obiala R., and Odenbreit C., (2015). Analyses of the loadbearing behaviour of deep-embedded
concrete dowels, CoSFB. Steel Construction, 8: 167–173. doi:10.1002 / stco.201510024.
Dubina D., et al. (2015). “High strength steel in seismic resistant building frames.” Final Rep. Grant No.
RFSR-CT-2009-00024, RFCS Publications, European Commission, Bruxelles, Belgium.
EN 1994-1-1 (2004). “Eurocode 4: Design of composite steel and concrete structures—Part 1: General
rules and rules for buildings”. Brussels, Belgium.
EN 1998-1-1 (2004). “Eurocode 8: Design of structures for earthquake resistance—Part 1: General rules,
seismic actions and rules for buildings.” Brussels, Belgium.
Hauf G., (2010). Trag- und Verformungsverhalten von Slim-Floor Trägern unter Biegebeanspruchung.
Malaska M., (2000). Behaviour of a Semi-Continuous Beam-Column Connection for Composite Slim Floors,
Ph.D. Thesis, Espoo, Finland, ISBN 951-22-5224-4.
Wang Y., Yang L., Shi Y., Zhang R., (2009). Loading capacity of composite slim frame beams, Journal of
Constructional Steel Research 65 (2009) 650–661.
Improved Beam-to-Column Connection for
Partially Encased Composite Structure
Yiyi Chen
Tongji University
Shanghai, China
yiyichen@tongji.edu.cn
Guanghong Chuan
Tongji University, China
rabbit_chuan@163.com
Jie Li
Tongji University
Shanghai, China
tongjilijie@163.com
ABSTRACT
For partially encased composite (PEC) structural frames the connection is one of the key issues.
In Eurocode, there are recommended configurations of the connection which can be considered
as rigid or semi-rigid, but is more apt to fit the requirement under monotonic loads. In this paper,
an improved beam-to-column connection based on Eurocode configurations is suggested.
Loading tests on the proposed connection specimens are carried out. The test designs two
basic situations, one is the failure of connection, to observe the damage mechanism and to
evaluate the resistance of the connection, the other is the failure of PEC beam, to check
whether the proposed connection can guarantee the bending strength and ductility of the beams.
Test results show that the proposed connection possesses expected load capacity, satisfied
deformability and enables the connected beam to develop fully plastic deformation, therefore,
the improved connection is suitable for resisting cyclically reversed moments in earthquake.
1 Introduction
Partially encased composite (PEC) structure possesses many advantages, one of which is the
possibility to pre-fabricate members in workshop. This advantage has been paid more attention
in recent several years, and PEC structure is considered a potential structural system for
modern construction industrialization in China.
For pre-fabricated structures, the connection is a key issue. In Eurocode [1], there are several
recommended configurations of the beam-to-column connections by bolt joint. However, those
connections are typical pin connection or the one with asymmetrical moment resistant capacity.
With the configuration of upper extended end plate of beam, the connection can afford the
resistance against the moment induced mainly by gravity load, but relatively weak capacity to
the moment reversed. If the designer tends to make the connection resist beam end moment in
two directions, which is the demand of moment resisting frames, the connection details
recommended by Eurocode need to be improved.
PEC members including columns, beam-columns and beams have been extensively studied,
covering variety of load patterns [2-10] in recent three decades, however, few public literatures
concern the behavior of PEC beam-to-column connections, neither does the hysteretic behavior
of them.
In this paper, an improved bolt beam-to-column connection based on Eurocode configurations is
suggested. The target of the improvement is to enable the connection resist cyclically reversed
moment during earthquake. An extended end-plate out of the beam section in upper and low
edges with the possible arrangement of bolts is detailed designed. Loading tests are carried out.
The tests aim at two targets, one is to investigate the failure mechanism of connection, the other
is to check the capacity of the connection to allow development of full plastic moment at beam
end. The test results are reported in the paper.
2 Configuration and constructional detail of improved PEC connection

2.1 Conventional configurations of PEC beam-to-column connection
Figure 1 shows the conventional configurations of PEC beam-to-column connection. All of them
are with bolt joint, which makes construction on site free of weld. However, as for the connection
performance, three of them could be taken as pin connection if only PEC beam is considered
(Figure 1a, b and d). The else one using beam end plate (Figure 1c) can resist moment induced
mainly by gravity load. Though all the connections together with the above slab have the
capacity to resist beam end moment, for reversed bending action in earthquake the resistance
may not be deemed enough.
Bolt
Bolt
Bolt
Contact Think steel

End member
plate plate
Bracket
a b c d
Fig 1 – Conventional configurations of PEC beam-to-column connection
2.2 Objectives of the connection improvement and its details

The next aspects are set as the objectives of the connection improvement:
(1) To make the connection possessing the capacity against cyclically reversed moment at
beam end.
(2) To facilitate the assembling operation on construction site.
(3) To maximum the precast concrete part in PEC members.
As the result, the improved connection as shown in Figure 2 is adopted.
The end plate on beam is extended in two sides, instead of the one as shown in Figure 1c
where the plate only extends in upper side. Furthermore, the extended part is reinforced by pair
of triangular ribs so that it has enough stiffness and resistance against moment.
High-strength bolts are to be used for convenience in site assembling. Two possible
arrangements of the bolts are considered. In case of the beam with narrow flange, bolts in
middle shall arranged outside the PEC beam section as shown in Figure 2b, while in case of the
wide beam, bolts in middle can be within the profile of PEC beam section.
In the latter case, a separating plate near beam end (Figure 2d) is designed to leave operation
space for bolt installation on site. For the other part of the beam except for the operation space
concrete is able to be cast in workshop in advance. The same countermeasures are also taken
for PEC column. A pair of horizontal separating plates is welded to the H steel above and
beneath the stiffeners of panel zone, as referred in Figure 2a. Within the scope of two
separating plates and steel flanges, concrete shall be cast after the bolts are tightened in
erection. For the main body of the column, however, the concrete shall be precast in workshop,
so the casting concrete work on site is remained in a small amount.
Bolt Bolt Bolt

Rib
Separating
plate
End plate
Separating End plate End plate

plate
a b c d
Fig 2 – Configuration of improved connection
3 Experimental programs
3.1 Design and fabrication of specimens
To realize the test purpose, i.e., to investigate the failure mechanism of connection and to check
whether the connection allows beam end to develop fully plastic deformation, the specimens
were designed in two expected failure modes, one is failure of connection and the other is
failure of beam.
The specimen is formed in crucial configuration as shown in Figure 3. There are total six
specimens. The specimen code is composed of 2 letters and 2 numbers. The letters indicate the
expected failure mode and load condition in test. The first letter J or M refers to connection
failure mode or member failure mode, respectively, and the last letter C refers to cyclic load
pattern. The digits in specimen code indicate the configuration details. The first digit number 1
or 2 refers to the middle bolts between H steel flanges inside or outside the concrete of PEC
beam (Figure 2b&2c), while the second digit number 1 or 2 denotes whether horizontal
reinforcement bars in panel zone exist or not. The parameters of the specimens are shown in
Table 1.
There are two basic types of PEC sections. One with the relatively stocky steel element and
shear stud, longitudinal rebar and stirrup is recommended by current Eurocode (Figure 4a), and
the other is featured with relatively thin steel element and free from stud and stirrup (Figure 4b).
In this research, the latter is adopted. Built-up steel section is used. For the specimens designed
as connection failure mode (J series), the steel column web in joint region is replaced with 4 mm
thick plate, and its steel plate is shifted into Q235 with normal yield strength of 235 MPa lower
than the column body adopting Q345 with normal yield of 345 MPa.
Jack 千斤顶
1000kN
千斤顶
Bracing
柱顶支撑
+3.330
200
Articulated device
柱顶铰装置 y上Up
Reaction frame
x东East
W E Specimen
试件
2000 +1.920 东
西
Actuator
100kN 作动器 Actuator
200kN作动器
反力架 Articulated
880
device
柱底铰装置
350
Base
钢基座 ±0.000
Fig 3 – The test specimen and loading set up
Table 1 – Parameters of specimens

Code J11-C J12-C J21-C J22-C M11-C M22-C
Section of steel beam B1 B2 B3 B4
Bb/tbf (steel beam flange) 4.6 5.6 12 14.5
Steel grade (steel beam) Q345 Q235
Section of steel column C1 C2
Bc/tcf (steel column flange) 16.2 16
Steel grade (steel column) Q345 Q345
α, (panel aspect ratio) 100 50
Steel grade (panel zone) Q235 Q345
Note: The section of B1 through B4 is H200×100×8×10, H200×120×8×10, H200×100×4×4, H200×120×4×
4, respectively, and section of C1 and C2 is H200×200×6×6, H200×200×8×6, respectively. The panel zone
aspect ratio is defined as (hc+hb)/tcw, where, hc, hb and tcw are the depth of column and beam section, as well as
column web thickness, respectively.
I Steel Thin-walled
I Steel Concrete
Concrete
Shear Link
stud Stirrup
Longitudinal
rebar
a b
Fig 4 – Typical PEC sections
Steel beam is connected to the steel column flange by 10.9 grade high strength bolts through
extended end plate. Eight bolts are used in one joint face, four of them outside the beam flanges
are M22 with shank diameter of 22mm, and remained four between the beam flanges are M16
with shank diameter of 16mm.
In joint regions including the area surrounded by separating plates in beam and its end plate,
and the area surrounded by horizontal separating plates in column together with column flanges,
the concrete shall be cast on site in the practical engineering construction. For the test
specimens, however, after the assembling of column and beam steel by pre-tensioned high
strength bolts, the whole specimen is cast from one side to the other side, thus no formwork is
needed.
3.2 Loading set up and protocol

The loading set up is shown in Figure 3. The base and top of specimen column are pin
connected. The base is on a ground reaction beam and the top links to a horizontal bracing. An
oil jack on top applies axial load on specimen column, and keeps constant load equal to 0.2
times of yield force of PEC column during the test. A pair of actuators is set on the two beam
ends to produce cyclic moment and shear. Figure 5 shows the loading protocol which is referred
to ATC-24 [11]. In the figure, ∆ refers to the beam tip displacement, while ∆ y is the yield
displacement which is determined by the previous monotonic loading tests. After the capacity
peak, three cycles are enforced for each displacement amplitude, in order to study the effect of
gradual concrete damage and local buckling of steel flanges.
9
8
7
Displacement ratio ∆/∆ y
6
5
4
3
2
1
0
-1 0 2 4 6 8 10 12 14 16 18 20 22 24 26
-2
-3
-4
-5
-6
-7
-8
-9
Cycling
Fig 5 – Loading protocol
3.3 Arrangement of transducers and strain gauges

The bending deformation of PEC beam and shear deformation of panel zone are mainly
concerned. To obtain the deformation precisely, rigid movement and incline of the specimen are
removed from the measured data by arrangement of the monitoring transducers.
Strain gauges are arranged in panel zone and the beam flanges near its fixed end.
4 Test results
4.1 Mechanical properties of steel and concrete used for specimens
Mechanical properties of steel and concrete used for specimens are obtained by tensile steel
coupons and compressive concrete cubes respectively. The data of steel which relate to
specimen nonlinear behavior are listed in Table 2, where each datum is the average value of
three coupons. And the compressive strength of concrete is listed in Table 3. Each datum is
also the average value of three concrete cubes. By the table, the strength of concrete cast in
one batch can be represented by the average value.
Table 2 – Mechanical properties of steel material

Steel Plate thickness or fy fu
Applied element ∆ (%)
grade rebar diameter (mm) (MPa) (MPa)
4 Panel zone, beam flange 297 445 31.0
Q235
20 End plate of beam 274 436 31.5
6 Column flange and web 355 514 30.5
8 Column and beam web 373 532 29.0
Q345
10 Beam flange 379 555 30.0
18 End plate of beam 387 566 28.0
HPB300 6 Link between flanges and rebar 596 657 -
Table 3 – Compressive strength of concrete (in MPa)

Standard
Specimen code J11-C J12-C J21-C J22-C M11-C M22-C Average
deviation
Side A cast firstly 38.8 41.3 39.1 43.2 43.6 41.2 40.6 2.39
Side B cast secondly 31.5 37.2 37.9 36.9 41.6 39.1 37.5 2.09
4.2 Damage evolution observed

Table 4 briefly summarizes the observed phenomena during the tests.
Table 4 – The observed phenomena of damage of specimen

∆/∆y J11-C J12-C J21-C J22-C M11-C M22-C
±0.50 CB1 CB1 CB1, CB2 CB1
±0.75 CJ1 CB2, CJ1 CB1, CJ1 CB1, CJ1
±1.00 CJ2 CJ2
±2.00 CJ3 CB3, CJ3 CJ2 CJ2,CJ3 CB3, CJ1 CB2, CJ1, SB1
±3.00 SJ1, U1, U2 CJ4, U1, U2 CJ3, CJ4, U1 U1, U2, SB1, SB2 CB3, CB4
±4.00 SJ2, CJ4 SJ1 SJ1, U2 SJ1, CJ4 CB4, U1,U2 U1, U2
SB3, SB4 SJ6,
±5.00 SJ2 SJ2
Loading stopped Loading stopped
±6.00 SJ2
SJ3, SJ4, SJ3, SJ4, SJ5, SJ3, Loading
±7.00 SJ4
loading stopped loading stopped stopped
SJ3, SJ5,
≥8.00
Loading stopped
The following symbols are used to represent the typical damage: CB1, Concrete tiny flaw in
beam at or near its fixed end; CB2, Concrete flaw through whole beam depth; CB3, Concrete
cracking through whole beam depth; CB4, Concrete dropped off from beam; SB1, Slightly
buckling of beam flange; SB2, Severe buckling of beam flange; SB3, Visible buckling of beam
web; SB4, Fracture of beam flange; CJ1, Concrete tiny flaw in joint zone; CJ2, Declined
concrete cracking appearing in joint zone; CJ3, Concrete cracked severely in joint zone; CJ4,
Concrete dropped off from/squeezed out of joint zone; SJ1, Severe shear deformation of panel
zone; SJ2, Opening largely between beam end plate and column flange; SJ3, Penal zone steel
buckling visible; SJ4, Link welding spot cracked; SJ5, Panel zone plate cracking visible; SJ6,
Fracture of end plate rib weld; U1, Ultimate load; U2, Peak load in second cycle obviously below
that in first cycle.
Figure 6 and Figure 7 show the appearance of the typical failure modes.
(a) General view (b) Concrete filled in joint zone (c) Opening between (d) Buckling and fracture
dropped off column flange and of panel zone
beam end plate
Fig 6 – Typical damage of J-series specimens with connection failure mode
(a) General view (b) concrete cracking and steel (c) Steel flange fracture
flange buckling
Fig 7 – Typical damage of M-series specimens with beam failure mode
By Table 4, it is known that it is common the tiny flaw of beam concrete due to bending
appeared at first for all specimens when the relative deformation was about 0.5~0.75%, but the
damage of J-series specimens then shifted to the connection, they underwent the crack and
drop off of concrete in joint zone, the opening and closing cyclically of beam end plate and
column flange, the local flexural deformation of column flange and buckling of panel zone after
concrete squeezed out of. However, for the M-series specimens the main damage remained in
the beams, though the joint zone concrete developed small flaw and fine cracking in the later
process. The residual deformation shown in Figure 6a and Figure 7a exhibit the different
features of final status of these two series specimens.
4.3 Loading-deformation curves

Figure 8 and Figure 9 display the typical load-deformation curves of the J-series specimens
represented by J11-C which are designed as connection failure mode and M-series represented
by M11-C which are beam failure mode. In each figure, the left one shows beam shear force
versus its tip displacement while the right one is the beam shear force versus the panel zone
distortion. By the figures, we can read that (1) The non-linear deformation of specimen J11-C is
mainly due to the deformation of joint zone. It can be deduced by comparison the two cyclic
curves of Figure 8. On the contrary, non-linear deformation of specimen M11-C is contributed
almost by beam because the curve of shear force versus panel zone deformation keeps almost
linear as shown in Figure 9b. Of cause in this case no noticeable damage of column was
observed in the test. (2) The specimen J11-C reaches its ultimate when its beam slope is about
0.03, while in the case of specimen M11-C its ultimate is corresponding to beam slope of 0.045
and the resistance keeps a short plateau till to the rotation of 0.055. (3) Specimen resistance
degrades remarkable in the following cycle after the specimen reaches its ultimate, because the
concrete damage evolutes severely. The curves of other specimens show similar features as
these two specimens.
Beam shear force F (kN)

60 J11-C 60 J11-C
45 45
30 30
梁端剪力(kN)
梁端剪力(kN)
15 15
0 0
-15 -15
-30 -30
-45 -45
-60 -60
-200 -160 -120 -80 -40 0 40 80 120 160 200 -0.15 -0.12 -0.09 -0.06 -0.03 0.00 0.03 0.06 0.09 0.12 0.15
梁端位移(mm) Δ (mm)
Beam tip displacement Panel zone 节点域剪切角 distortion
(rad) γ (rad)
Fig 8 – Loading-displacement curves of J11-C

60 60
M11-C M11-C
45 45
30 30
梁端剪力(kN)
梁端剪力(kN)
15 15
0 0
-15 -15
-30 -30
-45 -45
-60 -60
-200 -160 -120 -80 -40 0
40 80 120 160 200 -0.15 -0.12 -0.09 -0.06 -0.03 0.00 0.03 0.06 0.09 0.12 0.15
Beam tip 梁端位移(mm) Δ (mm)
displacement Panel zone 节点域剪切角 distortion
(rad) γ (rad)
Fig 9 – Loading-displacement curves of M11-C
4.4 Ultimate, deformability and energy dissipation of specimens

Table 5 shows the ultimate recorded in tests. By figure 8 and 9, it is known that there are two
peak values in reversed cyclic loading for each specimen. However, the two peaks are quite
proximate with the diversity less than 5%, so the average of the two peaks denoted as FUT is
taken as the tested ultimate of specimen. The data are listed in Table 5.
As comparison, the predicted resistance FUJ,pre corresponding to the connection failure mode
and FUB,pre corresponding to beam failure mode are listed too. The predicted resistance of
connection is the minimum of those corresponding to end-plate failure, bolt failure and joint zone
failure. The former two can use the available equations [12], but the last one is suggested by
this study. The joint zone is mainly subject to shear force induced by beam end moment as well
as the axial load on column top. The resistance is computed based on the following
assumptions, (1) The shear stress on panel zone uniformly reaches yield strength of steel, i.e. in
fully plastic deformation, (2) The column flange, horizontal stiffener, separate plates and rebar if
existing act as rigid bar but with two end springs developing full plastic moment, (3) The
concrete in the joint zone plays a role of inclined compressive stub, and (4) The ratios of axial
stress due to axial load on column in steel and concrete is the same. There several possible
failure modes and the resistance of each failure mode is computed. The minimum of them is
determined as the resistance of the connection. By this resistance, the corresponding beam end
load FUJ,pre is predicted. The moment capacity of beam is computed too, according to the
equations recommended by Eurocode [1], and then converted into the beam tip load FUB,pre. It
can be seen by Table 5 that the computed resistances predict the capacity of specimens very
well, and thus by the prediction it is possible to identify the failure mode.
Two ductility factors of the specimen are presented in Table 5. One is by the beam flexural
deformation, denoted as δS, the other is by the joint zone shear deformation, denoted as δJ. In
both cases, the yield formation is identified in the same method: make the envelope curve of
cyclic loading-deformation loops, draw initial elastic stiffness line through original point, and
intersect the line to the horizontal level against ultimate load, then take the deformation
corresponding to the crossover point as yield deformation. And the ultimate deformation to
stipulate the ductility factor is the deformation when the resistance of specimen falls into 85% of
its ultimate. By Table 5, it can be seen that no matter what failure mode is predominant, the
ductility factor is quite good. For the J-series in which connection failure mode is predominant,
Table 5 lists the ductility factors based on joint zone shear deformation. It shows that joint zone
possesses relatively great deformability.
Table 5 – Summary of key data of the test

Key data J11-C J12-C J21-C J22-C M11-C M22-C
FUT (kN) 57.0 58.1 57.5 58.9 39.9 46.7
FUJ,pre (kN) 55.2 55.2 55.2 55.2 70.8 70.8
FUB,pre (kN) 74.0 74.4 85.4 85.3 29.4 33.0
δS 4.14 4.14 4.05 4.13 5.21 4.50
δJ 14.6 19.2 11.3 13.6 - -
Figure 10a shows the dissipated energy of specimens and joint zones for per cycle, presented
in the form of energy dissipating factor. The factor is computed by the surrounded area of load-
deformation loop per cycle, as shown in Figure 10b with shadow, divided by the double
triangular area (∆OBE+∆ODF in Figure 10b). The figure includes the factors computed by whole
specimen and by joint zone only for J-series. As general tendency, the energy dissipating
factors increase with the increase of cycle numbers. But with the resistance deterioration in the
second and third cycle in a given displacement amplitude, the factor should display a certain
decrease. However, during the loading process, the actuator could not be controlled precisely,
thus the followed deformation might exceed the previous amplitude. So the predicted 'decrease'
does not show clearly in the figure. On the other hand, we can find in the J-series that the
dissipating factor of joint zone is far greater than the specimen factor at the beginning. It is due
to the damage concentration on joint zone but no big nonlinear behavior of the specimen at this
stage.
3.0
2.8 Note: “Sp” in the legend
Energy dissipating factor
2.6
2.4 presents “Specimen”, and
“Jz” presents “Joint zone”.
滞回圈耗能系数
2.2
2.0
1.8 J11-C
J11-C试件Sp
1.6 J11-C
J11-C Jz
节点域
J12-C
J12-C试件Sp
1.4 J12-C
J12-C Jz
节点域
1.2 J21-C
J21-C试件Sp
1.0 J21-C
J21-C Jz
节点域
0.8 J22-C
J22-C试件Sp
0.6 J22-C Jz
J22-C节点域
0.4 M11-C
M11-C Sp
试件
0.2
0.0
M22-C
M22-C Jz
试件
The ith displacement amplitude
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Cycling
滞回圈次
a b
Fig 10 – The energy dissipating factors
5 Brief discussions
The ultimate of the specimens can be well predicted based on the assumption of completely
plastic capacity, which implies the fact that the beams and connections of specimens possess
satisfied plastic capacity. Considering J-series specimens with large panel aspect ratio of 100,
we know that the filled concrete in joint zone together with vertical rib strengthen the steel panel
effectively. Premature local buckling can be prevented and the plastic deformability can be
developed. The test also exhibits that the configuration of bolt connection with extended end
plate by rib stiffened guarantees the full plastic capacity and rotation deformability of beam
member, which indicates that the improved connection is appropriate to for moment resisting
frame in PEC structures.
The ductility factors of the specimens obtained by tests show that the improved connection itself
has quite good deformability, thus can be adopted in seismic design.
Construction details vary in the specimens, including the location of middle row bolts inside or
outside the filled concrete, with or without the horizontal reinforced rebar in joint zone to confine
the infilled concrete. Tests show that they have no substantial influence on the specimen
behavior, especially on the resistance, ductility factor and energy dissipating factor. It shall
provide more freedom for design.
6 Summary
The improved beam-to-column connection for PEC frames was proposed in order to resist
inverse moments on beam end in potential earthquake. The constructional details were
considered.
The cyclic loading tests on the proposed connection were carried out through six specimens to
check its performance in the situations of connection failure mode predominated or beam failure
mode predominated.
The test shows that the well designed connection can make the beam develops full plastic
capacity and deformability. On the other hand, the connection can develop its ultimate till large
non-linear deformation. Thus it can be concluded that the improved connection can be used in
PEC frame for seismic design.
More study is still expected to make the classification of connection performance corresponding
different seismic design requirement.
Acknowledgement
This research is partly supported by the Key Research and Develop Project of MST of China
(Grant No. 2016YFC0701603)
Reference
[1] European Committee for Standardization (2004). Design of composite steel and concrete
structures ， Part 1-1 ： General rules and rules for buildings(EN 1994-1-1).Retrieved from
https://www.cen.eu/
[2] Chicoine T，Tremblay R，Massicotte B，Ricles J，Lu L W. (2007). Behavior and strength
of partially encased composite columns with built up shapes．Journal of Structural Engineering，
128（3）：279~288. doi:10.1061/(ASCE)0733-9445(2002)128:3(279)
[3] Chicoine T，Massicotte B，Tremblay R． (2003). Long term behavior and strength of
partially encased composite columns with built up shapes．Journal of Structural Engineering，
129（2）：141~150. doi:10.1061/(ASCE) 0733-9445(2003)129:2(141)
[4] Oh M H, Ju Y K, Kim M H, et al. (2006). structural performance of steel-concrete composite
column subjected axial and flexural loading. Journal of Asian Architecture and Building
Engineering . 5(1): 153~160. Retrieved from https://www.jstage.jst.go.jp/article/jaabe/5/1/5_1_
153/_pdf.
[5] Chen Y Y，Wand T，Yang J，Zhao X Z. (2010). Test and numerical simulation of partially
encased composite columns subject to axial and cyclic horizontal loads．International Journal
of Steel Structures, 10(4)：385~393.doi: 10.1007/BF03215846
[6] Kindmann R, Bergmann R. (1993) Effect of reinforced concrete between the flanges of the
steel profile of partially encased composite beams. Journal of Constructional Steel Research.
27(1-3): 107~122. doi: 10.1016/0143-974X(93)90009-H
[7] Nakamura S, Narita N. (2003) Bending and shear strengths of partially encased composite I-
girders. Journal of Constructional Steel Research. 59(12): 1435~1453.doi:10.1016/S0143-
974X(03)00104-4.
[8] Nardin S D, Debs A L H C E1. (2009) Study of partially encased composite beams with
innovative position of stud bolts. Journal of Constructional Steel Research. 65(2): 342~350. doi:
10.1016/j.jcsr.2008.03.021
[9] Tsavdaridis K D，D’Mello C，Huo B Y. (2013). Experimental and computational study of
the vertical shear behavior of partially encased perforated steel beams．Engineering Structures，
65：805~822. doi: 10.1016/j.engstruct.2013.04.025
[10] Muise J L． (2004). Behavior of simple framing connectors to partially concrete encased H
section columns：(Master’s Thesis). University of Toronto, Toronto, Canada.
[11] Krawinkler H. (1992). Guidelines for cyclic seismic testing of components of steel structures.
Report No. ATC-24, Applied Technology Council. Redwood City, CA.
[12] American Institute of Steel Construction (2010). Specification for structural steel
buildings． (AISC．360-10). Retrieved from https://www.ansi.org/.
GIRDER-WALL CONNECTIONS USING CONCRETE ANCHORS IN
COMPOSITE CONSTRUCTION
Jian Zhao
University of Wisconsin
Milwaukee, WI, USA
jzhao@uwm.edu
Bo Yang
Chongqing University
Chongqing, China
yang0206@cqu.edu.cn
ABSTRACT
Composite high-rise buildings usually benefit from concrete core walls and steel frames.
Embedded connections are needed, which usually consist of a steel plate fixed to a number of
headed anchor bolts/studs that are embedded in concrete. These connections can be subjected
to complex loads in an earthquake; consequently, the seismic behavior of embedded connections
is critical for the structural behavior.
Detailed design methods have been proposed for anchor reinforcement that can help single cast-
in anchors to achieve ductile behavior under simulated seismic loading. Two reinforced anchors
were installed in the plastic hinge zone of concrete columns. One was tested under monotonic
loading while the other under cyclic loading. The tests indicated that well confined core concrete,
even within plastic hinge zones, can support anchor connections. With the recommended anchor
reinforcement, an innovative anchor connection between steel girders and concrete core walls is
introduced for use in steel-concrete composite construction.
INTRODUCTION
Embedded anchor connections are used in composite structures with a concrete shear wall and
a steel frame as shown in Figure 1. Embedded connections usually consist of a steel plate
attached to several headed anchor bolts/studs or deformed bars that are embedded in concrete.
These connections can be subjected to complex loads in an earthquake; therefore, the seismic
behavior of embedded connections is critical for the structural behavior. Shake table tests of
structure models in the literature have shown that the embedded connections are susceptible to
damage in earthquakes, which in turn disturbs the desired structural performance (Gong et al.
2004; Zhou et al. 2012).
Seismic tests of cast-in anchors in plain concrete have shown unreliable seismic behavior
(Petersen et al. 2011). The most common modes of anchor failure encountered in practice are
anchor steel failure and concrete breakout failure. Anchor steel failure is caused by fracture of
the anchor shaft in tension or shear while typical concrete breakout failure is shown in Figure 2:
a concrete cone is broken away from the base concrete in which the connection is located.
Concrete breakout failure occurs when anchors are located close to an edge or with small
embedment depths. Breakout cracks usually originate from the anchor and propagate towards
the reaction points in laboratory tests. A 35-degree crack propagation angle is assumed in design
(Fuchs et al. 1995), as shown by the regions encased in dashed lines in Figure 2.
Concrete breakout is a brittle failure mode and thus not preferred for anchors in seismic zones
(Petersen et al. 2013a). Engineers can use steel reinforcement to improve the behavior of
anchors likely controlled by concrete breakout. The existing anchor shear reinforcement
recommended in ACI 318-14 code, mainly V-shaped hairpins wrapping around anchor shafts,
may not be practical (Petersen et al. 2013b). In addition, tests have indicated that the existing
anchor tensile reinforcement, in form of U-shaped hanger steel placed close to anchor bolts, can
restrain concrete breakout failure; however, the desired anchor steel failure may not be achieved
unless the concrete around the anchors are properly confined (Lin et al. 2013). In this paper,
recommended anchor reinforcement was briefly introduced and applied to the anchor connections
between concrete core walls and steel girders.
LITERATURE ON ANCHOR TENSION REINFORCEMENT

Existing studies
Hasselwander et al. (1974) conducted a series of tests on large-diameter (e.g., 1.75 in.), high-
strength anchor bolts embedded near edges of concrete piers with edge distances ranging from
one to six inches. The concrete piers contained No. 9 longitudinal bars at a spacing about five
inches- and No. 4 stirrups. The anchor bolts were installed close to the longitudinal bars with an
embedment depths of 15da (da is the embedment depth of the anchors); therefore the closest four
to six longitudinal bars may be considered as the anchor reinforcement. The tests indicated that
the longitudinal bars as anchor reinforcement did not ensure the load transfer from the anchor to
the bars, and the anchors in only two specimens developed their yielding strength. Typical failure
of the specimens started with cover concrete being pushed out near the anchor head and the
concrete crushed above the large-size anchor heads.
Kotani et al. (2006) tested 0.5-in. [13-mm] bolts embedded 6.7 in. [170 mm] in narrow footing
beams typically used for residential houses. No. 3 stirrups with a spacing of 12 in. [305mm] were
used in four out of sixteen specimens. The closest two stirrups located roughly 0.9hef from the
anchor did not restrain the breakout failure though an increase of 24 percent in anchor capacities
and 100 percent increase in peak displacements were observed. Note that the concrete between
the anchors in tension and the compression flange of the steel column was not protected with
stirrups and surface reinforcement; therefore, the desired load transfer mechanism was not
achieved.
Lee et al. (2007) tested five groups of four anchors with large diameters and deep embedment
depths. One group of anchors were reinforced with four No. 8 U-shaped hairpins and another
group with eight No. 8’s. The vertical hairpins were divided into two groups, one group located
within 0.2hef and the other 0.35hef from the anchor bolt. The No. 8 bars were assumed fully
developed below the anchor head through a length of 14db with hooked ends, which is slightly
smaller than the code-specified development length. The vertical hairpins were not proportioned
to carry the anchor steel capacity in tension. Anchors with four hairpins developed two times the
capacity of unreinforced anchors while the tests of the anchors with eight hairpins were terminated
before the anchor capacity was clearly achieved.
Closed stirrups were provided in one of the two tension tests of shear studs for use in composite
construction by Saari et al. (2004). The reinforced specimen contained No. 3 stirrups at a spacing
of 3.5 in., which, along with longitudinal bars at all corners, formed a reinforcing cage around the
studs. Concrete failure occurred to the unreinforced anchor while steel fracture occurred to the
reinforced anchor with capacity increase about 100 percent.
Design Recommendations
Building codes and design guidelines allow engineers to use steel reinforcement to avoid concrete
breakout failure in design (ACI 318-14; Zhao and Lin 2017). The recommended anchor
reinforcement usually consists of U-shaped hairpins. The existing design methods assume that
the concrete breakout similar to that in plain concrete occurs before anchor reinforcement carries
the design load. The recommendations on the design of anchor tension reinforcement are
summarized in Figure 3. Note that the groups of two anchors in tension in Figure 3 can be the
part of a four-anchor connection subjected to a moment from the connected steel element. U-
shaped hairpins are typically recommended. In general, two design methodologies have been
adopted: 1) concrete breakout cone forms before reinforcement within a certain range from the
anchor bolts carry the tension force; 2) the load transfer from the anchor bolt to nearby
reinforcement can be treated as that between spliced deformed bars or visualized using strut-
and-tie models (STMs).
The existing anchor reinforcement recommended in ACI 318-14 code mainly consists of U-
shaped hanger steel, placed close to the anchor in tension. The reinforcing bars embedded in the
breakout cone, as shown in shaded areas in Figure 3, are assumed to provide tensile resistance.
The U-shaped hairpins are required to be within a distance equal to half of the embedment depth
(0.5hef) as illustrated in Figure 3c. This effective range is increased in CEN/TS (2009) to 0.75hef.
In addition, stirrups in beams within 1.0hef from the anchors can be counted as reinforcement for
anchors according to a CEB report (1997).
Proposed Anchor Tension Reinforcement
Zhao and Lin (2017) evaluated the existing anchor reinforcement using experimental tests.
Compared with the anchors embedded in plain concrete, a capacity increase ranging from 20%
to 130% was observed in the tests of reinforced anchors. However, the expected steel fracture
was not achieved mainly due to the lack of confining reinforcement. Splitting cracks developed,
and the concrete around the anchor head lost its confinement and crushed prematurely, resulting
in anchor pullout failure.
Based on these observations and other tests in the literature, recommendations for anchor
tension reinforcement were proposed (Petersen et al. 2013b). The proposed reinforcement, as
illustrated in Figure 4 includes: 1) load-carrying reinforcement in the direction of the anchors; 2)
crack-controlling reinforcement in all directions that has a limited edge distance; and 3) local
confining reinforcement near the anchor head if side-face blowout may control the failure. In
addition, the tensile load on the anchors eventually needs to be transferred to the rest of the
structure. Sufficient reinforcement, such as stirrups and surface bars, is needed to ensure proper
load transfer. Note that with the concrete confined around the anchor, it is expected that the
concrete will distribute part of the tensile force to the rest of structure.
The load-carrying reinforcement should be proportioned to carry a force equal to the design
tensile capacity of the anchor bolt. The nominal yield strength of steel should be used in the
calculation. The load-carrying reinforcement should be implemented using small-diameter closed
stirrups. The stirrup width should be small yet satisfy the bending radius requirements. The depth
of the stirrups should be large enough such that the vertical legs are fully developed for the tension
load at both sides of the anchor head. Two stirrups should be placed next to the anchor shaft,
where the crack in concrete may initiate above the anchor head under a tensile load. Rather than
placing all bars within a small distance from the anchor, closely spaced stirrups with a center-on-
center spacing of 2 to 3 in. [50-75 mm] can extend 1.0hef, where hef is the embedment depth of
the anchor, from the anchor bolt, as shown in Figure 4. Although the development length
requirements for the anchor reinforcement can be satisfied through the interaction between the
closed stirrups and corner bars, it is recommended that the closed stirrups are at least 8.0db from
the anchor head.
The crack-controlling reinforcement is needed to cross all possible splitting cracks. Considering
the load transfer from the anchor head to the load-carrying reinforcement: a strut-and-tie model
can be established with the struts conservatively assumed from the anchor head to the outmost
load-carrying stirrups. The rest crack-controlling reinforcement can be determined considering
the load transfer from the load-carrying reinforcement to the rest of the structure. All crack-
controlling reinforcement should be proportioned using a steel stress of 0.6fy, and implemented
using small diameter bars evenly distributed with a small and practical spacing in two orthogonal
directions. The crack-controlling reinforcement should be properly developed. In addition, the
closest four stirrups, as the load-carrying reinforcement, can be encased by three or four closed
hoops in the transverse direction near the anchor head if the anchors are close to a free side
surface (Figure 4). This is to better confine the concrete that is subjected to large compressive
stresses above the head.
EVALUATION OF REINFORCED ANCHORS IN TENSION

The key role of anchor reinforcement, in addition to carrying the forces from the anchors, is to
protect concrete around the anchors from splitting, breaking out, and crushing. This
understanding of the related behavior has led to alternative designs and detailing for anchor
reinforcement (Petersen and Zhao 2013). The tests in this study (Zhao 2014) were conducted to
verify the proposed reinforcement for anchors placed in plastic hinge zones.
A total of three column specimens were used for single anchors in tension. The column height (h)
was designed as 5 ft. [1.52 m] from the top face of the base block, which is close to that of a half
story height. The column base block had a dimension of 48x20 in. [1220x508 mm] and a height
of 17 in. [432 mm] as shown in Figure 5. Two tie-down holes were created using embedded PVC
tubes. The tie down points were 3 ft apart following the hole pattern in the strong floor of the UWM
Structures Laboratory. The horizontal loads were applied to the top of the column at 62 in. [1.57
m] from the base through a steel loading block.
Eight No. 5 bars (ASTM Grade 60) were provided as the longitudinal reinforcement for the
columns. The concrete cover was 1.5 in. [38 mm], typical for RC members. A section analysis
indicated that the column had a nominal moment capacity about 60 k-ft [81.3 kN-m]. This
corresponded to about 12 kips [2.7 kN] at the top of the column. The actual loading capacity can
be higher due to strain hardening of the bars. The shear design for the column for this load led to
No. 4 [13-mm] ties at a spacing of 6 in. [150 mm], as illustrated in Figure 3. Additional ties were
provided within the plastic hinge zone (the length of the plastic hinge zone lp=1.5h=18 in. [457
mm]), as shown by shaded regions in Figure 5. The plastic hinge length of 1.5h was used because
longitudinal bars yielded within this range (Lin et.al 2013).
The test anchors were made from 3/4 in. [19 mm] diameter ASTM A193 Grade B7 threaded rods
(fya=105 ksi [724 MPa] and futa=131 ksi [903 MPa]). The net tensile area (Asa,N) and the net shear
area (Asa,V) for the 3/4-in [19-mm] threaded rods are 0.334 in.2 [215 mm2]. The test anchor, if fully
developed, would take an ultimate tension load of 43.8 kips [195 kN], which is within the capacity
of the MTS Model 244.31, actuator used in the study. A plate washer (1.5×1.5 in. [38x38 mm])
and a hex nut were tack welded to the end. The net bearing area (Abrg) was 1.8 in.2 [1161 mm2],
sufficient to carry the tensile load without causing pulling out failure as observed previous tests
(Petersen et al. 2013b). The concrete compressive strength (fc') was assumed as 4000 psi [27.6
MPa] and the maximum bearing strength 6.0fc' was used in design of the head plate (ACI 318-
14).
The required anchor reinforcement for the 3/4-in. [19-mm] anchors in tension was found to be
0.73 in.2 [471 mm2] Two additional No. 4 [13-mm] ties (with four stirrups legs as anchor
reinforcement) were used, and located 2-in. [50-mm] from the test anchor, as shown in Figure 3.
The needed anchor shear reinforcement was found to be 0.44 in.2 [284 mm2] The two No. 4 [13-
mm] ties next to the anchors were deemed sufficient, considering other ties within the plastic
hinge zone.
The columns were expected to develop flexural cracks under lateral loads, especially in the plastic
hinge zones. The cracks could propagate rather arbitrarily, passing the anchors, when the column
develops inelastic behavior. In this case, concrete bearing strength can be significantly reduced.
Therefore, three No. 4 [13-mm] U-shaped hairpins were placed near the anchor in the vertical
plane, as shown in Figure 5. These hairpins were expected to confine the concrete from flexural
cracking near the anchor heads. This local crack-controlling bars were slightly more than the
calculation using a strut-and-tie model (ACI 318-14).
Ready mixed concrete with Wisconsin Department of Transportation Type A-FA mixture was
used. The target concrete compressive strengths for the normal strength concrete was 4000 psi
[27.6 MPa]. The hardened concrete had a compressive strength of 5800 psi at 28 days, and the
compressive strength went up 6900 psi [47.6 MPa] at about 84 days, when the tests were
conducted. The No. 5 [16 mm] reinforcing steel with a nominal yield strength of 60 ksi [414 MPa]
were used as longitudinal reinforcement. The measured yield strength was 65 ksi [448 MPa], and
the ultimate strength 101 ksi. [696 MPa].
The column specimen was fixed to the strong floor of the testing laboratory using two high-
strength threaded rods. One MTS Model 244.31 actuator was used to apply a reversed cyclic
displacement at the top of the column. The lateral load was applied to the column at 62 in. above
the base block. Another MTS Model 244.31 actuator was used to apply tensile/shear forces to
the anchors. The centerline of this actuator was located at the height of the test anchor. In order
to minimize the relative displacement between the reaction frame and the specimen, a shoring
frame was placed in between the reaction frame and the specimen. Strain gages and linear
potentiometers were used to monitor the behavior of the test columns and the anchors.
The columns were subjected to reversed cyclic displacements. The displacement history included
groups of three cycles with the following peak displacements: ±y. ±2y, ±3y, ±4y, ±6y, and
beyond ±8y. Loading rates for displacement cycles were kept at 0.24 in./min [6 mm/min]
throughout the tests. The anchors were loaded after the columns were subjected to the entire
cyclic loading history and developed damage in the plastic hinge zones.
The behavior of the two single anchors subjected to tension is shown in Figure 6. The load vs.
displacement behaviour of Specimen T2 demonstrated that the steel ductile failure was achieved.
Unlike normal sharp drop after peak load, the anchor steel had a relative long descending range.
It may have been attributed to slow energy release rather than sudden energy release in a typical
coupon tests, in which rigid loading devices are typically used.
The comparison of the load vs. displacement curve at the monotonic loading (T2 test) and cyclic
loading in this case (T3 test) in Figure 6 indicated that the load curve followed the monotonic path
closely till the failure. The small difference was mainly caused by upward orientation of the anchor
in this case (T3) that may cause an additional small bending as well as governing tension on the
anchor shaft when the anchor was subjected to tension, thus leading to the slightly low load path
and final capacity. The difference may also have come from variations built in experimental tests.
The anchor remained steady at the load level of 44 kips [196 kN] without too much additional
displacement. According to potential failure modes, the column will be able to take 61.8 kips [275
kN] before a flexural failure occurs. With the diagonal cracks path, the anchor tension force was
again transferred to the rest of the column through the dry friction. The sliding surface was along
a flexural crack above the base, which was opened before the tension force was applied to the
anchor. This was slightly different from the monotonic loading in T2. According to the monotonic
loading curve in T2, the anchor at a commanded displacement of 0.58 in. [14.7 mm] at 43 kips
[191 kN], however, was compatible to that of 0.6 in. [15.2 mm] at the same loading level. This
means that the column was stable without obvious degradation under cyclic loading and can still
carry more loads.
The load vs. anchor displacement for both T2 and T3 confirmed that the anchors can be installed
in the plastic hinge zone when suitable anchor reinforcement is provided, as stated in Zhao and
Lin (2017). These tests of anchors installed in the plastic hinge zones of RC columns also
indicated that well confined concrete core can support anchor connections. With the proposed
anchor reinforcement, the anchors can develop their full tensile capacities.
CONCEPT OF EMBEDDED CONNECTION AND RELATED LITERATURE

The proposed embedded connection is illustrated in Figure 7. The design is based on the existing
research in the literature and the results of this study.
Girder-wall connections have a limited embedment depth, potential prying action from the plate
deformation, and damaged concrete affect the seismic performance of embedded connections.
The proposed connection thus seeks to mitigate these issues. Specifically, the available
embedment length can be limited by the wall thickness, especially for the connections on the side
faces of a wall, as illustrated in Figure 1. The tensile capacity of the connection in such situations
is controlled by concrete breakout failure according to the current design regulations (ACI 318-
14) if anchor reinforcement is not properly provided. It is proposed to use the newly developed
anchor reinforcement consists of a group of closely spaced stirrups (rather than U-shaped
hairpins) with a goal to confine concrete near the anchor heads. In addition, crack-controlling
reinforcement was provided along all faces of concrete to restrain the concrete from splitting.
Girder-wall connections have a limited plate thickness as limited by practical applications. In most
connections, a shear tab is welded to the embedded plate, and the forces are transferred from
the shear tab to the anchor bolts/deformed bars through the embedded plate. Prying occurs when
the stiffness of the embedded plate is not sufficient, which causes an increase in the tensile force.
The bending deformation of the embedded plate also causes uneven distribution of tensile forces
among the anchor bolts/deformed bars. Premature failure occurs in both cases. It is proposed to
add stiffeners onto the back of the embedded plate to increase its stiffness, to minimize the prying
action on the anchor bolts, and to evenly distribute the resulting tensile forces among the anchor
bolts/deformed bars subjected to tension.
Girder-wall connections may be embedded in concrete experiencing extensive damage.
Concrete walls may develop cracks, especially in lower stories of a complex building structure (as
shown in Figure 1) when subjected to an earthquake. It is proposed to use the cracking controlling
reinforcement similar to that used in this study.
The proposed girder-wall connections require extensive experimental verification before being
used in practice. The design of anchor reinforcement may follow Appendix A. A prefabricated
reinforcement cage, as shown in Petersen (2011) may benefit the construction.
SUMMARY
Two tests were conducted to verify the proposed reinforcement for anchors in tension. The single
anchors were installed in the plastic hinge zone of concrete column specimens. Despite large
cracks and concrete spalling occurred to the concrete within the plastic hinge zones, ductile steel
fracture was achieved in all tests. The successful tests indicated that well confined core concrete,
even within plastic hinge zones, can support anchor connections. The test also confirmed that
the anchor reinforcement should confine concrete, restrain concrete from splitting cracks, and
distribute loads from the anchor to the rest of the structure/structural element.
A new girder-wall connection is proposed based on the anchor tension reinforcement verified in
this study. Special reinforcing cage will be used to confine the concrete around the girder-wall
connections, and transfer tensile loading from the connection to the concrete. Stiffeners will be
used to distribute the tensile loading from the connection to the anchors. The stiffeners will also
serve as shear keys such the anchors can be designed for tension only. This connection needs
experimental verification before being used in practice.
REFERENCES
1. American Concrete Institute (ACI), “Building Code Requirements for Structural Concrete (ACI
318-14).” Farmington Hills, MI, 2014.
2. European Committee for Standardization, “Design of fastenings for use in concrete - Part 4-
2: Headed Fasteners,” CEN/TS 1992-4-2, Brussels Belgium, 2009.
3. Comité Euro-International du Béton (CEB), “Fastenings to Concrete and Masonry Structures:
State of the Art Report.” Thomas Telford Service Ltd., London. 1997.
4. Fuchs, W.; Eligehausen, R.; and Breen, J., “Concrete capacity design approach for fastening
to concrete.” ACI Structural Journal, 1995, 92(1), pp.73-94.
5. Gong, Z., Lu, X., Lu, W., Li, P., Yang, S., Zhao, B. (2004) Shaking Table Model Test of a
Hybrid High-Rise Building Structure, Earthquake Engineering and Engineering Vibration, Vol.
24, No. 4. pp.99-105. (in Chinese).
6. Hasselwander, G.; Jirsa, J.; Breen, J.; and Lo, K., "Strength and Behavior of Anchor Bolts
Embedded Near Edges of Concrete Piers," Research Report 29-2F, Center for Transportation
Research, the University of Texas at Austin, TX, 1974.
7. Kotani, H.; Matsushita, K.; Kajikawa, H.; Wu, D., “Experimental Study on Pull out Behavior of
Anchor Bolt for Small Buildings (Footing With Singly Arranged Reinforcing Bars),” Summaries
of technical papers of Annual Meeting Architectural Institute of Japan. C-1, Structures III,
2006, pp. 99-100. (in Japanese)
8. Lee, N.; Kim, K.; Bang, C.; and Park, K., "Tensile-Headed Anchors with Large Diameter and
Deep Embedment in Concrete," ACI Structural Journal, Vol. 104, No. 4, 2007, pp. 479-486.
9. Lin, Z., Luke, B., Shahrooz, B., and Zhao, J. "Headed Anchors in Plastic Hinge Zone of
Reinforced Concrete Members." Final Report (Volume III) for Project - Behavior and Design
of Cast-in-Place Anchors under Simulated Seismic Loading. University of Wisconsin,
Milwaukee, WI, 2013.
10. Lin, Z.; Petersen, D.; Zhao, J. and Tian, Y., “Simulation and Design of Exposed Anchor Bolts
in Shear,” International Journal of Theoretical and Applied Multiscale Mechanics. 2011; 2(2),
pp. 111-29.
11. Petersen, D., “Seismic Behavior and Design of Cast-in-Place Anchors,” MS Thesis, University
of Wisconsin, Milwaukee, WI, 2011.
12. Petersen, D. and Zhao, J. "Design of Anchor Reinforcement for Seismic Shear Loads." ACI
Structural Journal. Vol. 110, No. 1, pp.53-62, 2013.
13. Petersen, D., Lin, Z., and Zhao, J. "Reinforcement for Headed Anchors." Final Report (Volume
I) for Project - Behavior and Design of Cast-in-Place Anchors under Simulated Seismic
Loading. University of Wisconsin, Milwaukee, WI, 2013a.
14. Petersen, D., Lin, Z., and Zhao, J. "Reinforcement for Headed Anchors." Final Report (Volume
II) for Project - Behavior and Design of Cast-in-Place Anchors under Simulated Seismic
Loading. University of Wisconsin, Milwaukee, WI, 2013b.
15. Saari, W.; Hajjar, J.; Schultz, A.; and Shield, C., “Behavior of Shear Studs in Steel Frames
with Reinforced Concrete Infill Walls,” Journal of Constructional Steel Research, Vol. 60, No.
10, 2004, pp. 1453-1480.
16. Shahrooz, B.; Tunc, G.; and Deason, J., "Outrigger Beam–Wall Connections. II: Subassembly
Testing and Further Modeling Enhancements," Journal of Structural Engineering, Vol. 130,
No. 2, 2004, pp. 262-270.
17. Zhao, J. and Lin, Z. "Design of Anchor Reinforcement for Seismic tension Loads." Engineering
Structures. (in review).
18. Zhao, J. 2014, "Seismic behavior of single anchors in plastic hinge zones of RC columns,"
Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake
Engineering July 21-25, 2014 Anchorage, Alaska.
19. Zhou, Y., Yu, J., Lu, X., and Lu, W. (2012) "Shaking table model test of a high-rise hybrid
structure building with steel frame-concrete core wall." Earthquake Engineering and
Engineering Vibration, Vol. 32, No. 2, pp. 98-105. (in Chinese).
Fig. 1 - Complex structures using embedded connections
35º
Fig. 2 - Typical concrete breakout failure of cast-in-place anchor bolts.
Fig. 3 - Existing reinforcement schemes for anchors in tension.
Fig. 4 - Proposed anchor tension reinforcement.
Fig. 5 - Dimension and reinforcement of tension specimens (1 in. = 25.4 mm)
45
40
35
Applied load (kips)

30
25
20
15 T2 T3
10
5 T3
T2
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Anchor Displacement (in.)
Fig. 6 - Anchor bolts in plastic hinge zone under tensile loading (1 in. =25.4 mm; 1 kip=4.5 kN)
Fig. 7 - Concept of proposed innovative embedded connection
Appendix: Proposed Revision for Anchor Tension Reinforcement
17.4.2.9 — Anchor tension reinforcement R17.4.2.9 — Anchor tension reinforcement
17.4.2.9.1 Anchor tension reinforcement shall R17.4.2.9.1 Two design approaches for
be designed either by 17.4.2.9.2 for anchors and proportioning anchor reinforcement are included in
anchor groups in mass concrete, or 17.4.2.9.3 for 17.4.2.9.1. The provisions of 17.4.2.9.2 are similar
anchors and anchor groups close to concrete to those of the 2011 Code. Section 17.4.2.9.3
edges. allows an alternative design, in which closed stirrups
and crack-controlling bars confine concrete around
the anchors. The confined concrete along with the
reinforcement transfers the tensile load to the
structure.
17.4.2.9.2 Where anchor reinforcement is R17.4.2.9.2 Same as the existing wording.
developed in accordance with 25.4 on both sides
of the breakout surface, the design strength of the
anchor reinforcement shall be permitted to be
used instead of the concrete breakout strength in
determining φNn. A strength reduction factor of
0.75 shall be used in the design of the anchor
reinforcement.
17.4.2.9.3 anchor tension reinforcement shall R17.4.2.9.3 With a goal to confine concrete in
include 1) load-carrying reinforcement in the front of anchors and to prevent concrete breakout,
direction of the anchors; 2) crack-controlling the anchor reinforcement consists of closely spaced
reinforcement in all directions that has a limited stirrups, corner bars, and crack-controlling bars
edge distance; and 3) local confining distributed along all concrete faces as illustrated in
reinforcement near the anchor head if side-face Fig. R17.5.2.9.3.
blowout may control the failure. The area of load-
carrying reinforcement shall be determined by
futa Ase,N
Asa = . (17.4.2.9.3)
fyt
where the limitation of 1.9fya on futa shall not be

applied. The value of fyt shall satisfy 20.2.2.4.
Fig. R17.4.2.9.3—anchor reinforcement for

tension.
The load-carrying reinforcement should be

proportioned to carry a force equal to the design
tensile capacity of the anchors, and implemented
using small-diameter closed stirrups. The closely
spaced stirrups with a center-on-center spacing of 2
to 3 in. can extend up to 1.0hef from the anchor bolt.
The design of the crack-controlling reinforcement Crack-controlling reinforcement are necessary

can use two strut-and-tie models: one describing crossing all possible cracks. All crack-controlling
the load transfer from the anchor head to the load- reinforcement should be proportioned using a steel
carrying reinforcement, and the other describing stress of 0.6fy, and implemented using small
the load transfer from the load-carrying diameter bars evenly distributed with a small and
reinforcement to the rest of the structure. practical spacing in two orthogonal directions. In
addition, the anchors and the load-carrying
reinforcement close to the anchor can be encased
by three or four closed hoops in the transverse
direction near the anchor head if the anchors are
close to a free side surface. With the anchor tension
reinforcement, anchor shaft fracture is expected if
other failure modes in 17.4 do not control the anchor
behavior.
The quasi-static cyclic tests of the reinforced
anchors in tension, conducted at the University of
Wisconsin, Milwaukee showed insignificant capacity
reduction. Therefore no capacity reduction is
needed for reinforced anchors subjected to cyclic
tensile loading
STEEL-TO-CONCRETE JOINTS WITH LARGE ANCHOR PLATES
UNDER SHEAR LOADING
Dipl.-Ing. Jakob Ruopp

jakob.ruopp@ke.uni-stuttgart.de

sekretariat@ke.uni-stuttgart.de
ABSTRACT
In industry and plant engineering especially, high flexi-
bility is required at joints between steel and concrete.
According to current standards, the maximum number
of fasteners is limited to an arrangement of 3 x 3 an-
chors on an anchor plate. The load-carrying behaviour
of large anchor plates under tension, shear and re-
straining forces was investigated within the scope of
the research project “Large Anchor Plates with Headed
Studs for Highly Stressed Constructions in Industry
and Plant Engineering”. This paper describes the re-
search results obtained at the University of Stuttgart for
large anchor plates under shear loading. Findings for
large anchor plates under tension and restraining
forces were mainly investigated at the University of
Kaiserslautern (Scholz et al., 2016).
The shear behaviour and distribution of forces within
the anchor plate are crucial issues in the development
of a design concept for steel-to-concrete joints with
large anchor plates. The influence of different parame-
ters, such as the dimensions of the anchor plate, the Fig. 1 – Test setup for shear loading
embedment depth of the headed studs or the eccen-
tricity of the shear force, have been studied in several test series. Supplementary reinforcement
was placed close to the headed studs in tension to strengthen the load-carrying capacity of the
joint. The distribution of the shear forces has been assessed by means of numerical investigations
together with the influence of further parameters such as concrete strength and reinforcement
ratio. With regard to the load distribution within the anchor plate, elastic and plastic design ap-
proaches have been taken into account. Based on these studies, an efficient design model has
been developed for steel-to-concrete joints with large anchor plates and a higher number of fas-
teners than originally permitted.
1. INTRODUCTION
Recently, hybrid building technology, where both materials, steel and concrete, are used accord-
ing to their properties and their most suitable application, has spread among buildings and indus-
trial practice, such as steel frames connected to column bases on foundations, or steel or com-
posite beams joined to the concrete walls of staircases. Recently, several design concepts have
been developed for steel-to-concrete joints within national and international research projects.
Design approaches based on the component method have gained acceptance, where concrete
and steel components are considered simultaneously.
In industry and plant engineering, large forces have to be transferred from structural steel ele-
ments to concrete elements such as walls or foundations. As a consequence of this, large anchor
plates are required which enable the activation of a sufficiently large concrete volume to support
these loads. However, current design rules limit load-carrying capacities due to the limitation of
nine fasteners per anchor plate and the elastic design concept, which is the dominant method for
the design of fastenings. Therefore, the resistance of an anchor plate is limited, which leads to
complicated load distribution arrangements over several single anchor plates or solutions in pure
concrete that avoid the “difficult” interface between steel and concrete. The scope of the research
project was the development of a design model for large anchor plates with more than nine
headed studs per anchor plate. Tests under normal, shear and restraining forces were conducted
within the scope of the project (Figure 1).
2. STATE OF THE ART AND DESIGN CONCEPTS

The design concept for large an-
chor plates was developed
based on the component method
(Jaspart, 1991; Wald et al., 2008;
Kuhlmann et al., 2012) which dis-
tinguishes between different fail-
ure modes based on a character-
ized structural behaviour of the
single components. In an overall
design concept, steel compo-
nents according to EN 1993-1-8
and concrete components ac- Fig. 2 – Elastic (left) and plastic (right) distribution of forces
cording to EN 1992-4 are consid-
ered simultaneously. In addition to the load-carrying capacity of the joint, the load-deformation
behaviour can be determined for the whole joint.
In the design of the joint, it is possible to strengthen certain, more flexible components, thus
achieving ductile behaviour of the joint. For example, if the thickness of the anchor plate is varied,
the flexible behaviour of the anchor plate, which is considered within the T-stub under tension and
compression according to EN 1993-1-8, can come into play. On the other hand, higher re-
sistances for some concrete failure modes can be reached if, for example, supplementary rein-
forcement is placed next to the headed studs in tension and the interaction between concrete and
reinforcement is considered according to recent approaches (Berger, 2015; Kuhlmann et al.,
2012; Kuhlmann et al., 2015).
The more detailed analyses of the load-carrying behaviour of the single components, their indi-
vidual stiffnesses and their interaction enable assumptions to be made regarding the loading side
for the distribution of loading which are significant for the design of anchor plates. Internal forces
such as normal, shear and bending forces are distributed among the rows of headed studs ac-
cording to an elastic design concept (EN 1992-4) if the anchor plate can be assumed to be suffi-
ciently rigid (Figure 2). As part of the elastic design concept, it is assumed that all fasteners have
the same stiffness. Load redistributions in the area of the anchor plate should not be considered,
as the fastener carrying the maximum load is critical for the load-carrying capacity of the anchor
plate. The elastic design concept is mainly used for the design of fasteners because it is especially
suitable for concrete failure modes. Due to the reduced deformation capacity of the pure concrete
components, load redistributions are neglected. Plastic redistributions among the rows of headed
studs may be taken into account if brittle concrete failure modes are excluded. In the case of
flexible anchor plates, a non-linear distribution of forces is possible, based on the requirement of
a technical report referenced by EN 1992-4.
A plastic design concept was derived for anchor plates with several rows of fasteners within the
scope of the first investigations of the ductile behaviour of steel-to-concrete joints (Cook et al.,
1989). Besides the non-linear distribution of normal forces, this concept allows shear forces to be
distributed according to the utilization factor based on the interaction equations of each row. This
is contrary to the elastic approach, where the shear forces are distributed equally among the rows.
Limiting cases for the distribution of shear forces may be derived from equilibrium conditions for
different eccentricities. With a large eccentricity of the load, for example, the whole shear force
can be carried by friction. For smaller eccentricities, shear loads are carried by friction and by a
contribution of the rows of fasteners on the loaded side, and in the case of very small eccentrici-
ties, the shear force is distributed equally among the rows of fasteners. An improved approach
was developed for plastic design in (Lotze et al., 1997). For a full utilization based on the interac-
tion of normal and shear forces, sufficient deformability should be available at the anchor plate
and its fasteners.
3. EXPERIMENTAL INVESTIGATIONS
3.1 GENERAL
Firstly, experimental investiga-
tions were carried out to gather in-
formation on the behaviour of
large anchor plates under shear
loading and to apply these results
as validation for further numerical
investigations. In a second step,
an analytical design model that
takes into account the load distri-
bution in the steel-to-concrete joint
was developed based on the ex-
perimental and numerical studies
(Kuhlmann et al., 2016; Kurz et al.,
2016). Fourteen tests were carried
out on large anchor plates under
shear loading without edge ef-
fects. Steel-to-concrete joints, e.g.
in walls, were considered for these
configurations as the full concrete Fig. 3 – Reinforcement plan of test specimen R2-3Q and configu-
volume can be activated without ration of anchor plates for the other test series
any influences of the edges.
The following parameters were varied within the specified limits:
 Eccentricity e (80 and 1000 mm)
 Thickness of anchor plate tap (15 and 40 mm)
 Length of headed stud hn (100 and 250 mm)
 Cracked and uncracked concrete
 Reinforcement ratio of supplementary reinforcement (1 ø8 mm and 2 ø8 mm)
By varying the parameters given in Table 1, concrete and steel failure modes were planned be-
forehand and achieved in testing. In order to identify the influence of the different parameters, a
testing matrix was chosen in a way that reference tests had been defined from which the other
tests differed by only one single parameter. Thus, the influence of the various single parameters
could be clearly identified.
Table 1 – Test parameters of shear tests
Test Eccentricity e Thickness of an- Supplementary reinforce- Concrete
Headed stud
[mm] chor plate [mm] ment per headed stud state
B3-Q SD16/100 80 40 1 ø8 mm cracked
R2-1Q SD16/100 1000 40 2 ø8 mm uncracked
R2-3Q(1) SD16/100 1000 40 2 ø8 mm cracked
R2-3Q(2) SD16/100 1000 40 2 ø8 mm cracked
R2-4Q(1) SD16/100 80 40 2 ø8 mm cracked
R2-4Q(2) SD16/100 80 40 2 ø8 mm cracked
R3-2Q(1) SD16/250 1000 40 2 ø8 mm uncracked
R3-2Q(2) SD16/250 1000 40 2 ø8 mm cracked
The tension loading on the headed studs on the unloaded side of the anchor plate was influenced
by the variation of the eccentricity. In tests with a large eccentricity, higher tension forces were
obtained due to the resulting higher bending moment at the anchor plate. For this reason, different
values of eccentricity were used. In addition, the thickness of the anchor plate was varied in order
to determine the influence of a flexible anchor plate.
Where a concrete cone developed due to tension loading on the headed studs, these forces could
be tied back by supplementary reinforcement, and, as a consequence, the load-carrying capacity
of the related concrete failure modes increased. The supplementary reinforcement was placed
next to the headed studs as shown in Figure 3 based on previous research results (Kuhlmann et
al., 2012). In test B3-Q, one leg of a stirrup was placed next to the headed stud, whereas one
stirrup with two legs per headed stud was considered in all other tests. In addition, several tests
were performed in cracked concrete. Metal sheets were placed near the headed studs in tension
and a crack width of approx. 0.3 mm was achieved on the axis of the two rows of headed studs
on the unloaded side (Figure 3).
3.2 LOADBEARING BEHAVIOUR AND FAILURE MODES

The tests under shear load were conducted at the Materials Testing Institute, University of
Stuttgart (Materialprüfanstalt Universität Stuttgart). To verify the analytical model, different failure
mechanisms, such as concrete or steel failure, were initiated by varying the parameters given
above (Figure 4).
Concrete pry-out failure
was observed in the series
with short headed studs, for
tests with small and large
eccentricity (Figures 4a and
4c). An elastic distribution
of the normal forces was
present in these tests, as
the deformation of the an-
chor plate itself was very
small. The cross-section of
the anchor plate remained
plane and no plastic defor-
mations occurred. The em-
bedment depth of the sup-
plementary reinforcement
in the concrete cone was
small in comparison to the
tests with long headed
studs. Nonetheless, the
supplementary reinforce-
ment was activated. In the
test with a smaller rein- Fig. 4 – Failure mechanisms of shear tests with large anchor
forcement ratio (B3-Q), the plates
180 1200
load-carrying capacity was
Force [kN]
Force [kN]
Test R2-1Q Test B3-Q

160 Test R2-3Q(1) Test R2-2Q
1000
20 % lower (Figure 5b) than 140 Test R2-3Q(2) Test R2-4Q(1)
Test R2-4Q(2)
120 800
in the test with high rein- 100
600
forcement ratio (R2-2Q). In 80
60 400
the test series with cracked 40
200
concrete, the load-carrying 20
capacities were reduced by 0

0 5 10 15 20 25 30 35 40 45 50 55 60 65
0
0 5 10 15 20 25 30 35
Displacement [mm] b)
20 % compared with the a) Displacement [mm]
reference tests (Figures 5a Fig. 5 – Load-deformation curve of test series 2 with a) large ec-
and 5b). centricity and b) small eccentricity
The tests with longer headed studs and a thicker anchor plate (Figure 4b) resulted in a pry-out
failure with a large concrete cone. The supplementary reinforcement was activated and the yield
stress of the reinforcement was reached. In the tests with a small eccentricity, high shear forces
occurred and the joint failed due to the headed studs shearing off. The influence of the pre-in-
duced cracks was negligible, as many radial cracks developed in cracked and uncracked con-
crete. Only small deformations of the thick anchor plate were measured in these tests. In the rows
of headed studs in tension, plastic redistributions
occurred, measured by the strain gauges, due to
the large uplift on the unloaded side of the anchor
plate.
In the tests with thin anchor plates, the influence of
the thickness of the anchor plate was observed by
means of the development of yielding mechanisms
(Figure 4d). In the tests with a large eccentricity,
yielding mechanisms developed in the tension Fig. 6 – Yielding mechanism within the an-
zone of the anchor plate. At maximum load, a chor plate at maximum load
mixed failure occurred due to yielding in the anchor plate and yielding in the headed studs (Figure
6). In the tests with small eccentricity and thin anchor plate, the headed studs sheared off at the
base of the shaft similar to the tests with a thick anchor plate.
3.3 LOAD DISTRIBUTION OF NORMAL FORCES

One of the main goals of the 1,1 1,5
Distribution of normal forces
Distribution of normal forces

experimental investigations 0,9
25% Fu,test
50% Fu,test 1,3
25% Fu,test
50% Fu,test
was to determine the distri- 0,7 75% Fu,test
100% Fu,test
1,1
0,9
75% Fu,test
100% Fu,test
0,5
bution of normal forces in the 0,3
0,7
different rows of headed 0,1

0,5
0,3
studs by varying the eccen- -0,1 0,1
-0,3 -0,1
tricity of the shear force. To -0,5 -0,3
do that, strain gauges were a)
Row 4 Row 3 Row 2 Row 1
Row of headed studs b)
Row 4 Row 3 Row 2 Row 1
Row of headed studs
attached to the shafts of the Fig. 7 – Load distribution of normal forces: a) test R2-1Q and b)
headed studs, to the supple- test R5-1Q
mentary reinforcement and
to the top of the anchor plate. The stresses were evaluated by means of strains. In the rows of
headed studs, the strain gauges were located according to a chessboard pattern. They were
attached to the middle of the shaft in order to avoid the influence of the bending. The forces in the
rows of headed studs were determined as mean values of the results measured by the strain
gauges based on the mechanical properties of the material. An elastic distribution of normal forces
was monitored in the tests with short headed studs and concrete failure modes. On the other
hand, in the tests with long headed studs, plastic redistribution was measured towards the inter-
mediate rows of studs. Compression forces were determined up to the maximum load in the first
row in all tests with large eccentricity (Figure 7). The values measured by the strain gauges for
large anchor plates and short headed studs comply with the definition of load distribution within
anchor plates according to (Eligehausen et al., 2006), which assumes an elastic distribution of
forces for brittle failure modes such as concrete pry-out failure. An elastic distribution of forces
was determined in the tests with short headed studs (Figure 7a). Plastic redistribution among the
rows of headed studs was observed for ductile failure modes such as steel failure in test R5-1Q
(Figure 7b).
4. NUMERICAL INVESTIGATIONS
The aim of the numerical 400 1200
Maximum numeric load (Fu,FE)
Maximum numeric laod (Fu,FE)
R2-1Q B3-Q
investigations of the shear R2-3Q(1) not conservative
R2-3Q(2)
R2-2Q not conservative
300 R2-4Q(1)
tests was to recalculate R3-1Q
R3-2Q(1)
1100
R2-4Q(2)
the maximum loads and 200
R3-2Q(2)
R5-1Q 1000
R3-3Q
R5-3Q
the failure modes of the R5-2Q
conservative conservative
tests. Based on the tests, 100 900
the numerical model was 0 800

validated and the load dis- 0 100 200 300 400 800 900 1000 1100 1200
a) Maximum load in test (F ) b) Maximum load in test (F )
tributions in the anchor u,test u,test
plate evaluated – specifi- Fig. 8 – Comparison of maximum loads of FE model and test results
cally, the distribution of for a) large eccentricity and b) small eccentricity
shear forces among the
rows of headed studs. In addition, parameter studies were conducted. The investigations of the
anchor plate considered various parameters, including thickness, headed stud embedment depth
and shear force eccentricity. The numerical results allowed also the enlargement of the database
for the validation of the P1-2: P1-5:
e = 1000 mm e = 80 mm
analytical model. The tap = 40 mm tap = 40 mm
non-linear FE program
MASA (Macroscopic
Space Analysis) was
used for the numerical
studies as it permits
modelling of the local- 0,9
P1-2:
0,9
P1-5:
ized damage behaviour 0,7 0,7
Distribution of shear forces

in the concrete of the 0,5 0,5
shear joints (Ožbolt 0,3 0,3
2014). Good correlation 0,1 0,1
was observed between -0,1

50% Fu
-0,1
50% Fu
the maximum loads -0,3 75% Fu
100% Fu
-0,3 75% Fu
100% Fu
from the tests and the -0,5
Row 4 Row 3 Row 2 Row 1 Friction
-0,5
Row 4 Row 3 Row 2 Row 1 Friction
numerical results. Fig-
P1-14: P1-17:
ures 8a and 8c com- e = 1000 mm e = 80 mm
pare large and small tap = 15 mm tap = 15 mm
eccentricities. The nu-

merical model was also
able to reproduce the
load-carrying behaviour
of all different failure 0,9 0,9
modes such as steel 0,7
P1-14:
0,7
P1-17:
and concrete failure for 0,5 0,5

the same assumptions 0,3 0,3
concerning boundary 0,1 0,1
and modelling condi- -0,1 -0,1
tions. The non-linear -0,3
50% Fu
-0,3
50% Fu
75% Fu 75% Fu
behaviour of the steel -0,5
100% Fu
-0,5
100% Fu
was considered by the Row 4 Row 3 Row 2 Row 1 Friction Row 4 Row 3 Row 2 Row 1 Friction
von Mises yield crite-

rion and the damage
behaviour of the con-
Fig. 9 – Principal compressive stresses in N/mm² and load distribu-
crete by the microplane
tion of shear forces within the different rows of studs for different ec-
model implemented in
centricities and anchor plate thicknesses
MASA.
A number of parametrical studies were performed in order to examine the load-carrying behaviour
of the anchor plates and to quantify the influence of different factors such as geometry and mate-
rial properties. The parameters were selected so that the upper and lower values were validated
by test results and the database was enlarged by parameter studies for values in between those
boundaries. In addition, parameter studies were also conducted beyond the configurations of the
test specimens. The concrete strength and the reinforcement ratio were varied in addition to var-
ying the parameters eccentricity and anchor plate thickness. The studies of the influence of ec-
centricity and anchor plate thickness are shown in the following. A detailed description of all pa-
rameter studies is given in the final research report (Kurz et al., 2016).
The distribution of shear forces among the different rows of headed studs was quantified by the
numerical investigations, because in the tests, it was not possible to determine the shear force
per stud in a similar way as for the normal forces. Therefore, the nodal forces, derived from the
numerical calculations, were summarized in a section through the base of the headed studs in
the different rows and compared with each other. The distribution of shear forces is illustrated in
Figure 9 by the principal compressive stresses in the concrete. The results of the different princi-
pal compressive stress plots are scaled equally to the range between 0 and -10 N/mm² in order
to be able to compare the plots with each other. Varying the eccentricity of the shear force has an
influence on the load distribution of the tensile forces on the unloaded side and in the distribution
of the shear forces, since large deformations occur due to the tension uplift of the anchor plate in
the area of the T-stub (P1-2 and P1-14). High shear forces were determined in the shafts of the
headed studs, especially with a thin anchor plate. For the studies with large eccentricities, a high
contribution by the friction forces was determined for the shear capacity due to the comparatively
compressive forces at the anchor plate. In these cases the shear force is mainly transferred by
friction and the first row of studs. The distribution of the shear forces, determined in numerical
studies with a small eccentricity, is approximately similar for thick and thin anchor plates. The
shear force is distributed equally among the rows of headed studs (P1-5 and P1-17). The obser-
vations of the load distribution of the shear force, based on the numerical studies, are generally
in line with the current plastic design concept for steel-to-concrete joints. However, high tensile
and shear forces may occur due to membrane effects in the tension zone of the anchor plate.
These effects require a common design concept for steel and concrete components.
5. COMPONENT MODE FOR STEEL-TO-CONCRETE JOINTS

5.1 GENERAL
In the design concept of the component method, the design resistances of the single components
are compared with their actual loading accordingly. In addition, interaction equations due to load-
ing in different directions have to be considered. Within the scope of a design concept for steel-
to-concrete joints under shear loading, simplifications have to be considered, especially for anchor
plates with a higher number of headed studs due to the high degree of static indeterminacy. A
simplified analytical design model is proposed for steel-to-concrete joints with large anchor plates.
In this model the tensile forces in the rows of headed studs on the unloaded side are assembled
into one resultant tensile force. The analytical design model for the tension component has been
calibrated and verified by test results under normal forces at the University of Kaiserslautern
(Scholz et al., 2016). On the loading side, an elastic design concept with brittle failure mechanisms
may be considered as well as a plastic design concept with load redistributions corresponding to
the large deformations of the steel components.
5.2 SINGLE COMPONENTS FOR TENSION AND SHEAR
The component model for the steel-to-concrete joints is based on existing design concepts for the
failure modes of the single components. The concrete capacity design (Fuchs et al., 1995; Eli-
gehausen et al., 2006; EN 1992-4) for the current concrete components is expanded by more
current approaches that consider the reinforcement near the fasteners (Berger, 2015; Kuhlmann
et al., 2012; Kuhlmann et al., 2016).The analysis of the concrete components have to consider,
on one hand, the fastener with the highest load and, on the other, the group failure mechanisms.
Simplified design approaches were developed (Kurz et al., 2016), where the resistances of the
supplementary reinforcement and the concrete are combined. For the steel components within
the anchor plate, the T-stub model for tension and compression loading (Wald et al., 2008; EN
1993-1-8) is considered, which has been extended by newer concepts for T-stubs with several
rows of bolts (Couchaux et al., 2015).
5.3 COMPONENT MODEL FOR LARGE ANCHOR PLATES UNDER SHEAR
In the following, an analytical design model is proposed for steel-to-concrete joints under shear
loading which considers the different requirements and assumptions concerning the load distri-
bution and the resistances depending on the different failure mechanisms. The eccentric loading
of a steel-to-concrete joint is split into a bending moment for the definition of the normal compo-
nents and into a pure shear force for the determination of the shear components, see Figure 10.
This subdivision is applied to all failure modes. Of course, both loading directions are considered
for the interaction equations.
Fig. 10 – Design approach for large anchor plates under shear loading: a) load distribution for
steel failure, b) load distribution for concrete failure and c) interaction formulas for tension and
shear forces
Furthermore, the steel and concrete failure mechanisms are differentiated. Therefore, it is possi-
ble to distinguish between a single failure of the most heavily loaded fastener for steel failure and
a group failure for concrete failure. In contrast to EN 1992-4, the components of the reinforcement
are associated with the group failure mode because the recent approach for the load-carrying
capacity of the reinforcement considers the interaction between steel and concrete. So the re-
sistance of the concrete cone is considered for the strength of the characteristic concrete compo-
nent which should only be determined by considering the group of fasteners. For the tensile com-
ponents, increases of the resistance are considered due to the overlap of the compression and
tension zones (Zhao 1994). If elastic distribution of normal forces is assumed, effects due to an
eccentric loading are included in the calculation of the tensile resistance for concrete failure (EN
1992-4). Within the scope of the plastic design concept, if brittle failure mechanisms can be ex-
cluded, the loading may be distributed equally among the rows of headed studs assuming plastic
deformations. The tensile loading may be assigned to the rows of headed studs on the unloaded
side for steel failure mechanisms, and the shear force that cannot be carried by the friction com-
ponents is assigned to the first row of studs (Figure 10).
The normal forces in the rows of headed studs on the unloaded side are determined based on
the equilibrium of vertical forces and moments around the compression zone (Figure 2). The in-
ternal forces can be distributed within the anchor plate according to Figure 10 for the different
failure modes. The plastic design concept can be applied, if steel failure is the governing failure
mode and the interaction equations for the plastic design given in Figure 10 are considered. Ini-
tially, an assumption is made for the depth and location of the compression zone and the number
of rows of headed studs in tension. Within an iterative calculation, the assumption is reviewed
and probably updated. A step-by-step procedure for designing the steel-to-concrete joints under
shear loading and a general procedure are given in the project report (Kurz et al., 2016). In addi-
tion to a simplified design model according to (Kurz et al., 2016), where the tension loading is only
assigned to the two rows of headed studs on the unloaded side, a more detailed model imple-
mented in design software is described. Consequently, load redistributions and the flexible anchor
plate can be considered based on the displacement method using the stiffness of the headed
studs (Kuhlmann et al., 2012).
5.4 VERIFICATION OF THE COMPONENT MODEL

For the tests with large 400 1200
Maximum load in model (Fu,model)
Maximum load in model (Fu,model)
R2-1Q
eccentricity, the results R2-3Q(1) not conservative
B3-Q
R2-2Q not conservative
from the analytical de- 300 R2-3Q(2)
R3-1Q
1000 R2-4Q(1)
R2-4Q(2)
sign model for the dif- R3-2Q(1)
R3-2Q(2)
R3-3Q
R5-3Q
200 800
ferent failure modes R5-1Q
R5-2Q
agree sufficiently well 100
conservative
600
conservative
with the maximum test
loads. The actual fail- 0
0 100 200 300 400
400
400 600 800 1000 1200
ure modes are also re- a) Maximum load in test (F ) b)
u,test
Maximum load in test (F
u,test )
produced by the de- Fig. 11 – Comparison of maximum loads of the analytical model and
sign concept (Figure the experimental results for a) large eccentricity and b) small eccen-
11a). The comparison tricity
between the results of
the design model and the tests with small eccentricity show a conservative approach to the load-
carrying capacities (Figure 11b). For the tests with small eccentricity and short headed studs, the
resistance of the concrete pry-out failure is the governing failure mode in the analytical design
model. As the supplementary reinforcement in the area of the headed studs cannot yet be con-
sidered within the model for pure shear, increases in the load-carrying capacity of this component
will be possible in the future development of an improved model. The database for validating the
analytical model was enlarged with the help of the numerical parameter studies. Further influ-
ences of different parameters were considered. In addition to the validation of the analytical model
in comparison to the 14 test results, approx. 90 additional numerical investigations were used for
verification (Kurz et al., 2016).
6. CONCLUSIONS
Results of a research project were presented, where design models for large anchor plates under
shear, tension and restraining forces had been generated. These models were developed based
on experimental and numerical investigations for normal and restraining forces (Scholz et al.,
2016) and shear forces. One aim of the project was to determine the load distribution of forces at
the different rows of studs on the anchor plate, measured in the tests by strain gauges on the
shafts of the headed studs and obtained by numerical calculations. Besides the embedment depth
of the headed studs, other parameters were varied such as eccentricity, thickness of anchor plate
and the reinforcement ratio of the supplementary reinforcement. In addition, tests were performed
in cracked and uncracked concrete. Extensive numerical parameter studies were performed in
order to enlarge the database of the tests and to determine the distribution of shear forces among
the rows of headed studs. The numerical models were validated by the test results. The distribu-
tion of shear forces among the rows of headed studs could be quantified based on the numerical
studies. The load-carrying behaviour and the influence of different parameters were examined
with the help of the experimental and numerical studies. It might be possible to expand existing
design concepts (EN 1992-4; EN/TR 1992-4) for large anchor plates if the requirements of an
elastic or plastic design are considered consequently. Therefore, the resistances of the different
failure modes of the single steel and concrete components (EN1993-1-8; EN 1992-4; EN/TR
1992-4) were used and extended by recent approaches to take into account the supplementary
reinforcement according to (Berger, 2014; Kuhlmann et al., 2012). Further studies of the load-
carrying behaviour of large anchor plates were performed (Ruopp 2018). Edge effects are not yet
considered in the design model and have to be conservatively estimated by applying all loads to
the row of headed studs close to the edge. Further numerical and experimental investigations
have to be carried out for large anchor plates with edge effects in order to estimate the load-
carrying behaviour and the load redistributions.
7. ACKNOWLEDGEMENTS
This IGF project of the Deutscher Ausschuß für Stahlbau (DASt) was funded by the AiF within the
scope of the cooperative Industrial Research Programme (IGF) by the Federal Ministry for Eco-
nomic Affairs and Energy based on a decision by the German Bundestag. The authors would like
to take this opportunity to express their gratitude to all supporters of their research. Special thanks
go to the industry partners Köster & Co. GmbH and Rau-Betonfertigteile GmbH & Co. KG for
supplying the test specimens. Moreover, the authors would like to thank the project partners from
the University of Kaiserslautern for the good cooperation and the Materials Testing Institute, Uni-
versity of Stuttgart (MPA), for their support regarding the tests.
REFERENCES
Berger, W. (2015). Trag- und Verschiebungsverhalten sowie Bemessung von Kopfbolzenverankerungen
mit und ohne Rückhängebewehrung unter Zuglast (Load-displacement behaviour and design of anchor-
ages with headed studs with and without supplementary reinforcement under tension load). PhD thesis,
Institute of Construction Materials, University of Stuttgart.
Cook, A.; Klingner, R. (1989): Behavior and Design of Ductile Multiple-Anchor Steel-to-Concrete Connec-
tions. Research Report CTR 1126-3, University of Texas at Austin.
Couchaux, M.; Demonceau, J.-F.; Weynand, K. (2015). Calcul d’un assemblage par platine comportant
quatre boulons par range. Revue Construction Métallique 2.
EN 1993-1-8 (Eurocode 3): Design of steel structures – Part 1-8: Design of joints, Dec 2005.
EN 1992-4 (draft, Eurocode 2): Design of concrete structures – Part 4: Design of fastenings for use in
concrete (not yet published).
EN/TR 1992-4: Design of fastenings for use in concrete – Plastic design of fastenings with headed and
post-installed fasteners, Oct 2013 (not yet published).
Eligehausen, R.; Mallée, R.; Silva, J. F. (2006). Anchorage in Concrete Construction, Ernst & Sohn, Ber-
lin.
Fuchs, W.; Eligehausen, R.; Breen J. E. (1995). Concrete Capacity Design (CCD) Approach for Fastening
to Concrete. ACI Structural Journal 92. pp. 73–94.
Jaspart, J.-P. (1991). Étude de la semi-rigidité des noeuds poutre-colonne et son influence sur la résis-
tance et la stabilité des ossatures en acier. PhD thesis, University of Liège.
Kuhlmann, U.; Hofmann, J.; Wald, F.; da Silva, L.; Krimpmann, M.; Sauerborn, N. et al. (2012). New Mar-
ket Chances for Steel Structures by Innovative Fastening Solutions between Steel and Concrete (IN-
FASO). Final report, Report EUR 25100 EN, European Commission.
Kuhlmann, U.; Hofmann, J.; Wald, F.; da Silva, L.; Krimpmann, M.; Sauerborn, N. et al. (2015). Valorisa-
tion of Knowledge for Innovative Fastening Solutions between Steel and Concrete (INFASO+). Final re-
port, EUR 27745 EN, European Commission.
Kuhlmann, U.; Ruopp, J. (2015). Anschlüsse zwischen Stahl und Beton – Anwendung in der Praxis
(Steel-to-Concrete Joints – Practical Application). Bauingenieur 90, No. 12, pp. 583–593.
Kuhlmann, U.; Ruopp, J. (2016) Large Anchor Plates with Headed Studs for Highly Stressed Construc-
tions in Industry and Plant Engineering under Shear Loading. Test report, research project AiF/IGF-Nr.
17654, Deutscher Ausschuß für Stahlbau (not published).
Kurz, W.; Kuhlmann, U.; Scholz, J.; Ruopp, J. (2016). Large Anchor Plates with Headed Studs for Highly
Stressed Constructions in Industry and Plant Engineering. IGF Project, Deutscher Ausschuß für Stahlbau
(DASt), final report.
Lotze, D. Klingner, R. E. (1997). Behavior of Multiple-Anchor Connections to Concrete from the Perspec-
tive of Plastic Theory. PMFSEL Report No. 96-4, University of Texas at Austin, pp. 10–17.
Mishaxhiu, N. (2014). Steel-to-Concrete Joints – Development of a FE-Model for Large Anchor Plates
with Concrete Failure Mechanisms. Master thesis, No. 2015-10X, Institute of Structural Design, University
of Stuttgart.
Ožbolt, J. (2014). MASA – Microplane Analysis Program, Institute of Construction Materials, University of
Stuttgart.
Ruopp, J. (2019). Studies on Steel-to-Concrete Joints under Shear and Moment Loading with Large An-
chor Plates or Various Edge Influences (working title). PhD thesis, Institute of Structural Design, Univer-
sity of Stuttgart.
Rybinski, M.: (2014). Komponentenmethode für Ankerplatten mit Kopfbolzen unter einachsiger
Beanspruchung (Component method for anchor plates with headed studs under uniaxial loading). PhD
thesis, Institute of Structural Design, University of Stuttgart.
Scholz, J.; Kurz, W. (2016). Anschlüsse zwischen Stahl und Beton mit Zwangsbeanspruchung (Steel-to-
concrete joints with restraining forces). Stahlbau 85, pp. 811–821.
Tschemmernegg, F.; Huber, G.; Huter, M.; Rubin, D. (1997). Component Method and Component Tests.
Development of Building Constructions in Mixed Structures. Stahlbau 66, pp. 624–639.
Wald, F.; Sokol, Z.; Steenhuis, M.; Jaspart, J.-P. (2008). Component Method for Steel Column Bases.
HERON 53, No. 1/2.
Zhao, G. (1994). Tragverhalten von randfernen Kopfbolzenverankerungen bei Betonbruch (Loadbearing
behaviour of fastenings with headed anchors at concrete failure remote from edge). PhD thesis, Institute
of Construction Materials, University of Stuttgart.
Seismic column base connections: experiments, simulations,
component and building models, and design implications
Amit M Kanvinde
Department of Civil and Environmental Engineering
University of California
Davis, California
kanvinde@ucdavis.edu
ABSTRACT
Column base connections in steel frames usually require attachment of a steel column to a
concrete footing, either through anchor rods attached to a base plate, or through embedment of
the column into the footing. In either case, their response is controlled by complex interactions
between the steel and concrete components of the connection. Results from numerous studies
on these connections over the last decade are presented. These studies, conducted primarily in
an earthquake-engineering context, include: (1) experiments on exposed base plate as well as
embedded column connections, (2) finite element simulations, (3) methods for strength and
stiffness characterization, (4) component hysteretic models, and (5) nonlinear time history
simulations of buildings incorporating base connection response. Cumulatively, these studies
have resulted in an improved physical understanding of base connection response and tools to
design as well as simulate them. Key results from each of these studies are summarized, along
with integrative insights from all of them. These insights have implications for connection detailing,
as well as philosophies for designing safe and economical structures in seismic regions that
leverage beneficial aspects of connection response while considering possible weaknesses.
Areas for future research are identified.
INTRODUCTION
Connections between steel columns and concrete footings transfer moments, axial, and shear
forces from the entire structure into the foundation. Their design and construction takes numerous
forms, depending on the structural system they belong to (e.g., moment vs braced frames), and
the magnitude or type of forces they must resist. Broadly, these may be classified as exposed
type (Figure 1a), in which a base plate is anchored to the footing, or embedded type (Figure 1b),
in which the lower portion of the column is embedded into the footing. The former is appropriate
for low-rise moment or braced frames, whereas taller frames usually require some degree of
embedment for force and moment transfer, since an impractical number (or size) of anchor rods
is required in an exposed type connection.
This paper describes research on various aspects of these connections conducted over the last
decade, focusing on the work done in the author’s research group. These aspects include (1) load
resisting mechanisms, and strength and deformation characteristics, (2) stiffness of these
connections, and (3) implications of connection response on structural performance. The research
described here includes a comprehensive program of testing (>30 experiments), finite element
simulations for connection response, model development of strength and stiffness, and finally
frame simulations to examine the effect of base connection flexibility on structural response. The
paper begins by briefly reviewing some of the early research on these connections, before
describing developments over the last decade. The paper concludes by summarizing research
needs for the future.
(a) (b) (c)
Steel Steel
Steel Overtopping
Column Column
Column slab
Concrete
footing
Figure 1 – Column base connections and force transfer mechanisms: (a) exposed with base plate and
anchor rods, (b) deeply embedded, and (c) shallowly embedded due to overtopping slab
PRIOR RESEARCH ON COLUMN BASE CONNECTIONS
The American Institute of Steel Construction’s (AISC) Steel Design Guide One (Kloiber & Fisher,
2006) along with the Seismic Design Manuals published by AISC (AISC, 2010) and the Structural
Engineers Association of California (SEAOC, 2012) are primary references for the design of base
connections. These documents represent the synthesis and culmination of research conducted
over the last five decades. This includes experimental research by (Burda & Itani, 1999), (Astaneh
& Bergsma, 1999), and analytical research by (Ermopoulos & Stamatopoulos, 1996), and
(Pertold, Xiao, & Wald, 2000). For an exhaustive list see (Grauvilardell, Lee, Hajjar, & Dexter,
2005). A key part of Steel Design Guide One is the “Drake-Elkin Method,” (Drake & Elkin, 1999)
for the design of exposed base plate connections under axial compression and flexure. The
method is based on assuming a pre-determined stress distribution (based on the concrete or
grout failure stress) under the base plate under applied loads (illustrated schematically in Figure
1a) and then performing design checks on various parts of the connections. These include
checking the base plate in flexure on the tension and compression side of the connection, and
checking the anchor rods for tensile yielding. The design guide also provides guidance for
designing base plates subjected to pure compression.
Research described in this paper focuses on details, and aspects of response not directly
addressed by Design Guide One and preceding work. Specifically, the following issues are
discussed:
1. Strength and deformation capacity of base connections: For exposed base connections, this
involves: (a) experimental verification of the methods described in Design Guide One in a
cyclic loading dominated seismic context, (b) examination of the response excluded from
Design Guide One, and yet commonly used in construction, i.e., exposed base connections
with more than 2 rows of anchor rods (see Figures 2b and c), and embedded base
connections (Fig 1b), and (c) examination of shear-transfer mechanisms in base connections.
2. Flexibility of base connections: Even if designed to meet strength requirements (i.e., designed
to remain elastic under imposed loading), the base connections have significant flexibility.
This paper described models developed to characterize this flexibility.
3. Implications for building performance and design: Flexibility in the base connections results in
force redistribution and change in building seismic performance. This paper describes findings
from time history simulations that assess these effects. Additionally, the paper examines
whether dissipative characteristics of the column bases may be used in an advantageous
manner for seismic design.
Each of the points above constitutes one section of the paper. Within each broad theme, different
details are discussed along with appropriate references.
STRENGTH AND DEFORMATION CAPACITY OF BASE CONNECTIONS

As discussed in the preceding section, the strength of column base connections was examined
for different types of details, and for different mechanisms (i.e., axial force and moment, or axial
force and shear).
Exposed Base Plate Connections: Figure 2 shows the different types of exposed base
connections tested by the author’s group over the last decade. These include the following:
1. Exposed base plate connections with a standard 4 bolt pattern similar to that addressed in
Design Guide One subjected to combinations of axial compression and shear – see Figure
2a.
2. Exposed base plate connections with two types of 8 rod patterns subjected to axial
compression and shear (Figure 2b and 2c); these are not addressed by Design Guide One,
yet commonly used. Specifically, the 8-rod pattern shown in Figure 2b is commonly used for
square HSS (Hollow Steel Section) cantilever columns for mezzanines. Testing for points 1
and 2 is summarized in (Gomez, Kanvinde, & Deierlein, Exposed column base connections
subjected to axial compression and flexure, 2010), and (Kanvinde, Higgins, Cooke, Perez, &
Higgins, 2014).
3. Exposed base plate connections subjected to axial tension, compression and shear; see
(Gomez, Kanvinde, Smith, & Deierlein, 2009).
In addition, Finite Element simulations were conducted to examine conditions not directly
evaluated by the experiments; these included base plates of different thicknesses, as well as base
connections subjected to biaxial bending and axial compression (see Figure 3).
(a) (b) (c)
Figure 2 – Alternate anchor rod patterns: (a) 4-rod pattern addressed by Design Guide One (b, c) 8-rod patterns
Figure 3 – Finite element model of base plate subjected to biaxial bending, indicating interaction of loading
directions and high stress near corner
(a) (b) (c)
Figure 4 – Shear transfer mechanisms (a) Friction (b) Anchor rod bearing and (c) Shear key bearing
The primary findings pertaining to the strength of exposed base plate connections are
summarized below:
1. Design Guide One is generally appropriate for characterizing the strength of exposed base
plate connections with a 4-rod pattern (or more specifically two rows of rods in the direction
of bending). Measurements indicate that the stress blocks assumed in the Drake and Elkin
method appropriately represent the true stresses. However, it is noted that a check of base
plate yielding on the compression side (see limit states discussed earlier) is not necessary,
since it is a highly ductile limit state that does not affect the strength of the connection.
Consequently, disregarding this design check reduces conservatism of the method.
2. Design Guide One is ill-equipped to characterize the strength of base plates with the 8-rod
pattern. Specifically, it cannot address the static indeterminacy that arises due to the presence
of multiple rows of anchor rods in tension. For some cases, such as the one shown in Figure
2c it is conservative, because it does not count the reserve strength provided by the additional
anchor rods. However, in some cases, such as the one shown in Figure 2b where three anchor
rods are present in a row, it may be unconservative, because it cannot characterize the
unequal distribution of tensile forces among the anchor rods within a row (the central rod
carries a disproportionately high fraction of the load). The work by Kanvinde et al., (2014),
Gomez et al., (2010) suggests methods to address these (Figure 2b and Figure 2c) situations.
3. Three shear transfer mechanisms for exposed base plates were examined; these include (1)
friction between base plate and grout, which is available when there is net compression, (2)
anchor rod bending, and (3) an embedded shear key. Figure 4 shown illustrates these
schematically. It was determined that currently used approaches (as outlined in Design Guide
One) are appropriate to characterize strength for the first two mechanisms, but may be quite
unconservative for the shear-key bearing mechanism, since is disregards the size-effect in
concrete. This may be addressed through other approaches, e.g., the Concrete Capacity
Design method (Fuchs et al., 1995) which explicitly incorporates this effect.
4. The finite element simulations (similar to the one shown in Figure 3) confirm the above points,
but additionally indicate that exposed column base plates are subject to an interaction in
bending moments acting in the major and minor axes of the column, since the anchor rods
carry tension corresponding to both these moments. Currently, no approaches are available
for characterizing this interaction.
Embedded Column Base Connections: Current design guidance for Embedded Column Base
connections is limited to the AISC Seismic Design Manual, which adapts test results for similar
connections (e.g., steel link beams embedded in concrete shear walls) to develop strength
equations for these connections. The professional practice has relied on these and similar (e.g.,
approaches using design equations for steel-concrete composite moment connections) for design
of such connections. However, embedded column base connections are quite different as
compared to any of these, for various reasons. These include the presence of axial force in the
steel column as well as a high degree of confinement for the concrete in the bearing region ahead
of the column flange. Grilli et al., (2017 – in press) tested 5 full-scale embedded base connections,
in which the lowest portion column was embedded into a footing. These connections were
subjected to combinations of axial compression, tension, flexure and shear. The tests reveal that
the base moment and shear is resisted through a combination of three mechanisms, which are
illustrated schematically in Figure 5: (1) bearing stresses between the column flanges and the
concrete footing – Figure 5a, (2) the development of a concrete panel, which along with the
column web resists the panel zone shear – Figure 5a, and (3) resistance to uplift of the embedded
base plate – Figure 5b. The former two mechanisms act in series with each other (i.e., carry forces
that are in equilibrium with each other), and together in parallel with the third (uplift mechanism).
Grilli and Kanvinde (2017) present a strength characterization method for these connections; the
method has two steps. First, the applied moment must be decomposed into the moment carried
by mechanisms (1) and (2) above, and the moment carried by mechanism (3). Second, limit states
associated with each mechanism must be identified, and the strengths corresponding to each
must be aggregated into the connections strength. This process is not unlike that for composite
steel beam-concrete column moment connections, as described in the design guide for composite
connections (ASCE, 1994).
Column shear transfer in embedded base connections is relatively facile, and unlikely to result in
a limit state by itself. However, the column shear is included into the calculation of the bearing
stresses carrying the moment.
Barnwell (2015) conducted experiments on shallowly embedded column connections, in which
the embedment of the base plate is incidental (rather than for strength), and arises from an
overtopping concrete slab – see Figure 1c. In these, it was determined that the overtopping slab
results in an increase in moment capacity, albeit the dominant resistance mode is still resistance
to overturning of the base plate.
Figure 5 – Force transfer mechanisms in embedded base connections: (a) bearing on column flanges and panel
zone shear (b) restraint to base plate uplift and skin friction
Figure 6 shows representative plots of moment-rotation curves for the exposed and embedded
base connections. Two points may be noted from the figures. First, the connections show
significant dissipative capacity, along with large ductility, with rotations in the range of 3-10% for
almost all the specimens tested. Second, the specimens show significant rotational flexibility –
this is uniformly observed even when these connections are designed to be stronger than the
attached column or provided with an embedment to achieve base fixity. The implications of both
these observations for design are discussed in upcoming sections.
220
(e) Test #6
Exposed base plate Embedded base
Base Moment (kN-m)
-220
-10 0 10
Column Drift (%)
Figure 6 – Representative load-deformation graphs from exposed and embedded base connections, showing
deformation capacity and dissipation
FLEXIBILITY OF BASE CONNECTIONS
Referring to the previous section, base connections (whether embedded or exposed) show
significant rotational flexibility, such that elastic rotations of the base connection may approach
0.01 radians (which is on the order of yield drift in moment frames). However, these connections
are usually modeled as fixed or pinned in structural simulation; this has the potential of
mischaracterizing structural response. To address this, recent research by authors as well as
others (Tryon, 2016; Trautner et al., 2015) has focused on the development of methods to
determine the flexibility of base connections. Some of these approaches are briefly outlined here:
1. Kanvinde et al., (2012) proposed a method for characterizing the rotational flexibility of
exposed base connections. This approach decomposes the deformations in the base plate
into those from the concrete, anchor rod, and base plates (under a given axial force and
moment), and then enforces compatibility to determine connection rotation. Since the method
leverages the Design Guide One approach for strength characterization to determine stresses
in concrete, anchor rods, and base plate, it requires only a modest amount of effort to compute
base flexibility once the base plate has been designed. Figure 7 below shows the assumed
deformations in this method.
2. Torres Rodas et al., (2017 – in press) present a method for characterizing the flexibility of
embedded column base connections. The approach reflects physical insights developed
during the tests on Embedded Column Bases by Grilli et al., (2017 – in press). The method
considers deformations of several components within the connection, including bending and
shear deformations of the embedded column, in addition to bearing deformations of the
concrete adjoining the embedment. To resolve the internal indeterminacies of stress
distribution within the connection, the method relies on stress distributions established for the
strength model (Grilli and Kanvinde, 2017) discussed earlier. As a result, in a manner similar
to the flexibility method for the exposed base connections, it retains a degree of simplicity and
commonality with the strength model. Tryon (2016) presents a similar method for shallowly
embedded base connections.
Figure 7 – Deformation contributions to exposed base plate connections
The models above may be used for simulating base connections in buildings, to estimate
forces/moments for design, or they may be used for parametrically examining the effect of base
flexibility on generic building response to inform general design approaches and philosophies.
The next section discusses the implications of base connection response (as observed in this and
the previous section) on the seismic response of moment frame structures.
IMPLICATIONS FOR BUILDING RESPONSE AND DESIGN
Zareian and Kanvinde (2013) conducted a parametric study to examine the effect of column base
flexibility on the seismic response of steel moment frames. To this end, the response of four
frames (2, 4, 8 and 12-story) designed as per current code was investigated through static
pushover simulations and sophisticated nonlinear response history simulations including collapse
simulation. For each frame, a range of base fixities was examined, including realistic values that
were determined for the as-designed base connections using the models summarized in the
previous section. The results indicate that a reduction in base stiffness (with respect to the fixed
base assumption, which is commonly used for mid- to high-rise frames, if the base is designed to
be stronger than the column), lowers the point of inflection in the first story columns, greatly
increasing moment demands at the top of the column. The results in concentration of
deformations in the first story (i.e., a soft-story mechanism), undermining the performance in three
ways: (1) significant increase in deformation demands, (2) decrease in system ductility, and (3)
significant reduction in collapse-capacity, i.e., the median ground motion intensity associated with
sidesway collapse. For the 4-, 8- and 12-story frames, this suggests that the expected response
of frames is significantly worse than is implied by simulations and design approaches that assume
a fixed base condition.
From a design perspective, the primary implication is that for mid- and high-rise buildings, the
characterization of base stiffness is critical, and simulating the bases as fixed results in estimates
of response that are significantly unconservative. On the other hand, for low-rise frames, the
bases are generally modeled as pinned. This assumption is conservative since it discounts the
base stiffness, moving the inflection point to the bottom and greatly increasing the column moment
demands (with respect to the true condition). This may result in oversized first story columns.
Referring to prior discussion, almost all base connections tested in literature have significant
rotation capacity (drifts in the range of 3-10%), even if not specifically detailed to be a ductile
connection. Even if it is argued that the rotation capacity may be artificially high in well-constructed
lab tests, it still is significant, especially when compared to rotation capacity requirements for other
“ductile” connections such as beam column connections. Despite the available rotation capacity
in base connections, they are often designed to remain elastic and force the plastic hinge into the
lower end of the column (SEAOC, 2012).This means that the bases are designed for moments
on the order of 1.1RyMp of the connected column. There are two significant issues with this:
1. Requiring the bases to remain elastic (i.e., designing for high moment) entails significant cost
in terms of detailing (additional anchor rods, deeper embedment, thicker plates, more
reinforcement). This may be unnecessary, given the potentially high deformation capacity of the
base connections.
2. Tests on columns and W-sections show that plastic hinges in columns actually may have lower
deformation capacities (~2% - Lignos and Krawinkler, 2013) even compared to bases, owing to
phenomena such as local or lateral torsion buckling.
In summary, the current design philosophy for column bases may be backwards, i.e., protecting
the potentially ductile base connection (often at high expense) while forcing yielding into a
potentially less ductile component (the column). As outlined in this article, there is significant
established knowledge regarding the component response of base connections, which have
resulted in the fortuitous finding about the ductility. This knowledge may be used to develop a
more rational design philosophy for frames that fully leverages this ductility (and dissipation) to
result in economical and safer design.
SUMMARY AND RESEARCH NEEDS
This paper summarizes recent studies on the response of column base connections and its effect
of structural performance. These studies have been focused on understanding strength limit
states (under a variety of loadings), and then developing design considerations, i.e., strength
models and detailing guidelines for these connections. A secondary outcome of these studies has
been the understanding of the flexibility and hysteretic/ductility characteristics of these
connections to inform the design of frames that use these connections. The research has focused
on three different types of details:
1. Exposed base plate connections, in which a base plate is anchored to a concrete footing using
anchor rods. These are the primary focus of Steel Design Guide One, and are feasible for
low-rise buildings.
2. Deeply embedded base connections, in which the column is embedded into the concrete
footing to achieve moment resistance. These are preferred for mid-high rise frames, and no
design guidance is available for them.
3. Shallowly embedded base connections, which arise when exposed base connections are
overlaid by a concrete slab on grade. These are common in low-rise frames for lateral systems
and in high rise frames for gravity systems, providing lateral resistance in each case.
The state of research on each of these is at a different stage in terms of its maturity,
documentation in design guides and documents, simulation methods, and adoption into design
and construction practice. Despite this unevenness within the research on base connections, the
main observation is that research on column bases (when considered as a whole) is significantly
less mature than that on other connections such as welded beam column connections (due to the
focus on these after the Northridge earthquake). Future research on these connections has the
potential to significantly enhance the economy, performance, and safety of steel framed
structures. Some particularly important research needs are outlined below:
1. Characterization of seismic performance, and the development of strength models for bae
connections subjected to biaxial bending.
2. The development of dissipative and replaceable/repairable base connection technologies to
enhance structural performance, and methodologies to design structures that take advantage
of these technologies.
3. To enable point 2 above, the development of hysteresis models that can simulate the
nonlinear/dissipative characteristics of these connections
4. Examination of configurations (especially of embedded base connections) that have not been
tested, noting that only 5 deeply embedded base connections (reflecting US practice) have
been tested so far.
ACKNOWLEDGMENTS
The author thanks the numerous organizations and individuals who contributed to this work. Major
sponsors include: The American Institute of Steel Construction, Charles Pankow Foundation,
National Science Foundation via the NEES Berkeley Equipment site. Material and fabrication
donations were obtained from Herrick Steel of Stockton, CA; Gayle Manufacturing of Woodland,
CA; and Transbay Steel of Napa, CA. The author also thanks the support staff at the NEES
Berkeley equipment site, and the UC Davis laboratory. Students working on these projects include
Ivan Gomez, David Grilli, Ryan Cooke, and Santos Jordan. Collaborators include Greg Deierlein
of Stanford University, Paul Richards of Brigham Young University, Rick Drake of Fluor
Corporation, Jim Malley of Degenkolb Engineers, Kevin Moore and Paul Cordova, both of
Simpson Gumpertz and Heger, and Geoff Bomba of Forell Elsesser Engineers.
REFERENCES
AISC. (2010). Seismic Design Manual. Chicago, IL: American Institute of Steel Construction.
American Society of Civil Engineers (1994), “Guidelines for Design of Joints Between Steel
Beams and Reinforced Concrete Columns,” Journal of Structural Division, ASCE, 1994,
120(8), 2330-2357.
Astaneh, A., & Bergsma, G. (1999). Cyclic Behavior and Seismic Design of Steel Base Plates.
ASCE Structures Congress (pp. 409-414). Irvine, CA: American Society of Civil Engineers.
Barnwell, N. (2015). “Experimental Testing of Shallow Embedded Connections Between Steel
Columns and Concrete Footings,” Masters Thesis, Brigham Young University.
Burda, J. J., & Itani, A. M. (1999). Studies of seismic behavior of steel base plates. Report No
CCEER 99-7, Center of Civil Engineering Earthquake Research. Reno, NV: University of
Nevada, Reno.
Drake, R., & Elkin, S. (1999). Beam Column Base Plate Design. Engineering Journal, 36(1) 29-
38.
Ermopoulos, J., & Stamatopoulos, G. (1996). Mathematical Modeling of Column Base Plate
Connections. Journal of Constructional Steel Research, 36(2) 79-100.
Fuchs, W., Eligehausen, R. and Breen, J.E. (1995), “Concrete Capacity Design (CCD) Approach
for Fastening to Concrete,” ACI Structural Journal, Vol. 92, No. 1, pp. 73–94.
Gomez, I. R., Kanvinde, A. M., & Deierlein, G. G. (2010). Exposed column base connections
subjected to axial compression and flexure. Chicago, IL: American Institute of Steel
Construction.
Gomez, I. R., Kanvinde, A. M., Smith, C. M., & Deierlein, G. G. (2009). Shear Transfer in Exposed
Column Base Plates. Chicago, IL: American Institute of Steel Construction.
Grauvilardell, J. E., Lee, D., Hajjar, J. F., & Dexter, R. J. (2005). Synthesis of Design, Testing,
and Analysis Research on Steel Column Base Plate Connections in High-Seismic Zones.
Minneapolis, MN: University of Minnesota.
Grilli, D.A., Jones, R., and Kanvinde, A.M., (2017 - in press). "Seismic Performance of Embedded
Column Base Connections Subjected to Axial and Lateral Loads," Journal of Structural
Engineering, ASCE.
Grilli, D.A., and Kanvinde, A.M., (2017). "Embedded column base connections subjected to
seismic loads: Strength model," Journal of Constructional Steel Research, Elsevier, 129,
240-249.
Kanvinde, A.M., Grilli, D.A., and Zareian, F. (2012). “Rotational Stiffness of Exposed Column Base
Connections – Experiments and Analytical Models,” Journal of Structural Engineering,
ASCE, 138(5), 549-560.
Kanvinde, A. M., Higgins, P., Cooke, R. J., Perez, J., & Higgins, J. (2014). Column base
connections for hollow steel sections: seismic performance and strength models. Journal
of Structural Engineering, ASCE.
Kloiber, L., & Fisher, J. (2006). Steel Design Guide One, Base Plate and Anchor Rod Design (2nd
Edition). Chicago, IL: American Institute of Steel Construction.
Lignos, D.G., Krawinkler, H. (2013). “Development and Utilization of Structural Component
Databases for Performance-Based Earthquake Engineering”, ASCE, Journal of Structural
Engineering, Vol. 139 (NEES 2), 1382-1394.
Pertold, J., Xiao, R. Y., & Wald, F. (2000). Embedded Steel Column Bases – I. Experiments and
Numerical Simulation. Journal of Constructional Steel Research, 56(3) 253-270.
SEAOC. (2012). Special Moment Frame Base Connection Design Example 8. 2012 IBC SEAOC
Structural/Seismic Design Manual Vol 4, Examples for Steel-Frame Buildings; 255-280.
Sacramento, CA: Structural Engineers Association of California.
Torres Rodas, P., Zareian, F., and Kanvinde, A.M., (2017 – in press). “Rotational stiffness of
deeply embedded column base connections,” Journal of Structural Engineering, American
Society of Civil Engineers.
Trautner, C., Hutchinson, T., Grosser, P., and Silva, J. (2015). "Effects of Detailing on the Cyclic
Behavior of Steel Baseplate Connections Designed to Promote Anchor Yielding." Journal
of Structural Engineering, 10.1061/(ASCE)ST.1943-541X.0001361, 04015117.
Tryon, J.E., (2016). “Simple models for estimating the rotational stiffness of steel column to footing
connections,” Masters Thesis, Brigham Young University.
Zareian, F., and Kanvinde, A.M.,(2013). “Effect of Column Base Flexibility on the Seismic Safety
of Steel Moment Resisting Frames,” Earthquake Spectra, Earthquake Engineering
Research Institute, 29(4), 1-23.
MOMENT-ROTATION CHARACTERISTICS FOR DECONSTRUCTABLE
FLUSH END PLATE COMPOSITE JOINTS
Mark A. Bradford
m.bradford@unsw.edu.au
Abdolreza Ataei
a.ataei@unsw.edu.au
Xinpei Liu
xinpei.liu@unsw.edu.au
Centre for Infrastructure Engineering & Safety, School of Civil & Environmental Engineering
UNSW Sydney, NSW 2052, Australia
Abstract
A 3-D finite element modelling of flush end plate semi-rigid beam-to-column composite joints is
described. The joints comprise of deconstructable post-installed friction-grip bolted shear
connectors. The moment-rotation characteristics for these prototype joints subjected to a hogging
bending moment with the influence of partial shear connection is investigated using ABAQUS,
including material and geometric non-linearities. The accuracy and reliability of the numerical
formulation are verified against experimental data, and a parametric study is conducted. Moment-
rotation models for deconstructable flush end plate beam-to-column composite joints are proposed
from these results.
INTRODUCTION
Steel-concrete composite beam-to-column partial strength joints are widely used in medium-sized
steel-framed buildings, because they provide rapid construction solutions and possess excellent
structural performance. The efficiency of a composite joint lies in ensuring composite action
between the concrete slab and the steel beam. This composite action between these two
components is normally achieved by using headed stud shear connectors welded to the top flange
of the steel beam and embedded into a cast in-situ concrete slab. However, the deconstruction of
this system and reuse of its components is almost impossible, and this produces environmentally-
intrusive waste and CO2 emissions.
One method of providing composite action between the concrete slab and steel beam to improve
the structural behaviour of composite joints is to connect these two components with post-installed
tensioned friction-grip bolted shear connectors (PFGBSCs). The innovative friction and bearing
mechanisms for this connection have been shown to enable the composite beams and joints to be
deconstructed easily at the end of their service life. In addition, the use of precast a concrete slab
instead of a cast in-situ concrete slab can decrease the construction time and improve the quality
of composite construction significantly. Therefore, using PFGBSCs in lieu of welded stud shear
connectors improves the sustainability of composite construction significantly.
Surprisingly little research has been reported on beams and joints with PFGBSCs, especially from
the perspective of deconstruction, and previous work has been reviewed by Ataei (2015). Whilst
necessary for validation purposes, physical experimental studies of composite connections are very
expensive and time-consuming, and accordingly, considerable research momentum is gathering on
using finite element (FE) simulations. A numerical model to simulate the behaviour of a composite
beam with PFGBSCs was developed by Kwon, Engelhardt and Klinger (2012) using a FE
technique. Liu, Bradford and Lee (2014) developed a 3-D FE model to study the ultimate strength
and load-slip characteristics of PFGBSCs in beams with geopolymer concrete slabs, while Liu,
Bradford, Chen and Ban (2016) proposed a 3-D FE model to investigate the behaviour of composite
beams with PFGBSCs.
This paper presents a 3-D FE model using ABAQUS software to investigate the structural behaviour
of semi-rigid beam-to-column composite joints with deconstructable PFGBSCs, precast concrete
slabs and a strengthened I-section steel column. The detailed FE model developed includes all the
pertinent connection components, in which material and geometrical non-linearities as well as the
non-linearity of the interface contacts are incorporated. The accuracy and the reliability of the model
are verified against experimental data reported elsewhere as well as test results obtained from full-
scale laboratory tests conducted in this study. An extensive parametric evaluation is conducted,
and existing prediction and design models are extended to elucidate the significant parameters for
a composite joint with PFGBSCs and a precast concrete slab, including the initial rotational stiffness,
rotation capacity and moment capacity.
FINITE ELEMENT MODELLING

Contact Modelling
There are several different components of the deconstructable beam-to-column composite joints:
the steel beam, steel column, bolts in connection zone, bolt shear connectors, concrete reinforcing
bars, flush end plate, steel beam and column stiffeners and the interaction between them should
be taken into consideration in the modelling. The contact between these components was
represented using the surface-to-surface contact interaction technique as shown in Fig. 1. One
surface was taken as a master surface and the other a slave. Normal and parallel to the interface
plane, HARD and PENALTY options were used respectively, with a friction coefficient of 0.45 (Ataei,
Bradford & Liu 2016) for the interface between the steel beam and precast concrete slab and a
friction coefficient of 0.25 for the other interactions. To be consistent with the experimental set-up,
the holes in the steel column and flush end plate were modelled with a 1 mm clearance and could
slip during loading. The TIE option was used to define the contact between the flush end plate-
steel beam, while the EMBEDDED option was used to embed the reinforcement into the slab.
Finally, non-linear springs were used to simulate the PFGBSCs, in which the results from push-out
tests on M16 and M20 PFGBSCs (were used to characterise the load-slip behaviour of the bolt
shear connectors and to represent their behaviour in the composite joints. The same location of the
shear connectors was considered for the position of the springs as shown in Fig. 3.
Meshing, boundary conditions and restraints
A 3-D element was employed to simulate all components of the composite joint, in which truss
elements (T3D2) with a linear approximation of the displacement and eight-node linear hexahedral
solid elements with reduced integration and hourglass control (C3D8R) were used for modelling of
the steel bars and the other components, respectively. The mesh used for each component of the
composite joint was chosen based on an extensive sensitivity analysis conducted beforehand. Fig.
3 shows the mesh for the components in the joint.
Due to the symmetrical geometrical properties and applied loading, only half of each specimen
about an axis of symmetry was modelled and the associated symmetric boundary conditions were
considered on the plane of symmetry, as illustrated in Fig. 4. All nodes along the middle of the
precast concrete slab, steel column web, column stiffeners and longitudinal reinforcing bars were
restrained from moving in the X direction and against rotation in the Y and Z directions. As the
bottom of the steel column was fixed and restrained from displacing and rotating, all nodes at the
bottom surface of the steel column were so restrained in the X, Y and Z directions.
column bolts end plate bolts
column-end plate beam-end plate

Fig. 1 – Contact surfaces between joint components
Fig. 2 – Simulation of PFGBSCs
beam half column half slab
end plate bolt

Fig. 3 – FE mesh for joint components
Fig. 4 – Half composite joint and boundary conditions
Material constitutive relationships
The relationship proposed by Carreira and Chu (1985) which has also been used in related studies
was employed to simulate the behaviour of the concrete in compression. For concrete in tension,
a linear–elastic representation followed by a linear softening branch was adopted. The tensile stress
was assumed to increase linearly to 10% of its compressive strength. After cracking in the concrete,
the stress declines to zero at strain of about 10 times its failure strain. The plasticity damage
concrete feature available in ABAQUS was employed for this study.
To simulate the properties of all steel components used in the FE modelling, a multi-linear elastic-
plastic curve was adapted. The values of the key parameters such as the yield strength, modulus
of elasticity, ultimate strength and ultimate strain were determined and calibrated based on the
laboratory material tests conducted in the experimental study (Ataei, 2015; Ataei, et al., 2016).
COMPARISON WITH EXPERIMENTAL RESULTS
Ataei et al. (2016) conducted three full-scale beam-to-column joint tests (CJ1 and CJ2 being
composite; SJ3 being plain steel) designed and constructed in a cruciform arrangement to simulate
the internal joint in a semi-rigid frame. All specimens consisted of a 460UB82.1 beam and a
250UC89.2 column. The reinforcement ratio in the precast concrete slab of the both composite
joints was 0.9%. A total of 12 post-installed bolted shear connectors of Grade 8.8 M20 bolts (6 bolts
per shear span) were used in pairs to connect the precast concrete slab to the steel beams. A 12
mm thick flush end plate and four Grade 8.8 M24 bolts were used to connect the composite beam
to the steel column. Both composite joints and were designed with full shear connection.
Subsequently, to investigate the influence of the partial shear connection on the behaviour of the
composite joints, specimen CJ4 was designed and constructed by the authors in this study with
partial shear interaction, in which a total of 8 post-installed bolted shear connectors of Grade 8.8
M16 bolts (4 bolts per shear span) were used in pairs. A 10 mm thick flush end plate was welded
to the ends of the steel girders and four M24 Grade 10.9 bolts were used for connecting the steel
girders to the columns. The purpose for this connection arrangement was to alter the failure mode
from fracture of the bolts in the tension zone of the connection to fracture of the longitudinal
reinforcement bars. Figs. 5 and 6 show the arrangement of the experimental setup and the failure
mode of specimen CJ4. The moment versus rotation responses predicted by the FE models
compared with the experimental results are shown in Fig. 7, and Table 1 compares the initial
stiffness, rotation capacity, moment capacity and the final slip at the end of the beam obtained from
both the FE modelling and the tests. Generally, good agreement between the experimental results
and the FE predictions at the global level is evident.
PARAMETRIC STUDY
Shear connection ratios (SCRs) of 34%, 54%, 67%, 101%, 108%, 162%, 170% and 162% were
considered to investigate their effect at the interface between the precast concrete slab and the
steel beam. The ratio of the shear connection between the steel beam and precast concrete slab
was obtained by changing the number and size of shear connectors and their spacing along the
composite beam length.
Fig. 5 – Experimental setup and instrumentation for CJ4
Fig. 6 – Failure mode of specimen CJ4

700
600
500
Moment (kNm)
400
300
200 Test- CJ1
100 FE model
0
0 10 20 30 40 50 60
Rotation (mrad)
300 600
250 500
400
Moment (kNm)
Moment (kNm)
200
150 300
100 200 Test- CJ4

Test- SJ3
50 100
FE model FE model
0 0
0 20 40 60 0 20 40 60 80
Rotation (mrad)
Rotation (mrad)
Fig. 7 – Comparison of tests and numerical model
The comparison of the moment-rotation relationships for different SCRs are shown in Table 2 and
Fig. 8. As the SCR increases above a certain level (54%), the initial stiffness increases significantly,
but, the increase in initial stiffness is accompanied with a significant reduction in the rotation
capacity. A composite joint using a shear connection ratio of 34% shows limited ductility and
strength, due to fracture of the bolted shear connector. In this case, the plastic strain may not be
mobilised in the longitudinal reinforcing bars and the shear connector may fracture before the
reinforcing bars yield. It can be concluded that a minimum level of shear connection ratio (about
50%) should be provided to prevent fracture of the bolted shear connectors.
Table 1 – Comparison of test results and FE predictions
600
500 2M16
400 2M20
Moment (kNm)
4M16
300 4M20
6M16
200
6M20
100 10M16
10M20
0
0 20 40 60 80
Rotation (mrad)
Fig. 8 – Effects of degree of shear connection on moment-rotation response
Table 2 – Effects of degree of shear connection
Further parametric studies were conducted on the effects of shear connector spacing, slab
reinforcement ratio, bolt sizes, end plate thickness and grade, column steel grade, beam steel
grade, concrete slab thickness and column flange thickness and the results are presented in Tables
3 to 11 respectively.
Table 3 – Effect of shear connector spacing
Table 4 – Effect of slab reinforcement ratio
Table 5 – Effects of bolt size
Table 6 – Effect of end plate size
Table 7 – Effect of steel grade of end plate
Table 8 – Effect of steel grade of column
Table 9 – Effect of steel grade of beam
Table 10 – Effect of slab thickness
Table 11 – Effect of column flange thickness
DESIGN MODEL
A design model was proposed by Ahmad and Nethercot (1997) for determining the initial stiffness
of a composite joint, which for a stiffened column web reduces to
1  1 1 
hbschrb   + hb2  + 
(1)
 kb   krb kbsc 
Sj = ,
 1 1  1 
 +  
 krb kbsc  kb 
where kb = stiffness of top row of bolts, krb = stiffness of reinforcing bars, kbsc = stiffness of shear
connectors, with other parameters being defined in Fig. 9. The parametric study was used to
evaluate this design model, with 53 analyses producing an average ratio of Eq. (1) to the FE results
of 0.93 and a standard deviation of 0.38.
Anderson and Najafi (1994) proposed an expression for the moment capacity of a joint as
(2) M j = Frb (hrb − 0.5t fb ) + Fb (hb − 0.5t fb )
,
if Frb +Fb ≤ Ffb, and from
(3) M j = Frb (hrb − 0.5t fb ) + Fb ( hb − 0.5t fb ) − ( Frb + Fb − F fb )(0.5 yc + 0.5t fb )
if Frb +Fb > Ffb, in which the dimensions and actions are shown in Fig. 9. In addition, yc = (Frb + Fb –
Ffb)/(tfbfyb) denotes the depth of the web in compression. The parametric study was again used to
evaluate the predictions of this design mode, with 53 analyses producing an average ratio of the
model to FE results of 1.00 and a standard deviation of 0.13.
CONCLUSIONS
A 3-D FE model was developed using ABAQUS to simulate the structural performance of beam-
to-column composite joints with deconstructable PFGBSCs and precast concrete slabs. The FE
models were verified against experimental data reported elsewhere. In addition, a full-scale
laboratory test specimen having deconstructable flush end plate composite semi-rigid joint and
precast concrete slab was tested to investigate the influence of the partial shear connection on the
behaviour of the composite joints. The FE results were also compared with the test results obtained
from this test. An extensive parametric study was then conducted using the validated FE modelling.
Finally, some existing prediction and design models were extended to predict the fundamental
parameters for a composite joint with PFGBSCs and a precast concrete slab including the initial
rotational stiffness, rotation capacity and the moment capacity. The FE modelling was shown by
comparison with test results to be accurate. A parametric appraisal of design models for the initial
stiffness and moment capacity of two design procedures was carried out using the FE model, and
these were shown to be relatively accurate for the deconstructable joints proposed.
Fig. 9 – Actions and dimensions for design model
REFERENCES
Anderson, D., & Najafi, A. A. (1994). Performance of composite connections: major axis end plate
joints. Journal of Constructional Steel Research 31:31–57.
Ataei, A. (2016). A low-carbon deconstructable steel-concrete composite framed system with
recyclable beam and slab components. PhD Thesis. School of Civil and Environmental
Engineering, The University of New South Wales, Sydney, Australia.
Ataei, A., Bradford, M. A., & Liu, X. (2016). Experimental study of flush end plate beam-to-column
composite joints with precast slabs and deconstructable bolted shear connectors. Structures
7, 43-58.
Carreira, D. J., & Chu, K. H. (1985). Stress-strain relationship for plain concrete in compression.
ACI Structural Journal 82, 797-804.
Kwon, G., Engelhardt, M. D., & Klinger, R. E. (2012). Parametric studies and preliminary design
recommendations on the use of postinstalled shear connectors for strengthening
noncomposite steel bridges. Journal of Bridge Engineering, ASCE 17(2), 310-317.
Lee, S. S. M., & Bradford, M. A. (2013). Sustainable composite beam behaviour with deconstructable
bolted shear connectors. Composite Construction VII, Palm Cove, Queensland, Australia.
Liu, X., Bradford, M. A., & Lee, M. S. S. (2014). Behavior of high-strength friction-grip bolted shear
connectors in sustainable composite beams. Journal of Structural Engineering, ASCE.
http://dx.doi.org/10.1061/(ASCE)ST.1943-541X.0001090.
Liu, X., Bradford, M. A., Chen, Q.J., & Ban, H. (2016). Finite element modelling of steel–concrete
composite beams with high- strength friction-grip bolt shear connectors. Finite Elements in
Analysis and Design108, 54-65.
THE INFLUENCE OF A COMPOSITE SLAB ON THE SEISMIC BEHAVIOR OF MOMENT RESISTING
FRAMES
Roberto Tartaglia1,
Naples, Italy
roberto.tartaglia@unina.it
Mario D’Aniello2,
Naples, Italy
mdaniel@unina.it
Gian Andrea Rassati3,

gian.rassati@uc.edu
James Swanson4,
james.swanson@uc.edu
ABSTRACT
Moment resisting frames (MRFs) are widely used for steel structures in seismic areas. However,
MRFs are characterized by large lateral deformability that can impair the structural performance.
In the case of all steel MRFs, the influence of composite slabs is generally disregarded when
structural analyses are carried out at the design stage. However, it can be convenient to take into
account their stiffening effect on the steel beams of the MRFs, even though the beam-to-column
assemblies are designed with disconnected slabs. Therefore, in this paper the influence of
composite slabs on the seismic response of MRFs has been investigated by means non-linear
analyses.
Keywords: Moment resisting frame, composite, finite element analysis, beam-to-column joints,
extended end-plate connection.
INTRODUCTION
Moment resisting frames (MRFs) are largely used for steel multi-story low and medium rise
buildings in seismic areas. However, this structural system can be characterized by large lateral
deformability, thus being highly sensitive to second-order P-Delta effects. When the slab is
disconnected from the joint zone, MRFs are designed and calculated as all-steel, namely
disregarding the stiffening role of the composite slab, which can be beneficial to reduce the lateral
deformability. In addition, even in the case of dual systems where lateral forces are shared
between a moment frame and a braced frame, the presence of the slab can have a beneficial
effect on the global structural behavior. Finally, the presence of the composite slab can positively
influence the joint behavior (Aribert 2006, Leon 1998 and Thermou 2004).
Moreover, a recent numerical study (Tartaglia et al. 2018a) on composite joints demonstrates that
even though a protected zone around the column is maintained, there is a beneficial influence of
the slab to the joint behavior.
The main object of this work is investigating by means of finite element analysis the influence and
the advantages coming from considering the presence of the concrete slab in the MRF structural
behavior even if there is no direct connection with the steel joints. To this end, in the FE model,
developed in Seismostruct (SeismoStruct 2016), the real beam to column joint behavior was
introduced by properly calibrated nonlinear links. The link behavior was calibrated on the
experimental tests for the all steel joint (AS) while the calibration of the composite joint was done
with respect to previous Abaqus (Dassault 2014) simulation results (Tartaglia et al. 2018a).
The paper is organized into three main parts: the definition of the designed structures and their
geometrical characteristics, the introduction and the calibration of the nonlinear link in the
Seismostruct model and the presentation of both the results of the static and the dynamic
nonlinear analysis.
DESIGNED STRUCTURES
Eight MRFs were designed according to both EN1993-1-1 and EN1998-1-1 to investigate the
influence of the concrete slab on the structural response. All structures have six floors with a story
height of 3.5m, with the exception of the first floor, which is 4.5m. The same square plan with
three bays for both directions and with two MRF systems along the X and four concentrically
braced systems in the Y direction were considered (see Fig. 1).
The structures were alternatively designed considering two beam span lengths (6 and 8 meters)
and two peak ground accelerations (0.35g and 0.45g). Furthermore, two structural typologies
were investigated: an all steel solution (AS) and a partially composite solution (CS). In the former
case (AS), the presence of the slab was completely neglected at design stage and only the steel
structure was considered in the calculation and verification of the frames. In the second case
(CS), the presence of the collaborating slab was accounted for even if the joints were still
considered as all-steel, namely the hierarchy between the beam and the column was evaluated
neglecting the contribution of the slab. To be consistent with this assumption, the portion of the
ends of the beam at a distance equal to 2beff (effective length as defined in EN1994-1-1) from the
column face was considered without shear stud. Therefore, the stiffening effect of the slab was
accounted for only in the inner segments of beams were fully composite action was guaranteed
by the presence of shear studs (see Fig. 2).
Ly
Ly
3.5m
Ly
Lx Lx Lx
Lx Tot
Fig. 1: The reference structure and the joint components descriptions.
Beam
AS 2beff 2beff
AS CS AS
Column
Column
CS
Figure 2: Cross sections considered in the design phase.
The cross sections of the designed frames are summarized in Table 1, where only the profiles of
the first two bays are reported due to the structural symmetry.
For all the investigated cases, the most restrictive limitation was the deformability at the
serviceability limit state. Indeed, the maximum interstorey drift ratio permitted by EN1998-1-1 is
equal to 1% for buildings having nonstructural fixed elements or that cannot interfere with the
structural deformation. This limitation significantly influenced the response of the frames
designed, as also shown in Tartaglia et al. (2018b).
The presence of the slab, even if it is not collaborating with the steel elements in the protected
zone (EN1994-1-1), increases the structure’s lateral stiffness. Indeed, as shown in Table 1,
accounting for the contribution of the slab allows the designers reducing the size of steel profiles.
For instance, 6-5-8-0.45 structures (the buildings with 8m of span and designed for 0.45g) show
a great reduction of the dimensions of the columns along the entire height of the structure with
the only exception of the external column (first bay) at fifth and sixth storey.
Table 1: Structural elements cross section
First bay Second bay
Storey
6-5-6-0.35
Column Beam Column Beam
All Steel Composite All Steel Composite All Steel Composite All Steel Composite
I HE500M HE400M IPE550 IPE550 HE550M HE450M IPE550 IPE550
II HE500M HE400M IPE550 IPE550 HE550M HE450M IPE550 IPE550
III HE450B HE360B IPE500 IPE500 HE500M HE400M IPE500 IPE500
IV HE450B HE360B IPE500 IPE500 HE500M HE400M IPE500 IPE500
V HE360B HE340B IPE400 IPE400 HE450B HE360B IPE400 IPE400
VI HE360B HE340B IPE400 IPE400 HE450B HE360B IPE400 IPE400
6-5-6-0.45
I HE500M HE500M IPE600 IPE600 HE600M HE600M IPE600 IPE600
II HE500M HE500M IPE600 IPE600 HE600M HE600M IPE600 IPE600
III HE500B HE500B IPE550 IPE550 HE500M HE500M IPE550 IPE550
IV HE500B HE500B IPE550 IPE550 HE500M HE500M IPE550 IPE550
6-5-8-0.35
I HD400X393 HE550M IPE750x173 IPE750x173 HD400X422 HD400X393 IPE750x137 IPE750x137
II HD400X393 HE550M IPE750x173 IPE750x173 HD400X422 HD400X393 IPE750x173 IPE750x173
III HE550B HE550B IPE600 IPE600 HD400X393 HE550M IPE600 IPE600
IV HE550B HE550B IPE600 IPE600 HD400X393 HE550M IPE600 IPE600
6-5-8-0.45
I HD400X744 HD400X678 IPE750x173 IPE750x173 HD400x818 HD400x744 IPE750x173 IPE750x173
II HD400X744 HD400X678 IPE750x173 IPE750x173 HD400x818 HD400x744 IPE750x173 IPE750x173
III HD400x551 HD400x467 IPE750x147 IPE750x147 HD400X744 HD400X678 IPE750x147 IPE750x147
IV HD400x551 HD400x467 IPE750x147 IPE750x147 HD400X744 HD400X678 IPE750x147 IPE750x147
V HE450M HE450M IPE600 IPE600 HD400x551 HD400x467 IPE600 IPE600
VI HE450M HE450M IPE600 IPE600 HD400x551 HD400x467 IPE600 IPE600
Table 2: Interstorey drift displacement

Storey 6-5-6-0.35 6-5-6-0.45 6-5-8-0.35 6-5-8-0.45
AS CS AS CS AS CS AS CS
I 0.62% 0.68% 0.64% 0.69% 0.82% 0.82% 0.69% 0.72%
II 0.85% 0.85% 0.88% 0.90% 0.96% 0.96% 0.95% 0.95%
III 0.91% 0.92% 0.97% 0.95% 1.00% 0.95% 0.95% 0.97%
IV 0.86% 0.86% 0.93% 0.90% 0.99% 0.94% 0.84% 0.88%
V 0.89% 0.96% 1.00% 0.97% 0.96% 0.94% 0.82% 0.85%
VI 0.71% 0.72% 0.88% 0.84% 0.75% 0.73% 0.62% 0.61%
JOINT PERFORMANCE
Extended stiffened beam-to-column joints were adopted in all the designed structures (see Fig.
3a). Indeed, this type of joints can guarantee an adequate resistance and ductility in seismic areas
allowing the formation of plastic hinges at the beam extremities, while remaining in the elastic
range, even though the beam plastic hinges may experience large overstrength, as shown by
Tartaglia et al. 2018c. All joints were designed according to the recent procedure developed within
the EQUALJOINTS research project.
Two types of joints were considered in the modelling phase: an all steel solution (modelling only
the steel joint) and a composite solution (modelling both the steel joint and the concrete slab (see
Fig. 3 a and b respectively)).
Column
Rib
Rib
Secondary Concrete slab
Beam Secondary
beam
Primary Beam Primiry beam
Continuity Rib Continuit Rib

Plate Supplementary y plate
(CP) Supplementary web plate
web Panel (SWP)
a) b)
Figure 3: Introduced joint typologies and geometry.
Therefore, in order to simulate the realistic behavior of the beam-to-column joint, a nonlinear link
was introduced in the structural model. In the case of all steel solution the link behavior was
calibrated on the results obtained within the EQUALJOINTs research project. In particular, as
show, in Fig 4 a and b, the experimental test on a full strength joint was calibrated using the
Seismostruct software modelling the real joint dimension by rigid elements and concentrating all
the joint nonlinearity in a link. A smooth nonlinear spring element was used to simulate the joint
response. As it can be noted this model can satisfactorily overlap the experimental test result.
In the composite structure, the composite slab gives an appreciable contribution to the joint
behavior. Indeed, as reported by Tartaglia et al. (2018a) the slab influences both the elastic
stiffness and the ultimate joint capacity (by 15% and 20% respectively).
Column Applied
displacement
Link
Rigid
elements
Beam
Column
a)
ES – Full Strength
300
200
Force [kN] 100
0
-100
-200 Experimental
ES1-TS-F-C1
F
Numerical
-300
-400 -200 0 200 400
Displacemets [mm]
b)
ES – Composite - Full Strength
300
200
100
Force [kN]
0
-100
-200 Abaqus
Seismostruct
-300
-400 -200 0 200 400
Displacement [mm]
c)
Figure 4: the Seismostruct model (a) and the link calibration with respect to: (b) the
experimental test, for the all steel joint and (c) the Abaqus prediction, for the composite solution
RESULTS
Nonlinear static analysis
Nonlinear static analyses were performed considering two force distributions: proportional to the
first vibration mode (FM) and uniform (U). Figure 5 shows for all the investigated structures the
pushover curves in terms of base shear versus top displacement. As it can be observed, despite
the structures were designed to have a comparable lateral stiffness (see Table 2), the presence
of the composite slab increases the elastic stiffness. On the other hand, some differences can be
noted in terms of maximum capacity. Indeed, the maximum structural capacity depends not only
on the link resistance, but also on the distribution of the plastic hinges.
For instance, CS-6-5-6-0.35 (a structure with the composite slab) shows a lower capacity with
respect to AS-6-5-6-0.35 (the corresponding all steel structure) (see Fig. 5a). In both structures,
the first plastic hinges appear at the base of the columns at the first floor. It is worth noting that
the EN1998-1 design procedure, with the stringent drift limitation imposed often results in
structures forming plastic hinges at the base, although the strong-column weak-beam principle is
respected. In this case, the AS structure has a larger column with a corresponding larger
resistance. A similar behavior can be observed in Fig. 5b, indeed in this case both the AS and CS
structures have the same steel profile; therefore, no appreciable differences can be pointed out
in terms of ultimate capacity. Differently from these cases, the CS-6-5-8-0.35 structure shows an
increase not only in terms of lateral stiffness, but also from the capacity point of view (see Fig.5c).
6-5-6-0.35 6-5-6-0.45
7000 8000
6000 7000
Base shear [kN]

Base shear [kN]
5000 6000
5000
4000
AS-6-5-6-0.35-U 4000 AS-6-5-6-0.45-U
3000
AS-6-5-6-0.35-FM 3000 AS-6-5-6-0.45-FM
2000 CS-6-5-6-0.35-U 2000 CS-6-5-6-0.45-U
1000 CS-6-5-6-0.35-FM 1000 CS-6-5-6-0.45-FM
0 0
0.00 0.50 1.00 0.00 0.50 1.00
Displacement [m] Displacement [m]
a) b)
6-5-8-0.35 6-5-8-0.45
7000 16000
6000 14000
12000
Base shear [kN]
Base shear [kN]
5000
4000 10000
AS-6-5-8-0.35-U 8000 AS-6-5-8-0.45-U
3000
AS-6-5-8-0.35-FM 6000 AS-6-5-8-0.45-FM
2000 CS-6-5-8-0.35-U 4000 CS-6-5-8-0.45-U
1000 CS-6-5-8-0.35-FM 2000 CS-6-5-8-0.45-FM
0 0
0.00 0.50 1.00 0.00 0.50 1.00
Displacement [m] Displacement[m]
c) d)
Figure 5: Pushovers results in terms of base shear top displacement.
Nonlinear dynamic analysis

The structural seismic performances were also evaluated by means of nonlinear dynamic
analyses considering three limit states according to EN1998-1 : damage limitation (DL), signiﬁcant
damage (SD) and near collapse (NC). A set of 14 natural earthquake acceleration records were
considered (see Table 3 and Fig. 6) to perform the dynamic time history analyses on the examined
cases. The records were obtained from the RESORCE ground motion database (Akkar et al.,
2014); all the spectra were scaled to match the elastic acceleration spectrum provided by EN
1998-1 (see Fig. 6). Moreover, since all the structures were analyzed in plane, the acceleration
scaling factor was increased by a factor equal to 1.3 in order to take into account possible torsional
effects. Furthermore, to evaluate the residual deformations, all records were fictitiously extended
by 10 seconds with zero acceleration.
The transient interstorey drift ratios along the frame height presented in Fig. 7 reveal that for all
three EC8 limit states the requirements of EC8 are fully satisfied with an overall response mostly
in the elastic range for each limit state. Indeed, for all design methodologies, at the DL limit state,
the structures remain elastic, as expected. Considering that, for SD and NC limit states, transient
interstorey drift limits are 2.5% and 5% respectively, results show significantly lower values, within
ranges between 0.35% and 0.43% for SD, and 0.48% and 0.57% for the NC limit state. This might
be due to the highly restrictive design controlled by drift limits and stability criteria for all the cases.
Table 3: Basic data of the selected ground motions
Earthquake Date Station Name Station Magnitude Fault
name Country Mw mechanism
Alkion 24.02.1981 Xylokastro-O.T.E. Greece 6.6 Normal
Montenegro 24.05.1979 Bar-Skupstina Opstine Montenegro 6.2 Reverse
Izmit 13.09.1999 Yarimca (Eri) Turkey 5.8 Strike-Slip
Izmit 13.09.1999 Usgs Golden Station Kor Turkey 5.8 Strike-Slip
Faial 09.07.1998 Horta Portugal 6.1 Strike-Slip
L'Aquila 06.04.2009 L'Aquila - V. Aterno - Aquila Italy 6.3 Normal
Park In
Aigion 15.06.1995 Aigio-OTE Greece 6.5 Normal
Alkion 24.02.1981 Korinthos-OTE Building Greece 6.6 Normal
Umbria- 26.09.1997 Castelnuovo-Assisi Italy 6.0 Normal
Marche
Izmit 17.08.1999 Heybeliada-Senatoryum Turkey 7.4 Strike-Slip
Izmit 17.08.1999 Istanbul-Zeytinburnu Turkey 7.4 Strike-Slip
Ishakli 03.02.2002 Afyon-Bayindirlik ve Iskan Turkey 5.8 Normal
Olfus 29.05.2008 Ljosafoss-Hydroelectric Power Iceland 6.3 Strike-Slip
Olfus 29.05.2008 Selfoss-City Hall Iceland 6.3 Strike-Slip
Figure 6: Comparison between natural records and EC8 design spectrum
6-5-6-0.35 6-5-6-0.45
6 6
DL DL
5 5
Strorey 4 4
Strorey
3 3
2 2
Damage limitation (DL)
1 AS AS
CS 1 CS
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
InerStorey [%] InerStorey [%]
6-5-8-0.35 6-5-8-0.45
6 6
DL DL
5 5
4 4
Strorey
Strorey
3 3
2 2
AS AS
1 CS 1 CS
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
6-5-6-0.35 6-5-6-0.45
6 6
SD AS SD
5 CS 5
4 4
Strorey
Strorey
3 3
Significant Damage (SD)
2 2
1 1 AS
CS
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
6-5-8-0.35 6-5-8-0.45
6 6
SD AS SD
5 CS 5
4 4
Strorey
Strorey
3 3
2 2
1 AS
1 CS
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
6-5-6-0.35 6-5-6-0.45
6 NC 6
AS NC AS
5 CS 5 CS
Near Collaps (NC)
4 4
Strorey
Strorey
3 3
2 2
1 1
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
6-5-8-0.35 6-5-8-0.45
6 6
AS NC AS
5 CS 5 CS
Strorey 4 4
Strorey
3 3
2 2
1 1
NC
0 0
-2.00% -1.00% 0.00% 1.00% 2.00% -2.00% -1.00% 0.00% 1.00% 2.00%
Figure 7: Interstorey drift at DL, SD and NC for all the investigated structures
The composite structures showed a more rigid global behavior than the all-steel cases (see Fig
5). However, no appreciable differences can be noted either in terms of displacement distribution
along the height or in terms of values, with respect to the all steel solution. Indeed, observing the
results reported in Fig. 7, it is not possible to recognize an overall difference between the
responses of the AS and CS structures.
Comparing the pushover analyses with the IDA curves obtained considering the maximum drift in
correspondence of the maximum base shear, it can be observed that the structures, for most of
the fourteen ground motions, remain elastic (see Fig.8). In all the investigated cases, the stiffness
results of IDAs are bounded within the two pushover curves (considering the two force
distributions previously introduced) and show an elastic behavior of the structures even for the
larger scaling factor (1.73).
AS-6-5-6-0.35 CS-6-5-6-0.35
7 000 R1 7 000 R1
6 000 R2 6 000 R2
R3 R3
Base shear [kN]
Base shear [kN]
5 000 R4 5 000 R4
4 000 R5 4 000 R5
R6 R6
3 000 R7 3 000 R7
2 000 R8 2 000 R8
R9 R9
1 000 R10 1 000 R10
0 R11 0 R11
0.00 0.50 1.00 R12 0.00 0.50 1.00 R12
R13 R13
Displacement [m] R14 Displacement [m] R14
AS-6-5-8-0.35 CS-6-5-8-0.35
10 000 R1 10 000 R1
R2 R2
Base shear [kN]
8 000 R3 8 000 R3
Base shear [kN]
R4 R4
6 000 R5 6 000 R5
R6 R6
4 000 R7 4 000 R7
R8 R8
R9 2 000 R9
2 000 R10
R10
R11 0 R11
0 R12 R12
0.00 0.50 1.00 0.00 0.50 1.00 R13
R13
Displacement [m] R14 Displacement [m] R14
Figure 8: Comparison between pushover and IDAs curves for both the AS-6-5-6-0.35 and CS-6-
5-6-0.35 structures.
CONCLUSIONS
In light of the results presented and discussed in the previous paragraphs, the following
conclusions are drawn:
• A refined structural model considering also the presence of the concrete slab in the middle
part of the beam allows properly accounting for the lateral stiffness of the structure. Indeed,
if the presence of the slab is accounted for in the design phase, the size of the element
cross section can be reduced as well as the costs.
• The non-linear link element defined in Seismostruct can adequately calibrate the real joint
behavior. Indeed, by the introduction of a zero-length link between the joint and the beam
it is possible to replicate the results of experimental tests and the ones coming from more
sophisticated models.
• The concrete slab has an important role in the definition of the local beam-to-column joint
behavior, even if all the EC1994-1-1 prescriptions for decoupling the steel joint from the
slab are respected.
• All investigated composite structures show a larger lateral stiffness with respect to the all
steel solution; however, from the resistance point of view, only when the plastic hinges are
active at the beam end, a consistent increment of capacity can be observed.
• The highly restrictive limitation lead to a design controlled by drift limits, indeed, all the
investigated structures show, for all the three limit states required by the EC1998-1-1, an
elastic behavior, also for a scaling factor equal to 1.73.
REFERENCES:
Akkar S., Sandıkkaya M.A., Şenyurt M., Azari Sisi A., Ay B.Ö., Traversa P., Douglas J., Cotton
F., Luzi L., Hernandez B., Godey S. (2014). Reference database for seismic ground-motion
in Europe (RESORCE). Bulletin of Earthquake Engineering. 12(1):311-339.
Aribert J, Ciutina A, Dubina D. (2006). Seismic response of composite structures including actual
behaviour of beam-to-column joints. Composite Construction Steel Concrete. V:708–17.
Corp.
EQUALJOINTS – European pre-QUALified steel JOINTS: RFSR-CT-2013-00021. Research
Fund for Coal and Steel (RFCS) research programme.
EN 1993:1–1, (2005) Design of Steel Structures - Part 1–1: General rules and rules for buildings.
CEN.
EN 1998-1, (2005) Design of Structures for Earthquake Resistance - Part 1: General Rules,
Seismic Actions and Rules for Buildings. CEN.
EN 1994:1–1, (2005) Design of composite steel and concrete structures - Part 1–1: General rules
and rules for buildings. CEN.
EN 1993:1–8, (2005) Design of Steel Structures - Part 1–8: Design of Joints. CEN.
Leon RT., (1998). Analysis and design problems for PR composite frames subjected to seismic
loads. Engineering and structures 20(4-6):364-71.
“SeismoStruct (2016) – A computer program for static and dynamic nonlinear analysis of framed
structures,” available from http://www.seismosoft.com.
Tartaglia R., D’Aniello M., Rassati G.A., Swanson J., Landolfo R. (2018a). Influence of composite
slab on the nonlinear response of extended end-plate beam-to-column joints. Key Engineering
Materials 763:818-825.
Tartaglia R, D’Aniello M., Di Lorenzo G., De Martino A., (2018b). Influence of EC8 rules on p-
delta effects on the design and response of steel MRF. Ingegneria sismica 35(3):104-120.
Tartaglia R., D’Aniello M., Zimbru M., Landolfo R., (2018c). Finite element simulations on the
ultimate response of extended stiffened end-plate joints. Steel and Composite
Structures,27(6):727-745.
Thermou GE, Elnashai AS, Plumier A, Doneux C., (2004). Seismic design and performance of
composite frames. Journal of Constructional of Steel Research. 60:31–57.
CONCRETE-FILLED STEEL TUBE
DUAL LATERAL-FORCE RESISTING SYSTEM
Johnn P. Judd, Ph.D., S.E.
Assistant Professor, Department of Civil and Architectural Engineering
University of Wyoming
Laramie, Wyoming
jjudd5@uwyo.edu
Nipun Pakwan
Graduate Research Assistant, Department of Civil and Architectural Engineering
University of Wyoming
Laramie, Wyoming
npakwan@uwyo.edu
ABSTRACT
A dual lateral-force resisting system consisting of a primary lateral-force resisting system and
secondary concrete-filled steel tube (CFT) gravity column system is presented in this paper. The
dual CFT system concept relies on the primary system to supply the main lateral strength, while
the inherent lateral strength in the gravity framing system, augmented with the additional
robustness of CFT columns, then provides additional lateral strength. To illustrate the concept,
the seismic performance of 4-story office buildings with perimeter steel moment frames was
predicted and compared to the performance of the buildings employing a dual CFT system using
square HSS gravity columns. The analyses predicted that buildings with the dual CFT system
were 22–68% less susceptible to seismic collapse depending on the ductility of the primary
moment frame and the orientation of the gravity columns, compared to buildings with only the
moment frame.
INTRODUCTION
Dual lateral-force resisting systems have two key characteristics that may enhance seismic
performance compared to single systems. First, dual systems explicitly provide increased
redundancy, allowing load to be transferred through alternate paths that are intended as part of
the design, instead of unintentional redistribution of load. Second, dual systems may provide
added safety against collapse and added resistance to structural and non-structural damage. The
latter characteristic is the motivation for the dual system explored in this study.
The potential for the gravity framing system to act as a dual or “reserve” lateral-force resisting
system has been recognized for several years. The effectiveness of the dual approach using the
gravity framing system depends on several factors, such as building height, structural
configuration, location of column splices, and beam-to-column connection details (Flores et al.,
2014; Judd et al., 2016). Depending on these factors, gravity framing may help reduce residual
story drift (Eatherton and Hajjar, 2011), improve collapse safety (Flores et al., 2014; Elkady and
Lignos, 2015), and improve serviceability (Judd and Charney, 2014). For a dual system that
involves a steel moment frame (MF), a previous study (Judd et al., 2016) suggested that the
gravity system needs to be capable of resisting at least 10% of the prescribed seismic forces in
order to be effective. In many cases, the lateral capacity contribution of a conventional gravity
framing system can be modest, on the order of 10–30% percent of the lateral-force resisting
system. Furthermore, soft-story mechanisms can limit the global collapse resistance.
Concrete-filled steel tube (CFT) gravity columns may provide a pragmatic approach to increase
the lateral strength of the gravity framing system and to improve the global collapse mechanism.
Prior research (e.g. Hajjar, 2000; Denavit et al., 2016) has demonstrated that CFT columns can
have significant flexural strength, axial compression strength, significant strength when subjected
to combined axial and flexure loads, and may contribute to cost-effective construction. Perhaps
more important in the context of a gravity framing system, CFT columns have excellent resistance
in two orthogonal directions, compared to wide-flange (W-shape) sections.
The objective of this analytical study was to explore the potential of a dual CFT system concept.
The dual CFT system concept relies on the primary system to supply the main lateral strength
and stiffness. The inherent lateral strength in the gravity framing system, augmented with the
additional robustness of CFT columns, then provides additional lateral strength. In this study, two
types of primary lateral-force resisting systems were considered: (1) a non-ductile steel MF, and
(2) a ductile steel MF. One type of CFT gravity column was considered: square HSS with a steel
minimum yield stress, Fy equal to 317 MPa (46 ksi) and a concrete compressive strength, fc ′ equal
to 34.5 MPa (5 ksi). Rectangular and circular HSS and high-strength steel and concrete were not
considered in this study. Gravity beam to-column connections use a single plate (shear tab)
welded to flange or CFT wall. The seismic performance of 4-story office buildings with perimeter
steel MFs was then predicted and compared to the performance of the buildings employing a dual
CFT system using an adaptation of the FEMA P-695 methodology (FEMA, 2009a).
METHODOLOGY
The example building plan (Figure 1) was based on steel MF archetype buildings designed for
the ATC-76 project (NIST, 2010). The building has perimeter moment frames, a 6.10-m (20-feet)
bay length, a 4.57-m (15-feet) first story height measured to the top of the beam, and a 3.96-m
(13-feet) upper story height (Figure 1a). Two types of MFs were evaluated for the primary lateral-
force resisting system. First, a non-ductile MF designed for the highest spectral accelerations
corresponding to Seismic Design Category (SDC) B for stiff soil (“site class D”, shear wave
velocity equal to 183-366 m/s) was evaluated. The non-ductile MF used fully-restrained directly-
welded flange connections. Second, a ductile MF designed for the highest spectral accelerations
corresponding to SDC C for stiff soil was evaluated. The ductile MF was a Special Moment Frame
(SMF) with reduced beam sections. Both types of MFs used columns fixed at the base.
The gravity framing system was designed for a 4.31 kN/m2 (90 psf) dead load and a 2.39 kN/m2
(50 psf) live load. The roof and floor of the buildings used a 140-mm (5.5-inch) composite slab
and gravity framing beams spaced at 3.05 m (10 feet) on center connected to girders with shear
tab connections. The gravity framing wide flange (W-shape) columns were pinned at the base
and spliced 1.22 m (4 feet) above the third floor. The W-shape columns were oriented with the
weak axis in the same direction as the moment frames (Figure 1b), and with the strong axis in the
moment frame direction (Figure 1c). In this way, the influence of the gravity system column
orientation was bounded for this building layout. Member sizes for the lateral-force resisting
system and for the gravity framing system are provided by Judd (2015). Gravity columns varied
from a W14X30 to a W14X90. For the dual CFT system (Figure 1d), W-shape columns were
replaced with HSS14X14X5/16 steel tubes with ASTM A500 Gr. B steel, Fy equal to 317 MPa (46
ksi), and filled with un-reinforced concrete, with fc ′ equal to 34.5 MPa (5 ksi).
Gravity columns (W14X30 - W14X90)
Primary LFRS (moment frame),
6.10-m bay width
Gravity framing
3.96 m
3.96 m
3.96 m
4.57 m
a) Elevation view of building frame b) W-shape gravity columns weak orientation
Moment frame
Gravity columns (HSS14X14X5/16)
Gravity framing
c) W-shape gravity columns strong orientation d) CFT gravity columns
Fig. 1 – Building plan.
A model of the building was made using OpenSees (PEER, 2015). A perspective view of the
model is shown in Figure 2. Beams and columns were modeled using an assembly of linear
elastic elements and zero-length nonlinear springs, to simulate the formation of plastic hinges in
the reduced section of the beams, plastic hinges at the top or bottom of columns, and shear
yielding of the column panel zone. Second-order effects were included by modeling the gravity
framing explicitly (instead of using a leaning column), and by using a “corotational” coordinate
transformation (where the local element coordinate system continuously rotates).
The moment-rotation behavior of non-ductile MF beam-to-column connections (Figure 2a) and
shear tab beam connections (Figure 2b) was modeled using the Pinching4 and MinMax uniaxial
materials in OpenSees, with load-deformation parameters based on empirical and test data, e.g.
FEMA P-440A (FEMA, 2009b) and FEMA 355D (FEMA, 2000). (See Judd and Charney, 2016.)
The moment rotation behavior of ductile MF beam-to-column connections was modeled using the
bilin material implementation of the modified Ibarra-Medina-Krawinkler bilinear model (Ibarra
and Krawinkler, 2005), with load-deformation parameters based on a statistical analysis of
experimental test data of bare steel RBS connections. The effect of composite action between
the moment frame and slab was not included, but was considered in a related study (Judd, 2016).
W16x40 Monotonic W16x31
envelope
Cyclic
response
a) Non-ductile beam-to-column connection b) Composite beam shear tab connection
CFT column
c) Perspective view of building frame plastic hinge
(Figures 3, 4)
W18x76 W18x76
d) W-shape column plastic hinge e) W-shape column panel zone

Fig. 2 – Analytical model of building frame
W-shape column plastic hinge behavior (Figure 2d) was modeled similarly to ductile MF
connection behavior, except column strength was reduced to account for axial load interaction.
The reduction in flexural strength was approximated using AISC 360-16 Equations H1-1a and H1-
1b (AISC, 2016) and a constant axial load determined resulting from the gravity load plus half the
lateral load calculated in a nonlinear static “pushover” analysis.
W-shape column panel zones were modeled using a form of the Krawinkler model that combines
the contributions of the column web and the column flanges (Charney and Marshall, 2006).
Shear-distortion behavior (Figure 2e) was modeled using the hysteretic material (a tri-linear
envelope curve and bilinear hysteresis behavior). Column splices were idealized as pinned
connections with no rotational constraint. The effect of the splice location and strength has been
considered in a related study (Flores et al., 2014).
CFT column plastic hinge behavior (Figure 3) was modeled using the BoucWen material, with
load-deformation parameters based on a parametric study of square CFT columns subjected to
cyclic loading (Skalomenos et al., 2014). This modeling approach was selected over other models
(e.g. Varma et al., 2005; Tort and Hajjar, 2010; Denavit and Hajjar, 2014), primarily due to the
computational efficiency of the concentrated plasticity model and the numerical robustness of the
BoucWen material when subjected to cyclic load and load reversal. In the BoucWen material,
based on the Baber and Noori (1985) implementation, the force, σ versus deformation, ε
relationship is defined as the sum of a linear part and a hysteretic part (Equation 1).
σ = α ko ε+(1-α) ko z (1)
In Equation 1, z represents the hysteretic deformation, ko is the elastic stiffness, and α is the ratio
of the post-yielding to elastic stiffness. The hysteretic deformation variable z accounts for
degradation using Equation 2. For square CFT sections, α is equal to 0.02 and ko is calculated
based on the effective stiffness, EIeff (AISC 360-10, Equation I2-13).
A - |z|n [β sgn(ϵ̇ z)+ γ] ν

ż = [ ] 𝜖̇ (2)
η
In Equation 2, the parameters γ and β control the shape of the hysteretic loop, n controls the
transition from linear to nonlinear behavior. For square CFT sections, γ and β are equal to 0.25/εy
where εy is the yield deformation accounting for axial-flexure interaction (Figure 4) based on the
plastic stress distribution method (Method 2 in Commentary, AISC 360-16). The parameters A,
ν, and η control the stiffness and strength degradation via Equations 3 to 5, respectively.
A = (Ao - δA e) (3)
ν = (1 + δν e) (4)
η = (1 + δη e) (5)
In the preceding equations, Ao is an amplitude factor, and δA , δν , and δη are strength and
stiffness degradation parameters. For square CFT sections, Ao is equal to 1/εy , δA is equal to 0.0,
and δν and δη are calculated using equations 6 and 7, respectively, where b and t are the width
and thickness of the HSS section in meters, respectively, and Fy and fc ′ are expressed in MPa
(Skalomenos et al., 2014). Finally, the rate parameter, e is defined by Equation 8.
δν = Ao exp [-4.5008 - 11.2683 b + 4.9987×10-4 ( b⁄t )2 - 50.8341⁄√Fy ] (6)

3.9012 730.0289
δη = Ao exp [-17.7959 + - +0.2583 √Fy ⁄fc '] (7)
√b ( b⁄t )2
ė = (1 -α ) ko ϵ̇ z (8)
HSS14X14X5/16
Cyclic response
Fy = 317 MPa (46 ksi)
fc ′ = 34.5 (5 ksi) Monotonic
envelope
Fig. 3 – CFT column plastic hinge
HSS14X14X5/16
Plastic stress distribution method

(Method 2 in Commentary to AISC 360-16)
Pure axial
(Point A)
Intermediate
(Point C)
Fy = 317 MPa (46 ksi)
fc ′ = 34.5 (5 ksi) Maximum flexure
(Point D)
Pure flexure (Point B)
Fig. 4 – Interaction diagram for CFT column
The seismic performance of the buildings with the dual CFT system compared to “conventional”
buildings (with W-shape columns in a strong or weak orientation) was evaluated using a sequence
of analyses. An initial pre-load (gravity) analysis was followed by an Eigen-value analysis to
determine the periods and modes of vibration of the building. The computed fundamental period
of vibration was used for both referencing spectral accelerations. The gravity and vibration
analyses were then followed by two pushover analyses: a first pushover analysis to estimate axial
loads in columns, and a second pushover analysis to account for axial-flexure load interaction.
Incremental nonlinear dynamic response history analyses (IDA) were conducted using the FEMA
P-695 Far-Field normalized ground motion set. Damping was explicitly modelled by the
component hysteresis. Artificial “Rayleigh” damping was not added. The model was subjected
to the set of ground motions, incrementally scaled in intensity with respect to the target SDC
spectrum, 𝑆aMCEl until the ground motions caused the building model to collapse in a side-sway
mechanism. The collapse fragility was determined by fitting individual collapse points (spectral
acceleration at the reference period and corresponding probability of collapse for a one scaled
ground motion record) with a curve, assuming a lognormal cumulative distribution function. The
median collapse margin ratio (ACMR) was increased to account for spectral shape (“SSF” factor)
and the total dispersion in collapse capacity, 𝛽Total based on record-to-record dispersion
measured in the incremental dynamic analyses, 𝛽RTR and epistemic dispersion due to design, test
data, and modeling uncertainty (Judd, 2015).
RESULTS
The results are summarized in Table 1, relative to the maximum spectral accelerations in SDC D
for purposes of comparison with other seismic systems. As in prior studies (e.g. Judd, 2016), the
computed fundamental period, T1 was higher than empirical equations used to ASCE 7, but in
line with the values calculated in the ATC-76 project. The buildings with a non-ductile MF were
7–19% stiffer compared to the conventional buildings, depending on the orientation of the gravity
columns. Buildings with a ductile MF were 6–13% stiffer.
The effective base shear coefficient, Cs is given relative to SDC Dmax. Since the MFs were
designed for SDC Bmin and SDC Dmin, the building “over-strength,” Ω is less than 1.0. The dual
CFT system increased Ω observed in the pushover analyses (Figure 5) by up to 40%. The CFT
columns also provided more over-strength compared to orienting W-shape gravity columns in the
strong direction. The period-based ductility, μT of the building with the CFT columns increased.
Table 1. Collapse safety evaluation, relative to SDC Dmax.

Building Static Pushover Dynamic IDA
Gravity T1 SaMCE ACMR Pc |
column (s) Cs  T (g) SSF SaDmax RTR Total SaDmax
Non-ductile moment-frame buildings designed for SDC Bmin
W-shape weak 2.32 0.21 0.35 3.01 0.39 1.33 0.81 0.34 0.63 63%
W-shape strong 2.01 0.21 0.41 3.72 0.45 1.38 1.01 0.34 0.63 49%
CFT 1.87 0.21 0.49 4.60 0.48 1.27 1.08 0.34 0.63 38%
Ductile moment-frame buildings designed for SDC Dmin
W-shape weak 1.73 0.08 1.54 4.75 0.52 1.44 1.40 0.30 0.43 22%
W-shape strong 1.60 0.08 1.80 4.92 0.56 1.45 1.73 0.32 0.44 11%
CFT 1.51 0.08 2.11 5.40 0.60 1.48 1.85 0.29 0.42 7%
CFT
CFT
W-shape strong
W-shape weak
W-shape strong
W-shape weak
a) Non-ductile MF designed for SDC Bmin b) Ductile MF designed for SDC Dmin
Fig. 5 – Nonlinear static pushover response
Figure 6a shows a representative response spectrum for the FEMA P-695 Far-Field ground
motion set, normalized, scaled and anchored to the computed fundamental period of vibration of
the building. Note that in the FEMA P-695 procedure, there is record-to-record dispersion at all
periods, including the target period. Figure 6b shows a representative behavior of a CFT column.
The results of the IDA are shown in Figure 7 for the non-ductile MF buildings and in Figure 8 for
the ductile MF buildings, relative to the maximum spectral accelerations in SDC D. A comparison
of the results for the weak orientation (Figure 7a and Figure 8a) and the strong orientation (Figure
7b and Figure 8b) show that the dual CFT system is more effective at increasing the median
collapse capacity than orienting all the W-shape columns in the strong direction. Furthermore, it
should be noted that such an arrangement would produce a building with weak columns in the
orthogonal (transverse) direction of the building. As a result of added stiffness due to the CFT
columns, buildings with a dual CFT system were subjected to slightly higher spectral accelerations
(see Table 1, column 6 for example). Notwithstanding the increase in spectral accelerations
caused by the increased stiffness, by employing the dual CFT system the probability of collapse
given the MCE-level ground motions (P𝑐 | 𝑆aMCEl ) was 22–40% lower for the non-ductile MF
buildings and 36–68% lower for the ductile MF buildings.
FEMA P-695 Far-Field

ground motion set,
scaled and normalized
Median response
MCE Design Story 1
Frame line 3,3
target MUL009 record
1.4x intensity factor
a) Scaled response spectrum b) CFT moment-rotation behavior

Fig. 6 Representative response for non-ductile SDC Bmin with CFT gravity columns
Collapse
points
Collapse fragility
MCE SDC Dmax including
uncertainties
MCE SDC Dmax
Median collapse
a) W-shape gravity columns oriented in weak direction
b) W-shape gravity columns oriented in strong direction
c) CFT gravity columns
Fig. 7 – Non-ductile MF designed for SDC Bmin IDA response
MCE SDC Dmax
Collapse
MCE SDC Dmax points
Collapse
fragility
including
Median collapse uncertainties
a) W-shape gravity columns oriented in weak direction
b) W-shape gravity columns oriented in strong direction
c) CFT gravity columns
Fig. 8 – Ductile MF designed for SDC Dmin IDA response
CONCLUSIONS
The potential for using a primary MF system to supply the main lateral strength and stiffness, and
the gravity framing system augmented with the additional robustness of CFT columns to provide
additional lateral strength was explored in this study. The results indicated that the 4-story
buildings with a non-ductile MF, and HSS14X14X5/16 columns with Fy equal to 317 MPa (46 ksi)
and filled with 34.5-MPa (5-ksi) concrete, had a 22–40% lower probability of seismic collapse,
compared to the conventional buildings with W-shape gravity columns, depending on the
orientation of the W-shape column. Buildings with a ductile MF and the CFT columns had a 36–
68% lower probability of seismic collapse.
Future research is underway to expand the scope of the study to consider a range of HSS
sections, high-strength steel and high-strength concrete, and to examine the effect of the dual
CFT system in a broader range of building heights, both low-rise and tall buildings. In this study,
the CFT columns did not significantly change the global collapse mechanism of the building.
Thus, one goal of the ongoing research is to reconfigure the CFT columns in an effort to develop
a “stiffened spine” that inhibits story-drift concentrations that can lead to a soft-story mechanism,
and thereby increase the global collapse resistance of the buildings. Another objective of the
ongoing research is to examine the seismic performance of the dual CFT system in terms of repair
costs, repair time, and business downtime by using the FEMA P-58 (FEMA 2012) framework.
Finally, the implications for construction of the proposed system and the potential for application
of the concept to building retrofit and rehabilitation need to be evaluated before implementation
of the dual CFT system in actual buildings.
REFERENCES
AISC (2010). Specification for structural steel buildings. ANSI/AISC 360-10, American Institute
for Steel Construction (AISC), Chicago, Illinois.
AISC (2016). Specification for structural steel buildings. ANSI/AISC 360-16, American Institute
for Steel Construction (AISC), Chicago, Illinois.
Baber, T.T., Noori, M.N. (1985). “Random vibration of degrading, pinching systems.” Journal of
Engineering Mechanics, Vol. 111, No. 8, 1010–1026.
Charney, F.A., Marshall, J.A. (2006). “Comparison of the Krawinkler and Scissors models for
including beam–column joint deformations in the analysis of moment resisting frames.”
Engineering Journal, Vol. 43, No. 1, 31–48.
Denavit, M.D., Hajjar, J.F. (2014). Characterization of behavior of steel-concrete composite
members and frames with applications for design. Report No. NSEL-034, Department of Civil and
Environmental Engineering, University of Illinois at Urbana-Champaign.
Denavit, M.D., Hajjar, J.F., Perea, T., Leon, R.T. (2016). “Seismic performance factors for moment
frames with steel-concrete composite columns and steel beams.” Earthquake Engineering and
Structural Dynamics, Vol. 45, No. 10, 1685–1703.
Eatherton, M.R., Hajjar, J.F. (2011). “Residual drifts of self-centering systems including effects of
ambient building resistance.” Earthquake Spectra, Vol. 27, No. 3, 719–744.
Elkady, A., Lignos, D.G. (2015). “Effect of gravity framing on the overstrength and collapse
capacity of steel frame buildings with perimeter special moment frames.” Earthquake Engineering
and Structural Dynamics, Vol. 44, No. 8, 1289–1307.
FEMA (2000). State of the art report on connection performance. FEMA-355D, Federal
Emergency Management Agency, Washington, D.C.
FEMA (2009a). Quantification of building seismic performance factors. FEMA P-695, Federal
Emergency Management Agency (FEMA), Washington, D.C.
FEMA (2009b). Effects of strength and stiffness degradation on seismic response. FEMA P-440A,
Federal Emergency Management Agency (FEMA), Washington, D.C.
FEMA (2012). Seismic performance assessment of buildings, volume 1—methodology. FEMA P-
58-1. Federal Emergency Management Agency (FEMA), Washington, D.C.
Flores, F.X., Charney, F.A., Lopez-Garcia, D. (2014). “Influence of the gravity framing system on
the collapse performance of special steel moment frames.” Journal of Constructional Steel
Research, Vol. 101, No. 10, 351–362.
Hajjar, J.F. (2000). “Concrete-filled steel tube columns under earthquake loads.” Progress in
Structural Engineering and Materials, Vol. 2, No. 1, 72–81.
Ibarra, L.F., Krawinkler, H. (2005). Global collapse of frame structures under seismic excitations.
Technical Report No. 152, The John A. Blume Earthquake Engineering Research Center,
Department of Civil Engineering, Stanford University, Stanford, California.
Judd, J.P. (2015). Multi-hazard performance of steel moment frame buildings with collapse
prevention systems in the central and eastern United States. Ph.D. dissertation, Civil Engineering.
Virginia Polytechnic Institute and State University, Blacksburg, Virginia.
Judd, J.P. (2016). “Seismic collapse risk of steel-concrete composite moment-frame buildings.”
International Colloquium on Stability and Ductility of Steel Structures, Timișoara, România.
Judd J.P., Charney, F.A. (2014). “Seismic performance of buildings designed for wind.” Structures
Congress, Boston, Massachusetts, 2342–2354.
Judd J.P., Charney F.A. (2016). “Seismic collapse prevention system for steel-frame buildings.”
Journal of Constructional Steel Research, Vol. 118, 60–75.
Judd J.P., Charney, F.A., Flores, F.X. (2016). “The influence of gravity framing on the
performance of steel buildings subjected to seismic loads.” International Workshop on
Connections in Steel Structures (CONNECTIONS VIII), May 24–26, Boston, Massachusetts.
NIST (2010). Evaluation of the FEMA P-695 methodology for quantification of building seismic
performance factors. NIST GCR 10-917-8, National Institute of Standards and Technology
(NIST), Gaithersburg, Maryland.
PEER (2015). Open Systems for Earthquake Engineering Simulation (OpenSees), version 2.4.6.
Pacific Earthquake Engineering Research Center (PEER), University of California, Berkeley,
California.
Skalomenos, K.A., Hatzigeorgiou, G.D., Beskos, D.E. (2014). “Parameter identification of three
hysteretic models for the simulation of the response of CFT columns to cyclic loading.”
Engineering Structures, Vol. 61, 44-60.
Tort, C., Hajjar, J.F. (2010). “Mixed finite-element modeling of rectangular concrete-filled steel
tube members and frames under static and dynamic loads.” Journal of Structural Engineering,
Vol. 136, No. 6, 654–664.
Varma, A., Sause, R., Ricles, J.M., Li, Q. (2005). “Development and validation of fiber model for
high-strength square concrete-filled steel tube beam-columns.” ACI Structural Journal, Vol. 102,
No. 1, 73–84.
An Introduction to Coupled Composite Core Wall Systems
for High-Rise Construction
Amit H. Varma
Professor, Lyles School of Civil Engineering, Purdue University
ahvarma@purdue.edu
Zhichao Lai
Postdoctoral Research Associate, Lyles School of Civil Engineering, Purdue University
laizhichao@gmail.com
Jungil Seo
Research Engineer, Lyles School of Civil Engineering, Purdue University
seo2@purdue.edu
ABSTRACT
This paper introduces the latest innovation in steel-concrete composite structures, namely,
coupled composite core wall systems that are being developed for high-rise building
applications as an efficient alternative to address issues associated with conventional concrete
construction. The system consists of concrete filled – composite plate shear walls (CF-CPSWs)
that are coupled together using steel beams or composite members. The CF-CPSWs consist of
a concrete wall sandwiched between two steel faceplates. The faceplates are anchored to the
concrete using stud anchors, and connected to each other using tie bars. The steel modules are
pre-fabricated in the shop, and shipped to the site for erection, assembly, and concrete casting.
Construction schedule is expedited by the pre-fabricated steel modules (from the shop), which
also serve as the stay-in-place permanent formwork during casting.
A research program focusing on the lateral load (wind and seismic) behavior, analysis and
design of the coupled core wall system and its individual components including the core walls,
coupling beams, and beam-to-wall connections is currently ongoing. This paper summarizes
preliminary results from the first sub-task of the research program, which focuses on the testing
and numerical analysis of the individual core walls. The paper includes the proposed tentative
test matrix, specimen details, and results from pre-test numerical analysis of the specimens
using detailed 3D finite element models. The paper discusses the expected behavior (stiffness,
strength, and ductility of the specimens), the effects of parameters such as the flange-to-web
plate thickness ratio, and the axial load ratio. These results are used to finalize the test matrix.
INTRODUCTION
Coupled reinforced concrete (RC) shear walls have been widely used around the world in high-
rise buildings as the primary lateral force resisting system for both wind and seismic applications.
Some of the challenges associated with coupled RC core wall systems include rebar congestion
(for high rise buildings), and construction schedule due to formwork placement, removal, and
concrete hardening, and the delays associated with discontinuity between the concrete core
wall and the steel outrigger and belt truss system (when used). Recently, there has been
increasing interest in the use an innovative coupled composite plate shear wall (CPSW) system
as an alternative to address these issues associated with coupled RC core walls.
A concrete filled-composite plate shear wall (CF-CPSW) consists of a concrete wall sandwiched
between two steel faceplates. The faceplates are connected to each other using regularly
spaced tie bars, and anchored to the concrete infill using the ties or additional stud anchors. The
steel faceplates serve as the primary reinforcement for the concrete wall, and no other
additional rebar is needed or provided. Additionally, the steel faceplates serve as stay-in-place
formwork for construction activities and concrete casting. The steel modules consisting of the
steel plates, ties, stud anchors etc. are fabricated in the shop, and shipped to the field for
erection, assembly of several such modules, and eventually casting of concrete.
Concrete-filled composite plate shear walls (CF-CPSWs) offer a potential alternative to
reinforced concrete walls for high rise buildings because of the following reasons (Varma et al.
2015): (i) The steel modules can be pre-fabricated in the shop, which reduces the construction
schedule associated with assembling and tying rebar cages. (ii) The steel faceplates serve as
permanent formwork, which further reduces construction schedule. (iii) High reinforcement
ratios such as 2% or more can be achieved relatively easily using steel faceplates. (iv) Rebar
congestion issues are reduced by eliminating rebar cages. (v) Better concrete placement can be
achieved using conventional or self-consolidating concrete.
In high rise buildings, these CF-CPSWs can be coupled together using steel beams or
composite members (concrete filled steel tubes). As compared to conventional coupled RC
shear walls, coupled CPSW have several potential advantages including: (i) higher stiffness and
strength due to the use of steel faceplates that can achieve higher reinforcement ratios without
congestion issues, (ii) better seismic performance due to the synergistic interaction between the
steel and concrete materials, namely, confinement of the concrete by the steel, and local
buckling restraint to the steel plate provided by the concrete infill, and (iii) construction schedule
efficiency due to modular construction techniques, prefabrication and erection of steel modules,
and continuity during construction of the steel belt truss and outriggers when needed.
A research program focusing on the lateral load (wind and seismic) behavior, analysis and
design of the coupled core wall system and its individual components including the core walls,
coupling beams, and beam-to-wall connections is currently ongoing. This paper summarizes
preliminary results from the first sub-task of the research program, which focuses on the testing
and numerical analysis of the individual core walls. The paper includes the proposed tentative
test matrix, test setup, and pre-test numerical analysis of the specimens using detailed 3D finite
element models.
This monotonic and cyclic flexural behavior of the proposed CPSW specimens are analyzed
using detailed 3D nonlinear finite element (FEM) models built and analyzed using ABAQUS.
The models are used to evaluate the effects of flange-to-web plate thickness ratio and axial load
ratio, which were used to finalize the details of the specimens in the matrix. The detailed FEM
models include various behavioral complexities such as: (i) yielding, local buckling, and fracture
of the steel plates, (ii) concrete cracking and crushing, and (iii) cyclic stiffness degradation and
recovery.
BACKGROUND
The lateral load behavior of CF-CPSWs, including their lateral force-displacement behavior and
ductility, has been evaluated recently by many researchers. However, the focus has been on
the in-plane shear behavior of squat walls for nuclear structures (Seo et al., 2016), and the in-
plane flexural behavior of wall piers with or without boundary elements (Ramesh, 2013, Alzeni
and Bruneau 2014, Kurt et al. 2016). Research on the lateral load behavior of coupled CF-
CPSW core wall structures is somewhat limited. Figure 1 shows an example of a typical coupled
CF-CPSW core wall structure. As shown, it consists of two individual core walls connected by
coupling beams along the height of the structure. The individual core walls consist of two or
more CF-CPSWs connected together to make either C, U, T, or I shapes.
There are some unique challenges that need to be addressed and resolved for coupled CF-
CPSW core wall structures. For example, (i) the lateral load behavior including the stiffness,
strength and drift capacity of the individual core walls, (ii) the effects of axial force on the lateral
load behavior of individual core walls, (iii) the detailing of the connections between the individual
CF-CPSWs comprising the core wall, (iv) the design of the coupling beam and the detailing of
its connections to individual core walls, (v) the influence of coupling ratio on the lateral load
behavior and design optimization of the coupled core wall system. Some of these unique
challenges have been investigated and addressed for concrete core wall structures (El-Tawil et
al., 2010) and included in AISC 341-16 (2016)for hybrid coupled core wall structures.
Fig.1 - Schematic of coupled CF-CPSW core wall structure: (a) structural plan, (b) elevation,
and (c) simulated deformed state
Figure 2 shows an idealized coupled shear wall system deformed under lateral loading, which
causes a global over turning moment (OTM). The coupled system resists the OTM by
developing an axial force couple (ΣVbeam,j over the lever arm L) as well as flexural moment
resistances in the individual walls (m1 and m2). The contribution of the axial force couple
(ΣVbeam,j) to resisting OTM is defined as the coupling ratio (CR) and it can be calculated using
Eq. (1). CR=0 implies that no coupling beams are present or the two individual shear walls are
disconnected. CR=100% is the theoretical case where the coupling beam length is zero.
CR =
L åV beam, j
=
L åV
beam, j (1)
L åV
beam, j + m1 + m2 OTM
Fig. 2 - Deformed coupled shear wall system
The axial force couple (ΣVbeam,j) is applied to the individual shear wall (in addition to gravity
loads) that vary during an earthquake. The axial compression force acting on the individual
shear wall can potentially reduce the ductility of the wall due to concrete crushing failure and the
axial tension force acting on the individual shear wall can reduce the stiffness and the shear
capacity of the wall.
To evaluate (i) the structural performance of CF-CPSW and (ii) the effects of the axial force, six
scaled tests will be conducted. The six CF-CPSW specimens (see Figure 3) will be subjected to
cyclic quasi-static lateral loading as well as axial loading. The test specimens are 1/3-1/4 scale
models of the bottom four-stories of the prototype walls. The test specimens have a wall aspect
ratio of approximately 3. The proposed tentative test matrix for the CF-CPSW specimens is
presented in Table 1. As shown, the primary parameter is the degree of axial compression force
(10, 20, and 30% of the axial load capacity, Agf’c). The influence of axial compression force on
the structural behavior (stiffness, strength, and drift capacity) of the uncoupled CF-CPSW core
wall can be investigated by comparing the test results of the four specimens. In addition, the
influence of the plate slenderness ratio (sstud/tp and sstud/tp) will be considered.
Table 1 – Tentative Test matrix

Specimen H, in L, in P, %Agf’c tsc, in tp, in sstud/tp stie/tp ρt
CW-42-55-10 108 36 10 9 0.1875 - 24.0 0.55%
CW-42-55-20 108 36 20 9 0.1875 - 24.0 0.55%
CW-42-55-30 108 36 30 9 0.1875 - 24.0 0.55%
CW-42-14-20 108 36 20 9 0.1875 - 48.0 0.14%
CW-42-14-20 108 36 20 9 0.1875 - 48.0 0.24%
CW-42-24-20 108 36 20 9 0.1875 24 48.0 0.14%
Figure 4 shows a schematic view of the CF-CPSW core wall test setup with the specimens. As
illustrated in the figure, the cyclic lateral loading will be applied at the top of the specimens using
a set of hydraulic actuators. The compressive axial force will be applied to the specimens using
another set of hydraulic actuators. The lateral loading history will be developed based on the
validation testing requirements prescribed in AISC 341-16 (2016).
Push
Web
Bottom flange
Top flange
Fig. 3 - Typical test specimen Fig. 5 - Typical FEM model
Fig. 4 - Schematic view of test setup
DETAILS OF SCALED SPECIMEN AND FEM MODELS

The height of the scaled CPSW is 120 in, the width is 40 in, and the thickness is 10 in. Tie bars
were spaced 8 in. to develop composite action between the steel plates and the concrete infill.
The diameter of the tie bars is 0.5 in. The resulting tie bar reinforcement ratio is 0.31%. The
yield stress of the steel plates and tie bars is 50 ksi, and the compressive strength of the
concrete infill is 6 ksi.
The FEM models were developed using ABAQUS (2016). Details of the FEM models include
the element types, contact interaction, steel and concrete material models (including steel
fracture and concrete cyclic damage), geometric imperfections, boundary conditions, and the
analysis method. These details are similar to those developed by the authors for CFT members
in Lai and Varma (2017), and are briefly summarized as follows. Figure 5 shows the typical FEM
model used in the analysis.
The steel face plates were modeled using four-node shell elements with reduced integration
(S4R). The concrete infill was modeled using eight-node solid elements with reduced integration
(C3D8R). The tie bars were modeled using two-node beam elements (B31). The connections
between the tie bars and steel face plates were modeled using connector elements
(CONN3D2). The axial and interfacial shear behavior of the connections were defined by
specifying the corresponding force-displacement behavior of the connector elements. In this
research, the axial behavior was assumed to be linear elastic (stiffness = EsAt, where At is the
cross-section area of the tie bars). The shear behavior was defined using the model proposed
by Ollgaard et al. (1971).
EFFECT OF FLANGE-TO-WEB THICKNESS RATIO

The developed FEM models were used to evaluate the effect of flange-to-web thickness ratio on
the monotonic and cyclic behavior of the CPSW specimen. The flange-to-web thickness ratios
were varied by changing the flange thickness (0.1875, 0.375, and 0.5 in) while keeping the web
thickness constant at 0.1875 in. No axial load was applied during this evaluation.
Monotonic analysis and behavior

Three monotonic analyses were conducted to investigate the effect of flange-to-web thickness
ratio on the strength and stiffness of CPSW. Figure 6 shows the resulting moment-displacement
curves. The displacement reported in this figure represented the lateral deflection at the free
end. As shown, the flexural stiffness and strength increase as the flange thickness increases.
This figure also shows that the flexural strengths of all three members are greater than the
corresponding plastic moment strengths Mp calculated based on AISC 360-16 (2016). This
indicates that all three members can develop the plastic moment capacity.
30000
t f = 0.5 in.
25000
Mp
Moment (kips-in)
20000
Mp
15000 Mp
t f = 0.375 in.
10000 t f = 0.1875 in.
5000
0
0 1 2 3 4 5
Displacement (in.)
Fig. 6 - Effect of flange thickness on the monotonic moment-displacement curve from analysis
Cyclic analysis and behavior

Three cyclic analyses were conducted to investigate the effect of flange-to-web thickness ratio
on the cyclic behavior and ductility of CPSW. Cyclic lateral loading was applied at the free end
of the wall using the loading history shown in Figure 7. As shown, three elastic cycles were
applied each at 0.5My and 1.0My, where My is the moment corresponds to the first yield. After
that, three inelastic cycles were applied each at 1.0Δy, 2.0Δy, 3.0Δy, and 4.0Δy, where Δy
represents the lateral displacement at the loading locations corresponding to the plastic moment
(Mp). Δy, My, and Mp were obtained from corresponding monotonic analysis.
The analyses indicated that the failure modes for all three members are similar, i.e., local
buckling of the flange and web at the base occurred first, followed by the fracture at these
locations. Table 2 summarizes the loading cycles during which local buckling and fracture
occurred. As shown, local buckling first occurred at 1.0Δy push cycle. When the flange thickness
is 0.1875 in., local buckling occurred at the flange first; while when the flange thickness is 0.375
or 0.5 in., local buckling occurred at the web first. Table 2 also indicates that increasing the
flange thickness has negligible influence on the ductility (fracture occurred during the first 4.0
push and pull cycle for all specimens).
0 10 20 30 40
5
4 Push
3
2
Displacement, ∆y
1
0
-1
-2
-3
-4 Pull
Time
-5
Fig. 7 - Loading history
Table 2 Summary of loading cycles during which local buckling and fracture occurred (CPSW
with different flange thickness)
Flange First appearance of local
No. Initial Fracture
thickness buckling
Top flange at the base, Top flange at the base,
1 0.1875
first 1.0Δy push cycle first 4.0Δy pull cycle
Web at the base close to Bottom flange at the

2 0.375 the top flange, first 1.0Δy base, first 4.0Δy push
push cycle cycle
Web at the base close to

Top flange at the base,
3 0.5 the top flange, first 1.0Δy
first 4.0Δy pull cycle
pull cycle
Figure 8 shows the cyclic load-displacement behavior of the walls. As shown, the cyclic
behavior of all three walls was very good during the elastic cycles. There were no local buckling
distortions, bulging or fracture during elastic loading cycles. During the first 1.0Δy push cycle,
local buckling developed at the base in the top and bottom flanges. The local buckling resulted
in the about 9% loss of strength. After that, local buckling propagated into the webs. The local
buckling increased during the 3.0Δy cycles and resulted in fracture of the flanges in the first
4.0Δy push and pull cycle. The fracture developed and spread to the webs during the second
and third 4.0Δy push and pull cycle. Figure 8 indicates that the flange-to-web thickness ratio has
negligible influence on the ductility (all three walls have a ductility of 3, the wall is assumed to
remain its ductility if its strength loss is less than or equal to 20%). Figure 8 also includes the
monotonic load-displacement behavior from the monotonic analysis. As shown, the monotonic
load-displacement curve is similar to the envelope obtained from the cyclic analysis. Based on
all these results, the flange thickness of 0.1875 in. was selected for all further evaluation and the
design of the specimens identified in the tentative test matrix of Table 1.
30000 30000
Push Push
20000 20000
Moment (kips-in)
Moment (kips-in)
10000 10000
0 0
-10000 -10000
-20000 Cyclic -20000 Cyclic

Pull Monotonic Pull Monotonic
-30000 -30000
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Displacement (in) Displacement (in)
(a) Flange thickness = 0.1875 in (b) Flange thickness = 0.375 in.

30000
Push
20000
Moment (kips-in)
10000
-10000
-20000 Cyclic
Pull Monotonic
-30000
-6 -4 -2 0 2 4 6
Displacement (in)
(c) Flange thickness = 0.5 in.

Fig. 8 - Effect of flange thickness on the cyclic moment-displacement behavior from analysis
EFFECTS OF AXIAL LOAD RATIO

The developed FEM models were used to evaluate the effect of axial load ratio on the
monotonic and cyclic behavior of CPSW. Five axial load ratios were used (-0.1, 0, 0.1, 0.2, and
0.3). The axial load ratios were varied by changing the applied axial load. The flange and web
plate thickness was kept constant at 0.1875 in.
Monotonic analysis and behavior

Five monotonic analyses were conducted to investigate the effect of axial load ratio on the
strength and stiffness of CPSW. Figure 9 shows the resulting moment- displacement curves.
This figure indicates that as the axial load increases, the flexural stiffness and strength increase,
and the post-peak response becomes more negative. This figure also includes the plastic
moment strengths Mp [calculated based on AISC 360-16 (2016)] of the wall without any axial
load applied. As shown, when the axial load ratio is greater or equal to 0, the CPSW have
flexural strengths greater than Mp. When the axial load ratio is equal to -0.1 (in tension), the
CPSW have flexural strength close to Mp (98%).
Figure 10 compares the beam-column strength of CPSW with the beam-column interaction
curve calculated using AISC 360-16 (2016) for filled composite columns. As shown, the beam-
column strength of CPSW can be reasonably estimated using these AISC design equations.
25000
Mp
20000
Moment (kips-in)
15000
10000 P = 0.1
P = 0.2
5000 P = 0.3
P=0
P = -0.1
0
0.0 1.0 2.0 3.0 4.0 5.0
Displacement, in
Fig. 9 - Effect of axial load ratio on the monotonic moment-displacement curve from analysis
1.2
P = 0.1
1 P = 0.2
0.8 P = 0.3
M/M n
0.6
0.4
0.2
0
0 0.5 1 1.5
P/Pn
Fig. 10 - Comparisons of the analytical beam-column strength of CPSW with the beam-column
interaction curve calculated using AISC 360-16
Cyclic analysis and behavior

Five cyclic analyses were conducted to investigate the effect of the axial load ratio on the cyclic
behavior of CPSW. The loading history as shown in Figure 7 was applied. The analyses
indicated that the failure modes of all walls are similar, i.e., local buckling of the flanges and
webs occurred at the base, followed by the fracture at these locations. Table 3 summarizes the
loading cycles during which local buckling and fracture occurred. As shown, increasing the axial
load ratio does not have significant effect on the initiation of local buckling; however, it delays
the initial fracture. When the axial load ratio is less than or equal to 0.1, initial fracture occurred
at the first 4.0Δy push and pull cycle; when the axial load ratio is 0.2, initial fracture occurred at
the third 4.0Δy push cycle; when the axial load ratio is 0.3, there is no fracture.
Figure 11 shows the cyclic load-displacement behavior of the walls. As shown, the cyclic
behavior of all walls was very good during the elastic cycles. There were no local buckling
distortion, bulging or fracture during elastic loading cycles. During the first 1.0Δy push and pull
cycle, local buckling developed at the base in the top and bottom flanges. After that, local
buckling propagated into the webs. The local buckling increased significantly during the 3.0 and
4.0 Δy loading cycles.
25000
30000
20000 Push
15000 20000
Moment (kips-in)
10000
Moment (kips-in)
10000
5000
0 0
-5000
-10000 -10000
-15000 Cyclic Cyclic
-20000
-20000 Pull
Monotonic Monotonic
-25000 -30000
-6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 -6 -4 -2 0 2 4 6
(a) Axial load ratio = -0.1 (b) Axial load ratio = 0

25000 25000
20000 Push 20000 Push
15000 15000
Moment (kips-in)
Moment (kips-in)
10000 10000
5000 5000
0 0
-5000 -5000
-10000 -10000
-15000 Cyclic -15000 Cyclic
-20000 Pull Monotonic -20000 Pull Monotonic
-25000 -25000
-6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0
(c) Axial load ratio = 0.1 (d) Axial load ratio = 0.2
25000
20000 Push
15000
Moment (kips-in)
10000
5000
0
-5000
-10000
-15000 Cyclic
-20000 Pull Monotonic
-25000
-6.0 -4.0 -2.0 0.0 2.0 4.0 6.0
Displacement (in)
(e) Axial load ratio = 0.3

Fig. 11 - Effect of axial load ratio on the cyclic moment-displacement behavior from analysis
When the axial load ratio is less than or equal to 0.2, the local buckling resulted in the fracture of
flange and web plate in the 4.0Δy loading cycles. After 3.0Δy loading cycles, the local buckling
resulted in the loss of strength of 3.4%, 10%, 17.2%, 29%, and 33.7% for walls with the axial
load of -0.1, 0, 0.1, 0.2, and 0.3, respectively. This indicates that ductility decreases as the axial
load ratio increases (the wall is assumed to remain its ductility if its strength loss is less than or
equal to 20%). Figure 11 also includes the monotonic load-displacement behavior from the
monotonic analysis. As shown, the monotonic load-displacement curve is similar to the
envelope obtained from the cyclic analysis.
Table 3 - Summary of loading cycles during which local buckling and fracture occurred (CPSW
with different axial load ratio)
Axial load First appearance of local
No. Initial Fracture
ratio buckling
Bottom flange at the base, Bottom flange at the
1 -0.1 base, first 4.0Δy push
first 1.0Δy pull cycle
cycle
2 0

3 0.1

4 0.2
first 1.0Δy push cycle third 4.0Δy push cycle
Top flange at the base,

5 0.3 No fracture
first 1.0Δy push cycle
CONCLUSIONS
Concrete filled – composite plate shear walls (CF-CPSW) are being evaluated as an alternative
to reinforced concrete shear walls in high rise building construction. This paper presented the
preliminary results from the first sub-task of an extensive research program focusing on the
lateral load (wind and seismic) behavior, analysis and design of coupled CF-CPSW core wall
structures for high rise buildings. A tentative test matrix focused on evaluating the lateral load
behavior of individual CF-CPSWs was proposed and pre-test analyses were conducted using
3D finite element models accounting for the complexities of behavior including local buckling,
concrete crushing, steel yielding and fracture, etc.
The results from the pre-test analyses were used to detail the specimens in the tentative test
matrix. These results indicate that: (i) the lateral load behavior and strength of CF-CPSW
specimens will be governed by their in-plane flexural response and plastic moment capacity. (ii)
The plastic moment capacity and the axial load-moment capacity (P-M) interaction of CF-CPSW
specimens can be estimated using provisions for filled composite members in AISC 360-16. (iii)
Increasing the thickness of the flange plate increases the flexural strength, but does not have a
significant influence on ductility. (iv) Increasing the axial load level increases the flexural
stiffness and strength, but reduces the ductility, particularly for axial load levels greater than
20%. (v) The failure modes of specimens subjected to cyclic loading will be similar, i.e., local
buckling of the flange and web at the base, followed by steel plate fracture at these locations.
ACKNOWLEDGMENTS
The research project discussed in this paper has been funded by the Charles Pankow
Foundation and American Institute of Steel Construction. The opinions and findings discussed in
this paper represent the preliminary thoughts of the authors alone.
REFERENCES
ABAQUS. (2016). ABAQUS Version 6.16 Analysis User’s Manuals. Providence, RI, USA:
Dassault Systemes Simulia Corporation.
AISC 341-16. (2016). Seismic provisions for structural steel buildings. Chicago, IL, USA: AISC.
AISC 360-16. (2016). Specification for structural Steel buildings. Chicago, IL, USA: AISC.
Alzeni, Y., & Bruneau, M. (2014). Cyclic inelastic behavior of concrete filled sandwich panel
walls subjected to in-plane flexure. In 10th National Conference on Earthquake
Engineering (pp. 1–7). Anchorage, Alaska, USA: Frontier of Earthquake Engineering.
El-Tawil, S., Harries, K. a., Fortney, P. J., Shahrooz, B. M., & Kurama, Y. (2010). Seismic
design of hybrid coupled wall systems: state of the art. Journal of Structural Engineering,
136(7), 755–769.
Kurt, E. G., Varma, A. H., Booth, P., & Whittaker, A. S. (2016). In-plane behavior and design of
rectangular SC Wall piers without boundary elements. Journal of Structural Engineering,
142(6), 4016026.
Lai, Z., & Varma, A. H. (2017). Seismic behavior and modeling of circular concrete partially filled
spirally welded pipes (CPFSWP). Thin-Walled Structures, 113, 240–252.
Ollgaard, J. G., Slutter, R. G., & Fisher, J. W. (1971). Shear strength of stud connectors in
lightweight and normal weight concrete. AISC Engineering Journal, (Aprl).
Ramesh, S. (2013). Behavior and design of earthquake–resistant dual–plate composite shear
wall systems [Dissertation]. Purdue University.
Seo, J., Varma, A. H., Sener, K., & Ayhan, D. (2016). Steel-plate composite (SC) walls: In-plane
shear behavior, database, and design. Journal of Constructional Steel Research, 119,
202–215.
Varma, A.H., Malushte, S., Lai, Z. (2015). “Modularity and Innovation Using Steel-Plate
Composite (SC) Walls for Nuclear and Commercial Construction.” Proceedings of the 11th
International Conference: Advances in Steel-Concrete Composite Structures (ASCCS), At
Beijing, China, Dec. 2015, http://dx.doi.org/10.13140/RG.2.1.4665.4804
IN-PLANE CYCLIC TESTING OF CONCRETE FILLED SANDWICH STEEL PANEL WALLS
WITH AND WITHOUT BOUNDARY ELEMENTS
Michel Bruneau
SUNY Distinguished Professor, University at Buffalo
Department of Civil, Structural, and Environmental Engineering, 212 Ketter Hall,
Buffalo, 14260, NY, USA
Email: bruneau@buffalo.edu
Yasser Alzeni
Assistant Professor, Alexandria University
Department of Structural Engineering
Alexandria, Alexandria Governorate 11432, Egypt
Email: yasser_alzeni@azizmi.com
ABSTRACT
This paper presents results from an experimental study investigating the in-plane cycling
inelastic flexural behavior of Concrete Filled Sandwich Steel Panel Walls (CFSSP-Walls) that
are composed of two steel skin plates interconnected by tie bars, with the space between the
skin plates filled with concrete. CFSSP-Walls with and without circular boundary elements were
tested. All of the walls were able to exceed their theoretical plastic moment capacity, calculated
assuming a full plastic stress distribution of the cross section. The tested specimens exhibited
stable ductile behavior up to 3% drift (and beyond in some conditions). Local buckling of the
steel skin led to minor degradation in flexural strength. Fracture of the skin plates eventually
developed as the ultimate failure mode.
INTRODUCTION
Concrete-Filled Sandwich Steel Panel Walls (CFSSP-Walls) are being considered by practicing
engineers for construction in seismic regions of the USA, including as ductile flexural walls in
high-rise applications. Their appeal is that they are envisioned to be highly ductile, redundant,
of high strength, and easy and rapid to construct (overcoming the congestion of reinforcement
details that can be encountered in ordinary reinforced concrete walls, and because the steel
shell can be used as formwork). Beyond accelerated construction and ductile seismic
performance, use of a CFSSP-Wall instead of a conventional reinforced concrete wall in
building applications can translate into thinner walls with resulting greater leasable space.
However, at the time this research was conducted, there was a critical lack of knowledge on the
in-plane cyclic inelastic behavior of such CFSSP-Walls, which, in spite of all the positive
attributes of the structural system, was an impediment to implementation in seismic regions.
To help understand the in-plane flexural behavior of such walls, in the current study, four
specimens with and without boundary elements were subjected to cyclic inelastic loading.
Performance was evaluated in terms of ductility and lateral load carrying capacity. The results
of this experimental work are presented here.
Note that at the time this research was conducted, the terminology “walls with boundary
elements” was used by the authors to refer to the presence of a distinct structural shape wider
than the rest of the wall (namely, a full HSS here), by opposition to the case where a simple
closure plate (square or half-circular) is used at the end of the wall (referred to here as “walls
without boundary elements”). This terminology is kept here to ensure consistency with the
information contained in the full research report (Alzeni and Bruneau 2014), because changing
the names of specimens and terminology here would only serve to confuse the readers
referring to this report. However, the AISC 341-16 seismic provisions published more recently
use the term “boundary elements” to encompass both types of walls tested here, and the term
walls “without boundary elements” refer to those that do not have such cap plates (built, for
example, using formwork at their ends to contain the concrete within the wall during the pour).
LITERATURE REVIEW
The type of composite wall considered here consists of sandwich steel panels filled with
concrete. These have been used in many applications. Past research on that structural system
has largely focused on walls subjected to out-of-plane loading and/or non-cyclic behavior (e.g.,
Wright (1998), Hossain and Wright (1998)). Sener et al (2015) summarized the experimental
data base of out-of-plane flexural tests conducted on SC walls under different loading
configurations, and different design parameters. Also, some of the past research has been
conducted in relation to the Bi-Steel sandwich system that has been developed as flooring
system, beams, columns, and walls system used in some non-seismic regions (Bowerman et
al. 1999).
Limited research has been conducted on the in-plane inelastic cyclic response of CFSSP-Walls.
Eom et al. (2009) conducted in-plane cyclic testing of three individual flexural walls and two
coupled walls having rectangular and T-shaped cross-sections, representing 1/3 and 1/4 scale
models of the first stories of a 30 story prototype. The tested specimen showed premature
failure in the corner welding of the rectangular section and different methods were used to
strengthen the walls, which were retested and (in some cases) able achieve estimated peak
loads and develop reasonable ductile performance. Epackachi et al (2014) tested four CFSSP-
Walls having an aspect ratio of one, but having no cap plate at the cross-section ends. Under
in-plane cyclic loading, the tested walls were able to reach their yield moment (equal to My) and
failed in a flexural mode, while the hysteretic loops exhibited significant pinching. Kurt et al
(2016) tested eight CFSSP-Walls without boundary elements under cyclic lateral loads, where
the walls had an aspect ratio between 0.6 and 1.0; the lateral load capacity of the tested walls
were governed by the in-plane flexure capacity of wall cross section at the base. Seo et al.
(2016) used a mechanics based model (MBM) to present the fundamental in-plane shear force-
shear strain of the SC walls, the results were compared to the results of the experimental data
base and an estimated value for the strength reduction factor was introduced.
Most importantly, at the time the tests reported below were conducted (in 2012), there existed
only limited data available on the inelastic cyclic behavior of flexural CFSSP-Walls.
Four large scale CFSSP-Wall specimens were designed as cantilever walls fixed to a reinforced
concrete footing. The CFSSP-Walls tested as part of this research program were meant to be
representative of implementation in commercial buildings as cantilevering walls surrounded by
gravity frames, where the floor is tied to the walls through seismic collectors. Note that the
findings from this research cannot be directly extrapolated to designs for walls having T or C
plan cross-section, or other more complex configurations, for which additional research is
recommended.
The tested specimens were divided over two groups, with cross-section properties shown in
Figure 1. In the first group, identified by the label NB, the CFSSP-Walls have no boundary
elements and their cross section is composed of double web skin plates having thickness, t, of
8 mm (5/16”) and web width, b, of 1016 mm (40”), connected through circular tie bars spaced
equally in both horizontal and vertical direction at a spacing, S, equal to 203.2 mm (8”) and
304.8 mm (12”) for specimens CFSSP-NB1 and CFSSP-NB2 respectively, corresponding to S/t
ratio of 25.6 and 38.4. The cross-section is closed at its ends by half HSS sections; these were
used instead of square ends in order to avoid premature failure of the cross section’s corner
welds due to concentration of stresses that has been observed in prior research on composite
rectangular sections (e.g., Eom et al. 2009; El Bahey and Bruneau 2011; Ge and Usami 1996;
Usami and Ge 1994; Kawashima and Unjoh 1997). Both specimens of Group NB have the
same outer dimensions, plate thicknesses, and total width and height. Specimen CFSSP-NB1
and CFSSP-NB2 have a total width, W, 1235.1 mm (48.625”) (skin plates plus two half-HSS
8.625×0.325) and total thickness, tt, of 219.1 mm (8.625”). For specimens in Group NB, the tie
bars were welded to the web skin plate using plug welding: tie bars having a total length equal
to (tt - ts) were positioned to span the distance between the steel web plates center lines, and
the remaining half thickness of the plate on each side was filled with welding material to create
the plug weld. The tie bars had a diameter of 25.4 mm (1”)and were designed to remain elastic
throughout the test.
(a) CFSSP-NB1 (b) CFSSP-NB2
(c) CFSSP-B1 (d) CFSSP-B2

Figure 1. Tested specimens of CFSSP Walls
The second group of tested specimens, identified by the label B, are CFSSP-Walls having
boundary elements (i.e., complete structural shapes) consisting of concrete filled round HSS
section. The walls’ cross section consists of HSS columns and of double web skin plates having
a width, L, of 762 mm (30”), a thickness, ts, of 8 mm (5/16”), and connected through tie bars
spaced equally in both horizontal and vertical direction at a spacing, S, of 203.2 and 304.8 mm
(8” and 12”) for specimens CFSSP-B1 and CFSSP-B2, respectively, corresponding to S/t ratios
of 25.6 and 38.4 (i.e., the same values used for the NB specimens). The Specimen CFSSP-B1
tie bars were connected to the web plate using plug welding as described for the Group NB
specimens. However, for specimen CFSSP-B2, tie bars were assembled differently. Bars
having a total length of (tt+2ts) where used. As such, when installed, the tie bars protruded a
distance of 8 mm (5/16”) beyond the steel web plate on each side, and were fillet welded to the
web steel plates.
The approach taken to determine the vertical and horizontal spacing of the tie bars, w1 and w2,
respectively, such as to similarly prevent the skin plates from experiencing local buckling before
yielding, was to consider the plate as similar to the case where a compression flange in
concrete filled section is subjected to uniform compression and have simply supported edges
for which Wright (1995) showed that the S/t ratio should be less or equal to 37 to ensure ductile
behavior. Also, as an alternative and simpler approach, considering the plate as an idealized
fixed ends column and setting the critical stress equal to the yield nominal yield stress of the
web plate, a S/t value of 43 was obtained. Note that the limit would be 25 according to Zhang et
al. (2014) based on their review of the effect of shear connectors design on composite action in
such walls and the maximum steel plate slenderness needed to prevent local buckling of the
steel plates prior to yielding.
For the CFSSP-Walls tested in this research, for a maximum 304.8 mm (12 in.) spacing of tie
bars and web plates thickness of 7.94mm (5/16 in.), the corresponding maximum S/t ratio used
for the specimens was 38.4. This is approximately equal to the above value for plates having
simply supported edges.
The diameter of the tie bars should be selected such that they can provide adequate stiffness to
control local buckling of the web plates, resist the shearing force transferred between the
reinforced concrete core and the steel skin plate, and have adequate strength to resist the
tensile force that develops during formation of the plastic mechanism created during inelastic
buckling of the web skin plate.
Tensile force in the tie was calculated from the yielding of the web skin plate during buckling
after the CFSSP-Wall has attained its plastic moment capacity. When the web plates buckle,
plastic hinging of the skin plates forms as shown in Figure 2. Due to formation of this plastic
mechanism, plastic moment capacity of the plates is reached at the plastic hinges locations.
Although no shear forces proper develop at the end of each buckled segment, and analogy can
be made to convert the axial forces in equilibrium into equivalent shear forces that induce
tension forces on the tie bars. Accordingly, using this approximation, the minimum diameter of
the tie bar was calculated by the following equation:
 w   Fy , plate 
d min = 1.59ts  2   

 w1   Fy ,tie⋅bars 
Ties diameter was then selected to be conservatively greater than the value obtained by this
equation, as no guidance existed at the time the specimens were designed and the authors
wished to prevent tie-failure in this experimental program (past experiments have showed the
undesirable rapid wall strength and stiffness degradation (e.g., Ramesh et al. 2013) that occurs
when ties fail). Note that an adapted version of this equation eventually provided part of the tie
design requirements specified in Section H7 of AISC-341-16. The welding between the tie bars
and the skin plates (either plug welds or fillet welds, depending on the specimen) was sized to
resist the same calculated tensile force in the tie bars.
The concrete used for the four specimens was self-consolidating regular strength (27.6 Mpa;
4 ksi) concrete (SCC), with a slump of 76.2 mm. The steel used for the HSS was ASTM A252
Gr 3 and for the steel web plates was ASTM A572 Gr.50. Detailed information on material
properties obtained from steel coupons and concrete cylinders are provided in Alzeni and
Bruneau (2014).
Figure 2. Forces exerted on tie bars due to web local buckling
TEST SET-UP AND LOADING PROTOCOL

The tested specimens were 3048 mm (120”) tall cantilever type walls (providing an aspect ratio
of 2.5) fixed to a reinforced concrete base, itself connected to the strong floor of the lab using
pre-tensioned DYWIDAG bars. Details on the wall connection to the footing and actuator
connection details are provided in Alzeni and Bruneau (2014). Lateral loading was applied at
the top of the specimens by a servo-controlled actuator. Two bracing trusses were used to
avoid out-of-plane motion of the wall. Specimens were subjected to quasi-static displacement
cycles following the ATC 24 loading protocol, with three displacement cycles at each
displacement step, up to a displacement equal to three times the yield displacement, and two
cycles at each displacement magnitude after that. The yield displacement, δy, was initially
defined here as the displacement at which the extreme steel fibers of the CFSSP-Wall skin
plate starts to yield. The values of δy were obtained from push over analyses done using the
finite element models (in ABAQUS), considering a fixed base model. Information on calibration
of the finite element model is provided in Alzeni and Bruneau (2014). The specimens were
subjected to cyclic displacements per this protocol until substantial fracture in the steel web
plates and the HSS part of the CFSSP-Wall specimens occurred.
Figure 3: Test set-up
EXPERIMENTAL RESULTS
Only minor differences in response were observed in the cycling inelastic response of the 4
specimens. Observations are summarized here for specimen CFSSP-NB1, representative of
typical behavior. Thorough description of the detailed hysteretic response of all specimens is
presented in Alzeni and Bruneau (2014, 2017).
At 0.6% drift, strain gages on the specimen indicated the onset of yielding. Some flaking of
white wash paint was noted, especially in parts of the HSS. There was no visual evidence of
local buckling. At 1.2% drift the entire base of the wall seemed to be yielding, but there was still
no observed local buckling. Local buckling was first observed at 1.8% drift. This buckling was
located on the web plate of the specimen, on one side of the wall, between the first and the
second row of tie bars from the base. The half HSS part of the section did not buckle at this
drift level. At 2.4% drift, the half HSS end of the cross-section developed local buckling at a
location equivalent to between the first and the second row of ties above the footing (Figure 3).
Minor fractures appeared in some of the plug welds on the first row of tie bars.
At 3%, some of the tie bars plug welds fractured around the perimeter of the tie bars. During
cycling, fractures also started to propagate from the circumference of the first tie bar in the first
row towards the half HSS. The crack propagated by about 3/4 of an inch. At 3.6% drift, a first
crack developed in the HSS part of the cross-section, on the face of the buckled wave in the
half HSS. Testing stopped after a few cycles at this 3.6% drift value, as the wall fractured over
18 inches of its base, in a plane passing through the center line of the first row of tie bars
(Fig. 4).
Figure 3: The Local Buckling of the HSS Part at 2.4% Drift
Figure 4: Fracture of the CFSSP-NB1 Specimen East Side at 3.6% Drift
The experimentally obtained lateral load versus top displacement relationships for all
specimens are shown in Figure 5. The corresponding yield displacement, δy, maximum
displacement, maximum strength compared to Mp calculated based on coupon values, and
dissipated energy are summarized in Table 1. The displacement ductility reported in that Table
was calculated based on a definition of δy established by considering an equivalent bilinear
system. The maximum displacement was defined as the post-peak displacement where the
strength degrades to 80% of the peak value. Results indicate that the wall exhibited ductile
behavior up to significant drifts (considering that walls are stiff structural elements expected to
develop relatively small drifts during an earthquake). Experimental results (in Table 2) also
indicated that all wall tested were able to develop their plastic moment computed using cross-
section analysis and actual material properties.
Note that for the B-type specimens, larger drifts were required to produce the same limit states.
Significantly, for the specimen CFSSP-B2 having filled-welded ties: cracks initiating along the
plug welds was first observed at 3.3% drift, with major fracture and progressive loss of strength
developing around 4% drift.
(a) CFSSP-NB1 (b) CFSSP-NB2
(c) CFSSP-B1 (d) CFSSP-B2
Figure 5: Experimentally Obtained Hysteretic Behavior of Specimens
Table 1. Evaluation of Test Data
Max Max.drift Yield.displ. Ductility µ,(1st yield) µ,(bi-linear) Dissipated
Energy
Specimens Δu Δu/h (%) Δy (mm) Δu/Δy Φmax/Φy1 Φmax/Φyb (kN-m)
CFSSP-NB1 104.1 3.4% 20.3 5.13 14.44 7.74 1349
CFSSP-NB2 93.2 3.0% 25.4 3.67 12.88 6.54 1821
CFSSP-B1 121.9 4.0% 22.9 5.33 14.98 8 1595
CFSSP-B2 127 4.16% 22.9 5.50 11.06 5.55 1965
* At the point beyond Mmax where strength has dropped to 0.8Mmax
Table 2. Load Carrying Capacity of the CFSSP-Walls

Max Estimated, Ratio Drift % Drift %
Specimens Mmax (kN- (kN-m) Mp Mmax
CFSSP-NB1 41330.9 36200.2 1.13 1.20 1.80
CFSSP-NB2 41352.2 36200.2 1.13 1.20 1.80
CFSSP-B1 38042.1 30998.2 1.22 1.0 2.0
CFSSP-B2 38749.2 30998.2 1.25 1.0 2.67
LOCAL BUCKLING AND FAILURE MODES

All specimens developed a similar ultimate behavior, with local buckling developing during
ductile response while sustaining a lateral load with minimal strength degradation up to large
drifts. Fracture in all specimens developed upon repeated cycles of local buckling of the steel
web plate and of the HSS, and accelerated by fracture of the welding between the tie bars and
skin plate. However, in all cases, the vertical weld between the steel webs and the (half or full)
HSS sections remained intact. Note that, in all cases, the cracks that developed in the metal
after the development of local buckling only propagated upon development of larger drifts. This
made it possible for the specimens to survive up to the large drifts recorded. Arguably, whether
the crack propagation behavior would be different under dynamic loading is unknown at this
time and could be the subject of future research. However, evidence collected during large-
scale experiments of moment-resisting connections that fractured during the Northridge
earthquake (Bruneau et al. 2011) suggested that result from dynamic response might not be
significantly different from that those from pseudo-static testing. Review of the videos recorded
during the experiments showed that for Specimen CFSSP-B2, fracture started independently on
the HSS at a different location, and that the wall would have failed at the same drifts even if
fracture had not developed slightly earlier at the ties. This indicates that using a different
welded or bolted detail for the tie bar to skin plate connection would not have significantly
improved the behavior of the tested walls beyond what was obtained for specimen CFSSP-B2
(provided that the tie bars in all cases are designed to have adequate strength and stiffness).
DUCTILITY EVALUATION
The four of the tested specimens were able to sustain their strength up to drifts higher than 3%.
Displacement ductility ratio, µD, was calculated based on the ratio between the maximum
displacement, Δmax (corresponding to 80% of the peak load) and the yield displacement, Δy
(calculated from the idealized elastic-perfectly plastic envelope curve). Table 1 presents the
ductility ratio values for the tested specimens. Displacement ductility exceeded 5.0, except for
specimen CFSSP-NB2 which reached a value closer to 4 (i.e., 3.67). Also, curvature ductility
for the tested specimens was also calculated from the experimentally recorded rotations, θ.
Two curvature ductility values were calculated; one based on yield curvature defined from the
idealized elastic-plastic envelope of the resulting moment curvature hysteretic curves; another
based on yield curvature defined by first yield at the extreme fiber of the HSS boundary
element. Further details on curvature calculation are presented in Alzeni and Bruneau (2014,
2017).
LOAD-CARRYING CAPACITY
The plastic moment capacity of the CFSSP-Walls was calculated using simple plastic theory
principles, considering that the steel parts of the cross-section has fully yielded on both the
tension and compression sides of the plastic neutral axis, and that the concrete on the
compression side has reached fc’ (neglecting any reduction factors), as shown in Figure 6. The
closed-form expressions that were derived to calculate the corresponding plastic moments
(Alzeni and Bruneau 2014, 2017) have been included in AISC-341-16, and match the stress
blocks below. No factors were used to modify this calculated nominal plastic strength.
(a) CFSSP-NB cross-section (b) CFSSP-B cross-section
Figure 6. Schematic Diagram for stress distribution used to calculate Mp
Four cantilever CFSSP-Walls, with and without boundary elements, were tested in order to
evaluate their seismic performance. In walls without boundary elements, the cross-section was
closed at its ends by half HSS sections; full HSS columns were used at the ends of the cross-
section for walls with boundary elements. HSS sections met the AISC compactness
requirements for moderate ductility, and spacing of ties was limited to 38 times the web plate
thickness.
The tested walls were able to attain their plastic moment capacity, Mp, and all exhibited a ductile
behavior through cyclic excursions with limited strength degradation up to 3% lateral drift. Local
buckling of the steel web plates and/or the round HSS part of the cross section occurred only
after the tested walls reached their Mp. Behavior of the specimens having a closer tie spacing
of 25 times the web plate thickness was not significantly different. Note that Mp here was
calculated by assuming that the steel has reached Fy in tension and compression on respective
sides of the wall’s neutral axis, and that the concrete over the entire compressed part of the wall
has reached fc’ (using values from steel coupons and concrete cylinders).
In all of the tested walls, fracture started in the steel web, after local buckling, typically initiating
at the location of the connection between the tie bar and the web plate and propagating to the
rest of the wall. In specimen CFSSP-B2, that had tie bars fillet welded to the skin plates and
that exhibited an improved ductility, fracture simultaneously initiated in the HSS at a location
independent from the steel web, which suggests that methods to connect the tie bars that
would further delay fracture there would not, alone, enhance ductile behavior beyond that
observed for specimen CFSSP-B2.
ACKNOWLEDGMENTS
This work was supported by the AISC (American Institute for Steel Construction), and Egyptian
Ministry of Higher Education. However, any opinions, findings, conclusions, and
recommendations presented in this paper are those of the authors and do not necessarily
reflect the views of the sponsors.
REFERENCES
AISC-341 (2016). "Seismic Provisions for Structural Steel Buildings." American Institute of Steel
Alzeni, Y., Bruneau, M. (2014). “Cyclic Inelastic Behavior of Concrete Filled Sandwich Panel
Walls Subjected to in plane Flexure.” Technical report MCEER, 14-009, Buffalo, NY.
Alzeni, Y., Bruneau, M., (2017). “In-Plane Cyclic Testing of Concrete Filled Sandwich Steel
Panel Walls with and without Boundary Elements”, ASCE Journal of Structural Engineering,
Vol.143, No.9.
Bruneau, M., Uang., C.M., Sabelli, R. (2011). “Ductile Design of Steel Structures - 2nd Edition”,
McGraw-Hill, New York, NY, 921p.
Bowerman, H., Gough, M., and King, C. (1999). “Bi-Steel Design and Construction Guide”,
British Steel Ltd, Scunthorpe, U.K.
El-Bahey, S., and Bruneau, M. (2011). "Bridge piers with structural fuses and bi-steel columns.
I: Experimental testing." Journal of Bridge Engineering, 17(1), 25-35.
Eom, T.-S., Park, H.-G., Lee, C.-H., Kim, J.-H., and Chang, I.-H. (2009). "Behavior of double
skin composite wall subjected to in-plane cyclic loading." Journal of Structural Engineering,
135(10), 1239-1249.
Epackachi, S., Nguyen, N.H.,Whittaker, A.S.,and Varma, A.H.(2014).”In-plane seismic behavior
of rectangular steel-plate composite wall piers” Journal of Structural Engineering, 04014176(9),
1-9.
Ge, H. and Usami, T. (1996). ”Cyclic Tests of Concrete-Filled Steel Box Columns.” J. Struct.
Eng., 122(10), 1169–1177.
Kawashima, K., Unjoh , S. (1997). “The Damage of Highway Bridges in the 1995 Hyogo-Ken
Nanbu Earthquake and its Impact on Japanese Seismic Design”, Journal of Earthquake
Engineering, Vol. 1, No. 3, Pp. 505-541
Kurt, E.G., Varma, A.H., Booth, P.N., and Whittaker, A., (2016). "In-plane Behavior and Design
of Rectangular SC Wall Piers Without Boundary Elements." ASCE Journal of Structural
Engineering, J. Struct. Eng. Vol. 142, Issue 6, 10.1061/(ASCE)ST.1943-541X.0001481,
04016026.
Hossain, K. A., and Wright, H. D. (1998). "Performance of profiled concrete shear panels."
Journal of Structural Engineering, 124(4), 368-381.
Ramesh, S., Kreger, M.E., Bowman, M.D., (2013). “Behavior and Design of Earthquake-
ResistantDual-Plate Composite Shear Wall Systems”, Report submitted to Charles Pankow
Foundation, http://www.pankowfoundation.org/grants.cfm?prodonly=1 (project RGA#06-06).
Sener, K., Varma, A.H., and Ayhan, D. (2015). "Steel-Plate Composite SC Walls: Experimental
Database and Design for Out-of-Plane Flexure." Journal of Constructional Steel Research, Vol.
108, May, pp. 46-59, Elsevier Science.
Seo, J., Varma, A.H., Sener, K., and Ayhan, D. (2016). "Steel-Plate Composite (SC) Walls: In-
Plane Shear Behavior, Database, and Design." Journal of Constructional Steel Research,
Elsevier Science, Volume 119, Pages 202-215.
Usami, T. and Ge, H. (1994). ”Ductility of Concrete‐Filled Steel Box Columns under Cyclic
Loading.” J. Struct. Eng., 120(7), 2021–2040.
Wright, H. (1995). "Local stability of filled and encased steel sections." Journal of Structural
Engineering, 121(10), 1382-1388.
Wright, H. (1998). "Axial and Bending Behavior of Composite Walls." Journal of Structural
Engineering, 124(7), 758-764.
Zhang, K., Varma, A.H., Malushte, S., and Gallocher, S. (2014). "Effects of Shear Connectors
on the Local Buckling and Composite Action in Steel Concrete Composite Walls." Nuclear
Engineering and Design, Special Issue on SMiRT-21 Conference, Vol. 269, pp. 231-239,
Elsevier Science.
Cyclic Shear Behavior of Steel-Plate Composite Walls for High-Rise
Buildings
Xiaodong JI1, Xiaowei CHENG2, Xiangfu JIA3
Tsinghua University, Beijing 100084, CHINA ,E-mail: jixd@mail.tsinghua.edu.cn
1
2
Tsinghua University, Beijing 100084, CHINA, E-mail: cxw15@mails.tsinghua.edu.cn
3
Tsinghua University, Beijing 100084, CHINA, E-mail: jiaxiangfu@foxmail.com
ABSTRACT
Steel-plate composite walls are composed of steel plate with reinforced concrete encasement on
both sides (referred to as “single-plate composite (SPC) walls”) or a concrete infill sandwiched in
between two steel faceplates (referred to as “double-skin composite (DSC) walls”). This paper
presents a series of experimental tests, including 2 SPC and 3 DSC wall specimens, conducted
to investigate cyclic shear behavior of the composite walls used for high-rise buildings. All wall
specimens had a shear-to-span ratio of 1.2, and were designed with a reinforcement ratio of
approximately 6.0%. The specimen was designed with flange walls to enforce in-plane shear
failure in the web wall. The test results indicate that the DSC walls had the shear deformation
capacity 20% larger than the counterpart SPC walls, as the faceplates prevented spalling of
crushed concrete. In addition, shear strength design formulas specified in various codes are
verified using a large volume of test data.
Keywords：Composite walls; Seismic applications; Experimental; Cyclic shear behavior; Shear

strength; Design formulas
INTRODUCTION
With quick urbanization in past two decades, a large number of high-rise buildings have been
constructed in China. Most of the super-tall buildings adopt core wall-frame-outrigger system or
core wall-mega frame system, where the core walls carry a large portion of base shear forces and
overturning moments induced by earthquake actions. To meet the extremely large shear force
demand and ensure the flexural ductility of the walls in the lower stories, steel-plate composite
walls are developed as a promising alternative to reinforced concrete (RC) walls.
Two types of steel-plate composite walls have seen increasing use for high-rise buildings in China.
As shown in Figure 1, the first type is composed of steel plate with reinforced concrete
encasement on both sides, referred as “single-plate composite (SPC) walls”, and another type
consists of a thick concrete infill sandwiched by two steel faceplates on the exterior surfaces,
referred as “double-skin composite (DSC) walls”. Use of steel plate significantly increases the
shear strength of the walls, and thus translates into thinner walls with resulting greater usable
space. A statistical analysis of steel-plate composite walls for a large number of high-rise buildings
indicates that most composite walls are designed with a relatively high steel reinforcement ratio
of 6 to 8%. Besides, those walls are usually subjected to a high axial force ratio due to large
gravity force of super-tall buildings.
Distributed rebar Boundary longitudinal rebar
Boundary Steel Headed Crosstie Profiled Tie bar Headed stud Steel plate
tranverse plate stud steel
rebar
(a) Single-plate composite (SPC) wall (b) Double-skin composite (DSC) wall
Fig. 1 Typical cross-section of steel-plate composite walls
Recently, a number of experimental tests and numerical studies (e.g., Eom et al. 2009, Ji et al.
2013, Nie et al. 2013, Wang et al. 2017) have been conducted on the steel-plate composite walls
for high-rise buildings. While these studies focus on the in-plane flexural behavior of slender
composite walls, experimental data of the cyclic in-plane shear behavior of the composite walls
used for high-rise buildings is limited.
This paper presents a series of quasi-static tests used to examine the cyclic shear behavior of
steel-plate composite walls. Section 2 describes the experimental program. The test results are
detailed in Section 3. Finally, this study evaluates design equations for calculating the in-plane
shear strength of the composite walls from various codes by comparing them with test data.
TEST SPECIMENS
The test specimens were designed to represent the structural walls in the lower stories of a high-
rise building, and were fabricated at approximately one-third scale to accommodate the capacity
of the loading facility. A total of five wall specimens were tested, including two single-plate
composite walls (labeled as SPCW1 and SPCW2) and three double-skin composite walls (labeled
as DSCW1 through DSCW3). The specimens were designed with an I-shaped section, where the
web wall primarily resists in-plane shear and the flange walls resist the overturning moment. The
in-plane flexural strength of the wall specimen was designed to significantly exceed its in-plane
shear strength, which enforces pure in-plane shear behavior and failure in the web wall. Figure 2
shows the overall geometry, structural layout, and reinforcement details for the specimens. The
specimens were cast integrally with a large foundation beam and a loading beam. As shown in
Figure2(a), the flange walls were intentionally disconnected from the top beam to limit the possible
shear contribution induced by the secondary bending moments of the flange walls and for easy
installation of loading threaded bars. All steel plates and rebar were securely anchored to the
foundation beam, and those in the web wall were extended into the top beam.
As shown in Figure 2(b), the SPC web wall have steel plate thickness, tp, equal to 6 mm for the
specimens. The steel reinforcement ratio, tp/T, of the SPC web wall is equal to 4.9%, where T
denotes the thickness of web wall. The steel plate was connected to the I-shaped steel embedded
at its two edges by complete-joint-penetration (CJP) groove welds. Headed studs (diameter = 8
mm, length = 40 mm) were used to provide shear transfer between steel plate and RC
encasement. D6 (diameter = 6 mm) steel rebar were placed as horizontally and vertically
distributed reinforcement of the wall web with a spacing of 100 mm, corresponding to a
reinforcement ratio of 0.9%. Thus, the total steel reinforcement ratio of SPCW web wall was 5.8%.
Heavy longitudinal reinforcement was intentionally designed in the boundary elements and flange
walls to ensure large flexural strength of the wall specimens. The boundary transverse rebar and
web crossties were threaded through holes on the steel plate.
1
1000 350
Vertical rebar
Top beam 12D28
350
200
Distributed Rebar Tie bar
Vertical rebar D8@200
D6@100
4D16
200
850
Flange wall Web wall Shaped steel
120
120 I 83×45×8×8
Steel web Headed stud
Steel hoop plate t6 D8 @100
200
D8@100
Foundation beam
800
140 570 140

875 850 875 800
1 1-1
(a) Overall geometry dimension (b) Cross section of SPCW1 & SPCW2
Tie plate Horizontal stiffener Tie batten Horizontal stiffener
B20-t4@100 B15-t4@100 B20-t4@100 B15-t4@100
200
Headed stud Vertical stiffener Tie batten 200

Tie bar
D8@200 D8 @100 B25-t4 B20-t4@80
Fillet weld Fillet weld
120
120
Flange U shaped bar Web Flange U shaped bar Web

200
200
faceplate t8 D8@120 faceplate t4 faceplate t8 D8@120 faceplate t4
120 610 120 120 610 120
(c) Cross section of DSCW1& DSCW2 (d) Cross section of DSCW3

Fig. 2 Geometry and reinforcement of wall specimens
The DSC web wall had faceplate thickness, tp, equal to 4 mm for those specimens. The steel
reinforcement ratio, 2tp/T, of the DSC web wall was equal to 6.4%. As shown in Figure 2(c),
Specimens DSCW1 and DSCW2 used a typical combination of shear studs and tie bars, to
develop composite action between steel faceplates and concrete infill, and to prevent local
buckling of steel faceplates. Table 1 presents the details of the connections. The faceplate
slenderness ratio (i.e., the ratio of the stud spacing s over the steel faceplate thickness tp) was
equal to 25, which satisfies the AISC N690s1-15 slenderness limit of 1.0 E / fy for non-slender
steel faceplates to prevent local buckling before yielding. The steel faceplates of the DSC flange
walls were intentionally thickened to 8 mm to ensure large flexural strength of the overall wall
specimen. The web faceplates were welded to the flange faceplates using CJP groove welds. U-
shaped rebar was used to provide the bond shear strength along the interface between the steel
plate of the flange wall and the concrete infill of the web wall by the shear-friction mechanism (Ji
et al. 2013).
Specimen DSCW3 was identical to Specimens DSCW1 and DSCW2, with the exception that it
used a novel connector for developing composite action and for tying the faceplates together, as
shown in Figure 2(d). Vertical stiffeners were welded to the faceplates using fillet welds and they
were connected to each other using a series of tie battens (Nie et al. 2013). The ratio of the
stiffener spacing (s) over the steel faceplate thickness (tp) was 37.5, which is greater than the
AISC N690s1-15 slenderness limit.
Table1 Design parameters of wall specimens
Plate Steel
Stud/stiffener Tie bar/plate Total axial Axial force
Spec. no. slenderness reinforcement
spacing (mm) spacing (mm) force (kN) ratio n
ratio ratio
SPCW1 100 — 16.7 5.8% 1243 0.16
SPCW2 100 — 16.7 5.8% 2486 0.29
DSCW1 100 200 25 6.4% 1389 0.16
DSCW2 100 200 25 6.4% 2779 0.31
DSCW3 150 80 37.5 6.4% 2779 0.31
The strength grade of the concrete in the wall specimens is C50, with the nominal cubic
compressive strength fcu equal to 50 MPa. The cubic strength of concrete fcu,t, measured on the
day of specimen testing, was equal to 56.0 and 63.8 MPa for Specimens SPCW1 and SPCW2,
and 47.6, 52.7 and 50.5 MPa for Specimens DSCW1 through DSCW3, respectively. The axial
compressive strength of the concrete was taken as fc,t = 0.76fcu,t according to the Chinese code.
All the steel plates used for the specimens had a strength grade of Q235 (the nominal yield
strength fy = 235 MPa). All rebar had a strength grade of HRB400 (fy,d = 400 MPa). Table 2
summarizes the properties of the steel plate and rebar measured from standard coupon tests and
standard rebar tensile tests.
Table 2 Material properties of steel plate and rebar
Plate or Plate thickness tp or Yield strength fy,t Ultimate strength Elongation δ
rebar diameter D(mm) (MPa) fu,t (Mpa) (%)
6 281.3 424.9 30.2
Plate
8 302.1 450.8 37.3
8 424.0 636.0 9.5
Rebar 16 442.2 610.7 10.0
28 445.0 623.2 10.6
Note: The elongation of steel plate was measured after rupture along a 50-mm gauge length
including the fracture zone, while the elongation of rebar is the uniform elongation (i.e., the
measured strain corresponding to the peak stress of the rebar).
In accordance with the Chinese technical specification for concrete structures of tall building JGJ
3-2010, the axial force ratio for composite walls can be calculated using Eq. (1).
N
(1) n=
f c Ac + ∑ f y As + ∑ f p Ap
where, n denotes the axial force ratio, N denotes the axial force applied on the wall, fc denotes
the axial compressive strength of the concrete, fy and fp denote the yield strength of embedded
steel profile and steel plate, respectively, and Ac, As and Ap denote the cross-sectional areas of
concrete, steel profile and steel plate, respectively.
As the flange walls of the specimens were disconnected from the top beam, the axial force was
applied primarily to the web wall in the upper portion of the specimen. However, the axial force
could generally spread and transfer to the flange walls in the lower portion of the specimen.
Preliminary finite element analysis indicated that the web wall resisted an average of
approximately 77% of the applied axial force. The axial force ratio for the web walls was calculated
using Eq. (1), and the approximate proportion (77%) of the applied axial force, the measured
dimensions of the web wall and the actual strength of steel and concrete materials. Specimens
SPCW1 and DSCW1 had a moderate axial force ratio of 0.16, while Specimens SPCW2, DSCW2
and DSCW3 had a high axial force ratio of 0.29.
TEST SETUP AND INSTRUMENTATION
Figure 3 shows the test setup. The axial force was first applied and maintained constant during
the test. The lateral cyclic loading was applied in displacement-control using two actuators. The
first three cycles were applied in the elastic range of response. Three levels of drifts were included,
i.e. 0.05%, 0.1% and 0.2%, and one cycle was performed at each level. After the specimen
reached the estimated yield drift of 0.4%, the lateral displacement was increased with increments
of 0.4% drift, and two cycles were repeated at each drift level. The test was terminated when the
specimen failed completely due to crushing of the concrete.
Loading frame
Horizontal Vertical actuator

actuator
d1
d5 d2 d7
b
d3 d4
d9 d6 d8 d10
d11
Foundation beam
(a) Sketch of test setup (b) Photograph of test setup

Fig. 3 Test setup
Instruments were used to measure loads, displacements and strains in the specimen. Figure3 (a)
shows the layout of linear variable differential transformers (LVDTs) mounted on the specimens.
Strain-gauge rosettes were mounted on the steel plate and steel profile to measure the strains.
The strains in a few horizontally and vertically distributed rebar, boundary longitudinal rebar, shear
studs and tie bars are also monitored.
HYSTERETIC RESPONSES
Using the data measured by a pair of diagonal LVDTs d3 and d4 (see Figure 3(a)), the in-plane
shear deformation Δsh and average shear strain γ of the web wall were estimated as follows
(2) Δsh =(δ 3 - δ 4 ) a 2 + b 2 / 2ab
(3) Δsh b
γ =/
Where, δ3 and δ4 denote the deformation measured by LVDTs d3 and d4, and a and b denote the
clear length and height of the web wall as shown in Figure 3(a).
Figure 4 shows the hysteresis curves of in-plane shear force versus the average shear strain
calculated by Equations (2) and (3). Figure 5 shows the envelope curves of the specimens. As
the faceplates of the DSC wall specimens can prevent the spalling of the concrete, they showed
more ductile behavior than the SPC wall counterparts. As Specimens SPCW2 and DSCW2 were
subjected to higher axial force, they showed a faster decrease of the post-peak strength than
Specimens SPCW1 and DSCW1, respectively. Specimen DSCW3 that used vertical stiffeners
and tie plates for connecting two faceplates exhibited stable hysteretic loops up to very large
shear strain of approximately 3%.
2500 2500 2500
Cracking Cracking Yielding
Yielding Yielding Buckling
1500 1500 1500
Shear force (kN)

Shear force (kN)
Shear force (kN)
500 500 500
-500 -500 -500
-1500 -1500 -1500
-2500 -2500 -2500

-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
In-plane shear strain In-plane shear strain In-plane shear strain
(a) SPCW1 (b) SPCW2 (c) DSCW1

2500 2500 2500
Yielding Yielding SPCW1
Buckling Buckling SPCW2
DSCW1
In-plane shear force (kN)

1500 1500 1500
DSCW2
DSCW3
Shear force (kN)
Shear force (kN)
500 500 500
-500 -500 -500
-1500 -1500 -1500
-2500 -2500 -2500

-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03
In-plane shear strain In-plane shear strain In-plane shear strain
(d) DSCW2 (e) DSCW3 Fig. 5 Envelope curves of

lateral force versus in-plane
Fig. 4 Hysteretic loops of lateral force versus in-plane shear
shear strain of specimens.
strain of specimens.
DAMAGE AND FAILURE MODE

Analysis of the data from the strain-gauge rosettes indicated the yielding of steel plates occurred
at approximately 0.20 to 0.25% shear strain. For the SPC wall specimens, a large number of
crisscross diagonal cracks induced by cyclic shear loads developed and subdivided the web wall
into a series of concrete segments separated by inclined cracks. The cyclic reversal led to spalling
of some concrete segments at approximately 0.7%, followed by buckling of steel rebar and steel
plate. For the DSC wall specimens, local buckling of faceplates was observed at 1.8% shear strain
for Specimen DSCW1 and at 1.2% for Specimens DSCW2 and DSCW3. In the end, the concrete
infill crushed at the location where the faceplates buckled. More details of the damage and failure
of the DSC wall specimens can be found in Ji et al. 2017. Figure 6 shows the failure mode of the
wall specimens. Note that the damage of concrete was mainly concentrated in the upper portion
of the web wall where the axial force was larger than the lower portion.
(a) SPCW1 (b) SPCW2
(c) DSCW1 (d) DSCW2 (e) DSCW3

Fig. 6 Failure photographs of the wall specimens
SHEAR STRENGTH AND DEFORMATION CAPACITIES

Table 3 summarizes the shear strength and deformation capacities of the specimens. The yield
load Vy corresponds to the yielding of steel plates as measured by the strain-gauge rosettes. The
yield shear strain γy is the average shear strain measured by diagonal LVDTs corresponding to
the yield load Vy. The ultimate shear strain γu is defined as the post-peak shear strain
corresponding to the lateral load that is 85% of the peak load. As Specimen DSCW3 did not show
obvious strength degradation until complete failure, its ultimate shear strain was defined to be the
maximum level of shear deformation sustained for at least one full cycle of loading prior to failure
of the wall. The ductility factor is calculated as μγ＝γu/γy.
The following observations are made from Table 3. (1) The ultimate shear strain of Specimens
DSCW1 and DSCW2 was 36% and 34% larger than that of Specimens SPCW1 and SPCW2,
respectively. The increased shear deformation capacity of the DSC walls is attributed to the fact
that the faceplates can prevent the spalling of the concrete and offer somewhat confinement to
the infilled concrete. (2) The increase of the axial force ratio from 0.16 to 0.30 led to a slight
increase of the maximum shear strength by 5 to 10%, while it resulted in a significant decrease
of the ultimate shear strain by 27% for both SPCW and DSCW specimens. (3) The vertical
stiffeners and tie plates appeared to provide more effective restraint to the faceplates, and
consequently DSCW3 had a significantly larger ultimate shear strain than other specimens.
Table 3 Shear strength and deformation capacities of wall specimens
Ultimate
Spec. Yield shear Yield shear Peak shear Peak shear Ductility
shear
No. strainγy (%) strength Vy (kN) strainγp /% strength Vp (kN) ratio μγ
strainγu /%
SPCW1 0.21 1435 0.63 2021 1.13 5.4
SPCW2 0.17 1676 0.55 2260 0.83 4.9
DSCW1 0.27 1878 0.93 2212 1.54 5.7
DSCW2 0.22 1797 0.68 2306 1.12 5.1
DSCW3 0.26 1735 1.25 2387 3.09 11.9
DESIGN FORMULAS OF SHEAR STRENGTH

DESIGN FORMULAS FOR SINGLE-PLATE COMPOSITE WALLS
Per the AISC 341-10 provisions, the shear strength of SPC walls is estimated as the yield shear
strength of the steel plates only, neglecting the shear contribution provided by RC encasement,
which is given by
(4) V = 0.6f p Ap
Where, fp denotes yield strength of steel plate, Ap denotes cross-section area of steel plates. Note
that, the web of boundary steel profiles is taken as a portion of steel plate in the calculation of
shear strength.
Chinese technical specification for concrete structures of tall buildings (JGJ 3-2010) estimates the
in-plane shear strength of SPC walls by superposing the in-plane shear strength of steel plates
Vs and shear strength of RC encasement Vc, given by
(5-a) V =+
Vs Vc
1 A
(5-b) Vc = ( 0.5f tbwhw0 + 0.13N ) + f yv sh hw0
λ - 0.5 s
(5-c) 0.6
Vs = f A
λ - 0.5 p p
In these equations, ft denotes tensile strength of the concrete; fyv denotes yield strength of
horizontally distributed rebar; fp denotes yield strength of steel plate; bw denotes thickness of the
web wall; hw0 denotes effective depth of the wall; Ash denotes cross-sectional area of the
horizontally distributed rebar within spacing s; Ap denotes cross-sectional area of the steel plate
(the web of boundary steel profiles is taken as a portion of steel plate in this calculation); λ denotes
shear-to-span ratio of the wall (the value of λ is assumed to be equal to 1.5 if it is smaller than
1.5); and N denotes axial compressive force applied to the wall web.
A number of data from past tests (Sun et al. 2008; Yang et al. 2010) and from this test program
was used for validation of shear strength design formulas. A total of 12 SPC wall specimens are
considered, of which the steel reinforcement ratio varies from 4.1 to 6.2%, and the axial force ratio
varies from 0.16 to 0.29. Figure 7 shows the comparison between the calculated values per the
design equations and the test results. It indicates that the JGJ 3-2010 code equations provide a
reasonable estimation on the in-plane shear strength of SPC walls. The mean and standard
deviation for the ratio of Vtest/Vn equal to 1.10 and 0.21, respectively. However, the AISC 341-10
equation significantly underestimates the in-plane shear strength capacity of SPC walls, as it
neglects the shear contribution of RC encasement.
8000 8000
Sun et al. Vtest /VnAISC: Mean= 2.40 Sun et al. Vtest /Vn JGJ: Mean= 1.10
Yang et al. Standard deviation=0.59 Yang et al. Standard deviation=0.21
6000 This study 6000 This study

AISC code Vn AISC(kN)
JGJ code VnJGJ(kN)

4000 4000
2000 2000
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
Test value VTest(kN) Test value VTest(kN)
(a) AISC 341-10 (b) JGJ 3-2010

Fig. 7 Verification for design formulas of shear strength for SPC walls
DESIGN FORMULAS FOR DOUBLE-SKIN COMPOSITE WALLS

The design equations in U.S. code AISC N690s1-15 are based on the mechanics-based model
(MBM) theory, and further simplified as shown in Eq. (6-a). In this equation, κ is a calibration
factor calculated using Eq. (6-b). The values of ρ in Eq. (6-b) are calculated using Eq. (7-c), and
they vary from 0.01 to 0.04. In these equations, fc and fp denotes the concrete compressive
strength and the yield strength of steel faceplates in MPa, respectively.
(7-a) VnAISC =
κfp Ap
(7-b) κ =1.11- 5.16 ρ ≤ 1.0
1 f p Ap
(7-c) ρ=
83 Ac f c
The JGJ 3-2010 formulas for estimating the shear strength of DSC walls are similar as those for
SPC walls (Eq. 5), except for removing the shear contribution of horizontally distributed rebar.
Test data on DSC walls are collected from past tests (Akiyama et al. (1991), Takeda et al. (1995);
Takeuchi et al. (1998), Fujita et al. (1998), Ozaki et al. (2001) and (2004), Cao et al. (2013)) and
from this test program. A total of 42 test specimens are considered, of which the steel
reinforcement ratio varies from 1.3 to 6.4%, and the axial compressive force ratio varies from zero
to 0.31. Figure 8 shows the comparison between the calculated values per the design equations
and the test results. The figure indicates both AISC N690s1-15 and JGJ 3-2010 code equations
provide reasonable and conservative assessment of the in-plane shear strength of DSC walls.
8000 8000
Akimaya Akimaya
Cao Cao
This study This study
6000 6000
AISC code Vn AISC (kN)
Ozaki, 2004 Ozaki, 2004
JGJ code VnJGJ(kN)

Takeda Takeda
Takeuchi Takeuchi
4000 Ozaki, 2001 4000 Ozaki, 2001
Fujita Fujita
2000 2000
Vtest /Vn AISC: Mean= 1.29 VTest /VnJGJ: Mean= 1.36
Standard deviation=0.26 Standard deviation=0.18
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
Test value VTest(kN) Test value VTest(kN)
(a) AISC N690s1-15 (b) JGJ 3-2010

Fig. 8 Verification for design formulas of shear strength for DSC walls
CONCLUSIONS
This paper presented a series of quasi-static tests on cyclic shear behavior of steel-plate
composite walls for high-rise buildings. The following conclusions are drawn from this study.
(1) DSC wall specimens have approximately 35% larger shear deformation capacity than SPC
wall specimens, as the faceplates of DSC walls prevent the spalling of the concrete.
(2) Increasing the axial ratio from 0.16 to 0.30 slightly increased the shear strength of the steel-
plate composite walls, with it led to over 20% decrease in the shear deformation capacity.
(3) The DSC wall specimen with vertical stiffeners and tie plates connecting the two faceplates
had the ultimate shear strain reaching 3%, significantly larger than the DSC wall specimens with
headed shear studs and tie bars as connectors.
(4) Analysis of the test data of 12 SPC wall specimens indicates that the JGJ 3-2010 design
formulas provides reasonable estimates of the shear strength of SPC walls, while the AISC 341-
10 formula significantly underestimates the shear strength.
(5) Analysis of the test results of 42 DSC wall specimens indicates that the design formulas of
AISC N690s1-15 and JGJ 3-2010 can provide reasonable and conservative estimates of the
shear strength of DSC walls.
REFERENCE
AISC (American Institute of Steel Construction). (2015). “Specification for safety-related steel
structures for nuclear facilities including supplement no. 1.” AISC N690s1-15, Chicago.
ANSI/AISC (American Institute of Steel Construction). (2010). Seismic provisions for structural
steel buildings. ANSI/AISC 341-10, Chicago.
Akiyama, H., Sekimoto, H., Fukihara, M., Nakanishi, K., & Hara, K. (1991). A compression and
shear loading test of concrete filled steel bearing wall. In Transactions of the 11th international
conference on structural mechanics in reactor technology.
CMC (China Ministry of Construction). (2010). “Technical specification for concrete structures of
tall building.” JGJ 3-2010, Beijing. (in Chinese).
Cao, W., Yu C., Dong, H., Qiao, Q., Li, H., & Zhang, Y. (2013). Experimental study on seismic
performance of composite shear walls with double steel plates under different constructions.
Journal of Building Structures, 34, 186-191 (in Chinese).
Eom, T. S., Park, H. G., Lee, C. H., Kim, J. H., & Chang, I. H. (2009). Behavior of double skin
composite wall subjected to in-plane cyclic loading. Journal of Structural Engineering, 135(10),
1239-1249.
Fujita, T., Funakoshi, A., Akita, S., Hayashi, N., & Matuso, I. (1998). Experimental Study on
concrete filled steel structure: Part16 Bending Shear Tests (effect of bending strength).
Summaries of Technical Papers of Meeting Architectural Institute of Japan,1121-1122.
Ji, X., Jiang, F., & Qian, J. (2013). Seismic behavior of steel tube–double steel plate–concrete
composite walls: Experimental tests. Journal of Constructional Steel Research, 86, 17-30.
Ji, X., Cheng, X., Jia, X., & Varma, A. H. (2017). Cyclic In-Plane Shear Behavior of Double-Skin
Composite Walls in High-Rise Buildings. Journal of Structural Engineering, 143(6): 04017025.
Nie, J., Hu, H. H., Fan, J., Tao, M., Li, S., & Liu, F. (2013). Experimental study on seismic behavior
of high-strength concrete filled double-steel-plate composite walls. Journal of Constructional Steel
Research, 88, 206-219.
Ozaki, M., Akita, S., Niwa, N., Matsuo, I., & Usami, S. (2001). Study on steel plate reinforced
concrete bearing wall for nuclear power plants part1; shear and bending loading tests of SC walls.
Transaction of the 16th International Conference on Structural Mechanics in Reactor Technology,
Washington DC, Paper ID: 1554.
Ozaki, M., Akita, S., Osuga, H., Nakayama, T., & Adachi, N. (2004). Study on steel plate reinforced
concrete panels subjected to cyclic in-plane shear. Nuclear Engineering and Design, 228(1), 225-
244.
Sun, J., Xu, P., Xiao, C., Sun, H., & Wang, C. (2008). Experimental study on shear behavior of
steel plate-concrete composite wall. Building Structure, 38(6), 1-7 (in Chinese).
Takeda, T., Yamaguchi, T., Nakayama, T., Akiyama, K., & Kato, Y. (1995). Experimental study on
shear characteristics of a concrete filled steel plate wall. In Transactions of the 13. International
conference on structural mechanics in reactor technology.
Takeuchi, M., Narikawa, M., Matsuo, I., Hara, K., & Usami, S. (1998). Study on a concrete filled
structure for nuclear power plants. Nuclear Engineering and Design, 179(2), 209-223.
Wang, B., Jiang, H., & Lu, X. (2017). Seismic performance of steel plate reinforced concrete shear
wall and its application in China Mainland. Journal of Constructional Steel Research, 131, 132-
143.
Yang, X. (2010). Experimental research on shear behavior of high-strength concrete steel
composite walls. (Master Dissertation). China Academy of Building Research, Beijing, China. (in
Chinses)
Experimental Study on the Seismic Performance of New Hybrid
Coupling Wall Subassemblies with Replaceable Steel Coupling Beam

Xiamen, China
Hao Lin
Postgraduate of Structural Engineering, College of Civil Engineering, Huaqiao University
Xiamen, China
E-mail: linhao425@163.com
Zixiong Guo
Xiamen, China
Hongsong Hu
Xiamen, China
E-mail: huhs@hqu.edu.cn
Bahram M. Shahrooz
a
b
Xiamen, China
E-mail: shahrobm@ucmail.uc.edu
ABSTRACT
Coupled shear wall is an efficient and promising lateral load resistant system. Significant
accomplishments have been achieved in understanding the seismic performance and load
carrying mechanism of coupled wall systems with different components, e.g., reinforced concrete
(RC) walls with RC coupling beams, or RC walls with steel coupling beams. To further improve
seismic performance and rapid function restoration after earthquakes, a new hybrid system
consisting of steel plate composite structural wall piers and replaceable steel coupling beams
was investigated. Three half-scale subassemblies having different details were designed,
fabricated, and tested under cyclic loading. All three subassemblies used steel coupling beams
with the same span/depth ratios, but had different details. Two subassemblies with replaceable
steel coupling beam components were repaired in situ after being loaded up to 2% chord rotation,
and were subsequently loaded up to 10% chord rotation to study the seismic performance of the
specimens after repair. The results showed that the new hybrid subassemblies exhibited
satisfactory performance in terms of achieving the required strength and stiffness, excellent
ductility, and stable hysteretic behavior. The performance of the repaired specimens was nearly
the same as their counterparts before repair, indicating acceptable fuse details that can be
replaced.
INTRODUCTION
Various experimental programs have been conducted to address the interaction between steel
coupling beams and reinforced concrete structural walls in a hybrid coupled wall system. Past
research showed that properly designed and detailed steel coupling beams provided outstanding
strength, stiffness, ductility, and appreciable energy dissipation. In spite of excellent performance
of steel coupling beams, post-event repair would be costly or not feasible because of significant
inelastic deformations to which coupling beams are subjected. An innovative coupling beam with
a replaceable mid-span “fuse” was studied in a past project [Fortney et al. 2006]. This design
concept is based on concentrating the damage due to seismic loads in specially detailed
components, which can be removed and replaced after earthquakes allowing in situ repair of the
structure. In response to seismic excitation, steel coupling beams are expected to dissipate
energy in a manner similar to the response of shear links in eccentrically braced frames (EBFs).
The idea of a mid-span fuse is analogous to a replaceable link in EBFs [Fortney et al. 2007]. The
basic premise for coupling beams with a mid-span fuse coupling is that all inelastic deformations
are concentrated in the fuse.
The main objective of the research presented in this paper is to further enhance the performance
of steel coupling beams by providing post-event replaceability without costly repair of the wall
pier and the coupling beam connections. This objective was achieved by designing the beam
such that all of the inelastic damage would be concentrated in the middle of the steel coupling
beam, i.e., where the fuse is located. The goal is to restore the resistance of the structures after a
seismic event. This paper is focused on the experimental component of the reported research.
TEST SPECIMENS
Five specimens were fabricated and tested (Table 1). One specimen without a replaceable fuse
was tested as benchmark. The reinforcement details of all the specimens were the same, as
shown in Figure 1. The test variables were the connection details between the fuse and the rest
of the coupling beam that was detailed to remain elastic, see Figure 2. The composite wall piers
consisted of structural steel boundary elements welded to “shear plates”. The steel boundary
elements were H125×125×6×9. The elastic segments of the steel coupling beams were
embedded in the wall piers and welded to encased steel boundary elements. As seen from Table
1, the thickness of flanges, webs, and end plates in the elastic components were made thicker
than those in the fuse in order to prevent damage in the portions that are not meant to be
replaced. High strength 16-mm diameter bolts (Grade 10.9) were used to connect the end plates
of the “fuses” and the elastic components of the coupling beams. The ultimate strength of the
bolts was 1000MPa with the yield strength to ultimate strength ratio of 0.9. Shear stud
connectors were used on both sides of the steel plates and the flanges of the boundary elements
to ensure sufficient bond between the steel and concrete. The diameter and nominal length of
the studs were 16mm and 60mm, respectively.
MATERIAL PROPERTIES
All the specimens were cast at the same time using concrete from a single ready-mix truck. The
average 28-day compressive strength was 45.7MPa, as determined by testing cubes. The
maximum aggregate size was 10mm, selected to ensure good consolidation around the
embedded components. The horizontal reinforcement of the wall piers and stirrups in the loading
beam and pedestal were 6-mm diameter deformed bars. The diameter of the longitudinal
reinforcement in the wall piers was 14mm. The thickness of the steel plate and stiffeners were
6mm. The measured material properties were shown in Table 2.
(a) Elevation view
(b) Plan view

Fig. 1 - Reinforcement details of specimens (all dimensions are in mm)
Elastic components “Fuse” Elastic components “Fuse”

(H-250×125×12×20) (H-125×125×6×9) (H-250×125×12×20) (H-125×125×6×9)
End-plate
stiffener End-plate
stiffener
150 150 150 150

300 300 Flange
stiffener
M16, 10.9 @54 mm M16, 10.9 @54 mm
(a) Replacement detail A (b) Replacement detail B

Fig. 2 - Replacement details of the steel coupling beams
Table 1 - Parameters of specimens

Fuse or Elastic Replacement
Specimen I.D. Span/depth
coupling beam component detail
SCB-2 H250x125x6x9 --- 2.0 ---
FCB-1 H250x125x6x9 H250x125x12x20 2.0 A
FCB-1-1* H250x125x6x9 H250x125x12x20 2.0 A
FCB-2 H250x125x6x9 H250x125x12x20 2.0 B
FCB-2-1* H250x125x6x9 H250x125x12x20 2.0 B
* with a new fuse
Table 2 - Material properties of reinforcement steel
d(mm) fy(MPa) fu(MPa) Es(MPa) εy (×10-6)
6 625 630 2.0×105 3125
14 455 595 2.0×105 2275
6 *
295 595 2.0×10 5
1475
Note: d is the diameter of the steel bar; the thickness of the steel plate; fy is the yield strength;
*
fu is the ultimate strength; Es is the elastic modulus; εy is the yield strain.
EXPERIMENTAL SETUP
The test setup was shown in Figure 3. The specimens consisted of four parts: steel coupling
beam, composite wall piers, loading beam, and pedestal. For loading convenience, the
specimens were rotated 90 degrees such that the coupling beam was vertically placed, and the
loading beam was bolted to a rigid L-shaped loading girder. Three actuators were used to
simulate the loading and boundary conditions expected in coupled wall buildings. The specimen
was loaded horizontally with a 1000kN hydraulic actuator to create a lateral drift between the
ends of the coupling beam. Two 500kN hydraulic actuators were attached to the steel L-shaped
girder to ensure lateral drift without any rotation in the coupling beam. To guarantee lateral
stability of the loading system, the L-shaped loading girder was laterally braced at three points.
(a) Front view (b) Side view

Fig. 3 - Test setup
All the specimens were loaded similarly following a series of cyclic displacements with increasing
amplitudes. The selected loading protocol is shown in Figure 4. Specimen SCB-2 was loaded up
to a drift angle of 0.1 rad. Specimens FCB-1 and FCB-2 were first loaded up to a drift angle of
0.02 rad. After conducting 3 cycles at this displacement amplitude, loading was stopped with
zero horizontal load. The small drift angle and termination of loading simulated in situ
replacement of the fuse following a moderate earthquake. The procedure for replacing the fuse
was shown in Figure 5. The residential displacements of FCB-1 and FCB-2 with zero lateral load
were maintained with the three actuators described above. After installing new fuses, specimens
FCB-1-1 and FCB-2-1 were loaded up to drift angle of 0.1, with the ending states of the previous
loading history for specimen FCB-1 and FCB-2 taken as their starting point from 0 drift,
respectively.
0.12
0.1
0.10 For SCB-2, FCB-1-1
0.08 & FCB-2-1, stop here.
3 cycles at each
0.06
Displacement angle
displacement level
0.04
0.02
0.02
0.00
0 5 10 15 20 25
-0.02
Cycles
-0.04 1 cycle at each
displacement level
-0.06 Stop loading,
-0.08 for FCB-1 & FCB-2.
Fig. 4 - Horizontal loading history
(a) Remove bolts (b) Disassemble “fuse” (c) Replace “fuse”

Fig. 5 - In situ replacement process
DAMAGE AND CRACK PATTERN

The typical failure modes of the specimens were shown in Figure 6. For specimen SCB-2, some
minor cracks on the wall piers were observed at small drift angles (around 0.003 rad.). The
flanges of the steel coupling beam yielded during the first cycle of drift angle of 0.01 rad. The
flanges of the steel beam began to separate from the concrete of the wall piers at drift angle of
0.02 rad. Buckling of the flanges was observed at the first cycle of drift angle of 0.05 rad, and
became progressively severe during the following loading stages. Localized spalling of concrete
at the edge of the top wall pier occurred at the first cycle of drift angle of 0.1 rad. Local buckling of
the web was observed at the last cycle of drift angle of 0.1 rad, and diagonal tension fields began
to propagate at the bottom end of the steel coupling beam. A slight strength degradation was
detected only during the last loading cycle.
For specimens FCB-1, no discernable cracks were detected until the drift angle reached 0.01 rad.
Some minor cracking occurred at the beam-wall interface, and the web of the fuse began to yield
due to diagonal tension. At drift angle of 0.02 rad, there was a 1.5mm gap between the end
plates, and the maximum tension strain in the fuse web was 17251με (11.7εy), suggesting
considerable plastic deformation in the fuse.
For small drift angles, there was no major difference between FCB-1-1 and FCB-1. As the
displacement amplitudes increased, the gap between the end plates and the plastic strain in the
fuse web became larger. The developing gap and the reversed loading weakened the resistance
of the coupling beam, as evident by several sudden drops of the horizontal load and loud noises
due to slippage between the end plates. This phenomenon was clearly seen from the hysteretic
curve shown in Figure 7(b). For safety reasons, FCB-1-1 was monotonically loaded from drift
angle 0.05 up to 0.1 rad., which was the maximum intended drift angle. No weld fracture or
strength degradation was observed. It should be noted that Detail “A” used in FCB-1 is not
expect to provide significant moment transfer.
For specimens FCB-2, no major issues were observed except for a few minor cracks in the wall
piers. The maximum tension strain of the “fuse” web was 9701με (6.6 εy).
For specimen FCB-2-1, no gap between the end plates was observed during the entire loading
history, mainly because of the connection details (Figure 2b) that could transfer moment between
the fuse and the rest of the coupling beam. Initiation of weld fracture in the flange stiffener was
detected at drift angle of 0.05 rad. The fuse flanges buckled during the first cycle of drift angle of
0.05 rad, and the fuse web-stiffener weld began to fracture at the third cycle of drift angle of
0.067 rad. After completing three cycles of 0.1 drift angles, there was a sudden drop in the
horizontal load due to web fracture at the web-flange interface and propagation of fracture in the
web-stiffener weld, at which loading was stopped. The final failure mode of specimen FCB-2-1 is
shown in Figure 6(c).
(a) SCB-2 (b) FCB-1-1 (c) FCB-2-1

Fig. 6 - Failure modes of specimens
HYSTERETIC RESPONSE
The load (P)-drift angle(θ) hysteretic curves of all the specimens were plotted in Figure 7. The
curves of specimen SCB-2 were stable and showed very good energy dissipation characteristics.
Only slight strength degradation was detected for the second cycle of 0.1 rad drift angle at which
the web began to buckle. The hysteretic performance of specimen FCB-1-1 below drift angle of
0.02 rad was almost identical to that of FCB-1 (Figure 7(d)), suggesting that replacing fuses
would be feasible after moderate earthquakes, and the performance would be satisfactory. For
specimen FBC-2, equipment failure negatively impacted the hysteretic performance. As a result,
the responses of this specimen and FCB-2-1 were slightly different (Figure 7(e)). It was,
nevertheless, possible to demonstrate that the seismic performance of the coupling beam could
be recovered after replacing the fuse. The effects of excessive opening of the end plates in
FCB-1-1 was evident by a number of sudden load drops (Figure 7(c)). For specimen FCB-2-1,
the connection between the fuse and the elastic components was modified by extending the
ends plates above and below the flanges and using flange stiffeners. The hysteretic performance
of FCB-2-1 was, accordingly, improved remarkably in comparison to specimen FCB-1-1 and was
nearly similar to that of specimen SCB-2. As stated before, specimen moment transfer in
specimen FCB-1-1 was not expected to be large. The initial stiffness of specimen FCB-2-1 was
higher than SCB-2, because of the thicker web and flange of the elastic components in FCB-2-1.
Displacement Δ (mm)
-75.0 -50.0 -25.0 0.0 25.0 50.0 75.0
500
400
300
Load P (kN) 200 SCB-2
100
0 Δ
-100
-200
θ
-300
-400
-500
-15 -10 -5 0 5 10 15
Drift angle θ (Radians×10-2)
(a) SCB-2
Displacement Δ (mm) Displacement Δ (mm)
-75.0 -50.0 -25.0 0.0 25.0 50.0 75.0 -75.0 -50.0 -25.0 0.0 25.0 50.0 75.0
500 500
Web was
400 Sudden drop of load due to 400 fractured
slippage of end-plate
300 300
200 200 FCB-2
Load P (kN)
Load P (kN)
100 100
0 Δ 0 Δ
-100 -100
-200 -200
θ θ
-300 -300
-400 FCB-1 -400
-500 -500
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15
Drift angle θ (Radians×10-2) Drift angle θ (Radians×10-2)
(b) FCB-1-1 (c) FCB-2-1
Displacement Δ (mm) Displacement Δ (mm)
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15
400 400
300 300
200 200
Load P (kN)
Load P (kN)
100 FCB-1 100 FCB-2

0 Δ 0 Δ
-100 -100
-200 θ -200 θ
-300 FCB-1 -300 FCB-2

FCB-1-1 FCB-2-1
-400 -400
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
(d) FCB-1 (replaced) (e) FCB-2 (replaced)
Fig. 7 – Hysteretic curves of load – displacement angle.
SKELETON CURVE
Skeleton curves of all the specimens were shown in Figure 8, and the characteristic points of the
curves were summarized in Table 3. The yield point of the skeleton curve was determined
graphically. The strength of the specimen maintained an upward trend during the loading
process. The ultimate capacity of specimen SCB-2, FCB-1-1, and FCB-2-1 was 332.6kN,
370.4kN, and 424.2kN in forward loading, respectively. The repaired specimens exhibited nearly
identical strength with their former counterpart specimens, which indicate the reasonable details
of replacement for the steel coupling beams. The maximum strength of specimen FCB-1-1 and
FCB-2-1 were 11.4 % and 27.5% higher than that of SCB-2, respectively. The higher capacities
are attributed to the smaller span-to-depth ratio of the fuses in FCB-1-1 and FCB-2-1.
500
400
300
200
Load P (kN)
100
0
-100 SCB-2
FCB-1
-200 FCB-1-1
-300 FCB-2
-400 FCB-2-2
-500
-10 -8 -6 -4 -2 0 2 4 6 8 10
Drift angle θ (Radians×10-2)
Figure 8 - Skeleton curves of specimens
Table 3 - Load and displacement angle at characteristic points of the skeleton curves
No. Py/kN θy /rad Pm/kN θm /rad
SCB-2 231.3 0.00139 332.6 0.00936
FCB-1-1 275.5 0.00157 370.4 0.00965
FCB-2-1 325.3 0.00147 424.2 0.00551
Note：Py and Pm are yield load and peak load, respectively; θy and θm are drift angle at yield load
and peak load, respectively.
ENERGY DISSIPATION CHARACTERISTICS

As seen from Table 4, the dissipated energy, expectedly, increased with an increase in the drift
angle. Specimens FCB-1-1 and FCB-2-1 dissipated more energy than SCB-2 up to 0.05 rad.,
which was not the maximum drift angle for cyclic loading of SCB-2. These dissipated energies
suggest the adequacy of the selected details.
Table 4 - Dissipated energy (kJ) at different story drift angles

Specimen story drift angles Total dissipated energy
No. 0.010 0.013 0.020 0.029 0.050 before θ=0.05
SCB-2 1.89 4.35 10.93 21.44 54.88 93.50
FCB-1-1 1.38 3.49 9.90 20.45 51.81 87.03
FCB-2-1 1.34 3.97 11.33 23.67 66.05 106.37
The equivalent viscous damping coefficient mainly reflects the “fullness” of the hysteresis curve,
calculated according to Eq. 1.
1 S ABC + SCDA
he = (1)
2π SOBE + SODF
Where, SABC, SCDA are the areas surrounded by the hysteresis loops filled with vertical lines in
Figure 9, and SOBE, SODF are the areas of the corresponding triangles.
Table 5 summarized the equivalent viscous damping coefficients at different drift angles. As the
drift angle increased, the equivalent viscous damping coefficient, expectedly, increased,
indicating that the plastic energy dissipation of the specimen became larger. As mentioned
before, there were several sudden slippages of the end plates in specimen FCB-1-1 during the
loading process. These slippages caused some “up and downs” in the hysteretic curves; hence,
the equivalent viscous damping coefficient of FCB-1-1 was slightly lower than the other
specimens. However, in general, the equivalent viscous damping coefficients of all specimens
were not significantly different.
F A
O C E
D
Fig. 9 - Hysteretic loop
Table 5 - Equivalent viscous damping ratios at different story drift angles

story drift angles
No.
0.010 0.013 0.020 0.029 0.050
SCB-2 0.16 0.22 0.29 0.35 0.33
FCB-1-1 0.11 0.17 0.25 0.31 0.39
FCB-2-1 0.09 0.17 0.27 0.33 0.43
STIFFNESS CHARACTERISTICS
The degradation of peak-to-peak stiffness against drift angle was plotted in Figure 10. The initial
100 100
SCB-2
FCB-1
80 FCB-1-1 80
FCB-2
K (kN/mm)
K (kN/mm)
60 FCB-2-1 60
40 40
20 20
0 0
0 2 4 6 8 10 0 1 2 3
Fig. 10 - Stiffness degradation curve
stiffness of all specimens, expectedly, degraded with an increase in the drift angle. No significant
differences in the trend of the data were observed in general. The initial stiffness of specimen
SCB-2 was lower than the other four specimens, because the thickness of flanges and web in
this specimen were smaller than their counterparts in the specimens with a mid-span fuse.
CONCLUSIONS
Subassemblies consisting of two steel plate composite wall piers and a steel coupling beam with
a mid-span fuse were tested to evaluate the seismic performance of this system, and to examine
the feasibility of replacing the fuse. The following conclusions can be drawn based on this
research:
1) The specimens exhibited satisfactory performance, including: high strength and stiffness,
excellent ductility, and stable hysteretic behavior.
2) Based on the experimental results, both of the selected fuse details performed well in terms
of recovering the seismic performance after subjecting the specimen to a maximum drift
rotation angle of 0.02 rad. and replacing the fuse. The initial stiffness of the specimens with
fuses was higher than that with a single steel coupling beam, because of the thicker flanges
and webs in the elastic components of the former specimens.
3) The steel coupling beam with replacement detail “A” experienced some sudden slippages
between the end plates under large rotation angles, which led to temporary loss of load
carrying capacity, and degradation of energy dissipation capacity. It should be noted that the
selected detail is not expected to offer a large level of moment transfer. This problem was
successfully resolved using replacement detail “B”.
4) No significant damage was observed in the wall piers, especially for the specimens with
fuses. It is concluded that the selected replacement details offer an effective method to limit
the entire plastic damage within the fuse.
ACKNOWLEDGMENTS
Ph.D. Candidate H. Chen from Huaqiao University gave a great help on the modification of the
pictures in this paper. The author sincerely appreciates his help. This work was funded by the
National Natural Science Foundation of China (Grant No. 51378228 and 51878304), which is
greatly acknowledged.
REFERENCES
Gong, B., & Shahrooz, B. M. (2001). Concrete-steel composite coupling beams. i: component
testing. Journal of Structural Engineering, 127(6), 625-631.
Gong, B., & Shahrooz, B. M. (2001). Concrete-steel composite coupling beams. ii: subassembly
testing and design verification. Journal of Structural Engineering, 127(6), 632-638.
Harries, K. A. (2001). Ductility and deformability of coupling beams in reinforced concrete
coupled walls. Earthquake Spectra, 17(3), 457-478.
Wu, Y., Xiao, Y., & Anderson, J. C. (2009). Seismic behavior of pc column and steel beam
composite moment frame with posttensioned connection. Journal of Structural Engineering,
135(11), 1398-1407..
El-Tawil, S., Fortney, P., Harries, K., Shahrooz, B., Kurama, Y., Hassan, M., & Tong, X. (2010).
Recommendations for seismic design of hybrid coupled wall systems. American Society of Civil
Engineers, Virginia.
Mansour, N., Christopoulos, C., & Tremblay, R. (2011). Experimental validation of replaceable
shear links for eccentrically braced steel frames. Journal of Structural Engineering, 137(10),
1141-1152.
Farsi, A., Keshavarzi, F., Pouladi, P., & Mirghaderi, R. (2016). Experimental study of a
replaceable steel coupling beam with an end-plate connection. Journal of Constructional Steel
Research, 122, 138-150.
Li, X., Lv, H. L., Zhang, G. C., & Ding, B. D. (2015). Seismic behavior of replaceable steel truss
coupling beams with buckling restrained webs. Journal of Constructional Steel Research, 104,
167–176.
Eljadei, A. A., & Harries, K. A. (2015). On the use of fixed point theory to design coupled core
walls. Engineering Structures, 102, 61–65.
Bengar, H. A., & Aski, R. M. (2016). Performance based evaluation of rc coupled shear wall
system with steel coupling beam. Steel & Composite Structures, 20(2), 337-355.
Ji, X., Wang, Y., Zhang, J., & Okazaki, T. (2017). Seismic behavior and fragility curves of
replaceable steel coupling beams with slabs. Engineering Structures, 150(2017), 622–635.
Ji Xiaodong, Wang Yandong, Ma Qifeng, & Okazaki Taichiro. (2017). Cyclic Behavior of
Replaceable Steel Coupling Beams. Journal of Structural Engineering, 143(2), 04016169.
Fortney, P. J., Shahrooz, B. M., & Rassati, G. A. (2006). The Next Generation of Coupling Beams.
International Conference on Composite Construction in Steel and Concrete (pp.619-630).
Fortney, P. J., Shahrooz, B. M., & Rassati, G. A. (2007). Large-scale testing of a replaceable
“fuse” steel coupling beam. Journal of Structural Engineering, 133(12), 1801-1807.
NUMERICAL INVESTIGATION OF COMPOSITE DETACHABLE SHORT LINKS
Mariana Zimbru1,
Naples, Italy
mariana.zimbru@unina.it
Mario D’Aniello2,
Naples, Italy
mdaniel@unina.it
Aurel Stratan3,
Politehnica University of Timișoara, Department of Steel Structures and Structural Mechanics
Timișoara, Romania
aurel.stratan@upt.ro
Raffaele Landolfo4
Naples, Italy
landolfo@unina.it
Dan Dubină5
Politehnica University of Timișoara, Department of Steel Structures and Structural Mechanics
Timișoara, Romania
dan.dubina@upt.ro
ABSTRACT
Eccentrically braced frames (EBF) equipped with replaceable short links are effective seismic
resisting systems due to their ductility and ease of repair in the aftermath of an earthquake.
Several existing studies revealed that short links can develop shear overstrength (i.e. Vu/Vp,
where Vu is the ultimate shear strength and Vp the corresponding plastic resistance) larger than
the value recommended in EC8 (i.e. Vu/Vp =1.5). In particular, the presence of axial restraints is
one of the factors responsible for the higher shear overstrength owing to the development of
catenary actions in the link at large rotational demands. Another source of shear overstrength is
the composite slab that can resist the shear distortion together with the short link. Full scale
pseudo-dynamic experimental tests on a 3D EBF carried out within the DUAREM project (Ioan
et al, 2016) allowed investigating the response of detachable links with two different
arrangements: (i) steel solution – the link was uncoupled from the slab (ii) composite solution –
the slab runs over the link and is connected to the beam. In this paper, the second set of
composite detachable links are analyzed by means of finite element simulations with the aim to
evaluate the shear overstrength and the level of axial force developed by the links.
Keywords: Seismic design, replaceable links, composite, finite element analysis.
INTRODUCTION
Structural systems that allow fast and simple repair solutions in the aftermath of important
seismic events often represent the design target in seismic areas. The research carried out in
the past decades led to the development and improvement of traditional structural systems and
design strategies that can nowadays be re-adapted into innovative systems that satisfy the
previously mentioned objective. Among the large variety of steel structural schemes,
eccentrically braced frames (EBFs) are generally considered the rational compromise between
the large stiffness of the braced structures and the ductility that is typically provided by moment
resisting frames (MRFs). Indeed, with rough simplification, the braces guarantee lateral stiffness
and the links provide ductility and dissipative capacity. Since the links are the dissipative fuses,
several studies have been carried out to investigate the possibility to replace this component
(Mansour et al., 2008; Ioan et al., 2013). These studies highlighted the key role of the link end
connections that should guarantee the possibility to detach and to replace the damaged links. It
is trivial to recognize that the link connections must be designed with adequate overstrength.
The presence of axial restraints can induce catenary actions developing into the links to which
corresponds shear overstrength larger than 1.5Vp, being Vp the plastic shear of the link (Azad &
Topkaya 2017). The possibility to replace damaged links also depends on the residual lateral
displacements after the seismic event. Dubina et al., (2008) and Bosco et al, (2017) highlighted
that dual-eccentrically braced frames (D-EBFs), namely EBFs combined with some MRF spans,
are effective to minimize the residual drifts. Buildings designed with composite links had an
overall good performance during the Christchurch earthquake sequence of 2011-2012, showing
limited residual drifts (Clifton et al., 2012). Indeed, early in the development of EBFs, Ricles and
Popov (1987) highlighted the influence of the slab on the strength and stiffness of the structural
typology. Furthermore, Ciutina et al. (2013) and Danku et al. (2013) demonstrated that simply
disconnecting the slab from the dissipative zones in D-EBFs does not lead to a response
governed solely by the steel element, on the contrary, the overall strength and stiffness are
improved while the ductility is preserved.
Within the DUAREM project, Ioan et al. (2016) carried out full scale pseudo-dynamic
experimental tests on D-EBFs with very short links that demonstrated the effectiveness of the
reparability of D-EBFs with detachable shear links. In particular, these tests also focused on the
link-to-slab interaction. In particular, the experimental mockup (see Figure 1a) was conceived
with two different types of link-to-slab details, one per side of the prototype. Indeed, one of the
perimeter frames was detailed as an all-steel solution without slab interaction (i.e. South frame
in Figure 1b) and the frame on the opposite side was detailed with full-composite girder with
Nelson studs distributed on the braced span except for the link (e.g. North frame in Figure 1b).
These arrangements allowed to examine the interactions between the link and the slab as well
as the feasibility of the repair in case of slab damage.
In order to investigate the influence of the slab at the local level on the damage pattern as well
as the development of shear and axial overstrength, finite element models of the DUAREM
composite links were developed and the numerical results are described and discussed in this
paper. The numerical simulations allowed to deepen the interpretation of the results obtained
experimentally, but also provided a basis for potential theoretical developments for the design of
composite links.
a) Experimental mock-up b) Sub-structuring of the numerical model
Fig. 1 – The DUAREM mockup
THE FE MODEL
Finite element analyses (FEAs) were performed using Abaqus/Explicit software. The geometry
of the FE model (Figure 2a and b) was extracted from the sub-structuring of the full-scale frame
tested within the DUAREM research project (Ioan et al., 2016). Although the structure was a 3
storey D-EBF, only two models are required because the geometrical characteristics of the 1st
and 2nd storey link and connection are identical, while all other parts (beams, braces) are
identical throughout the entire building. Table 1 and Figure 2c detail the main geometrical
features of the links. The links were designed with two web stiffeners, in accordance with the
EC8 rules. As it can be observed in Figure 1a and b, the South frame (the all-steel solution) has
the slab fully disconnected from the link and the beam containing it, while the North frame (the
composite solution) has the slab running over the link and connected to the beam of the EBF
bay by means of shear studs. Figure 1b schematically depicts the mockup and the substructure
used for the numerical model (highlighted in grey for the first-floor model S1).
Table 1 – Link geometrical properties

Link length Assembly length Height Width Flange thickness Web thickness
Specimen
(elink) (Lassembly) (h) (bfl) (tfl) (tw)
[Mm] [mm] [mm] [mm] [mm] [mm]
S-1 350 450 230 170 12 8
S-2 350 450 230 170 12 8
S-3 350 450 230 120 12 4
The entire span containing the link and half transverse bay are modelled as depicted in Figure
2a. The model includes the link connected to the adjoining beams, the braces and the
secondary beams perpendicular to the plane of the frame. The concrete slab, steel deck,
reinforcement and Nelson studs were modelled with their corresponding experimental
geometrical features and material properties (Ioan et al., 2016).
y p0
x
z
u(t)
Hstorey
Lspan
a) Static scheme b) The FE model
c) Links and connections geometrical characteristics

Fig. 2 – Sub-structure geometry and boundary conditions
In order to reduce the computational effort, the following modelling assumptions were adopted:
• The reinforcement bars, the Nelson studs and the braces were modelled by means of
beam elements (B31 a 2-node linear beam in space, with 6 DOF per node) with their
respective cross-section. The brace segment in the immediate vicinity of the beam was
modelled as a solid part, for a better representation of the transfer of forces between the
two elements.
• The steel deck was modelled by means of continuum shell elements (SC8R an 8-node
quadrilateral in-plane general-purpose continuum shell, reduced integration with
hourglass control, finite membrane strains, with 3 DOF per node) with the thickness of
the real steel sheet (0.8 mm).
The other parts (steel beams, link, bolts, plates, stiffeners and concrete slab) were modelled
using C3D8R (8-node linear brick, reduced integration with hourglass control, with 3 DOF per
node) solid finite elements.
The material properties were defined for each element individually, function of the material
(steel, concrete) and the type of finite element adopted (beam, shell or solid). S355 steel is used
for steel beams, braces, end-plates and stiffeners. The steel density (7850 kg/m3) and nominal
elastic properties in terms of Young’s modulus (210 GPa) and Poisson ratio (0.3) were
considered. The nonlinear stress-strain curves were obtained from the tensile coupon tests
sampled from the members constituting the mockup. The plasticity was modelled with the
combined isotropic-kinematic hardening law. S235 steel was used for the links and different
properties were assigned to their flanges and webs, corresponding to the results of coupon
tests. The reinforcement bars, the Nelson studs and steel deck properties have similar elastic
properties but for the post-yielding behavior, a simplified linear branch was considered, with the
respective strength and strain limits. Table 2 summarizes the adopted properties of the
elements.
The bolts of the flush end-plate connections at the link ends were M27 HV grade 10.9. Their
mechanical behavior was simulated in accordance with Tartaglia et al. (2018, 2019). The slab
material property was simulated using the concrete damage plasticity model proposed by
Hillerborg et al. (1976). This model takes into account the reduction of the elastic stiffness
caused by plastic straining both in tension and compression and it accounts as well for the
stiffness recovery effects under dynamic loading conditions (Abaqus User’s Manual, 2014). The
characteristics of C25/30 concrete were defined in accordance with Eurocode 2, i.e. density of
2500 kg/m3, Young’s modulus of 31GPa, and Poisson’s ratio of 0.2.
Table 2 – Material properties

Yield Ultimate
Type Material Elongation
strength strength
[MPa] [MPa] [%]
Link Flanges S235 260 453 31
Web link S1 S235 296 473 29
Web link S3 S235 296 460 23
Reinforcement 610HD mesh B450C 450 540 15
Studs SMD19105 S235 235 360 15
Steel deck A55 – P600 G5 S250 260 450 15
Explicit analyses were carried out. Hence, a General Contact interaction was defined for the
entire model. The global property in terms of Normal and Tangential Behavior was defined as
‘Hard Contact’ and using the ‘Penalty’ friction formulation, respectively. Multi-point constraints
(MPCs) were used to define the connection between the shear studs and the beams, the beam
and solid part of the braces and the secondary and main beams. The steel sheet was tied to the
concrete slab. The reinforcements and the shear studs were embedded in the concrete.
The nature of the analysis dictates also the way the bolt preloading is defined. The clamping
action of the bolts was applied by inducing artificially the shrinkage of the shanks through a
temperature variation in the bolt. The permanent load was applied as distributed pressure on
the entire top surface of the slab, with a magnitude of 5 kN/m2 (which is the value applied during
the tests).The boundary conditions were defined based on the scheme in Figure 2. The lateral
actions were imposed as monotonically increasing displacement applied at the main beam
ends, along the slab edge, and except for the displacement along the y-axis, all other DOFs are
restrained. The braces were fully restrained while the boundary conditions at the end of the
composite slab (in the section at the mid-bay) allow the displacement along the y-axis (the
direction in which the lateral action is applied) and restrain all other DOFs. The mesh was
adequately defined for each element type (solid, continuum shell and beam) and considering
seed sizes that can provide optimum time/quality ratios. In total, each model had approximately
100000 solid elements, 20000 shell elements and 6000 beam elements.
Previous analyses on all-steel link assemblies were conducted in order to calibrate the
numerical model. As depicted in Figure 3, good agreement between the experimental and
numerical response was obtained (Zimbru et al., 2017).
500 500 500
400 400 400
300 300 300
200 200 200
100 100 100
Shear [kN]
Shear [kN]
Shear [kN]
0 0 0
-100 -100 -100
-200 -200 -200
-300 -300 -300
-400 -400 -400
-500 -500 -500
-0,040 0,000 0,040 0,080 -0,040 0,000 0,040 0,080 -0,040 0,000 0,040 0,080
Total Rotation [rad] Total Rotation [rad] Total Rotation [rad]
Numerical EXPERIMENTAL Numerical EXPERIMENTAL Numerical EXPERIMENTAL
S1 S2 S3
Fig. 3 All-steel (South Frame) Experimental vs Numerical Shear Force-Rotation Curves
RESULTS
The numerical results of the composite link assemblies were compared with the output obtained
in previous investigations of all-steel links (Zimbru et al., 2017) in order to assess the accuracy
of the current model. The comparison is made in terms of link forces (Vlink, Nlink), thus
disregarding the slab contribution for the composite model. Figure 4 shows that the shear
strength (a) and axial force (b) developed in the link of the composite model (North frame) have
very close values to the ones of the forces in the all-steel link (South frame). Furthermore, the
numerical model is able to replicate the initial stiffness and plastic shear capacity.
500 500
S1 South [NUM]
S1 North [NUM]
Link Shear force [kN]
400
Link Axial force [kN]

300
300
100
200
All-steel link assembly
-100 (South Frame)
100 S1 North [NUM]
S1 South [NUM]
0 -300 Vlink Vlink
0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15
Link Rotation [rad] Link Rotation [rad]
a) S1 assembly Link-slab composite assembly
(North Frame)
500 500
S3 North [NUM] VRC VRC
400 S3 South [NUM]
Link Shear force [kN]
Link Axial Force [kN]
400
300 Vlink Vlink
300
200
200 Vtotal = Vlink + VRC
100
100 S3 North [NUM] 0
S3 South [NUM[
0 -100
0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15
Link Rotation [rad] Link Rotation [rad]
b) S3 assembly
Fig. 4 – Comparison between the shear (Vlink) and axial force (Nlink) developing into the all-steel
(South) and the composite (North) links
The link-to-slab assembly develops larger shear forces, as it can be seen in Figure 5a and b,
and in particular for S1, the contribution of the slab can be clearly distinguished as the
difference between the total shear and the link shear. Overstrength after reaching the plastic
strength can also be noticed for both links. Indeed, as depicted in Figure 5, the link-to-slab
assembly indicates shear overstrength values larger than those exhibited by the all-steel links.
The numerical results confirm that the slab contributes to the shear strength of the link by
resisting shear in parallel. In particular, the peak shear overstrength at 0.08 rad is about 1.7 for
both link-to-slab assemblies. This finding, in line with those of Ciutina et al., 2013, highlights the
need to revise the capacity design rules in order to guarantee the effectiveness of the capacity
design rules.
Figure 5c and d depict the equivalent plastic strain (PEEQ) distribution in the two link-slab
assemblies at 8% of link rotation. It is possible to observe the large concentration of damage in
both link and slab, confirming thus the development of the composite action.
a) S1 assembly b) S3 asssembly
c) S1 assembly d) S3 assembly
Fig. 5 – Normalized link and total shear force (a and b) and PEEQ distributions (c and d)
In order to evaluate analytically the contribution of the RC slab to the response of the link-slab
assembly, it was assumed that the two elements work in parallel (given that no shear studs are
placed on the link, but only on the beams of the frame), and the ultimate bending capacity of the
slab was evaluated. Under this assumption, two possible mechanisms for achieving the shear
capacity of the slab were alternatively considered: (i) the shear capacity corresponding to the
flexural failure of the slab and (ii) the shear capacity of the RC slab. For the first scenario, Figure
6 briefly describes the static scheme of the models and the cross-section of the slab for the
evaluation of the bending moments (the slab flexural capacity at the left - Mleft - and right - Mright -
end of the link differ as the section is not symmetric under sagging and hogging bending) and
the corresponding shear (Vslab,M). Meanwhile, for the evaluation of the RC shear capacity
(Vslab,Rd,c), the recommendations of Eurocode 2 were used.
Table 3 shows the values analytically calculated, while Figure 7a and b depict the comparison
with the values obtained from the FE analysis. It can be observed that for assembly S1, the link-
to-slab assembly shows an increase of shear capacity with respect to the bare steel profile and
that the analytical prediction matches perfectly the numerical curve, i.e. evaluating the shear
capacity as the sum of the link shear strength and slab shear capacity gives a good
approximation. However, for the case of S3 (i.e. the shorter link), the initial plastic shear
strength is achieved by means of the steel link alone, while the shear capacity developed at 8%
of link rotation (the value that is used for the capacity design of non-dissipative members) is also
influenced by the slab.
Fig. 6 – Increase of shear capacity due to the flexural hinging of the slab
Table 3 Summary of resistance contribution of the RC slab

beff Vslab,M Vslab,Rd,c Vy,link VA VB
m kNm/m kN kN kN kN
S1 0.40 54 49 290 350 499
S3 0.40 54 49 145 194 266
Where:
beff is the effective width of the RC slab calculated according to EC4
Vslab,M is the shear contribution of the RC slab considering its ultimate flexural resistance (see
Figure 5)
Vslab,Rd,c is the shear resistance of the RC slab evaluated according to Eurocode 2
Vy,link is the shear capacity of the steel link
VA is the shear capacity of the link-slab assembly obtained by summing up Vy,link and the
minimum between Vslab,M and Vslab,Rd,c
VB is the shear capacity of the link-slab assembly at 8% link rotation obtained considering 1.5
Vy,link (where 1.5 is the shear overstrength coefficient according to Eurocode 8) and the
minimum between Vslab,M and Vslab,Rd,c.
a) S1 b) S3
Fig. 7 – Shear from the all-steel link and the link-to-slab assembly
Figure 8 depicts the tensile damage distribution in the RC slab of S1 model for varying link
rotation levels. As it can be expected, the damage increases as the rotational demand
increases. In particular, when the link deforms inelastically it pushes the floor slab out of plane.
The distribution of the tensile damage into the slab at 0.08 rad link rotation is far from the
damage recognizable into a composite beam, but it transversally involves the deck with the
formation of yield lines oriented with the imposed deformation. This finding is in line with the
observations presented by Clifton et al. (2012).
2% 4%
x
6% 8%
y
Fig. 8 – Top view of the tension damage distribution in the slab of S1 assembly
CONCLUSIONS
In light of the results of the finite element simulations presented and discussed in the previous
paragraphs, the following conclusions are drawn:
• The shear and the axial forces developing into the steel links of the North frame (the link-
to-slab system) do not differ from those exhibited by the all-steel assemblies. This result
confirms that the design strength of link end-connections does not depend on the slab.
• The reinforced concrete slab brings a significant contribution to the dissipative
mechanism but with the downside of significant damage in the slab, not only in the area
above the link but also spreading along the beam containing the link.
• The link-to-slab assemblies may develop larger shear strength than the all-steel links.
This implies that this increase of strength should be properly accounted for the capacity
design of braces.
• During the experimental pseudo-dynamic tests, the level of damage observed was
reduced, at the maximum link rotation of 6%. However, it can be observed from the
numerical analyses that damage increases in the affected areas and spreads to new
ones when the link rotation exceeds the rotation of 8%. The significant level of damage
in the slab leads to a proportional increase in repair costs in the aftermath of the design
earthquake.
• The mechanisms leading to additional shear capacity owed to the RC slab need to be
further investigated and proper rules for the definition of the effective width (beff) need to
be developed.
REFERENCES:
Azad, K.S.. & Topkaya, C., (2017). “A review of research on steel eccentrically braced frames”
Journal of Constructional Steel Research, 128, 53-71
Bosco, M., Marino, E.M., Rossi, P.P. (2017). A design procedure for dual eccentrically braced-
moment resisting frames in the framework of Eurocode 8. Engineering Structures, 130, 198-
215.
Ciutina, A., Dubina, D. and Danku, G. (2013) Influence of steel-concrete interaction in
dissipative zones of frames: I – Experimental study. Steel and Composite Structures. 15(3),
281-000. http://dx.doi.org/10.12989/scs.2013.15.3.281
Clifton G.C., Nashid H., Ferguson G, Hodgson M., Seal C., Bruneau M., MacRae G.A. and
Gardiner S. (2012). Performance of Eccentrically Braced Framed Buildings In The
Christchurch Earthquake Series of 2010/2011. Proceedings. of 15th WCEE, Lisbon,
Portugal, September 24-28.
Danku, G., Dubina, D., and Ciutina, A. (2013). Influence of steel-concrete interaction in
dissipative zones of frames: II - Numerical study. Steel and Composite Structures. 15(3),
305-000. http://dx.doi.org/10.12989/scs.2013.15.3.305
Corp.
Dubina, D., Stratan, A., Dinu, F. (2008). Dual high-strength steel eccentrically braced frames
with removable links. Earthquake Engineering and Structural Dynamics, 37, 1703–1720.
EN 1998-1, Design of Structures for Earthquake Resistance - Part 1: General Rules, Seismic
Actions and Rules for Buildings. CEN, 2005.
EN 1993:1–8, Design of Steel Structures - Part 1–8: Design of Joints. CEN, 2005.
EN 1992:1–1, Design of Concrete Structures - Part 1–1: General rules and rules for buildings.
CEN, 2005.
Hillerborg, A., Modeer, M., and Petersson, P-E. (1976). Analysis of crack formation and crack
growth in concrete by means of fracture mechanics and finite elements. Cement and
Concrete Research. 6, 773-782. Pergamon Press, Inc. Printed in the United States.
Ioan, A., Stratan, A., Dubina, D. (2013). Numerical simulation of bolted links removal in
eccentrically braced frames. Pollack Periodica 8(1), 15-26.
Ioan, A., Stratan, A., Dubină, D., Poljanšek, M., Molina, F.J., Taucer, F., Pegon, P., Sabău, G.
(2016). Experimental validation of re-centring capability of eccentrically braced frames with
removable links. Engineering Structures, 113, 335–346.
Mansour, N., Shen, Y., Christopoulos, C., Tremblay, R. (2008). Experimental evaluation of
nonlinear replaceable links in eccentrically braced frames and moment resisting frames. The
14th Conference on Earthquake engineering, Beijing, China, October 12-17.
Ricles, J. M. and Popov, E. P. (1987). Experiments on eccentrically braced frames with
composite floors. Berkeley, California, Earthquake Engineering Research Center, University
of California: 324 pp
Tartaglia, R., D’Aniello, M., Zimbru, M., Landolfo, R. (2018). Finite element simulations on the
ultimate response of extended stiffened end-plate joints. Steel and Composite Structures,
27(6), 727-745. DOI: 10.12989/scs.2018.27.6.727
Tartaglia, R., D’Aniello, M., Rassati, G.A. (2019). Proposal of AISC-compliant seismic design
criteria for ductile partially-restrained end-plate bolted joints. Journal of Constructional Steel
Research 159, 364-383.
Zimbru, M., D’Aniello, M., Stratan, A., Landolfo R., and Dubina, D. (2017) Finite element
modelling of detachable short links. Eccomas Proceedia COMPDYN (2017) 790-801, doi:
10.7712/120117.5457.17470
DIRECT ANALYSIS FOR STRUCTURES MADE OF HIGH
STRENGTH STEEL
Si-Wei Liu1, Siu-Lai Chan1* and Tian-Ji Li2

1 Department of Civil and Environmental Engineering,
The Hong Kong Polytechnic University, Kowloon, Hong Kong.
2 Shanghai Construction No.5 (Group) Co., Ltd. Shanghai, China
*Corresponding Author: Email: ceslchan@polyu.edu.hk
A S RAC
Buckling curves for high strength steel are not easily obtainable because of limited resources
in testing the members under high loads, resulting in difficulty for engineers to design
structures made of high strength steels. This paper proposes a method based on direct
analysis (DA) method for design of high-strength steel members and frames allowing explicit
modeling of residual stresses such that numerical simulation analysis and design can be
conducted as substitutes of full-scale physical tests. A section analysis technique based on
quasi-Newton iterative scheme is adopted to take into account the effects of residual stress
with the geometrical imperfections measured. Residual stress for the welded box-sections
and H-sections fabricated by high-strength steel are measured and adopted for illustration. To
complete the consideration of buckling and second-order effects, the initially crooked with
arbitrarily-located-hinge (ALH) element is introduced. A fiber hinge model using the sectional
strength-iteration surfaces is used for modeling of material yields. Several examples are
employed to verify the accuracy and validity of the proposed approach. This paper is to
contribute to the formulation of a versatile numerical procedure to account for residual
stresses and geometric imperfections separately for design of high-strength steel members
and frames where physical tests are expensive.
e o ds Direct analysis, high strength steel, imperfections
I R DUC I
Direct analysis (DA) for steel structure design is a new trend for design of modern structures
without assumption of effective length. However, most design codes only contain clauses
for use steel with strengths ranging from 235MPa to 460MPa, the current design
formulae based on the linear assumption is not directly applicable to design of high-
strength steel members and frames.
DA allows the use of plastic hinge or plastic zone. Generally speaking, the former is efficient
but less accurate than the latter. This paper proposes a method of using plastic zone to
generate yield surface for plastic hinge so the advantages of both methods are retained.
Further, the suggested method allows separated considerations of residual stress so it can
deal with DA of high strength steel from the basic principle.
The applications of high-strength steel constructions is becoming more important and was
initiated since 1990s such as projects of Sony Center in Berlin, Latitude Tower in Sydney, Star
City in Darling Harbour of Sydney, NTV Tower in Tokyo, JR East Main Office in Tokyo,
Landmark Tower in Yokohama, Tokyo Skytree and Beijing National Stadium and CCTV
New Building. Many benefits were observed in practice regarding the structural efficiency,
constructional speed and so on.
The use of high-strength steels can significantly reduce member size and increase safety
margin, but the stability problems could be more acute. Geometrical initial imperfections and
the residual stress are two major factors for structural and member design. Initial geometric
imperfections exist in all steel members and frames and they are critical in stability design.
In addition, residual stress could reduce the ultimate capacity of compression members via
the degradation of stiffness and yielding, and decrease the fatigue endurance of members
under cyclic loading. Further, residual stresses have an obvious influence on the ultimate
load capacity for flame-cut welded box and H-columns (Ziemian 2010)
Analyses and design by the beam-column elements are considered to be the most efficient
in practice, and this approach has been extended to practical design in recent years. In the
present study, the curved
arbitrarily-located-hinge (ALH) element, proposed by Liu et al. (2014a, 2014b), is employed
and the initial member curvature associated with P-δ and P-Δ effects are directly modeled
in an incremental-iterative numerical procedure.
Material yielding is another major factor controlling the ultimate load and overall stability of a
framed structure when one uses DA for design. An approach using the plastic fiber hinge
model is proposed in the present study. The material yielding on hinge locations is
modeled by a gradually softening spring. Contrary to the traditional or refined plastic
hinge model where the empirical and simplified linear interaction equations are commonly
adopted for yield surface of sections, the present sectional yield surface is constructed by
a plastic fiber hinge model which could consider residual stresses directly and
analytically rather than empirically.
To generate the sectional yield surfaces, the cross-section analysis technique is
required and this method has been extensively studied since the 1960s. Numerous
researchers have proposed equations for yield surfaces and the present study uses a
more rigorous procedure of quasi-Newton interactive scheme, where the steel sections are
automatically divided into several fibers in calculating the stress resultant of a cross
section.
The residual stress models are vital and available for the analysis and have been
investigated by experiments [see, for examples, Usami and Fukumoto (1984) and
Rasmussen and Hancock (1995)]. In this paper, two distributed patterns of residual stress
for the welded box-sections and H-sections fabricated by high-strength steel are
discussed. The section analysis technique based on the Quasi-Newton iterative scheme
with explicit consideration of the residual stresses is elaborated. To consider second-order
effects associated with initial imperfections, the curved arbitrarily-located-hinge (ALH)
element is adopted. A plastic fiber hinge model using the sectional strength-iteration
surfaces is introduced for reflecting the yield behavior. Finally, two examples are given for
verifying the accuracy and validity of the proposed method.
ASSUMP I S
In the present study, the following assumptions are taken for analysis of the beam-column
members and their cross sections. (1) Strains in the element are small, but the large
deformations are achieved using the incremental tangent stiffness method; (2) Euler-
Bernoulli hypothesis is assumed and the strain distribution along the cross section is linear;
(3) Second-order effects including the P-∆ and P-δ deformations are considered; (4)
Material yielding on the critical section is reflected by the plastic fiber hinge; (5) Loads are
conservative and applied on element nodes; and (6) Warping and beam lateral-torsional
buckling are not considered.
EAM C LUM ELEME F RMULA I S
The important parameters such as initial imperfections, large deflection and material
yielding, are required to be simulated in the numerical procedure. In the present study, the
arbitrarily-located-hinge (ALH) element shown in Fig. . below proposed by Liu et al.
(2014a, 2014b) is used and should be referred for details.
Fig The beam-column element
PLAS IC E M DEL
Plastic analysis is required to form a yielding function to monitor the process of the
sectional plastification. The plastic fiber hinge method is derived from the refined
plastic hinge approach proposed by Chan and Chui (1997). Contrary to the traditional
lumped plasticity or refined plastic
hinge methods, which assume the simplified criteria to examine the material yielding
on hinge locations, the proposed model employs the analytical and rigorous sectional yield
surfaces to control the hinge stiffness.
YIELD SURFACES
For any steel sections under a given axial-load, the corresponding moments can be
computed by the presented sectional analysis method. The sectional yield surface,
adopted to describe the axial force and related moments of section, is outlined below (see
Fig. 2).
The outermost boundary is a failure surface shown in Fig. 2. The ultimate state of a section is
described by these data points that induce failure of steel sections. The surface is
necessary for determining the individual load-bearing capacity in traditional design of steel
structures.
Initial yield surface is also a basic strength control criterion for plastic analysis. Within this
surface generating from a specific load combination, the corresponding section
maintains elastic. When generating initial yield surfaces, all fibers of a section will be
monitored and the onset of strains beyond the elastic limit is also detected. The internal
force and moments are calculated through the integration of stresses along the cross-
section. Subsequently, sectional capacity can be obtained from the initial yield condition.
Fig.2 Initial yield and failure surfaces
Yield and Plastic Moments
The whole strength interaction space is divided into failure and initial yield surfaces in three
zones as shown in Fig. 3. These zones are named as the elastic, elasto-plastic and plastic
hardening zones. The hinge stiffness is determined according to the current loading
position in the strength interaction space. When the load point lies inside the elastic zone,
the section reminds elastic and the hinge stiffness is infinitely large, which means no plastic
deformation is formed at the hinge location. Once the loads go into the elasto-plastic zone,
the partially yielding on the critical section begins and the hinge stiffness decreases. If the
load exceeds the failure surface, the corresponding section will be fully yielded and the
hinge stiffness is set to a small value. The elastic and plastic moments are defined as
minimum moment causing yield and maximum moment capacity, which are further used to
update the hinge stiffness.
Fig.3 Initial yield and failure surfaces
Hinge stiffness
As aforementioned, the degradation in hinge stiffness reflects the material yielding on hinge
sections. The formulations for calculating the spring stiffness S, are written as,
EI 0 (1)
S  1010 , for M i  M er
L
EI ( ) (2)
S  M pr  M i  /  M i  M er  for M er  M i  M pr
L
EI ( ) (3)
S  , for M i  M er
L
where EI(ζ)/L is the bending stiffness in the current iteration; the elastic and plastic moment
capacities at the current axial force are given by M  and M pr.

er
Cross section analysis
The sectional yield surfaces, i.e. yield and plastic surfaces, are generated by the some
rather standard procedures as locations of neutral axis, determination of axial force and
moment resistance by a summation process of stresses which can include residual
stresses directly below.

 i = f  i  (4)
 i =  ai   ri  i vi (5)
in which i is the direct stress, i is the total strain, εai is the axial strain; εri is the residual
strain, φi is the curvature and i is the distance.
And the bending moments can be obtained from Eqs. (4) and (5) through the following
summation process.
n n

M z  M u cos n  M v sin  n     i Ai vi  cos n   i Au
i i  sin  n  (6)
 i 1 i 1 
n n
M y  M u sin n  M v cosn   i Av
i i  sin n   i Au
i i  cos n
(7)
i 1 i 1
n
Nx   i Ai (8)
i 1
in which Mu, Mv and Nx are the bending moments and axial force and Mz and My are the global
moments referred to the geometrical centroid.
EXAMPLES
Two examples are selected to demonstrate the accuracy and applicability of the proposed
method. Example 1 - High-strength welded box columns of Q460
A box-section column fabricated by Q460 high-strength steel under axial compression are
analyzed and compared the analysis results with the test data reported by Wang et al.
(2014). The geometric dimensions of the box-column are shown in Figs. 4 (a) and (b)
below.
Component plates: P
f y=505.8 MPa
E=207800 MPa
11.43
3260
133.6
156.5
11.43
156.5
(a) Cross section (b) Evaluation
Fig. 4 Box-section column
The data of material properties and initial geometric imperfections are measured and given by
Wang et al. (2014) with the actual yield strength 505.8 MPa, the elastic modulus 207.8 GPa and the
initial deflection is L/666. Distributed model of residual stresses on the section selects the
rectangular shape pattern recommended by Wang et al. (2012). The ratio of tensile residual stress
to yield stress is 0.678 and of compressive residual stress to yield stress is 0.195.
Fig. 5 illustrates the comparison of the results obtained in the proposed analysis method
with the test results. The ultimate capacity of the column derived by the presented method
is 2701 kN that is close to the experimental value (2591 kN) but the deflection or stiffness is
less accurate with the theory larger. The numerical result indicates an acceptable
performance of this method in tracing the nonlinear behavior of high-strength steel box-
columns. Thus, the proposed method accounting for residual stress and geometrical
imperfections separately is further employed in DA of the following example.
Fig.5 Comparison with test results
Example 2 - Two-story spatial steel frame with different steel grades
In the example, a two-story and two-bay spatial steel frame with I-sections for both beams
and columns is introduced in Fig. 6, which is originally presented and analyzed by the Finite
Element Method (FEM) by Cuong et al. (2007). The sectional dimension of beams and
columns in the frame is H150×160×10×6.5. The yielding strength, elastic modulus and
shear modulus are 350 MPa, 221 GPa and 85 GPa. The concentrated load P acting on the
nodes is 80 kN. The geometrical imperfections of
the frame from the original publication is simulated by nodal coordinates. The imperfections
of individual members are excluded in the analysis. In order to explore the behavior of high-
strength steel in frames, grade Q460 and Q690 are adopted in this spatial frame. The
nominal yield stress for Q460 and Q690 are 460 MPa and 690 MPa, respectively.
Roof
(
(
2nd
( floor
level
(
Z Base
Y level
X
Fig.6 The studied 3-dimensional frame
Fig. 7 shows load-displacement curves at the roof and 2nd floor levels for three steel grades.
It can be seen that the ultimate load factors obtained by the present study and the FEM are
similar and equal to 0.92 and 0.94 respectively. It suggests that the presented analysis
method possesses high accuracy for direct analysis of spatial steel frames. The failure load
factors for steel grade Q460 and Q690 are 1.13 and 1.5, respectively. It is concluded that
the ultimate capacity of high-strength steel members is considerably higher than ordinary
steel members under the same sectional dimension and loading cases. It also indicates that
utilizing high-strength steel can significantly reduce the section size, material
consumption and construction costs because of lighter weight as well as increase in usable
floor area which is an important factor in cities with high land cost. All these make a
significant contribution to economical benefits and resource-saving.
Fig.7 The results using different steel grades
Conclusions
In this paper, a new direct analysis method is proposed for advanced analysis and design
of high-strength steel members and frames. The sectional analysis technique with quasi-
Newton iterative scheme is exploited to consider the residual stress effects, and suitable for
arbitrary steel sections. The second-order effects related to initial geometrical imperfections
can be modeled by the curved ALH element. A plastic fiber hinge method is used to reflect
the yielding behavior of members. The major advantage is to abandon the needs utilizing
the approximate buckling curves in specifications. Additionally, the consideration of residual
stresses by constructing interaction curves could avoid the scenario that their effects are
overestimated or underestimated in practical design owing to the simplified clauses in
codes for residual stress consideration.
Acknowledgements
The authors are grateful to the financial supports by the Research Grant Council of the Hong Kong
SAR Government on the project “Second-order and Advanced Analysis of Arches and Curved
Structures (PolyU 152012/14E)” and “Second-Order Analysis of Flexible Steel Cable Nets Supporting
Debris (PolyU 152008/15E)” and support by Innovation and Technology Fund of the Hong Kong SAR
Government for the project “Development of an energy absorbing device for flexible rock-fall barriers
(ITS/059/16FP)”.
References
Chan, S. L. and P. P. T. Chui (1997). "A generalized design-based elastoplastic analysis of steel
frames by section assemblage concept." Engineering Structures 19(8): 628-636.
Liu, S.-W., Y.-P. Liu and S.-L. Chan (2014a). "Direct analysis by an arbitrarily-located-plastic-hinge
element—part 1: planar analysis." Journal of Constructional Steel Research 103: 303-315.
Liu, S.-W., Y.-P. Liu and S.-L. Chan (2014b). "Direct analysis by an arbitrarily-located-plastic-hinge
element—part 2: spatial analysis." Journal of Constructional Steel Research 103: 316-326.
Ngo-Huu, C., S.-E. Kim and J.-R. Oh (2007). "Nonlinear analysis of space steel frames using fiber plastic
hinge concept." Engineering Structures 29(4): 649-657.
Rasmussen, K. J. R. and G. J. Hancock (1995). "Tests of high strength steel columns." Journal of
Constructional Steel Research 34(1): 27-52.
Usami, T. and Y. Fukumoto (1984). "Welded box compression members." Journal of Structural Engineering
110(10): 2457-2470.
Wang, Y.-B., G.-Q. Li, S.-W. Chen and F.-F. Sun (2014). "Experimental and numerical study on the
behavior of axially compressed high strength steel box-columns." Engineering Structures 58: 79-91.
Wang, Y. B., G. Q. Li and S. W. Chen (2012). "The assessment of residual stresses in welded high
strength steel box sections." Journal of Constructional Steel Research 76: 93-99.
Ziemian, R. D. (2010). Guide to stability design criteria for metal structures, John Wiley & Sons.
PERFORMANCE OF COMPOSITE BEAMS IN FIRE: PRE-TEST
ANALYSIS OF FULL-SCALE EXPERIMENTS
Joseph A. Main, Fahim Sadek, Jonathan M. Weigand, Jian Jiang,

Lisa Choe, Selvarajah Ramesh, Matthew S. Hoehler, Mina Seif, John Gross
Engineering Laboratory, National Institute of Standards and Technology
100 Bureau Drive, Gaithersburg, MD 20899
joseph.main@nist.gov, fahim.sadek@nist.gov, jonathan.weigand@nist.gov, jian.jiang@nist.gov,
lisa.choe@nist.gov, selvarajah.ramesh@nist.gov, matthew.hoehler@nist.gov,
mina.seif@nist.gov, john.gross@nist.gov
ABSTRACT
This paper presents results from pre-test analysis of a full-scale, 42 ft (12.8 m) span, composite
beam with double-angle connections, under uniform gravity loading simulated using six equally
spaced concentrated loads, and subjected to a 4000 kW compartment fire followed by a period
of cooling. This configuration corresponds to the first in a series of fire tests on composite
beams to be performed in the National Fire Research Laboratory, with varying steel gravity
connection types and slab restraint conditions. The beam, connections, and composite slab are
modeled using a reduced-order, shell-element-based modeling approach that enables coupled
thermal-structural analysis by using the same finite element mesh for the heat transfer and
structural analyses. This approach includes (i) realistic fire-induced thermal loads, (ii)
temperature-dependent thermal and mechanical material properties, (iii) layered shell elements
for the framing members and connections, and (iv) a newly developed reduced-order modeling
approach for the composite slab on steel decking.
INTRODUCTION
In current engineering practice in the United States, fire protection of structures is measured by
fire endurance duration and is achieved through prescriptive requirements specified based on
standard fire tests. Standard fire tests (ASTM E119, 2016 or ISO 834, 2014) typically
characterize fire-performance through furnace testing of isolated components or small
subsystems. These tests do not provide an accurate representation of structural performance in
actual fires (Phan et al. 2010). Steel buildings subjected to structurally significant fires
experience thermal assault comprising elevated temperatures and thermal gradients, which
may induce both temperature-dependent degradation of material mechanical properties and
loads in the composite slab, steel framing, and connections that were not accounted for in the
structural design. The framing members and connections in the structure may be subjected to
large compressive forces induced by restraint against thermal expansion, followed by large
deflections, rotations, tensile forces, and potential destabilization. Similarly, the composite slab
may also be subjected to large compressive forces due to restrained thermal expansion, which
might exacerbate the potential for heat-induced explosive spalling of concrete. In addition,
thermal contraction during cooling can lead to the development of large tensile forces. Properly
accounting for the complex interactions between the framing members, composite slab, and
connections presents significant challenges that require thorough examination from both
experimental and analytical perspectives. New computational tools and design approaches are
needed for evaluating the response of structural systems to realistic, uncontrolled fires.
The paper is focused on a recently-developed computational modeling approach for combined
thermal-structural analysis of composite floor systems that adequately accounts for the relevant
behaviors and failure modes of structural members, connections, and composite slab subjected
to fire loading. The reduced-order, shell-element-based, computational modeling approach
allows the same finite element mesh to be used for both the thermal and structural analyses.
This analysis approach is applied to a series of composite beam tests, to be conducted at the
National Fire Research Laboratory (NFRL), to provide pre-test predictions of the specimens’
response and ultimate capacity.
This paper presents preliminary results from pre-test analysis of a full-scale, composite beam
with welded-bolted double-angle connections (angles welded to the column flange and bolted to
the beam web), under uniform gravity loading simulated using six equally spaced concentrated
loads and subjected to a 4000 kW compartment fire. This configuration represents the first in a
series of composite beam specimens that will be tested in the NFRL, with varying gravity
connection types and slab restraint conditions. These pre-test predictions are an essential
aspect of the NFRL’s test program because they are used to: (i) help with planning of
experiments and instrumentation layout, (ii) estimate when failure may occur to help ensure
safe execution of tests, and (iii) evaluate accuracy of current modeling approaches. Due to the
innumerable uncertainties in the analyses, the pre-test analyses will provide the “mean-case”
response based on best-estimate values of the models’ input parameters, along with a range of
expected responses when accounting for the various sources of uncertainties.
COMPOSITE BEAM TESTS

The NFRL at the National Institute of Standards and Technology (NIST) (Bundy et al., 2016) is
a unique facility that enables research on the response of real-scale structural systems to
realistic fire and mechanical loading under controlled laboratory conditions. Using this new
facility, NIST is conducting full-scale structure-fire experiments on composite floor systems in
multi-story steel framed buildings, that are intended to provide the technical basis and data
necessary to advance performance-based design of structures subjected to significant fires,
validate physics-based computational models to predict the fire-performance of structures, and
evaluate approaches to measure structural deformations in the composite floors. Composite
floor systems were selected for this experimental program due to their widespread use in U.S.
building construction and because of the significant challenges inherent in modeling the
complexities of their response under fire loading, such as the interactions between the gravity
connections, concrete slab, metal deck, and shear studs.
As part of this testing program, the NFRL is testing five 42 ft (12.8 m) span composite beam
subassemblies subjected to combined structural and fire loading. Each composite beam
specimen consists of a lightweight concrete slab cast on top of profiled steel decking acting
compositely with a steel beam via steel shear studs. The concrete slab has a thickness of
3.25 in (82.6 mm) above the top of the decking, with embedded 6×6 W1.4/1.4 welded wire
mesh reinforcement having a 6 in × 6 in (152 mm × 152 mm) grid spacing with wire cross-
sectional area of 0.014 in2 (9.0 mm2) placed at mid-depth of the 3-1/4 in (82.6 mm) topping
slab. The width of floor slab is 6 ft (1.83 m), see Figure 1. The minimum specified compressive
strength of the concrete is 4000 psi (27.6 MPa), and the measured 28 day compressive
strength was 6588 psi (39.2 MPa). Polypropylene fibers were used in the concrete mix to
mitigate potential for explosive spalling of the concrete slab under fire. The 20 gage galvanized
Vulcraft 3VLI20 steel decking (ASTM A653 Grade 33) has a thickness of 0.0358 in (0.91 mm)
with 3 in (76.2 mm) deep ribs spaced at 12 in (305 mm) on-center. Steel headed shear studs
(ASTM A108) with a diameter of 3/4 in (19 mm) and length of 5-3/16 in (132 mm), are installed
at 12 in (305 mm) intervals to develop composite action between the beam and the slab.
The centerline-to-centerline span between columns of the composite beam specimen is 42 ft
(12.8 m) and the distance between the bolt lines of the shear connections is 40 ft (12.2 m). The
floor beams are W18×35 sections and the columns are W12×106 sections; both are made of
ASTM A992 steel with a minimum yield strength of 50 ksi (345 MPa). The beams are thermally
insulated using 5/8 in (16 mm) thick Southwest 5MD sprayed fire-resistive material (SFRM),
resulting in a 2 hour fire rating of the floor system. The beams are connected to the supporting
steel columns using one of the following gravity connections:
1. Double angles bolted to the beam web and welded to the column (Figure 2(a)), with
three rows of 3/4 in (19 mm) diameter A325 high-strength bolts and L5×3×3/8 angles.
The angles are welded to a sacrificial plate, measuring 12-3/4 in × 18 in × 3/4 in
(324 mm × 457 mm × 19 mm), on the columns using 5/16 in (8 mm) fillet welds; or
2. Single plate shear connection with a 9 in × 5 in (229 mm x 114 mm) shear plate with a
thickness of 7/16 in (11 mm). The shear plate is welded to a sacrificial plate, measuring
12-3/4 in × 18 in × 3/4 in (324 mm × 457 mm × 19 mm), on the columns using 5/16 in
(8 mm) fillet welds and bolted to the beam web using three rows of 3/4 in (19 mm)
diameter A325 high-strength bolts (Figure 2(b)). The shear plate uses A36 steel.
3.
4. Fig. 1 – Composite beam specimen and floor slab geometry.
(a) (b)
5. Fig. 2 – (a) Welded-bolted double-angle connection; (b) single-plate shear connection.
The test setup is shown schematically in Figure 3. The composite floor slab extends to the
centerlines of the support columns where it is supported on girders which approximate rigid
vertical support. Two restraint conditions are considered along the short edges of the composite
floor slab: free and fixed. The long edges of the floor slab are braced to prevent twisting of the
test specimen while allowing for vertical displacement (i.e., the concrete slab is permitted to
deflect with the beam, but is not allowed to twist transverse to the axis of the floor beam). The
supporting columns are braced, as shown in Figure 3, to provide horizontal stiffness and
resistance to thermal expansion.
During the fire tests, uniform gravity loading (representative of the service gravity loading
condition) is applied via six equally-spaced concentrated loads along the centerline of the beam
using three water-cooled beams (shown in blue in Figure 3). The water-cooled beams are
loaded at their ends using hydraulic actuators located below the test floor and controlled to
maintain a constant and equal load in each actuator throughout the duration of the test.
Fig. 3 – Composite beam test setup.
Fig. 4 – Configuration of fire compartment and load-distributing frame.
The thermal load and resulting structural temperatures will be developed and controlled using
three 3.5 ft × 5 ft (1 m × 1.5 m) natural gas burners distributed along the centerline of the
composite beam specimen. The heat release rate (HRR) of each burner will be held constant at
1333 kW (for a total of 4 MW) for 3.5 hours to achieve an upper layer gas temperature of
900 °C. The heating will be followed by a 30 min period of passive cooling. Walls constructed
from gypsum board and lined with thermal ceramic blanket (Kaowool) will be used to confine
the fire below the test specimen, as shown in Figure 4. Ventilation for inlet air and outgoing
combustion products is provided through a 26.6 ft2 (2.47 m2) total area of openings distributed
along the bottom portion of the walls.
COMPOSITE SLAB MODELING AND VALIDATION

Prior to developing the pre-test model of the composite beam specimens, a significant effort
was made to develop and validate a reduced-order, shell-element-based modeling approach for
the composite slab (Jiang et al., 2017), in which a layered composite shell formulation with
distinct structural material, thermal material, and thickness was specified for each layer. Distinct
layers were specified for the steel decking and the welded wire mesh reinforcement, with
multiple layers representing concrete specified through the thickness of the slab. The periodic
geometry of the slab was captured using alternating strips of shell elements representing the
thick portions (Shell A) and thin portions (Shell B) of the composite slab, as illustrated in
Figure 5.
Heat input from the web of the steel decking had a significant influence on the temperature
distribution within the thick portion of the slab. In this reduced-order modeling approach, the
specific heat of concrete in the rib of the steel decking was modified to indirectly account for the
additional heat input through the decking web, since only thermal loading on the upper and
lower flanges of the decking can be explicitly modeled in the layered shell approach. Reducing
the specific heat increases the thermal diffusivity, thereby increasing the rate of heat flow
through the rib. An optimum ratio of the modified specific heat (c′p) in the rib to the actual value
(cp) was determined by Jiang et al. (2017), based on the thickness of the upper continuous
portion of the slab, h1, relative to the thickness of the rib, h2.
Shell A Shell B
l1 ∕ 2 l3 ∕ 2
l1 ∕ 2 l3 ∕ 2
Shell A Shell B
Concrete topping
(Shells A and B): h1
• Density:  0
h1 • Specific heat: cp Upper flange of
w4 ∕ 2 steel decking
Concrete in rib (Shell A):
• i  0 ( wi l1 ) h2 w3 ∕ 2
h2 “Dummy material”
• cp   c p w2 ∕ 2
(Shell B):
h2 w1 ∕ 2
Lower flange of • Low specific heat
steel decking l2 ∕ 2 • High through-
thickness thermal
l2 ∕ 2 Convection and radiation boundary conditions conductivity
(applied to lower surface of Shells A and B)
(a) (b)
Fig. 5 – (a) Representation of composite slab using alternating strips of shell elements, and (b)
layered-shell representation of thick and thin portions of composite slab.
1220 (48″)
184 ( ″) ∅3.4@152 (6x6W1.4xW1.4)

159 6.4 (0.25″) 83 ( ″)
( ″)
76 ( ″)
60 121 121
( ″) ( ″) ( ″)
(a) (b)
Fig. 6 – Exploratory composite slab tests: (a) setup and (b) geometry. Dimensions in mm (in).
Reduced-order model
Reduced-order model – Deck

High-f idelity model – 6 mm abov e deck
(a) (b)
Fig. 7 – Comparison of measured and computed temperatures of from the thick portion of
composite slabs: (a) steel decking; (b) reinforcement.
The accuracy of the reduced-order composite slab modeling approach was evaluated against
three exploratory fire tests on small composite slabs supported by two W5x19 steel beams
tested at the NFRL, shown in Figure 6. The slabs were 4 ft (1.22 m) long and 3 ft (0.91 m) wide,
and were heated from beneath the steel decking with a natural gas burner. Similar to the full-
length composite beam specimens, the slabs used Vulcraft 3VLI decking, with a 3.25 in
(83 mm) concrete topping on 3 in (76 mm) deep ribs. Lightweight concrete was used with a
measured density of 129 lb/ft3 (2068 kg/m3) and measured moisture content of 9.2 %. The
thermal conductivity and specific heat of concrete at ambient temperature were measured as
2.1 W/(m·K) and 938.7 J/(kg·K), respectively. The temperature-dependent thermal properties of
the concrete were based on Eurocode 4 (EN 2005), scaled per the measured values at ambient
temperature. The convection and radiation boundary conditions used in the heat transfer
analysis were consistent with those used by Jiang et al. (2017).
Figure 7 presents a comparison of measured and computed temperature histories of the steel
decking and reinforcement in the thick portion of the slab (Point M). The difference between the
measured and computed temperatures of the decking (nearly 200 °C difference at 60 minutes)
may be partially attributable to the mounting of the thermocouple on the top surface of the
decking, which resulted in the thermocouple being partially embedded in concrete. Some
temperature lag relative to the steel decking would thus be expected, because of the steep
temperature gradient in the concrete slab near the decking. To investigate this effect, high-
fidelity finite element analyses were performed using the approach presented by Jiang et al.
(2017), and Figure 7(a) shows the computed temperature history at 1/4 in (6 mm) above the
decking (the first solid-element node in the concrete above the decking), which agreed well with
the temperature measured by the thermocouple. Another factor that may have contributed to
the lower measured temperature of the steel decking is debonding of the decking from the
concrete slab, which was observed after the tests. This debonding would have disrupted the
heat transfer between the decking and the concrete and may have contributed to the lower
temperatures measured by the thermocouple on the top surface of the decking. The predicted
temperature of the reinforcement (Figure 7(b)) agreed well with the test results. Retardation of
heating due to evaporation at 100 °C can be observed in the measured results in Figure 7(b).
The effect of moisture content is accounted for in the models through an increase in the specific
heat between 100 °C and 200 °C (Jiang et al. 2017), which does not capture the plateau at
100 °C seen in the measured data. However, this discrepancy between the measured and
computed results is not significant.
The structural response of the reduced-order composite slab modeling approach has been
validated previously for composite floor systems subjected to column loss (Main and Sadek,
2012; Main et al. 2015). Further validation under combined thermal-structural loading is
ongoing. The concrete, steel decking, and welded wire reinforcement were all modeled using
temperature-dependent stress-strain curves and thermal expansion coefficients based on
Eurocode 2 (EN 1992-1-2, 2005). Spalling of concrete at elevated temperatures was not
considered in the modeling.
PRE-TEST COMPOSITE BEAM MODEL

The beam, connections, and composite slab on steel decking were modeled using a reduced-
order shell-based modeling approach that enabled coupled thermal-structural analysis by using
the same finite element mesh for the heat transfer and structural analyses (Figure 8). This
approach includes (i) realistic fire-induced thermal loads generated using the Fire Dynamics
Simulator (FDS), (ii) temperature-dependent thermal and mechanical material properties to
account for the degradation of material characteristics, strength, and stiffness at elevated
temperatures, (iii) temperature-dependent layered shell elements for the framing members and
connections, and (iv) the newly developed reduced-order shell-based modeling approach for
the composite slab on steel deck described above. The model used the measured concrete
strength and expected properties (yield strength of 55 ksi (379 MPa)) for the A992 steel. The
steel decking and welded wire mesh materials were modeled as elastic-perfectly plastic with
yield strengths of 45 ksi (310 MPa) and 75 ksi (517 MPa), respectively.
The NIST Fire Dynamic Simulator (FDS) was used to estimate the thermal loads that the
specimen would experience during the fire, using a computational fluid dynamics model of the
compartment and the composite floor. These thermal loads were transferred to the model for
thermal analysis via convective and radiative heat transfer boundary conditions, assuming a
uniform temperature in the upper gas layer. The maximum upper gas layer temperature
predicted using FDS was approximately 1650 °F (900 °C). The gas temperature history is
presented subsequently with the analysis results.
Fig. 8 – Details of composite beam shell-based modeling approach.
The boundary conditions in the model were specified to represent those of the composite beam
tests. In the analysis presented in this paper, the bases of two end-columns were modeled as
fixed; the column restraint conditions will be updated based on stiffness test of the column
bracing system prior to the experiments. The gravity loads were applied to six steel loading
plates in contact with the top surface of the composite slab (Figure 9), with the gravity loads
transferred to the steel beam through contact interfaces between the loading plates, the slab,
and the top flange of the beam. Consistent with the applied gravity loading scheme, the loading
plates were free to rotate with the slab, but were permitted to translate only in the vertical
direction. Lateral displacements of both slab edges were restrained through contact with vertical
steel plates at 1/6, 1/2, and 5/6 of the beam span.
Each shear stud was modeled using a solid hexahedral element (Figure 10(a)), which provided
the same load-slip behavior for relative shear of the top and bottom surfaces in any direction.
Rigid-body constraints were used to tie solid elements to the respective nodes on the beam
flange and the composite slab. The shear studs were heated through thermal contact with the
top surface of the beam’s top flange. The load-slip response of the solid elements (representing
the shear studs) at ambient temperature was calibrated to achieve a target load-slip behavior
based on Ollgaard et al. (1971). Degradation in the strength and stiffness of the shear studs at
elevated temperatures was determined based on data from high-strength bolts (Figure 10(b)).
The temperature-independent slip at shear stud failure, at peak load, was based on Dara
(2015).
Fig. 9 – Overview of 42 ft (12.8 m) composite beam model. Vertical arrows correspond to

locations of applied gravity loads.
120
20 °C
300 °C
100
400 °C
80
Load (kN)
500 °C
60
40
600 °C
20 800 °C
0
0 5 10 15
Slip (mm)
(a) (b)
Fig. 10 – (a) Solid hexahedral element representing shear studs, and (b) temperature-
dependent shear stud load-displacement behavior (1 kN = 0.225 kip; 1 mm = 0.0394 in).
The welded-bolted double-angle connections were modeled using a relatively coarse mesh of
shell elements. The behavior of the connections was modeled by calibrating the temperature-
dependent material models of the shell elements in the beam web and the angles to achieve
target capacities and load-displacement responses. Specifically, the stress-strain curves of the
shell elements in the beam web local to the connection (Figure 11(a)) were used to capture bolt
shear and bearing-induced deformations of bolt holes, and the stress-strain curves of the shell
elements in the angles (Figure 11(b)) were calibrated to model angle opening. Figure 11(c)
shows the tensile load-displacement response of a single bolt row of the welded-bolted angle
connection at ambient temperature. A triaxiality-dependent erosion criterion was calibrated to
represent tearing of the angles at their heel or welded edge. Heat transfer between angles,
beam, and column plate was modeled using surface-to-surface thermal contact.
40
Load per Bolt Row (kN)
30
20
10
0
0 10 20 30 40 50
Axial Displacement (mm)
(a) (b) (c)
Fig. 11 – Welded-bolted web angle connection: (a) shell elements in beam web local to the
connection, (b) shell elements representing the angles, and (c) ambient temperature tensile
load-displacement response of a bolt row of the connection (1 kN = 0.225 kip; 1 mm =
0.0394 in).
RESULTS FROM COMPOSITE BEAM PRE-TEST ANALYSIS

Preliminary results from the pre-test analysis of the full composite beam are presented herein,
with no sprayed fire resistive materials, to illustrate the capabilities of the modeling approach.
The first analysis was considered push-down loading at ambient temperature, with the gravity
loads increased until failure. The structural analysis was conducted using an explicit numerical
scheme. The results in this section present the mean-case response based on a best estimate
of the models’ input parameters. The thermal and structural analyses, however, are subject to
many sources of uncertainties, including uncertainties associated with the ambient and
elevated-temperature mechanical properties of the steel and concrete, input temperature time-
history, and restraint conditions. Analysis of uncertainties is currently underway to first identify
the most influential parameters that affect the response using design of experiments
methodology, and then to bound the analysis results by providing a range of responses for two
cases: most severe and least severe.
Figure 12 shows the results of the structural analysis under push-down loading. As the vertical
loading on the beam increased, the beam deflected, inducing rotations at the double-angle
connections. The abrupt increase in horizontal load at approximately 9 in (229 mm) of vertical
displacement at midspan corresponded to binding of the beam web against the column
(Figure 13(a)). The ultimate capacity of the composite beam was associated with failure of
shear studs and failure of the connections by tearing of the angles at their welded edges.
Failure occurred at a vertical displacement of 16.9 in (429 mm) at midspan and a peak vertical
load of 24.3 kip (108 kN) per loading point.
120 300
Horizontl Force at Column (kN)
100 250
Vertical Load per Point (kN)
80 200
60 150
40 100
20 50
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Vertical Displacement at Midspan (mm) Vertical Displacement at Midspan (mm)
(a) (b)
Fig. 12 – Analysis results for ambient-temperature push-down loading: (a) vertical load per
loading point; (b) horizontal force at exterior columns (1 kN = 0.225 kip; 1 mm = 0.0394 in).
Failure of shear studs
Tearing of angle
at welded edge
Binding of beam web
(a) (b)
Fig. 13 – Analysis results for ambient-temperature push-down loading: (a) Binding of beam web
on column flange; (b) failure of shear studs and tearing of angles at welded edge.
1000
Midspan Deflection (mm)
600
Horizontal Reaction (kN)

800
450
600
300
400
200
150
0 0
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Time (min) -150
Time (min)
(a) (b)
800
Temperature ( C)
600
400
Gas
Web: mid-height
200 Bottom flange
Top flange
Slab (thick part)
0
0 4 8 12 16 20 24
Time (min)
(c)
Fig. 14 – Thermal-structural analysis results: (a) midspan deflection, (b) horizontal force at
exterior columns, and (c) temperature distribution (1 kN = 0.225 kip; 1 mm = 0.0394 in).
In the second pre-test analysis, the gravity loading of 10 kip (45 kN) per loading point was
gradually increased prior to heating and then held constant during the heating. The thermal
loads were applied to the exposed surfaces using convective and radiative boundary condition,
based on the average temperature time-history for the upper gas layer obtained from the FDS
simulation of the fire (Figure 14(c)). The analysis was performed using implicit thermal and
explicit structural analyses. To overcome the computational costs associated with running
explicit structural analyses over a long duration, the thermal and structural analyses were
performed on different time scales, with a 4.4 hour (16,000 s) fire duration corresponding to a
duration of 16 s (16,000 ms) in the structural analysis. Thus, different base units were used in
defining properties for thermal materials (N, m, s) and structural materials (N, m, ms). This 16 s
duration for the structural analysis was sufficient to ensure a quasi-static response prior to
failure, with the kinetic energy remaining less than 5 % of the external work.
The results of the combined thermal-structural analysis are shown in Figure 14. The abrupt
increase in the horizontal reaction after about 4.5 minutes of heating (at a midspan deflection of
4.4 in (112 mm)) was due to binding of the beam web on the column flange. The binding was
due to the combined effects of temperature-induced thermal expansion of the beam and
rotation of the beam ends at the web angle connections. As the heating continued, the binding
force on the beam flange increased, initiating local buckling of the beam web after about 15.5
minutes of heating at a midspan displacement of 16.2 in (411 mm), after which the horizontal
reaction at the columns decreased and the midspan deflection increasing more rapidly. Failure
of the unprotected composite beam occurred after only about 20 minutes.
SUMMARY
This paper presented preliminary results from pre-test analysis of a full-scale 42 ft (12.8 m)
span composite beam with double-angle connections, subjected to (i) push-down analysis
under uniform gravity loading and (ii) constant uniform gravity loading and a 4000 kW
compartment fire. Results from the push-down analysis showed that failure of the shear studs,
along with nearly simultaneous rupture of the welded-bolted double-angle connections, limited
the capacity of the composite beam. Results under uniform gravity loading and heating showed
that beam binding due to thermal expansion in the beam was sufficient to induce localized web
buckling in the beam, which in turn decreased the horizontal resistance of the composite beam.
The unprotected beam failed after only about 20 minutes of heating. This paper also introduced
a reduced-order modeling approach for thermal and structural analysis of composite slabs on
steel decking, which represents the periodic profile of the composite slab using alternating
strips of layered shell elements. The performance of this reduced-order approach for heat
transfer analysis was validated against full-scale composite slab subassembly tests subjected
to real fires. Discrepancies between measured and computed temperatures of the steel decking
may have been influenced by debonding of the decking, but good agreement was observed
between the measured and computed temperatures of the reinforcement.
DISCLAIMER
Certain commercial entities, equipment, products, or materials are identified in this document to
describe a procedure or concept adequately. Such identification is not intended to imply
recommendation, endorsement, or implication that the entities, products, materials, or
equipment are necessarily the best available for the purpose.
REFERENCES
ASTM International (2016), “ASTM Standard E119-16a Standard Test Methods for Fire Tests of
Building Construction and Materials”, Standard E119-11a, ASTM International, West
Conshohocken, PA.
Bundy, M., Hamins, A., Gross, J., Grosshandler, W., Choe, L. (2016). “Structural fire experimental
capabilities at the NIST National Fire Research Laboratory”, Fire Technology, 52(4), 959–966.
Dara, S. (2015). “Behavior of the Shear Studs in Composite Beams at Elevated Temperatures.”
Ph.D. Dissertation, The University of Texas at Austin, TX.
EN 1992-1-2. (2005). Eurocode 2 Design of concrete structures: Part 1.2: General rules, Structural
fire design, European Committee for Standardization (CEN), Brussels.
EN 1994-1-2. (2005). Eurocode 4 Design of composite steel and concrete structures: Part 1.2:
General rules, Structural fire design, European Committee for Standardization (CEN), Brussels.
ISO (2014), “Fire-resistance tests – Elements of building construction” Standard ISO 834-11,
International Organization for Standardization, Geneva, Switzerland.
Jiang J., Main J.A., Sadek F., Weigand J.M. (2017). “Numerical Modeling and Analysis of Heat
Transfer in Composite Slabs with Profiled Steel Decking.” NIST Technical Note 1958, National
Institute of Standards and Technology, Gaithersburg, MD.
Main, J.A., Weigand, J.M., Johnson, E., Francisco, T., Liu, J., Berman, J.W., and Fahnestock L.A.
“Analysis of a Large-Scale Composite Floor System Test under Column Loss Scenarios.”
Proceedings of the 2015 ASCE/SEI Structures Congress, Portland, Oregon.
Main, J.A. and Sadek, F. (2012). “Robustness of steel gravity frame systems with single-plate shear
connections.” NIST Technical Note 1749, Gaithersburg, MD.
Ollgaard, J.G., Slutter, R.G., and Fisher, J.W. (1971). “Shear strength of stud connectors in
lightweight and normal weight concrete.” Engineering Journal, AISC, 8(2), 55-64.
Phan, L.T., McAllister, T.P., Gross, J.L., and Hurley, M.J. (2010). “Best Practice Guidelines for
Structural Fire Resistance Design of Concrete and Steel Buildings.” NIST Technical Note 1681,
National Institute of Standards and Technology, Gaithersburg, MD.
TOWARDS AN ENGINEERING METHOD FOR FIRE DESIGN OF CON-
CRETE-FILLED STEEL TUBE COLUMNS WITH SOLID STEEL CORE
Martin Neuenschwander
ETH Zurich
Institute of Structural Engineering
Zurich, Switzerland
neuenschwander@ibk.baug.ethz.ch
Markus Knobloch
Ruhr-Universität Bochum
Institute of Steel, Lightweight and Composite Structures
Bochum, Germany
markus.knobloch@rub.de
Mario Fontana
ETH Zurich
Institute of Structural Engineering
Zurich, Switzerland
fontana@ibk.baug.ethz.ch
ABSTRACT
Concrete-filled steel tube columns with solid steel core are innovative pre-fabricated composite
columns with outstanding structural fire behavior, allowing slender designs under high loading
without additional fire protection. Their singular structural behavior in fire stems conceptually
from their specific design of beneficial composite action between the different structural compo-
nents (steel tube, concrete infill and solid steel core). These structural, economic and architec-
tural advantages make these columns appealing for use in high-rise building practice. However,
due to their high load-bearing capacity, experimental investigations of their structural behavior in
fire are substantially hampered by test facility limits and cost. Based on the findings of a previ-
ous combined experimental and numerical study using cutting-edge engineering methods, the
paper develops the grounds for a novel approach towards a simplified fire design method that is
suitable for practitioners and needed for further dissemination of this innovative type of compo-
site column in building practice.
INTRODUCTION
Concrete-filled steel tube columns with solid steel core are pre-fabricated innovative composite
columns that consist of a circular outer steel tube, a concentrically placed solid steel core and
concrete infill in between. This type of composite column is most suitable for use in high-rise
building practice, because of its high load-bearing capacity and the fact that exceptional fire re-
sistances can be reached without additional fire protection. The singular structural fire behavior
is based on an innovative design of composite action between the different components of the
column (steel tube, concrete, steel core). A gradual load-redistribution process from the outer
more quickly degrading components (steel tube, concrete) to the inner thermally shielded steel
core enables the long-standing stability and strength of such composite columns in the case of a
fire. However, the advantage of the high load-bearing capacity in turn hampers experimental
Table 1 – Summary of specimen properties of fire test series [4].
Fire test Dtube t Dcore tc L 20C Npl,20°C NR,20°C P0 μ tf End Condition
[mm] [mm] [mm] [mm] [mm] [-] [kN] [kN] [kN] [-] [min]
1 133.0 4.0 60.0 32.5 3540 1.515 a) 1876.4 a) 579.6 85 0.16 94 Pinned-Pinned
2 219.1 4.5 110.0 50.1 3600 0.687 a)
4911.0 a)
3497.2 a)
2000 0.57 24 Pinned-Fixed
3 219.1 4.5 110.0 50.1 3600 0.687 a)
4911.0 a)
3497.2 a)
4 219.1 4.5 150.0 30.1 3600 0.652 a)
6638.5 a)
4972.5 a)
a) According to Eurocode 4, calculated with experimentally established mean values of material properties
investigations. The compilation of an extended fire test database that would be needed for the
development of simplified fire design methods for such columns, is therefore not only very costly
but also partially infeasible due to capacity limits of test facilities. Considering the potential of
today’s advanced numerical methods, it appears therefore appealing to strive to substitute cost-
ly fire tests of concrete-filled steel tube columns with solid steel core with numerical simulations,
and develop on the basis of such numerically compiled databases simplified fire design meth-
ods that are needed to further disseminate this innovative type of composite column in high-rise
building practice.
In the case of ordinary concrete-filled steel tube columns this strategy has been followed by
several researchers in the past few years. This lead to the development of several simplified fire
design methods [1-3], which were derived from databases of simulated fire tests that were com-
piled with numerical models considering temperature-dependent code material models and non-
varying nominal boundary conditions. However, the first comprehensive experimental and nu-
merical study on the fundamental structural fire behavior of concrete-filled steel tube columns
with solid steel core [4-8] showed that replacing fire tests using this approach is not suitable for
this column type. This paper presents and validates an alternative modeling approach that is
suitable to numerically extend a sparse experimental fire test database of concrete-filled steel
tube columns with solid steel core for the purpose of developing a simplified fire design method.
PREVIOUS RESEARCH
A combined experimental and numerical research project on the fundamental structural fire be-
havior of concrete-filled steel tube columns with solid steel core has been recently accomplished
at ETH Zurich [4-8]. The experimental part of the project consisted of a series of four full-scale
ISO Standard fire tests with pre-loaded columns of different cross-sectional type, nominal
boundary conditions and load ratios as summarized in Table 1. The test series was performed
in the structural fire lab of the Materials Testing Institute in Brunswick in Germany. A detailed
description of the test setup, the results of the fire tests and of the post-test examination of the
specimens is comprehensively documented in [8]. Moreover, the temperature-dependent consti-
tutive behavior of the different materials of the full-scale composite column specimens was es-
tablished with an extensive experimental program performed at ETH Zurich [5-7]. The program
encompassed strain-rate controlled steady-state tensile tests at elevated temperatures of the
tube and the core steels, and strain-rate controlled steady-state cyclic compression tests at ele-
vated temperatures of the concrete. The numerical part of the project consisted of a compre-
hensive Finite-Element-Method (FEM) modeling study on the simulation of the complex thermo-
mechanical behavior of this type of composite columns in fire and the feasibility of replacing
costly full-scale fire tests with such numerical simulations [8]. The study showed that a best-
calibrated model can accurately reproduce the behavior observed in full-scale fire tests,
Fig. 1 – Predictive capacity of available simplified fire design methods for ordinary concrete-
filled steel tube columns: fire design method of [1] against test data of [10] and [11] (a) and fire
design methods of [2] and [3] against NRCC data [12].
when considering experimentally established temperature-dependent material behavior and in-

fluences of the test-setup that are varying over the course of the test time. When neglecting the
varying influences of the test-setup by using a model with non-varying nominal boundary condi-
tions however the comparison with the test data becomes inconsistent and allows only for a
conditional validation of such a model. This observation is exactly the starting point of the follow-
up research project on simplified fire design methods for concrete-filled steel tube columns with
solid steel core that will be presented in this paper, because this observation impedes a straight-
forward application of the methodology that was used successfully in the case of ordinary con-
crete-filled steel tube columns.
SIMPLIFIED FIRE DESIGN METHOD

Simplified design methods in structural engineering traditionally idealize the load-carrying be-
havior of structural members by conceptual mechanical models that have been derived on the
basis of experimental observations. The effectiveness of such methods for engineering practice
lies within the relative simplicity of predicting the capacity of a given structural member for a de-
sign. Usually this is achieved by sacrificing some accuracy for the benefit of conceptual simplici-
ty and/or ease of use in engineering practice. Often, the general range of applicability of such
methods is however restricted by the extent of the experimental database on the basis of which
they have been developed.
In structural fire design, the extent of available experimental databases is especially limited.
This is because experimental investigations are costly, often restricted by the capacity of test
facilities, and very complex and difficult to conduct in order to give meaningful results. In order
to provide the grounds for the development of simplified fire design methods, it appears there-
fore promising, using advanced numerical methods to extend existing experimental databases
or to compile numerical ones instead. One strategy to achieve this goal consists of validating in
a first step a numerical model against selected tests from a given experimental database. Typi-
cally, a numerical model is used that considers temperature-dependent code material models
(e.g. Eurocode or ASCE) and non-varying nominal boundary conditions. In a second step, a
comprehensive numerical database is compiled by performing extended parametric studies with
the validated model. It is noteworthy to realize that this strategy implicitly assumes that such a
selectively validated model can be arbitrarily generalized, completely irrespective of specific
conditions of the underlying validation tests. Using this strategy, several research projects have
successfully proposed simplified fire design methods for ordinary concrete-filled steel tube col-
umns in the past few years [1-3], among which the most recent and comprehensive one [1] is
evaluated to eventually substitute the currently suspended simplified fire design method (Annex
H) of Eurocode 4 [9]. Figure 1 illustrates the predictive capacity of these simplified fire design
methods with respect to test data. Figure 1a shows with round markers the results obtained with
the method of [1] when predicting the tests of [10] and [11], whereas Figure 1b gives with trian-
gular and square markers respectively the results when using the methods of [2] and [3] to pre-
dict the NRCC data [12]. It follows from Figure 1a that the proposed new method which will
eventually be considered in Eurocode 4 gives very conservative results with the linear trend-line
almost coinciding with the -40%-deviation line and the markers of the four lowest predicted val-
ues almost lying on the -60%-deviation line. Despite the advantage of a wide range of applica-
bility of this method including many different cross-section types (square, rectangular, tubular
and elliptical), high slenderness ratios and very large eccentricities, the sheer conservativeness
of the method gives rise for debate.
It is a generally recognized issue that the results of fire tests with identical structural members
and nominally identical test conditions can vary greatly depending on the test facility. There are
two main sources of influence that can explain the different behavior. The first can be assigned
to deviations in the fire exposure leading to different heating histories of a test member and fi-
nally different structural behavior. However, considerable research effort has been undertaken
in this field that mitigated the problem and allowed achieving a high level of consistency among
different fire test labs through suitable standardization. The second source of influence is the
concrete implementation of a member's nominal test boundary conditions, which is not stand-
ardized and naturally varies, as every test lab has its own best solution. However, not only the
aspect of the concrete implementation of boundary conditions should be taken into account, but
also how they possibly vary over the course of a fire test. Despite that this is a generally accept-
ed issue, very little research effort has been undertaken thus far to understand this possible
source of influence and especially, to assess its relevance for the conception of simplified fire
design methods.
Just very recently, this problem has been addressed for the first time with the case of concrete-
filled steel tube columns with solid steel core [8]. The study showed that the complex behavior in
fire of such composite columns (varying in cross-section type, slenderness and load ratios) can
be accurately simulated when considering not only experimentally established temperature-
dependent constitutive material behavior but also test-setup specific influences via varying
boundary conditions. In contrast, when using non-varying nominal boundary conditions the sim-
ulations compared inconsistently and overall poorer with the test data. This is illustrated in Fig-
ure 2 that compares the model responses obtained with varying boundary conditions and non-
varying nominal boundary conditions (blue dashed and red dotted lines respectively) against the
deformation-time histories of the axial displacement, u(t), and the lateral deflection, w(t), of two
different fire tests of [4] (black solid lines). In some cases, as the one shown in Figure 2a, the
model yielded also with non-varying nominal boundary conditions robust and reasonably con-
servative predictions of the fire resistance, whereas in other cases, as the one illustrated in Fig-
ure 2b, the non-varying nominal boundary conditions triggered in the model a local failure
mechanism that was however never observed in the tests, and lead to excessively conservative
predictions of the fire resistance (30 minutes versus 169 minutes in the test).
Fig. 2 – Influence of boundary conditions when modeling concrete-filled steel tube columns with
solid steel core: model responses versus test data of fire test 4 (a) and fire test 3 (b).
In the framework of simplified fire design methods this finding shows two things: It suggests first
that a different strategy than the one using non-varying nominal boundary conditions is needed
when compiling a numerical database for concrete-filled steel tube columns with solid steel
core, and second that a suitable alternative strategy consists of considering test-setup specific
influences via varying boundary conditions. Such an alternative strategy adapts the spirit of the
well-established principle – “built into a structure as tested” – of traditional experiment-based
structural fire engineering to today’s situation, where combined experimental and numerical
methods are used, and especially advanced numerical methods partially allow replacing fire
tests by simulations. In the context of replacing fire tests with numerical simulations, the afore-
mentioned principle then rather reads – “simulated as tested” – which specifically includes also
the simulation of boundary conditions and their varying character over the course of a fire test.
This novel strategy allows compiling a numerical database that has the character of an extend-
ed experimental database of a specific test-setup and excludes potentially over-conservative
simulation results that are related to non-varying nominal boundary conditions. Finally, resulting
simplified fire design methods from such a numerical database are also conceptually more suit-
able for engineering practice, because the configurations in which composite columns are used
in real structures are closer to the varying boundary conditions of fire test-setups than they are
to the non-varying nominal boundary conditions of purely numerical simulations.
Using this novel approach, the present study aims at computing for different fire ratings the
slender buckling strength of concrete-filled steel tube columns with solid steel core of different
length, in order to extend the experimental database of [4] without the need of additional costly
fire tests in the same lab. A simplified cross-section independent fire design tool can then be
derived when the resulting database is transformed into the format of normalized slender buck-
ling strength curves.
NUMERICAL EXTENSION OF EXPERIMENTAL DATABASE

This section in detail presents, how a sparse database of fire tests of concrete-filled steel tube
columns with solid steel core can be extended to a numerical database, by using the novel
strategy that was outlined in the preceding section. In a first step, a numerical model is devel-
oped that simulates a reference test from the fire test database at hand (fire test 4 of [4] in the
Fig. 3 – Schematic illustration of numerical extension of experimental database.
present case). Test-setup specific influences on the reference test specimen and its interaction
with the test rig are considered in the model by means of varying boundary conditions that are
calibrated such that the model response compares reasonably well with the measured global
deformation-time histories of the reference fire test. The calibrated model is then used to com-
pute at slenderness ratios that are relevant for design practice the corresponding points on the
slender buckling strength curves of given fire ratings, as illustrated in the schematic graph of
Figure 3. For every such point, an iterative calculation procedure is necessary to find the pre-
load a column of the given slenderness ratio can withstand for the time of the fire rating. The
slenderness ratio is varied in the model by changing only the length of the reference test speci-
men (Figure 3), whereas the cross-sectional type and the boundary conditions remain un-
changed. Finally, the resulting relative slender buckling strength curves are then compared
against the results of two other fire tests that have not been used to calibrate the varying
boundary conditions (fire test 2 and 3 featuring different cross-sectional type but identical nomi-
nal test boundary conditions as the reference test), in order to validate the slender buckling
curve approach of the proposed simplified fire design method.
The section is organized in three subsections: The first subsection describes in brief the FEM
model that was used in order to simulate additional fire tests on the test rig on which the refer-
ence test had been performed. The second subsection presents the calibration procedure that
was used to develop the non-varying boundary conditions that incorporate the characteristics of
the simulated test rig. The third subsection finally presents the simulation results and discusses
the validation of the simplified fire design method that can be derived on the basis of the simula-
tions.
Table 2 – Summary of meshing parameters.

 20  C L Elements axial direction Elements tangential direction Elements radial direction
qore qconcrete qtube mcore mconcrete mtube ncore nconcrete nube
[-] [mm] [-] [-] [-] [-] [-] [-] [-] [-] [-]
0.50 2692 112 178 202 9 18 28 3 2 2
0.75 4038 168 266 302 9 18 28 3 2 2
1.00 5384 224 356 404 9 18 28 3 2 2
Fig. 4 – Schematic diagram of numerical model (a), column specimen after fire test 1 with a fire
resistance of 94 minutes (b), fire test 2 with a fire resistance of 24 minutes (c) and fire test 4 with
a fire resistance of 179 minutes (d).
Numerical model
A three-dimensional model of the column of the reference fire test (fire test 4) with geometric
cross-sectional properties according to Table 1 was build using the implicit FEM code ABAQUS
Standard. This subsection summarizes only the most relevant model features, as a detailed de-
scription is available in [8]. Exploiting the radial symmetry of the cross-section allowed modeling
only half of the column as shown in the schematic diagram of Figure 4a that indicates also the
orientation of the model with respect to the global axis system. The different components of the
column were modeled independently with continuum elements of reduced integration (C3D8R
elements of the ABAQUS library) in the case of the steel core and the concrete infill, and with
continuum shell elements (SC8R elements) in the case of the steel tube. The steel-concrete in-
terfaces between the components were idealized by means of separable contacting surfaces
with a temperature-independent Coulomb friction model as mechanical interaction (µf = 0.1). The
end plates at the bottom and the top of the column were represented by analytical rigid surfac-
es. The eccentric loading condition of the test setup (e = 10 mm) was considered by offsetting in
global X-direction the reference nodes of the analytical rigid surfaces, in which also the pre-load
as well as the boundary conditions were prescribed. The data of the elevated-temperature ma-
terial testing program performed at ETH Zurich [5-7] was considered as constitutive input data in
the concrete damaged plasticity model and the classical metal plasticity model of ABAQUS
Standard. Experimentally established temperature-dependent thermal expansion of the steel [7]
was considered, whereas for the concrete the relation according to Eurocode 4 [4] was used.
Moreover, residual stresses resulting from the cooling phase after hot-rolling were considered in
the steel core according to the computed distributions by [13], whereas residual stresses and
strains due to cold-forming of the steel tubes were neglected.
The simulation of the fire tests was performed as a sequentially coupled thermal stress analysis
consisting of two steps. In the first step, the temperature-time histories of the nodes in the col-
umn model were computed with a heat transfer analysis. The heat transfer from the gas in the
furnace following the ISO Standard temperature-time curve to the column specimen was mod-
eled with convective and radiative boundary conditions using a convection coefficient,
h = 25 W/m2K and a resulting emissivity, ε = 0.7. The heat transfer across the outer steel-
concrete interface between the steel tube and the concrete was modeled as a gap conductance
with an effective heat transfer coefficient, kgap = 100 W/m2K, to account for a thermally induced
air-gap, whereas perfect contact was assumed for the inner interface between the concrete and
the steel core. The predictive capacity of this heat transfer model has been validated with tem-
perature recordings from tests of columns of two different cross-sectional types and fire expo-
sure times up to three hours in [8]. In the second step, the fire tests were simulated by incre-
mentally applying to the nodes of the pre-loaded columns the temperature-time histories from
the previous heat transfer analysis while simultaneously ensuring static equilibrium of the heat-
ed column until eventual failure.
Calibration of varying boundary conditions

This subsection presents the simplifying mechanical modeling assumptions that were developed
to represent via varying boundary conditions the concrete implementation of the nominal
boundary conditions and their changing nature over the course of the reference fire test (fire test
4). The starting point for this development constituted the comparison between the simulation
results that were obtained with a model using non-varying nominal boundary conditions and the
test recordings. The two deformation recordings that appeared to be difficult to track with the
model were the lateral deflection, w(t), and the rotation of the bottom end plate, φ(t). Figure 5a
and 5b respectively show with dotted red lines the predicted courses of these two deformations
over the test time, t, in comparison to the test recordings that are given with black solid lines.
The graphs in Figure 5 show two principle differences between the model response and the be-
havior observed in the test. First, it follows that at the start of the fire exposure at t=0 minutes
the lateral deflection and the rotation of the bottom end plate are predicted significantly smaller
than actually measured. Second, it is apparent that the predicted rotation of the bottom end
plate starts increasing significantly after 60 minutes of fire exposure, whereas in the test the ro-
tation remained stable over a more extended period of time. The observed deviations in the
model response can be attributed to the boundary conditions, because the influence of the ma-
terial behavior is already accurately captured in the model through experimentally established
constitutive calibration data at ambient and elevated temperature. In the next two paragraphs,
these deviations will be discussed in relation to the test end conditions of the column specimen.
From the first observation of Figure 5 follows that the nominally fixed upper end of the column
was in reality compliant in the test-setup. Assuming that the test rig remains elastic and at am-
bient temperature during the fire test, the upper boundary condition was modeled as an elastic
rotational restraint with a constant spring stiffness, ktop = 1.85 kNm/mrad (Figure 4a). The stiff-
ness of the spring was calibrated by matching the initial lateral deflection, w0, and the bottom
end plate rotation, φ0, of the model with the test recordings. This modeling approach implicitly
considers that the relative degree of rotational restraint by a test rig varies during the course of a
fire test, because it depends on the stiffness ratio between the test rig and the specimen. The
longer a fire test takes, the greater a specimen’s temperature-induced degradation of stiffness
Fig. 5 – Calibration of boundary conditions with reference fire test: simulated lateral deflection-
time histories (a) and rotation-time histories of bottom end plate (b).
and strength and in turn the higher the rotational restraint exerted by the test rig that remains in
contrast elastic and at ambient temperature. Figures 4c and 4d clearly illustrate this behavior
with the measured residual rotation of the nominally fixed upper end of the specimen of fire test
2 (Figure 4c, fire resistance tf = 24 minutes) that is significantly greater than in the case of fire
test 4 (Figure 4d, fire resistance tf = 179 minutes) which lasted in contrast almost 3 hours. How-
ever, considering in the model only the nominally fixed upper end with an elastic restraint and
keeping at the bottom end of the column the nominally perfect hinge naturally gives too con-
servative predictions of the fire resistance, as shows Figure 5 with the green dash-dotted lines
obtained with this modeling approach of semi-nominal boundary conditions.
From the second observation of Figure 5 can be concluded that frictional effects must have
been present in the nominally pinned boundary condition at the bottom end of the column that
increased with increasing temperature. Additional support for this explanation gives Figure 4b
that shows the only test specimen of [4] that was tested with nominally pinned boundary condi-
tions at both ends (fire test 1). Despite the symmetric nominal boundary conditions, the residual
end plate rotation at the top was significantly smaller than at the bottom. Comparison with the
temperature recordings in close vicinity to the column’s bottom and top end showed that the
smaller residual rotation at the top correlated with higher temperatures and the greater residual
rotations at the bottom with lower temperatures. This observation indicates that the rocker-
bearing type of implementation of the hinge in the test-setup exhibited with increasing tempera-
ture higher frictional effects that can be associated with deformations in the rocker-bearing [4].
In order to capture this influence, the non-varying nominally pinned boundary condition at the
bottom of the column was replaced in the model by a rotational spring with temperature-
dependent stiffness, kbot,θ, (Figure 4a) that was calibrated such that the simulated deformation-
time history of the bottom end plate rotation compared reasonably well with the test readings.
The resulting temperature-dependent characteristics of the rotational spring are displayed in
Figure 6c.
Figure 4 finally shows that the readings from the reference fire test (black solid lines) can be
predicted reasonably well (blue dashed lines), when using an engineering modeling approach
that considers via varying boundary conditions the compliance of the test rig and temperature-
dependent frictional effects in the nominally pinned boundary condition of the concrete test-setup.
Results and validation of simplified fire design method
The model with calibrated varying boundary conditions of the preceding subsection was used to
simulate a series of fire tests with columns of identical cross-sectional type and nominal bound-
ary conditions as the reference fire test. Within the simplifying mechanical modeling assump-
tions about the test-setup, such a simulated fire test series replaces an extended costly program
on the test rig in the structural fire laboratory of the Materials Testing Institute of Brunswick in
Germany. The goal of this simulation series was to compile the data for a simplified tabular de-
sign tool that predicts the slender buckling strength of concrete-filled steel tube column with sol-
id steel core for a given fire rating. The simulation program was set up in such a way that it pro-
vided slender buckling strength curves at ambient temperature and for ISO fire ratings of 30, 60
90, 120 and 180 minutes that cover most design situations of building practice. Discrete values
of the slender buckling strength curves were computed at slenderness ratios 20C of 0.5, 0.75
and 1.00, covering the range that is relevant for building practice. The different slenderness rati-
os at ambient temperature were realized by changing in the model only the length of the col-
umn. Table 2 summarizes the respective column lengths and mesh parameters that were used
in the simulation program and Figure 6a displays the resulting simplified fire design tool on the
basis of normalized slender buckling strength curves, whose development will be discussed
next. Each discrete value on the slender buckling strength curves was calculated iteratively by
adjusting the pre-load. A pre-load was accepted as the slender buckling strength of a given fire
rating, NR,θ, when the column resisted not more than 30 seconds longer than the required
Fig. 6 – Results and validation of method: normalized slender buckling strength curves for given
fire ratings (a), iteration procedure (b), and temperature-dependent spring stiffness (c).
time of the respective fire rating. Figure 6b illustrates this iteration procedure with the specific
results of a column of slenderness ratio of 0.75 and a fire rating of 180 minutes. The resulting
slender buckling strengths were then normalized with respect to the plastic cross-sectional ca-
pacity at ambient temperature, Npl,20°C = 6683.5 kN (Table 1). Plotting the normalized slender
buckling strength, NR,θ /Npl,20°C , against the slenderness ratio at ambient temperature, 20C , as
shown in Figure 6a, finally provides a simplified fire design tool for concrete-filled steel tube col-
umns with solid steel core that is independent of the cross-sectional type and therefore suitable
for design practice. Moreover, the tool allows to determine for a given design of a column at
ambient temperature the slender buckling strength for a required rating of ISO fire exposure,
without involving any heat transfer calculations.
Figure 6a includes besides the computed normalized slender buckling strength curves also
those test results of [4], namely fire tests 2 to 4 (Table 1) that fall into the range of slenderness
ratios that are covered by the design tool. The results of fire test 2 and 3 are both given with a
square marker (filled and unfilled respectively), to indicate that their specimens were of identical
cross-sectional type, whereas the result of fire test 4 (the reference fire test) of different cross-
sectional type is indicated with a round marker. It is apparent in Figure 6a that the marker of fire
test 2, in which a fire resistance of, tf = 24 minutes was reached, almost coincides with the line-
arly interpolated slender buckling strength curve of the design tool for a fire rating of 25 minutes.
In the case of fire test 3 (tf=169 minutes) however, the tool’s prediction is conservative in a simi-
lar degree as in the case of fire test 4 (tf=179 minutes), which can be seen in Figure 6a by com-
paring the corresponding markers of the test results with the respective interpolated slender
buckling strength curves of the design tool. In the case of fire test 3, a better comparison could
be reached, by improving the calibration of the temperature-dependent rotational spring with the
reference fire test, without compromising the excellent predictive capacity of the design tool at
lower fire ratings.
For two reasons, fire test 2 and 3 are especially suitable to validate the simplified fire design
tool. First, because unlike the reference fire test they have not been used to calibrate the vary-
ing boundary conditions, and second, because they have been performed with column speci-
mens of another cross-sectional type than the one of the reference fire test. The comparison of
the results of fire test 2 and 3 with the tool’s predictions (Figure 6a) therefore confirm the
soundness of the proposed approach of modeling the test-setup specific influences on a speci-
men during a fire test by means of varying boundary conditions, and validate the suitability of
using a method that is independent of the cross-sectional type such as a normalized slender
buckling strength curve.
CONCLUSIONS AND OUTLOOK

A novel modeling approach to extend sparse experimental fire test databases of concrete-filled
steel tube columns with solid steel core has been presented that considers test-setup specific
influences on the specimen during a fire test by means of varying boundary conditions. Using
this novel approach normalized slender buckling strength curves for given fire ratings of 30, 60,
90 120 and 180 minutes were calculated at slenderness ratios of 0.5, 0.75 and 1.0 that can be
used as a preliminary tabular fire design tool (Figure 6a). From the tool’s validation by fire tests
that have not been used in the calibration of the varying boundary conditions, the following con-
clusions and recommendations for future research can be drawn:
- Normalized slender buckling strength curves are conceptually suitable as a simplified fire de-
sign method for this type of composite column.
- Varying boundary conditions constitute a valid engineering approach to appropriately consider
complex test-setup specific influences on a specimen during a fire test when compiling a nu-
merical database for the development of a simplified fire design method.
- Further successful dissemination of this innovative type of composite column in high-rise
building practice depends on the development of a simplified fire design method that is sup-
ported by a numerical database, which reliably excludes over-conservative model predictions
that are related to non-varying nominal boundary conditions.
REFERENCES
[1] V. Albero, A. Espinos, M.L. Romero, A. Hospitaler, G. Bihina, C. Renaud, Proposal of a
new method in EN1994-1-2 for the fire design of concrete-filled steel tubular columns, Eng.
Struct. 128 (2016) 237-255.
[2] A. Espinos, M.L. Romero, A. Hospitaler, Simple calculation model for evaluating the fire
resistance of unreinforced concrete filled tubular columns, Eng. Struct. 42 (2012) 231-244.
[3] C. Ibañez, J.V. Aguado, M.L. Romero, A. Espinos, A. Hospitaler, Fire design method for
concrete filled tubular columns based on equivalent concrete core cross-section, Fire Saf.
J. 78 (2015) 10-23.
[4] M. Neuenschwander, M. Knobloch, M. Fontana, ISO Standard fire tests of concrete-filled
steel tube columns with solid steel core, J. Struct. Eng. (2016) 10.1061/(ASCE)ST.1943-
541X.0001695, 04016211.
[5] M. Neuenschwander, M. Knobloch, M. Fontana, Suitability of the damage-plasticity model-
ling concept for concrete at elevated temperatures: Experimental validation with uniaxial
cyclic compression tests, Cem. Con. Res. 79 (2016) 57-75.
[6] M. Neuenschwander, M. Knobloch, M. Fontana, Generic model stress-strain relationship
for concrete in compression at elevated temperatures, ACI. Mat. J. 114 (2017) 3-14.
[7] M. Neuenschwander, M. Knobloch, M. Fontana, Elevated temperature mechanical proper-
ties of solid section structural steel, Constr. Build. Mat. in press.
[8] M. Neuenschwander, M. Knobloch, M. Fontana, Modeling thermo-mechanical behavior of
concrete-filled steel tube columns with solid steel core subjected to fire, Eng. Struct. 136
(2017) 180-193.
[9] CEN (European Committee for Standardization), Eurocode 4: Design of composite steel
and concrete structures – Part 1-2: General rules – Structural fire design, EN 1994-1-2,
Brussels, Belgium, 2008.
[10] A. Espinos, M.L. Romero, E. Serra, A. Hospitaler, Circular and square slender concrete-filled
tubular columns under large eccentricities and fire, J. Constr. Steel Res. 110 (2015) 90-100.
[11] A. Espinos, M.L. Romero, E. Serra, A. Hospitaler, Experimental investigation on the fire
behaviour of rectangular and elliptical slender concrete-filled tubular columns, Thin-Walled
Struct. 93 (2015) 137-48.
[12] T.T. Lie, M. Chabot, Experimental studies on the fire resistance of hollow steel columns
filled with plain concrete, Internal Report No. 611, Inst. for Res. in Constr., National Re-
search Council of Canada (NRCC), Ottawa, Canada, 1992.
[13] G. Hanswille, M. Lippes, Composite columns made of high-strength steel and high-strength
concrete, Stahlbau 77 (2008) 296-307 (in German).
FIRE BEHAVIOR OF BLIND-BOLTED CONNECTIONS TO CONCRETE
FILLED TUBULAR COLUMNS UNDER TENSION
Dr.-Ing. Ana M Pascual

ampascualpastor@gmail.com
Prof. Dr.-Ing. Manuel L Romero

Universitat Politècnica de València, Spain
mromero@mes.upv.es
ABSTRACT
The blind-bolts allow the connection of steel girders to tubular columns preventing the problems
derived from welding in construction and its maintenance. The behavior of this type of connections
at room temperatures has been covered in several researches. Nonetheless, its performance in
fire is unknown. The aim of this work is to gain insight into the behavior of the blind-bolted
connection exposed to the standard fire curve ISO834 and pull-out load.
With that purpose, a thermo-mechanical numerical model using the Finite Element Method has
been developed. The simulation is a sequentially coupled analysis. The study includes the
comparison between the fire capacity of the connection to a Hollow Steel Section (HSS) column
and to a Concrete Filled Tubular (CFT) column and the assessment of the performance in terms
of fire resistance time.
1. INTRODUCTION
The advantages of CFT columns have been widely demonstrated. Concrete infill enhances the
structural behavior of the HSS tubes, increases the strength of the column and prevents local
buckling of the steel tube. The steel tube provides a confinement to the concrete and improves
its strength and stiffness. Additionally, the insulation and heat sink effect of concrete make them
more resistant under fire conditions.
However, the use of CFT is restricted because of the design of the connection since there is a
lack of suitable simple methods for its structural analysis. In that respect, the component method
in Eurocode 3 Part 1.8 (EN 1993-1-8, 2004) permits to obtain analytically the main structural
characteristics of joints by assembling springs of the contributive parts. But, the components
involved in HSS and CFT columns are still not defined (Da Silva et al., 2003; Lopes et al., 2013;
Pitrakkos & Tizani, 2013).
The bolted connection to tubular columns requires the use of blind-bolts, able to be tightened from
one side of the tube. Up to now, several blind-bolt systems have been developed, as a rule, these
systems provide shear connections. However, some bending moment can be transmitted by the
connections under a suitable design, as it has been proved by many researches (France et al.,
1999a, 1999b, 1999c; Lee et al., 2010).
Fig. 1 – Hollo-bolt. Fig. 2 – Extended Hollo-bolt.
Among the different fastener systems, the present investigation deals with the Hollo-bolt system
by (Lindapter, 2012) (Figure 1) and the Extended Hollo-bolt (Figure 2). The latter is a modified
version of the Hollo-bolt developed in the University of Nottingham (Ellison & Tizani, 2004;
Pitrakkos & Tizani, 2013) in order to transmit bending moment. The Hollo-bolt consists basically
of three elements: a standard bolt, a sleeve with four slots and a cone with a threaded hole where
the bolt is screwed, Figure 1. It has an easy installation: first, the piece is inserted through
clearance holes of the elements to join, then, the bolt is tightened causing the cone to move
against the inner face of the tube and the sleeve’s legs expand until the total clamping force is
transmitted.
Fig. 3 – Connection of a single blind-bolt from tests (Pitrakkos & Tizani, 2013).
The Extended Hollo-bolt (EHB) is similar to the Hollo-bolt (HB) but with a longer shank ended in
a screwed nut (Figure 2) to reduce the stress concentration in steel tube and distributing it within
concrete in order to improve the stiffness connection. Tests in full scale connections (Tizani et al.,
2013) and pull-out tests (Pitrakkos & Tizani, 2013) have given evidence of the Extended Hollo-
bolts to provide a certain stiffness and a subsequent semi-rigid behavior.
Several investigations have been developed to characterize the blind-bolted connections to
tubular columns at ambient temperature, however, no researches have been found related to their
fire performance. Traditionally, connections under fire have been assumed to be colder than the
elements jointed, due to its lower section factor A/V (exposed area A of the element per unit of
length divided by V volume of the element per unit of length), the same protection as the rest of
elements is therefore used. But, in several catastrophic fires (NIST, 2008) the failure of the
connection has been pointed out as the cause of the structural collapse, hence the awareness of
the limited understanding of connection fire performance and the necessity of the oversight of the
forces caused by the beams during the fire.
First studies on the connection fire performance were on steel connections (Al-Jabri et al., 2006)
(Yu et al., 2011) (Huang et al., 2013). The fire behavior of connections beam to CFT in realistic
structures were studied by (Ding & Wang, 2007), who considered five different types of joints, and
(Elsawaf et al.,2011). They demonstrated the importance of the connection flexibility and strength
if the full capacity of the beam is aimed. Reverse channel connections highlighted due to their
desirable ductility, being the objective also of the research of (Lopes et al., 2013).
Some of the aforementioned studies on the fire performance of bolted connections assumed
concrete filled tubular columns but none of them included blind-bolted connections and their fire
behavior. The thermal analysis of blind-bolted connections was covered by the authors in previous
research (Pascual et al., 2015a). The aim of the work here reported is to assess the thermo-
mechanical response of blind-bolted moment-resisting connections, specifically on the tension
zone. For that, FE analysis of the connections to HSS and CFT are developed, with Hollo-bolts
and with Extended Hollo-bolts. Thus, the role of the concrete infill and the bolt anchorage within
the concrete can be observed by comparison between the different types of connections.
2. FE MODELLING OF THE CONNECTION UNDER FIRE

2.1 INTRODUCTION
Three-dimensional numerical models of the connection were elaborated using the FEA package
ABAQUS. In the absence of tests of the blind-bolted connections under fire and pull-out forces to
validate the FE model, the thermal and the mechanical parts were validated separately. The
research covers two connections: single blind-bolt and T-stub connection, and three cases for
each: using Hollo-bolt system to connect a plate with an unfilled tube column (HSS), the same
blind-bolt but in a filled tube (CFT) connection and the Extended Hollo-bolt in a CFT. The geometry
of the connections analyzed can be seen in Figure 3 and Figure 4.
2.2 VALIDATION OF THE NUMERICAL THERMAL MODEL
The thermal behavior of blind-bolted connections was studied by the authors in a previous work
(Pascual, 2015; Pascual et al., 2015a), through experiments and FEA that permitted the
calibration of the thermal FE model. Twelve unloaded small-scale specimens were tested under
Fig. 4 – Double T-stub connection to a HSS 200x200x10 from tests (Ellison & Tizani, 2004)
the standard fire curve ISO834 in a gas furnace. The program was organized in 4 series of 3
different sections. Each series included 3 types of connections: Hollo-bolt in HSS, Hollo-bolt in
CFT and Extended Hollo-bolt in CFT. The difference between the series was the dimension of
the column section: 150x150 and 8mm tube thickness (Series 1), 220x220 and 10mm tube
thickness (Series 2), 250x150 and 10mm tube thickness (Series 3) and 350x150 and 10mm tube
thickness (Series 4). Measurements of temperature were taken at different positions of the bolt
and across the concrete section. In parallel, three-dimensional numerical models for the
connections were developed to simulate the pure heat transfer problem. The thermal properties
of the material and interactions were calibrated through comparison with the experiments.
The accuracy of the FE simulations to represent the thermal behaviour of the connection was
assessed by comparing the temperature-time curves numerically calculated and the ones
measured by the thermocouples in the laboratory. A good correlation (Pascual, 2015; Pascual et
al., 2015a) was observed and consequently the FE model was set to study other parameters.
2.3 VALIDATION OF THE NUMERICAL MECHANICAL MODEL
Before modelling the two configurations (Figure 3 and Figure 4) that are representative of the
tension area of a connection I-beam to CFT column and due to the complexity of the interactions
and non-linear behavior, other connections were modelled. The objective was to ensure with a
large number of simulations the adequacy of the final models used for the thermo-mechanical
analysis, starting from simpler models with standard bolts and only steel elements and finishing
with complex models of the blind-bolted connection to concrete filled steel columns. Table 1
shows a summary of the calibration work, where the authors of the tests, the type of connection
and elements connected are specified, additionally the ratio maximum load from tests to
maximum load from the finite element analysis is in the last column of the table.
Table 1 – List of connections of the calibration work (Pascual, 2015)
Calibration test Type of ξ=Mu,test/Mu,FEM
Type of bolt Beam Beam/Column
(authors) connection (Nu,test/Nu,FEM)
Jaspart & Bursi, 1998 2 T-stub M12 grade 8.8 IPE 300 IPE 300 0.93
Janss et al., 1987 Flush endplate M16 grade 10.9 IPE 300 HEB160 1.01
Wang et al., 2010 2 T-stub M16 grade 8.8 I-section t=15 mm I-section t=15mm 0.98
Wang et al., 2010 2 T-stub HB16 grade 8.8 I-section t=25 mm I-section t=25mm 1.02
Mesquita et al., 2010 Flush endplate HB20 grade 8.8 IPE 330 SHS 200x200x8 0.90
CFT 200x200x12.5
Tizani et al., 2013 Flush endplate EHB16 grade 8.8 356x171x67 1.05
(fc=40N/mm2)
CFT 200x200x10
Tizani et al, 2013 Flush endplate EHB16 grade 8.8 457x152x52 0.99
(fc=40N/mm2)
Extended CFT 200x200x10
Tizani et al, 2013 EHB16 grade 8.8 356x171x67 1.05
endplate (fc=40N/mm2)
CFT 200x200x8
(fc=40N/mm2)
CFT 200x200x8
(fc=40N/mm2)
CFT 200x200x12.5
(fc=40N/mm2)
CFT 200x200x10
(fc=40N/mm2)
CFT 200x200x10
(fc=60N/mm2)
The adequate correlation with the experiments led to the modelling of the two connections, Figure
5 and Figure 6, whose definition was based on the experiments used for the validation (Ellison &
Tizani, 2004; Pitrakkos & Tizani, 2013).
The first connection under study consisted of a single blind-bolt clamping a loading frame plate
and the tube column, Figure 3. The tensile force was transmitted to the blind-bolts through the
rigid loading frame plate, 30mm thick. The tube column was not strictly a real commercial HSS, it
was an arrangement of plates that worked similarly and where the upper plate was 20mm thick.
Fig. 5 – FE model of the single blind-bolt Fig. 6 – FE model of the T-stub connection.
connection.
The thickness of the plates was designed in order to behave in the elastic range and in
consequence, all attention was focused on blind-bolts performance which controlled the behavior
of the connection. Eight specimens from (Pitrakkos & Tizani, 2013) were modelled, they covered
different types of bolt (standard bolt, Hollo-bolt and Extended Hollo-bolt), types of column (HSS
and CFT), bolt grade and concrete resistance.
The second connection involved two T-stubs bolted to opposite sides of a HSS 200x200x10 S375
taken from (Ellison & Tizani, 2004), Figure 4. As in the specimen with the single blind-bolt the
main interest was concentrated on the blind-bolt behavior, so thickness of T-stub flanges was 50
mm. Load was applied by pulling out the upper T-stub while the bottom one was fixed. Two
specimens were calculated and validated, considering the connection to a concrete-filled tubular
(CFT) column with the two different fastener systems: the Hollo-bolt and the Extended Hollo-bolt.
The definition of the FE model for both connections followed essentially the same criteria, Figure
5 and Figure 6 depict the model of each connection. They were symmetrical about the vertical
planes, so only quarter of the specimens was simulated in order to reduce the computational cost.
The fastener was modelled in the tightened state, i.e. assuming the sleeve shape and fastener
cone position after the torque application. Hollo-bolt was simplified into two parts: the first one
included the standard M16 grade 8.8 and the fastener cone, the second one represented the
sleeve in the expanded state. The shank length was 100mm in the single blind-bolted connections
and 120mm length in the double T-stub connections. The Extended Hollo-bolt was similarly
modelled (Figure 5), by means of the two parts, but taking into account the longer shank (150mm
long for both connections) and a nut attached at the end. For the sake of simplicity, the screw
thread in the bolt shank was not considered, nor the hexagonal shape of the bolt head and nut.
The analysis procedure attempted to reproduce the actual execution steps of the specimen. In
the first stage and before the concrete fill took place, the torque was applied using the ABAQUS
“bolt load” function. In the second step, the concrete was inside the tube and the tension load was
transmitted through the loading frame plate or the upper T-stub as a static displacement. The
failure took place when the velocity of the deformation increased abruptly, which coincided with
the lack of equilibrium in the system and the loss of convergence of the Newton’s method.
Three-dimensional eight-noded solid elements with reduced integration (C3D8R) were employed
for all the parts in the connection. Mesh density was finer in areas where higher stress gradient
occurred, i.e. around blind-bolts, where most interactions happened.
Due to the multiple and different parts in contact, the complexity of the numerical model resided
to a greater extent in the interactions definition. A total of 9 contacts were tackled:
 Contacts between steel surfaces: headbolt to plate/T-stub flange, sleeve to plate/T-stub
flange hole surface, sleeve to tube column hole surface, sleeve to fastener cone and
external face of tube column to plate/T-stub flange.
 Contacts between steel and concrete surfaces: internal face of the steel tube column to
concrete, sleeve to concrete, shank to concrete and nut to concrete (Extended Hollo-bolt).
Interactions were defined as surface to surface contact with finite sliding formulation. In normal
direction, “hard” contact behavior was used, whereas in the tangent direction the Coulomb friction
model was employed. By means of the friction coefficient (µ), coulomb friction law relates the
maximum shear stress in the surfaces with the pressure between them, it was assumed µ=0.25
in the interaction steel-concrete. Eventually, µ=0.25 was used in steel-steel contacts except for
sleeve to fastener cone interaction and sleeve to tube hole where was necessary to increase the
sticking area increasing the friction coefficient, otherwise slippage of surfaces introduced bolt
motion and made the convergence difficult.
The plasticity model for steel behavior was isotropic multiaxial with the Von Mises yield surface.
Linear behavior was defined through Young’s modulus (E) and Poisson ratio’s (ν). Standard bolts
that formed the fastener system were high strength steel M16 (16mm diameter) grade 8.8. Strain-
stress curve was defined by means of the data from tests used for the validation (Ellison & Tizani,
2004; Pitrakkos & Tizani, 2013). Sleeve specifications are provided by (Lindapter, 2012). Sleeve
length depends on the total thickness to fasten, i.e. the plate plus the tube thickness. In the single
blind-bolted connections (30+20= 50mm) 63mm sleeve length was prescribed and 84mm in T-
stub connections (10+50=60mm). Regarding sleeve strength, 430MPa are indicated in the
catalogue, although values of yield and ultimate strength from experimental data available in the
literature (Liu et al.,2012) were finally adopted. In a similar way, structural steel for tube section
was of grade S355 but again values of strength were extracted from data test on analogous tubes
(Tizani et al., 2013). The true stress-strain curve was applied.
The concrete damaged plasticity model was selected to define the yielding part of the concrete.
The parameter values that described the yield surface and plastic flow were obtained from
literature (ABAQUS). For the dilation angle (ᵩ), which defined the dilatancy in plasticity, 30º and
15º were used, due to the negligible effect of the confinement there was no difference between
the two values. The ratio (K), which informed about shape of the deviatory plane, was 0.66. The
other three parameters that complete the model definition were: eccentricity (є=0.1), ratio of initial
equibiaxial compressive stress to initial uniaxial yield stress (fb0/fc0=1.16) and viscosity parameter
(0.01). The damaged plasticity model permitted the definition of the plastic regimen under
compression and tension. For concrete compression, the stress-strain hardening and softening
curve was obtained from (EN 1992-1-1, 2004). The post failure tension behavior was defined by
stress-fracture energy law (FIB 2010. Model Code 2010). Concrete of grade C40 was used in
specimens with a single blind-bolt, and C50 in the double T-stub connections.
Eight single connections and two double T-stub connections were simulated and compared with
experimental tests with a good correlation of the curves force-plate separation, Figure 7.
160 700
140 600
120
500
100
Force (KN)
Force (KN)
400
80
300
60
EXP-HB16-100-8.8D-C40-1 (Pitrakkos and Tizani) 200 EXP-TSTUB-HB16-120-88-C50 (Ellison and Tizani)
40 EXP-TSTUB-EHB16-150-88-C50 (Ellison and Tizani)
EXP-EHB16-150-8.8D-C40-2 (Pitrakkos and Tizani)
100 NUM-TSTUB-HB16-120-88-C50
NUM-HB16-100-88D-C40-1
20
NUM-TSTUB-EHB16-150-88-C50
NUM-EHB16-150-88D-C40-2
0
0 0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10 11 Plate separation (mm)
δglobal (mm)
a) b)
Fig. 7 – Comparison of force-displacement curve between experiments and FEA (Pascual,
2015b; Pascual et al., 2015) for a) single connection; b) T-stub connection.
2.4 DESCRIPTION OF THE THERMO-MECHANICAL MODEL

Once the suitability of the mechanical and thermal numerical models to capture the connections
behavior was proved, the fire analysis was carried out. The two blind-bolted connections
representatives of the tension area in moment-resisting connections were studied under tensile
load and elevated temperatures: the single blind-bolted connection and the double T-stub
connection to the tube column, Figures 3 and 4. Three cases were conducted for each one: with
HB and HSS column, HB and CFT column, and EHB and CFT column. The geometry of the
connections defined for the analysis at room temperature was maintained.
Sequentially coupled thermal-stress analysis were carried out, following the recommendations of
(Espinos et al., 2010). In this type of analysis the stress-strain solution depended on the
temperature field but not the opposite. Conversely, in a fully coupled analysis, the mechanical
and thermal response affect each other, the stress and thermal analysis are simultaneous. The
latter procedure is closer to reality, but at the expense of high computational cost and
convergence problems, the accuracy improvement is not worth noting.
In a sequentially coupled thermal-stress analysis two finite element models are needed: a thermal
model and a mechanical model. First, the pure heat transfer model was computed. Temperature-
time curves were obtained for each node and kept to be applied to the mechanical model as a
prescribed thermal load after the axial loading. Afterwards, the mechanical model for the stress-
deformation was developed. Before tensile load application, the stress produced in the steelworks
by tightening the bolts was input as an initial state. In the following stage, and taking into account
that the tube is filled with concrete, load was applied. Finally, the temperature field kept was input
onto the mechanical model while the load was propagated.
3. FIRE PERFORMANCE OF THE CONNECTION
3.1 FAILURE MODE
The plate and tube thickness were designed in order to focus the attention on the fire resistance
consequence of the fastener system. Depending on the type of connection, the failure was
governed by the shank of the bolt or by the sleeve. In case of being dominated by the sleeve,
strength was lower but flexibility higher.
a) b)
Fig. 8 – Mises stress (MISES,T), ultimate steel strength (fu,T), and temperatures (T) along a)
shank for the three cases of the single blind-bolted connections at connection failure; b) sleeve
for the three cases of the single blind-bolted connections at connection failure.
Figures 8a and 8b compare Mises stresses along the shank bolt and the sleeve with their
respective steel strength capacity (fu), immediately before the collapse and for the three types of
connections. They show the part of the connection (shank or sleeve) that reached the ultimate
capacity and indicate the section where failure was assumed and which is the same for all of the
connections. For the shank bolt, the critical section was next to the bolt head (Figure 8a), where
temperature was highest and steel strength is consequently lowest. Meanwhile, in the sleeve, the
highest stresses were concentrated around the folded section, Figure 8b. Figures 8a and 8b
include a second axis that reflects the temperature along both parts. It can be observed that
fracture occurred around 500ºC, when steel strength decrease became more significant and was
reached by Mises stress.
In the HB connection to the HSS column, sleeve and shank failed at the same time, Figure 8a
and 8b. However, in connections to CFT columns (HB and EHB) the failure was dominated by
the shank fracture, due to the fact that stresses distributed through the concrete reducing sleeve
deformation. Finally, the ultimate strength capacity of the shank bolt was reached in all of the
three connection types.
The FE model of the double T-stub bolted connections permitted to gather more data to
understand the connection fire performance. Thickness of the T-stub flange was designed to
eliminate any influence on the behavior, however, the tube thickness was significantly lower than
in the single blind-bolted connections, so its effect should not be totally neglected.
Figure 9 shows the deformed shape at failure of the double T-stub connections at elevated
temperatures. Concrete prevented the deformation of the tube column that took place in the HSS,
as it occurred at room temperature. Plastic deformation is also depicted in Figure 9, which helped
for the detection of the failure mode for each type of connection. In the connection to the unfilled
column, stress on the sleeve governed the failure whereas the shank did not present plastic
deformations (Figure 9a). In the case of CFT columns, bolt shank reached its ultimate capacity
and dominated the connections collapse, although high plastic strains were also observed in the
sleeve and the concrete (Figure 9b and 9c).
a) HSS connection b) HB to CFT column c) EHB to CFT column

Fig. 9 – Plastic strain in double T-stub connections to tube column at failure.
3.2 FIRE RESISTANCE RATING
The Fire Resistance Rating (FRR) indicates the time in minutes that the connection is capable of
sustaining the loads before failure. The single blind-bolted connections were calculated under
different load levels with regard to the maximum load supported at room temperature, in order to
know its influence on the FRR. Table 2 shows that FRR increased by 11 minutes for the three
connections types when load level decreased from 50% to 20% load level. Expectedly, as the
load level was lower, the FRR improved. Moreover, it is worth noting that unprotected connections
to CFT columns reached 36 min of FRR when 20% loaded.
The effect of concrete on bolt temperature can be observed from Figure 8a by inspecting the area
of the bolt directly in contact with concrete, starting from 0.08m along the shank length. However,
the shank bolt fracture occurred next to the exposed area of the bolt, so concrete influence was
not as remarkable as it was expected. Connections to CFT gained 4-5min (16-20%) of FRR
compared with HSS connections, Table 2. The heat sink effect of concrete was assumed to be
more important in case of thinner tube and plate.
On the other hand, there were no differences in FRR between the Hollo-bolt and the Extended
Hollo-bolt connection to CFT, Table 2. Similarly, it was attributed to the fact that the fracture took
place next to the head of the bolt and the bolt anchorage within the concrete did not reduce stress
nor temperature in that section.
Table 2 – FRR for single blind-bolted connections and T-stub connections. Load level influence
for single blind-bolted connections.
Load
Specimen index FRR FRRHB/EHB-FRRUHB
level
% min min %
UHB16-100-8.8D-L50 50 20.55
UHB16-100-8.8D-L40 40 23.15
UHB16-100-8.8D-L20 20 30.78
HB16-100-8.8D-C40-L50 50 24.68 4.13 20.12
HB16-100-8.8D-C40-L40 40 27.01 3.87 16.71
HB16-100-8.8D-C40-L20 20 35.60 4.82 15.67
EHB16-150-8.8D-C40-L50 50 24.70 4.15 20.21
EHB16-150-8.8D-C40-L40 40 27.14 4.00 17.27
EHB16-150-8.8D-C40-L20 20 35.63 4.85 15.77
T-UHB16-120-8.8D 50 17.78
T-HB16-120-8.8D-C50 50 22.15 4.37 24.55
T-EHB16-120-8.8D-C50 50 22.05 4.27 23.99
4. CONCLUSIONS
This research is a numerical study on the fire performance of blind-bolted connections to CFT
and HSS in the tension zone of moment-resisting connections. In the absence of fire experiments
to validate the thermo-mechanical model, the mechanical part of the problem was corroborated
at room temperature with tests on the same connections, whereas the validation of the heat
transfer was accomplished in a preceding research by the authors (Pascual et al., 2015a).
The objective was to provide data to gain insight into tensile loaded blind-bolted connections in
fire and to assess the effect of concrete infill and the anchorage of the blind-bolt. Two types of
connections were analyzed, a single blind-bolt connecting a plate to a tube column and a double
T-stub connection to a tube column. The type of column (HSS and CFT) and the type of fastener
system (HB and EHB) were the variables considered for each connection. The fastener system
was the element that governed the connection failure due to the thickness of the plate or T-stub
flange and the tube, but depending on the connection and temperatures, the sleeve or the bolt
shank were determinant. The conclusions from the study are the following:
- Concrete filling the tube column resulted in 16-20% enhancement in FRR for the connection with
the single blind-bolt compared with connections to unfilled columns, and around 25% for the
double T-stub connections. In connections to HSS, the failure was shared by the sleeve and the
shank of the fastener. For the connections to CFT columns the shank bolt of the fastener system
decided the connections collapse, where stresses in the sleeve were lower due to the concrete.
- The anchorage of the shank bolt into the concrete provided by the EHB did not represent an
increase in the FRR compared with the HB in connections to CFT. Failure of the shank was
located next to the bolt head where neither temperature nor stress were directly affected by the
anchorage.
REFERENCES
Al-Jabri, K. S., Seibi, A., & Karrech, A. (2006). Modelling of unstiffened flush end-plate bolted
connections in fire. Journal of Constructional Steel Research, 62(1-2), 151-159.
Bursi, O. S., & Jaspart, J. P. (1998). Basic issues in the finite element simulation of extended end
plate connections. Computers & Structures, 69(3), 361-382.
EN 1992-1-1 (2004). Eurocode 2: Design of concrete structures. Part 1-1: General rules and rules
for buildings. Brussels, Belgium: Comité Européen de Normalisation.
EN 1993-1-8 (2004). Eurocode 3: Design of steel structures. Part 1-8: Design of joints. Brussels,
Belgium: Comité Européen de Normalisation.
Da Silva, L. S., Neves, L. F. C., & Gomes, F. C. T. (2003). Rotational stiffness of rectangular
hollow sections composite joints. Journal of Structural Engineering, 129(4), 487-494.
Ding, J., & Wang, Y. C. (2007). Experimental study of structural fire behavior of steel beam to
concrete filled tubular column assemblies with different types of joints. Engineering Structures,
29(12), 3485-3502.
Elsawaf, S., Wang, Y. C., & Mandal, P. (2011). Numerical modelling of restrained structural
subassemblies of steel beam and CFT columns connected using reverse channels in fire.
Engineering Structures, 33(4), 1217-1231.
Ellison, S., & Tizani, W. (2004). Behavior of blind bolted connections to concrete filled hollow
sections. The Structural Engineer, 82, 16-17.
Espinos, A., Romero, M. L., & Hospitaler, A. (2010). Advanced model for predicting the fire
response of concrete filled tubular columns. Journal of Constructional Steel Research, 66(8-9),
1030-1046.
FIB 2010. Model Code 2010. Lausanne, Switzerland: Fédération Interantionale du Béton.
France, J. E., Davison, J. B., & Kirby, P. A. (1999a). Moment-capacity and rotational stiffness of
endplate connections to concrete-filled tubular columns with flowdrilled connectors. Journal of
Constructional Steel Research, 50(1), 35-48.
France, J. E., Davison, J. B., & Kirby, P. A. (1999b). Strength and rotational response of moment
connections to tubular columns using flowdrill connectors. Journal of Constructional Steel
Research, 50(1), 1-14.
France, J. E., Davison, J. B., & Kirby, P. A. (1999c). Strength and rotational stiffness of simple
connections to tubular columns using flowdrill connectors. Journal of Constructional Steel
Research, 50(1), 15-34.
Huang, S. S., Davison, B., & Burgess, I. W. (2013). High-temperature tests on joints to steel and
partially-encased H-section columns. Journal of Constructional Steel Research, 80, 243-251.
Janss, J., Jaspart, J. P., & Maquoi, R. (1987). Experimental study of the non-linear behavior of
beam-to-column bolted joints. In: Proc. of a State-of-the art workshop on connections and the
behavior, strength and design of steel structures. Elsevier Applied Science Publishers, 26-32.
Lee, J., Goldsworthy, H. M., & Gad, E. F. (2010). Blind bolted T-stub connections to unfilled hollow
section columns in low rise structures. Journal of Constructional Steel Research, 66(8-9), 981-
992.
Lindapter (2012). Type HB- Hollo-Bolt. Cavity fixings 2, Product brochure. Lindapter International,
UK. 41-43.
Liu, Y., Málaga-Chuquitaype, C., & Elghazouli, A. Y. (2012). Response and component
characterisation of semi-rigid connections to tubular columns under axial loads. Engineering
Structures, 41(0), 510-532.
Lopes, F., Santiago, A., Da Silva, L. S., Heistermann, T., Veljkovic, M., & Da Silva, J. G. (2013).
Experimental behavior of the reverse channel joint component at elevated and ambient
temperatures. International Journal of Steel Structures, 13(3), 459-472.
Mesquita, A., Da Silva; L.S., & Jordao,S. (2010). Behavior of I beam SHS column steel joints with
Hollo-bolts: An Experimental Study. In: Proc. 13th International Conference on Tubular Structures
XIII. The University of Hong Kong.
National Institute of Standards and Techology (NIST). Final Report on the collapse of the World
Trade Center building 7. USA: National Institute of Standards and Technology (2008).
Pascual, A. M., Romero, M. L., & Tizani, W. (2015a). Thermal behavior of blind-bolted connectios
to hollow and concrete filled tubular columns. Journal of Constructional Steel Research, 107, 137-
149.
Pascual, A. M., Romero, M. L., & Tizani, W. (2015b). Fire performance of blind-bolted connections
to concrete filled tubular columns. Engineering Structures, 96, 111-125.
Pascual, A. M. (2015). Fire Behavior of blind-bolted connectios to concrete filled tubular columns
under tension. Dissertation. Department of Construction Engineering and Civil Engineering
Projects, Universitat Politècnica de València.
ABAQUS. ABAQUS User´s Manual. Providence. Rohde Island, USA. Dassault Systèmes. Simulia
Corporation. ABAQUS vs 6.6. ABAQUS/Standard Version 6.6.
Pitrakkos, T., & Tizani, W. (2013). Experimental behavior of a novel anchored blind-bolt in tension.
Engineering Structures, 49, 905-919.
Tizani, W., Al-Mughairi, A., Owen, J. S., & Pitrakkos, T. (2013). Rotational stiffness of a blind-
bolted connection to concrete-filled tubes using modified Hollo-bolt. Journal of Constructional
Steel Research, 80(0), 317-331.
Wang, Z. Y., Tizani, W., & Wang, Q. Y. (2010). Strength and initial stiffness of a blind-bolt
connection based on the T-stub model. Engineering Structures, 32(9), 2505-2517.
Yu, H., Burgess, I., Davison, J., & Plank, R. (2011). Experimental and Numerical Investigations
of the Behavior of Flush End Plate Connections at Elevated Temperatures. Journal of Structural
Engineering, 137(1), 80-87.
ROBUSTNESS DESIGN AND POSSIBLE COLLAPSE MECHANISMS
OF BUILDING FRAMES IN A FIRE
Kei KIMURA Fuminobu OZAKI

Nippon Steel & Sumitomo Metal Corporation Nagoya University
Futtsu, Japan Nagoya, Japan
kimura.ze8.kei@jp.nssmc.com ozaki@dali.nuac.nagoya-u.ac.jp
Ryoichi KANNO
Nippon Steel & Sumitomo Metal Corporation
Futtsu, Japan
kanno.kx4.ryoichi@jp.nssmc.com
ABSTRACT
For fire safety design, conventional prescriptive design has been rapidly changing to more
flexible and economical performance-based design. In this design approach, extensive research
has been conducted on robustness to make full use of inherent structural resistance. This paper
first reviews and summarizes the past and ongoing research activities on structural robustness
design for fire resistance of floors and frames. A frame collapse mechanism that provides
structural resistance even in the case of semi-rigid frames is proposed. In the proposed
mechanism, resistance against gravity loading can be provided largely by the flexural resistance
of floor systems above the localized fire zone. With a simple limit analysis technique, the
increased resistance of a multi-storied frame is calculated and demonstrated, revealing that a
substantial number of the columns may also be unprotected in addition to the floor joists by floor
robustness design already established. The resisting mechanism proposed in this study for
building frames has not been fully verified experimentally; therefore, further studies are required.
INTRODUCTION
Fireproofing materials such as sprayed vermiculite can increase the fire resistance of structural
steel members and their thickness can be determined easily based on the required fire resistant
rating. Recently, this conventional prescriptive design has been rapidly changing to
performance-based design that can enable more flexible and economical design. In this design
approach, there have been extensive research activities on “robustness” or “redundancy” to
make full use of the inherent structural resistance. One of the representative examples is a
series of research activities in Europe on the membrane action of a composite floor slab under
increasing temperature (Wang et al., 1996), (Bailey et al., 2000), (Nadjai et al., 2011). The
design guidelines for the membrane action have been proposed so that floor joists can remain
unprotected against fire (SCI, 2006), (Vassart et al., 2012). Another example is frame
robustness research that takes into account the whole frame resistance when some columns
lose their load resisting capacities due to fire (JSSC, 2005). This research is currently limited to
the frames designed to protect against earthquakes, whose beam-column connections are fully
rigid.
This paper first reviews and summarizes the past and ongoing research activities on structural
robustness design under fire from both floor and frame resistance perspectives. By focusing on
the latter frame case, one possible frame collapse mechanism that can provide structural
resistance of semi-rigid frames is proposed. In the proposed mechanisms, the resistance
against gravity loading may be provided largely by the flexural resistance of floor systems above
the localized fire zone. Applying a simple limit analysis technique, the increased resistance of a
multi-storied frame is calculated.
OVERVIEW OF FIRE RESISTANT DESIGN IN JAPAN

Before discussing the robustness design, an overview of fire resistant design of steel buildings
in Japan is presented to ascertain the technological advancements in fire design. In Japan, the
most common method of increasing the fire resistance is to use fireproofing materials. One
representative example is sprayed vermiculite having high thermal insulation performance;
however, there are disadvantages in using it. For example, the working environment is difficult
and unhealthy, and regular maintenance is required because of its quality deterioration. As a
result, the use of other materials such as gypsum board, blanket or intumescent material has
increased; however, there still is a strong need to reduce the fireproofing work in Japan due
mainly to the problem of the growing labor shortage.
In 1988, fire resistant (FR) steel was developed in Japan. FR steel is a carbon steel that
includes alloys such as Cr and Mo in order to increase the strength at high temperature
(Sakumoto et al., 1992), (Sakumoto et al., 1994). FR steel has a yield stress at 600°C that is
equal to or larger than 2/3 times the design strength at ambient temperature as shown in Figure
1. Due to this improved performance, FR steel has been applied in many car parks and atriums
in Japan as shown in Figure 2.
Strength (N/mm 2)
600 Tensile strength
500
400 Yield point
300
200
100 FR steel Temp.
Ordinary steel (˚C)
0
0 200 400 600 800
Fig. 1 – Mechanical properties of (a) A car park (b) An atrium
FR steel at high temperature Fig. 2 – Examples of FR steel application
A performance-based design method for fire safety has been developed in Japan since the
1980s as shown in Figure 3. A series of research activities on the structural fire resistance of
steel frames were conducted from 1982 to 1987 and the first design method was proposed by
the Building Center of Japan (BCJ, 1989). Since then, further active studies on the behavior of
steel members and frames at elevated temperature have been conducted. The Architectural
Institute of Japan (AIJ) published a recommendation for the fire-resistant design of steel
structures (AIJ, 1999) and then a verification method for fire resistant performance was issued
for design practitioners by the Japanese government in 2001. Note that in the same period,
Eurocode3 for the structural fire design of steel structures was published in European countries.
In line with the rest of the world, conventional prescriptive design has been rapidly changing to a
more flexible and economical performance-based design in Japan. Along with this trend, there
has been interesting research conducted recently on the methodology in which stress
redistribution is taken into consideration for steel frames in a fire. This research conducted in
Japan is intended largely to remove the fire protection in some of the columns. In Europe, there
has been similar research activity, but the stress redistribution has been considered not on the
frames but on the floor slabs.
General fire prevention designing method for buildings (BCJ, 1989)

First performance-based fire resistant design in Japan
Recommendation for fire resistant design of steel structures (AIJ, 1999)
Fire resistant design for ultimate strength of steel structures
1980 1990 2000 2010
Legal specification for fire resistant structure
Specification for 1 to 3 hours fire resistant performance (prescriptive design)
Verification method for fire resistant performance
(Ministry of Land, Infrastructure and Transport, 2001)
Notification of performance-based fire resistant design
Fig. 3 – Timelines of fire resistant design in Japan
ROBUSTNESS DESIGN
The structural behavior of a steel frame in a fire changes over time due to the reduction of the
material strength and the thermal expansion of the structural members. Figure 4 shows the
behavioral process (Stages 1 to 3) in the fire of a steel frame with no fire protection from the fire
initiation until the complete collapse.
Stage 1: Continuing heat input by fire increases the temperature of the structural members.
Stage 2: Decrease in material strength at high temperature reduces the member strength.
Stage 3: Increase in the number of damaged members reduces the stability of the whole frame.
Stage 1: Temperature increasing Insulation
Stage 2: Strength reduction of the member Material
Stage 3: Whole collapse of the frame Robustness

Fig. 4 – Process to total collapse of frame and preventing solutions
There are solutions to secure the safety of the frame in fire. Heat insulation such as sprayed
mineral fiber is the most common solution for Stage 1. Another solution is to use a special steel
material such as FR steel, which can be categorized as a solution for Stage 2. For Stage 3, one
has to consider the stress redistribution in the frame and structural elements, because in this
stage, the members and elements in the frame lose their strength. The design for this stage is
based on an observation that one member or element did not cause the entire collapse of the
frame. This solution for Stage 3 is called “robustness design” and may be defined as a design
that considers the ultimate resistance of a frame or a structural element until its collapse.
In Europe, there are design guidelines for floor robustness, through which fireproofing materials
of floor joists can be partially omitted as shown in Figure 5. This is possible by considering the
load-bearing capacity of floor slabs when the slabs undergo large deflections and exhibit a
membrane action. On the other hand, in Japan, a design methodology for frame robustness has
been proposed so that fireproofing materials of the columns can be partially omitted as shown in
Figure 6. This design is possible by considering the load redistribution over the entire frames
when some of the columns lose their strength. Both floor and frame robustness designs will be
discussed more in detail after the next sections.
Unprotected Unprotected
(a) At ambient temp. (b) In a fire (a) At ambient temp. (b) In a fire
Fig. 5 – Load-bearing mechanisms of Fig. 6 – Load paths in frames at ambient and high
floor at ambient and high temperatures temperatures
FLOOR ROBUSTNESS
The fire resistant performances of concrete slabs and steel beams (joists) are evaluated
separately by standard fire tests with small-scale specimens because these tests are simple
and safe. In reality, however, a composite floor slab consisting of a reinforced concrete slab and
steel joists shows higher performance than isolated standard fire tests. Historically, observations
after some real fires have deepened the understanding on how multi-story steel buildings
behave in a fire and the knowledge obtained has had a large impact on the design of the
composite floors in a fire.
In 1996, a series of fire tests on an eight-story steel building were conducted in the UK at the
Building Research Establishment’s Cardington Laboratory (Wang, 1996). It provided valuable
information on the behavior of a steel frame in a real fire. After this landmark test, focusing on
composite floors with unprotected steel joists at elevated temperatures, a series of fire tests and
numerical analyses were further conducted to evaluate the load-bearing capacity of the floors.
The results clarified that even when the unprotected steel joists lost their strength, the vertical
load was resisted through the reinforced concrete slab by so-called tensile membrane action at
high temperature. The resistant mechanism is shown in Figure 7 (Bailey and Moor, 2000). Both
tension and compression zones were formed in the slab as the slab deflection became large,
providing a stable tensile membrane action to gain resistance against the vertical load. Design
guidelines considering this membrane effect have been proposed with simple design models
such as the one shown in Figure 8 (Bailey and Moor, 2000). The guidelines enabled steel joists
underneath the floor slab to remain unprotected against fire. Bailey and Moor (Bailey and Moor,
2000) also indicated that the aspect ratio and the amount of deflection of the slab affected the
strength as shown in Figure 9. In addition, a trial evaluation of slab and beam resistances during
a fire was conducted as shown in Figure 10, showing that the beam had a quick loss of its
strength while the slab gained its strength gradually due to the membrane action as the fire
progressed. Depending on the beam and floor joist arrangement in frame design, this floor
robustness design would make it possible to omit around 30% of the fire protection applied to all
the beams and floor joists.
Compression C C
S S
Tension
T2 T2
Fig. 7 – Compression and tension zones for T1

tensile membrane action Fig. 8 – In-plane force equilibrium
50
Beam strength at high temp.
Total strength (kN/m 2)
4 40
Strength increase ratio
Slab strength at high temp.

3 30
2 20
Applied load
1 10
0 0
0 1 2 3 4 5 6 7 8 9 10 0 5 10 15 20 25 30 35 40 45 50 55 60
Displacement / Effective depth Time (min)
Fig. 9 – Strength increase ratio for a Fig. 10 – Trial evaluation for a composite slab with
composite slab unprotected beams in a fire
FRAME ROBUSTNESS
The loss of the strength of heated columns does not immediately trigger overall frame collapse
in a fire. In particular, for rigid steel frames, the high capacity of stress redistribution against
gravity loading can be expected. Considering this stress redistribution, the Japanese Society of
Steel Construction (JSSC) proposed a frame robustness design in Japan that rendered some of
the columns in a steel frame unprotected against fire (JSSC, 2005).
Figure 11 (a) shows the concept of the frame robustness design. Considering a steel frame with
fire protection where a fire occurs on the first floor and the exterior columns on the same floor
are rendered unprotected, the exterior columns could remain unprotected if the remaining
columns and the upper frame above the floor are stable and have sufficient strength to support
the entire gravity load. When the frame becomes unstable and collapsed, the exterior columns
must be protected against fire. Various column combinations can be checked through this way
to determine the best unprotected column combination that gives the largest unprotected area.
As shown in Figure 11 (b), when story fire compartments are fully effective and fire does not
spread to other stories, the columns on each floor can be evaluated separately from the second
story to the fifth story to determine the best unprotected column combination in the entire frame.
This is the idea and basic procedure for the frame robustness design.
Fire zone
Loss of strength
(a) Robustness design for the first floor (b) Robustness design for the other stories
Fig. 11 – Frame robustness design at high temperature
Figure 12 shows the collapse modes of a five-story steel building frame, assuming unprotected
exterior columns on the first floor lose their strength at high temperature (JSSC, 2005). There
are two types of modes: beam collapse mode (Figure 12 (a)) and column collapse mode (Figure
12 (b)). The following design formulae can be derived for both collapse modes:
(1)
(2)
where
: distributed load on the girders
: length of the girder
: full plastic moment of the girder at the fire story at high temperature
: full plastic moment of the girder at the upper stories at ambient temperature
: vertical load applied to a protected column at the fire story
: axial strength of a protected column at the fire story at high temperature
L q
Plastic hinges
at ambient temp.
(M p )
Plastic hinges
at high temp. P
(M p )
(a) Beam collapse mode (b) Column collapse mode

Fig. 12 – Two types of collapse mode
Eqn. (1) is a formula obtained assuming that the distributed gravity load above the first floor is
resisted by the remaining frame. In Japan, the seismic design requires that the beam be weaker
than the column; therefore, applying the principle of virtual work to the plastic hinges shown in
Figure 13 (a), Eqn. (1) can be obtained as a condition in which the entire frame is stable. On the
other hand, Eqn. (2) is a formula that checks if the remaining columns on the first floor have
sufficient strength to support the vertical load above the first floor. If both formulae of Eqns. (1)
and (2) are satisfied, the entire frame with unprotected exterior columns on the first floor is
stable. Thus, fire protection rendered to the exterior columns on the first floor could be omitted.
Various unprotected column combinations would be checked not only for the first floor but also
for the above floors one by one to obtain the best unprotected column combination in the entire
frame.
DESIGN EXAMPLE OF FRAME ROBUSTNESS

A design example for the column robustness is shown in Figure 13, which was reported by the
Building Research Institute (BRI) in Japan (BRI, 2008). The building was a typical five-story
office building in Japan, designed against earthquakes based on the seismic design code in
Japan. Tables 1 and 2 show the sections of the columns and beams together with material
types (JIS standard) and their material strengths F. In this building, beams were fully-rigidly
connected to columns with full penetration welding; therefore, the strength of the connections
was the same as or larger than the full plastic moment of the beam. Note in the strength
calculations through Eqns. (1) and (2) that the full plastic moment of the beams and the axial
strength of the columns were assumed to be given by those calculated at ambient temperature,
considering fire protection rendered to the members.
3500
X1 G2 G2 G2 3500
C1 C3 C3 C1
3500
G1 G1 G1 G1
X2 G2 G2 G3 3500 G2 G2 G2 G1 G1
C2 C4 C4 C2
G1 G1 G1 G1 3600 C1 C3 C3 C1 C1 C2 C1
G2 G2 G3 (unit :mm)
X3 C1 C3
C3 C1 7000 7000 7000 7000 7000
Y1 Y2 Y3 Y4 Y1 Y2 Y3 Y4 X1 X2 X3
(a) Plan (b) Elevation
Fig. 13 – Building for design example
Figure 14 shows a comparison between the beam and column collapse modes. The horizontal
axis is the ratio of the resistance to the external force, called the safety index in this study.
Referring to Eqns. (1) and (2), the left hand side terms correspond to the external force whereas
the right hand side terms provide the resistance. As Figure 14 shows, the column collapse
mode was more stable than beam collapse mode because the safety index became large. This
is because the building was designed primarily against earthquakes.
In Japan, the overall beam collapse mechanism is generally preferred to increase the energy
dissipating capacity while the local story collapse mechanism is avoided. As a result, steel
frames are generally designed so that column strength is higher than beam strength such as a
50% increase in the column. This is why the column collapse mode is more stable for
seismically designed frames. This means, in turn, that the beam collapse mode is the dominant
mode of failure for the frame robustness design and checking the column strength may not be
necessary. As shown in the beam collapse mode in Figure 14, the upper stories were less
stable than the lower stories, because the number of beams to be resisted became fewer as the
designed story moved upward.
Table 1 – Sections for column Table 2 – Sections for beam
C1, C2, C3, C4 G1 G2 G3
Story Story
(BCR295) (SN400B) (SN400B) (SN400B)
5 □-400x400x12 R H-500x200x10x16 H-450x200x 9x14 H-488x300x11x18
4 □-400x400x12 5 H-500x200x10x16 H-500x200x10x16 H-500x200x10x16
3 □-400x400x12 4 H-488x300x11x18 H-500x200x10x16 H-500x200x10x16
2 □-400x400x14 3 H-488x300x11x18 H-500x200x10x16 H-500x200x10x16
1 □-400x400x16 2 H-488x300x11x18 H-488x300x11x18 H-488x300x11x18
(BCR295 : F=295N/mm2) (SN400B : F=235N/mm2)
(Size : H-Depth x Width x Web thickness x Flange thickness (unit : mm))
Figures 15 and 16 show the effect of the location of

unprotected columns on the beam collapse mode.
Collapse No collapse
As shown in Figure 16, the exterior columns had
less effect on the entire stability (strength) of the 5 Beam
building frame. This is because the exterior collapse
4 mode
Fire story
columns supported smaller gravity loads than the
interior columns. This comparison indicates that the
3 Column
exterior columns become more preferable to omit
the fire protection than the interior columns. As a collapse
2
result of the frame robustness design for the mode
building considered, it is possible that the exterior 1
columns can be unprotected while the interior 0 5 10 15 20 25
columns remain protected. Overall, approximately Safety index (resistance / external force)
50% of the fire protection rendered to the entire
columns would be omitted in this design example. Fig. 14 – Comparison of robustness
L L q Beam collapse No beam collapse

q
5
4 (a) (b)
Fire story
3
2
1 1
 qL   2M
2
p  2M p  2 qL   2M
2
p  2M p 10 2
Safety index
(a) Loss of inner columns (b) Loss of outer columns (resistance / external force)
Fig. 15 – Difference due to unprotected column location Fig. 16 – Comparison of robustness
POSSIBLE FRAME COLLAPSE MECHANISMS FOR SEMI-RIGID FRAMES

In the case of seismic resistant buildings, some columns can be unprotected because of the
relatively large ultimate capacity of the building frame. This large capacity can be attributed
mainly to the rigid connections between the columns and beams as shown in Figure 17 (a). On
the other hand, in the case of non-seismic resistant buildings, the load-bearing capacity
becomes smaller because of the smaller strength of semi-rigid connections. In particular, the
strength of the beam collapse mode decreases to a large degree in the case of semi-rigid
connections. Therefore, buildings with semi-rigid connections do not expect much from the
frame robustness.
One possibility for increasing the frame robustness even in the case of semi-rigid frames might
be to consider the flexural resistance of floor slabs as shown in Figure 18. As the figure shows,
the floor slabs might play an important role in stress redistribution in a building in a fire. Figure
19 shows the possible simplified collapse mechanisms in the case of semi-rigid frames when
one exterior column in the fire zone lost its strength. Figure 19 (a) is the case where the floor
slab contribution is not considered whereas Figure 19 (b) is the case where the floor flexural
resistance is directly considered.
For the mechanism shown in Figure 19 (a), external works, , and internal works, , are
given by the following equations:
(3)
(4)
where
: distributed load applied over the floor
: floor area
x : virtual rotation angle of the beam in the x-direction
: beam length in the x-direction
: beam length in the y-direction
: resisting moment strength of beam-column connection in the x-direction
: resisting moment strength of beam-column connection in the y-direction
Box-shaped H-shaped
column column Floors
Diaphragm
Beams
H-shaped H-shaped (2-directions)
beam beam
( M bp ) ( M bp )
Welded flanges Bolted web
and web (M j )
( M j ) M bp  M j M bp  M j
(a) For seismic design (b) For non-seismic design Fig. 18 – Proposed mechanism
Fig. 17 – Typical beam-to-column connection details for non-seismic design
In Eqn. (3), the external work can be summed up in all the upper stories above the fire zone.
The internal work in Eqn. (4) is calculated by adding all the works conducted at plastic hinges
formed in the upper stories above the fire zone. When is satisfied, the mechanism
shown in Figure 19 becomes stable and the exterior column is unprotected against fire. The
mechanism in Figure 19 (a) is theoretically possible, but as described previously, the strength of
and is generally quite small. Therefore, in reality the without-floor mechanism may not
be stable and effective for frame robustness design.
w, A f M f w, A f Columns can
be unprotected
M jy M jy 5 Connections
x x
M jx + floors
M jx 4
Fire story
Effects on
3 hinge lines
Loss of  xlx  xlx of floors
Only
column 2 semi rigid
Fire z connections
1
story
lx ly lx ly 0 1 2
y x Safety index
(resistance / external force)
(a) Without floor (b) With floor
Fig. 19 – Simple design models for frames with semi- Fig. 20 – Difference in safety index
rigid connections due to floor resistance consideration
For the mechanism having the floor contribution shown in Figure 19 (b), plastic yield lines might
be formed in the upper floors at ambient temperature. Note that the yield line resistance may not
be considered in the floor directly exposed to a fire (the floor just above the fire zone). Instead,
the floor may have a different mode of collapse, that is, the membrane collapse that was
described in floor robustness. For the mechanism shown in Fig. 19 (b), Eqn. (4) can be modified
to the following, by adding a term related to internal works by floor bending:
(5)
where
: plastic bending strength of the floor per unit width
TRIAL DESIGN CONSIDERING PROPOSED FRAME ROBUSTNESS

A trial evaluation based on frame robustness design was conducted for a semi-rigid building
frame. The building considered was similar to the five-story building shown in Figure 14, except
that the beam-column connections were semi-rigid. Note that the floor slabs on each story are
assumed to be made of composite floor slabs, consisting of steel decks with a depth of 50 mm
and of lightly reinforced concrete with a thickness of 90 mm above the top of the decks. The
following assumptions were made in this trial design:
 The plastic strength of the connections is given by 0.2 times the beam plastic moment.
 The plastic bending strength of the floor is given for negative bending around the weak axis.
For the first assumption, the ratio of the resistant moment of the connections to the full plastic
moment of the beams generally varies depending on the type of connections, but conservatively
the value of 0.2 was assumed. For the second assumption, the anisotropy of the floor due to
composite slabs or the presence of floor joists could have an effect on the bending strength of
the floor but it was disregarded by using the weak axis properties.
Figure 20 shows the results of the trial evaluation for the semi-rigid frames shown in Figure 19.
The vertical axis is the story number subjected to fire and the horizontal axis means the ratio of
to (safety index). As for the no floor case shown in Figure 19 (a), the safety indices
were smaller than 1.0 for most of the stories, which means that the omission of the fire
protection for the corner columns would be quite limited. On the other hand, as for the floor case
(Figure 19 (b)), the safety indices became larger than 1.0 for all the stories except the fifth one,
meaning that the majority of the corner columns would be unprotected. Regarding the fifth story
shown in Figure 20, there was no strength increase due to the floor because the floor above the
story was directly exposed to the fire and the flexural resistance might be very small.
Figures 21 and 22 show the effects of the column locations on the safety index. There were
three types of column locations considered as shown in Figure 21. Case (c) was for a corner
and the results have already been shown in Figure 20. Case (b) was for a side column and
Case (a) was an inner column. Figure 22 shows all the results together and the side column
from the first to the third floor would be unprotected. In addition, the interior column from the first
to the second floor would be unprotected. The reason that the inner column had lower safety
indices was because the tributary area (shaded area in Figure 21) for the floor load was larger.
This resulting larger load lowered the safety of the column compared to the other column cases.
As seen in the above, the safety index is affected not only by the location but also by the applied
load of the column.
Columns can
be unprotected
(a) (b) (c)
5
(a) Inner column
4
Fire story
3 Connections
(b) Side column + floors
2
Only semi-rigid
connections
1
(c) Corner column 0 1 2
Fig. 21 – Considered locations Safety index (resistance / external force)
of unprotected column Fig. 22 – Effects of location of unprotected column
Figure 23 shows a possible unprotected column combination obtained by conducting the safety
check of the column one by one in the entire building frame, where the floor slab resistance was
considered. As Figure 23 shows, approximately 30% of the columns would be unprotected
against fire, resulting in more economical design than the design considering only floor
robustness. Note in this trial design that the omission of the fire protection would obviously be
increased if the flexural resistance of the floor slabs is increased. Overall, frame robustness
design considering floor flexural resistance could increase the number of unprotected columns
even for buildings with semi-rigid connections. It should be noted, however, that floor resistance
in the frame robustness design needs to be verified by experiments and/or numerical analyses.
This hypothetical design methodology thus requires further study in the future.
Unprotected columns
Fig. 23 – Possible unprotected column combination based on frame robustness design
CONCLUSIONS
This paper reviewed and summarized the past and ongoing research activities on structural
robustness design under fire from both floor and frame resistance perspectives. By focusing on
the latter frame case, one possible frame collapse mechanism that can provide structural
resistance even of semi-rigid frames was proposed. In the proposed mechanism, the resistance
against the gravity load might be provided largely by the flexural resistance of floor slabs above
the localized fire zone. Applying a simple limit analysis technique, the robustness resistance of a
multi-storied frame with semi-rigid connections was calculated and demonstrated, which
realized that approximately 30% of the columns might also be unprotected in addition to floor
joists due to the membrane action already proposed in the past. The resisting mechanism
proposed in this study for building frames is not fully verified at present by experiments and/or
numerical analyses; therefore, further studies are obviously required in the future.
REFERENCES
Architectural Institute of Japan (AIJ) (1999). Recommendation fire resistant design of steel
structures (in Japanese).Bailey, C.G. & Moore, D.B. (2000). The structural behavior of steel
frames with composite floor slabs subject to fire, Part1: Theory & Part2: Design. The
Structural Engineer, 78(11), 19-33.
Building Center of Japan (BCJ) (1989). General fire prevention designing method for buildings
(in Japanese).
Building Research Institute (BRI) (2008). Manual for design and fabrication of cold-formed
square steel tube (in Japanese).
Japanese Society of Steel Construction (JSSC) (2005). Guidelines for collapse control Design –
Construction of steel buildings with high redundancy-, Part1: Design & Part2: Research.
Nadjai, A., Bailey, C.G., Han, S.H., Vassart, O., Zhao, B., Hawes, M., Franssen, J.M. & Simms,
I. (2011). Full scale fire test on a composite floor slab incorporating long span cellular steel
beams. The Structural Engineer, 89(21), 18-25.
Sakumoto, Y., Yamaguchi, T., Ohashi, M. & Saito, H. (1992). High-temperature properties of
fire-resistant steel for buildings. Journal of Structural Engineering, 118(2), 392-407.
Sakumoto, Y., Yamaguchi, T., Okada, T., Yoshida, M. & Saito, H. (1994). Fire resistance of fire-
resistant steel columns. Journal of Structural Engineering, 120(4), 1103-1121.
Steel Construction Institute (SCI). (2006). Fire safe design: A new approach to multi-storey steel
framed buildings (P288).
Vassart, O. & Zhao, B. (2012). Membrane action of composite slab in case of fire: Background
document & Design guide (Edition 2012-1).
Wang, Y.C. (1996). Tensile membrane action in slabs and its application to the Cardington fire
tests. Proceeding of the second Cardington conference. (pp. 55-67). England.
PROGRESSIVE COLLAPSE ANALYSIS OF COMPOSITE STEEL FRAMES UNDER
ELEVATED TEMPERATURE
Hussam Mahmoud, Ph.D.

Department of Civil and Environmental Engineering, Colorado
State University,
Fort Collins, USA
Hussam.Mahmoud@colostate.edu
Chao Qin
Department of Civil and Environmental Engineering, Colorado
State University,
Fort Collins, USA
Chao.Qin@colostate.edu
ABSTRACT
Collapse analyses of buildings under fire are relatively lacking. Recent events, particularly
the 9/11 terrorists attack, have spiked interest in assessing the full collapse behavior of
buildings under fire. In this study, a 3D numerical model of a composite steel frame is
developed, and its performance evaluated under fire up to and including collapse. The frame
is a hybrid 6-story building with moment, gravity, and concentrically braced frames. The
building is subjected to the ASTM E119 standard fire curve for two different scenarios – first-
floor corner compartment fire and whole first-floor fire. The results show that the full structural
response including the collapse mechanism during fire events can be predicted with
reasonable accuracy using advanced numerical finite element analysis. The results also
show collapse initiation to be triggered following the failure of few columns, as opposed to
one column, highlighting the importance of the analysis for performance-based engineering.
INTRODUCTION
Since the Broadgate Phase 8 fire in London and the subsequent Cardington fire tests in the
1990s, researchers have begun to experimentally investigate and understand the behavior
of steel-framed building structures in fire (British Steel 1999, Newman et al. 2000). Interest
in full system analysis of structures under fire was re-energized in 2001 following the collapse
of the Twin Towers and World Trade Center 7 (WTC-7) due to terrorist attacks. These high
impact events and early tests have prompted extensive finite element (FE) modelling to
investigate the full behavior of steel framed buildings under fire loading. Due to the extensive
computational demand, resulting from the high level of material inelasticity and geometrical
nonlinearity, 2D FE modeling was viewed as an attractive alternative than 3D models.
However, 2D models are limited in that they dismiss key behavioral features including
participation of all system components in the load carrying mechanisms, particularly at large
deformations. Various studies have been conducted to evaluate the local behavior and global
performance of steel framed structures under fire (Memari and Mahmoud 2014, Liew and Ma
2004, Gu and Kodur 2011, Wang and Moore 1995, Huang et al. 2000). Agarwal and Varma
(2013) conducted a qualitative assessment of the importance of gravity columns on the
stability of a typical steel building subjected to corner compartment fire. The study, however,
included only early stage of failure without simulating progressive collapse. Jiang, et al.
(2014) studied the collapse mechanisms of 2D steel braced frames exposed to single and
multi-compartment fires, while capturing large deformations. Similar to previous studies, the
analysis provided valuable insight on the behavior but did not include progressive collapse
of the system.
In the above-mentioned analysis, collapse or global instability of the structure was not
evaluated through explicitly modeling progressive collapse of the frames, instead
assessment of stability was generally evaluated through either monitoring of simulation
convergence or by assessing the limits of column interaction equations. The lack of
convergence while to some extent could be related to system instability due to P-∆ effects,
this is not always the case as highlighted in Mahmoud et al. (2015). Therefore, modeling
progressive collapse explicitly can eliminate the speculations as to the reasons for the lack
of convergence in numerical FE models.
Currently there are two studies in the literature by Sun et al. (2012) and Gann (2008) where
collapse analysis of 2D and 3D frames, respectively, under fire was conducted. In a study by
Sun et al. (2012), a robust static–dynamic procedure to capture collapse mechanisms of
braced 2D steel-framed structures under was proposed. The study, while it provided notable
advances, did not include failure of connections or fracture of elements. The study by Gann
(2008) pertained to the National Institute of Standards and Technology (NIST) investigation
of the collapse of WTC-7 and included assessment of the full 3D progressive failure of the
building. This was realized through incorporating material inelasticity, geometric nonlinearity,
contact between the collapsing structural elements, and element erosion based on defined
failure criterion.
Clearly, previous studies provided substantial knowledge pertaining to the large inelastic
response of steel frames under fire. However, apart from the WTC-7 analysis (Gann 2008),
the studies stopped short of modeling the entire system progressive collapse, which requires
simulating element separation and fracture as well as contact between the fractured elements
and collapsed floors. To that end, this study aims to build upon existing knowledge to evaluate
the progressive collapse mechanisms of a 3D six-story steel building. In the simulations,
complete connection failures and interaction between all elements are modeled in order to
capture the progressive failure of the system.
STRUCTURAL CONFIGURATION
The structure is 2-bay (15 m) by 2-bay (15 m) square (7.5m each bay) six-story building with
a height of 22.38 m and is shown in Figure 1 (Foutch et al. 1986). The first story is 4.5 m in
height while the remaining stories are 3.4 m each. The girder-to-column connections in the
frame axis A, B, and C are fully welded moment connections. In addition, the frame at axis B
is a Concentrically Braced Frame (inverted V). The pin connection between the girders and
the columns and beams and girders at frame line 1, 2, and 3 are shear tab connections. The
exterior frames 1 and 3 contains X-braces in each span. The columns, girders, and beams
are wide-flange sections, sized as shown in Figure 1, and made with A36 steel material. The
brace members are rectangular HSS made with A500 Grade B steel. Shear studs are utilized
to achieve composite action between the slabs and the beams and girders. The slab is made
of lightweight concrete with nominal thickness of 90 mm and 165 mm. The concrete
reinforcement bars are 6 mm in diameter on a 100 mm grid with a nominal minimum cover
of 29 mm.
1 2 3
Column sections
7500 7500
Story # C1 C2 C3 C4 C5
G1 G1
C 6-5 W10x49 W10x33 W10x33 W10x33 W12x40
C3 C2 C1 4-3 W12x65 W12x53 W10x39 W10x60 W12x72
b1 b1 2 W12x79 W12x65 W12x50 W12x79 W12x106
1 W12x87 W12x87 W12x65 W12x106 W12x136
7500
G3
G4
G3
b1 b1
Frames beam sections
Floor # G1 G2 G3 G4
G2 G2
B 6 W16x31 W16x31 W18x35 W21x50
C4 b1 C5 b1 C4 5 W16x31 W18x35 W18x35 W21x50
4 W18x35 W18x35 W18x35 W21x50
3 W18x35 W18x40 W18x35 W21x50
G3
G4
G3
7500 b1 b1 2 W18x40 W18x40 W18x35 W21x50
Floor beam and bracing sections

G1 G1
A Story # b1 B1 B2
C3 C2 C1 5-6 W16x31 ST-4x4x3/16 ST-4x4x3/16
4 W16x31 ST-5x5x1/4 ST-4x4x3/16
G1: Moment frame G2: V-braced frame 2-3 W16x31 ST-6x6x1/4 ST-4x4x3/16
G3: X-braced frame G4: Gravity frame 1 W16x31 ST-6x6x1/4 ST-4x4x3/16
b1: Floor beam
1 2 3 A B C
Fig. 7500
1 - Plan view and
7500member designation of the modeled building
7500 7500
FINITE ELEMENT MODELING APPROACH
All models used in this study were developed and analyzed using the general-purpose FE
program ABAQUS (Simulia 2014). Line elements were utilized for the beams and columns
and shell elements for the slabs. The density of ASTM A36 steel was assumed to be 7800
kg/m3. The tensile strength of the A36 steel was set to 250MPa and the Poisson’s ratio to
0.26. The brace members were cold formed rectangular HSS tubes of ASTM A500 B steel
with a specified yield stress of 315 MPa. The density of ASTM A500 B steel is around 7850
kg/m3. The stress-strain properties of the steel were modeled as elastic perfectly plastic. The
Johnson-Cook constitutive damage model (1985) was selected for inclusion of damage
initiation in the material model. The temperature-dependent mechanical and thermal
properties were adopted from European code as shown in Figure 2 (Eurocode 2005). In this
study, the Young’s modulus of the concrete was set to 15 GPa and the density to 2400 kg/m3.
The conductivity of concrete was 0.5 W/mK. A yield stress of the 6 mm steel reinforcing bars
of 398 N/mm2 was used. Because concrete is a good thermal insulation material, the
temperature-dependent mechanical properties of concrete and reinforcing steel were
neglected in the study. For modeling the concrete slabs, the concrete damage plasticity
(CDP) constitutive model was implemented in Jankowiak and Lodygowski (2005).
Fig. 2 - Temperature-dependent mechanical and thermal properties of steel
For moment connections, a spring is used with rotational stiffness, 𝑘𝑘𝜃𝜃 , determined from the
4𝐸𝐸𝑇𝑇 𝐼𝐼
stiffness matrix of a fixed-fixed beam as 𝑘𝑘𝜃𝜃 = where 𝐸𝐸𝑇𝑇 is Young’s Modulus as a
𝐿𝐿
function of temperature. The plastic moment, 𝑀𝑀𝑝𝑝 , as a function of temperature (T) is 𝑀𝑀𝑝𝑝 (𝑇𝑇) =
𝐹𝐹𝑦𝑦 (𝑇𝑇) ∗ 𝑍𝑍𝑥𝑥 , which is used is used to represent the plasticity of the fully weld connection; where
𝐹𝐹𝑦𝑦 (𝑇𝑇) is the yield stress as a function of temperature and 𝑍𝑍𝑥𝑥 is the plastic section modulus.
Failure of the connection is specified when a rotation value of 0.06 radians. It is important to
note that while the rotation at peak capacity is constant, the peak strength is not and is
function of 𝐹𝐹𝑦𝑦 (𝑇𝑇) as noted above.
A component-based model for shear tab connection subjected to fire was developed by
Sarraj (2007). In this model, the whole connection is treated as a group of springs
representing each bolt row in addition to bearing springs representing the two connected
plates. The initial stiffness of a shear connection is denoted as, 𝐾𝐾𝑖𝑖 , which is composed of
three parts as noted in Rex and Easterling (2003) and modified in this study to include
temperature effect:
1
𝐾𝐾𝑖𝑖 = 1 1 1 (1)
+ +
𝐾𝐾𝑏𝑏𝑏𝑏 𝐾𝐾𝑏𝑏 𝐾𝐾𝑣𝑣
where,
Kbr is the bearing stiffness, Kb is the bending stiffness and Kv is the shear stiffness and can
be found using the following equations.
𝐾𝐾𝑏𝑏𝑏𝑏 = 𝛺𝛺(𝑇𝑇)𝑡𝑡𝑝𝑝 𝑓𝑓𝑦𝑦 (𝑑𝑑𝑏𝑏 /25.4)0.8 (2)
𝐾𝐾𝑏𝑏 = 32𝐸𝐸(𝑇𝑇)𝑡𝑡𝑝𝑝 (𝐿𝐿𝑒𝑒 ⁄𝑑𝑑𝑏𝑏 − 0.5)3 (3)
𝐾𝐾𝑣𝑣 = 6.67𝐺𝐺(𝑇𝑇)𝑡𝑡𝑝𝑝 (𝐿𝐿𝑒𝑒 ⁄𝑑𝑑𝑏𝑏 − 0.5) (4)
where,
𝛺𝛺(𝑇𝑇) is a parameter obtained by curve-fitting parameter as a function of temperature; 𝑡𝑡𝑝𝑝
and 𝑓𝑓𝑦𝑦 are plate thickness and yield strength, respectively; 𝑑𝑑𝑏𝑏 is the bolt diameter, 𝐸𝐸(𝑇𝑇) is
the elastic modulus of the plate, 𝐿𝐿𝑒𝑒 is the plate edge distance, and 𝐺𝐺(𝑇𝑇) is the shear
modulus of elasticity.
The bolt shear stiffness can be represented using the proposed temperature dependent
expression.
0.15𝐺𝐺(𝑇𝑇)𝐴𝐴𝑠𝑠
𝐾𝐾𝑏𝑏𝑏𝑏 = (5)
𝑑𝑑𝑏𝑏
where,
𝐺𝐺𝑇𝑇 is the temperature dependent shear modulus; 𝐴𝐴𝑠𝑠 is the cross-section area of the bolt;
and 𝑑𝑑𝑏𝑏 is bolt diameter.
FIRE SCENARIOS, LOADING AND RESULTS
Total dead loads of 4.3 kN/m2, 3.6 kN/m2, and 1.4 kN/m2 were used for the floor, roof, and
exterior wall areas, respectively. The live loads for the floor and roof areas were 2.8 kN/m2
for the slabs and the beams and 1.8 kN/m2 for the girders. The load combination for the fire
analysis is set to 𝐷𝐷 + 0.5𝐿𝐿, where D is dead load, L is live load, which is based on the work
by Ellingwood and Corotis (1991). Two fire scenarios were utilized in the study and included a
contained corner compartment fire on the first floor and a whole first floor fire. Figure 3(a) shows
a rendered view of the corner fire case while Figure 3(b) shows a rendered case of the whole
floor fire. The standard ASTM E-119 time-temperature curve, shown in Figure 3(c) was utilized in
this study.
(c)
Fig. 3 – (a) fire Case 1, (b) fire Case 1, and (c) ASTM E119 Standard fire curve
For Case 1 (single corner compartment fire in Figure 4), when the step time was at 20 min,
the temperature reached 773.5°C on the time-temperature curve, and the global response of
the entire structure at this stage was marked by substantial deformation of the concrete slab.
At this point, the entire corner bay collapsed and started to drag other floors with it, causing
notable twist in the entire structure combined with lateral deformation. At 40 min (860.6°C),
the building lost its ability to sustain any loading as marked by the additional lateral and
vertical downward deformation of the entire structure. The substantial unbalanced forces
resulted in several fractures of braces in the first two floors.
Step Time=0 sec Step Time=10 min Step Time=20 min
Temperature=20℃ Temperature=704℃ Temperature=773.5℃
Step Time=30 min Step Time=40 min Step Time=50 min

Temperature=843℃ Temperature=860.6℃ Temperature=909.4 ℃
Fig. 4 - Collapse Sequence for Case 1
For Case 2, (whole floor fire in Figure 5), when the step time was at 20 min, the temperature
reached 773.5°C on the time-temperature curve, and the global response of the entire
structure at this stage was marked by substantial deformation in the columns of the first floor.
At this point, progressive collapse of the entire frame was initiated and progressed as shown
in Figure 5.
Step Time=0 sec Step Time=10 min Step Time=20 min

Temperature=20℃ Temperature=704℃ Temperature=773.5℃
Step Time=30 min Step Time=40 min Step Time=50 min

Temperature=843℃ Temperature=860.63℃ Temperature=909.37℃
Fig. 5 - Collapse Sequence for Case 2
In addition to the global collapse sequences, the displacement of each column at the element
nodes at the first-floor height were extracted from the results as shown in Figure 6(a) and (b)
For Case 1, at 700°C, the columns of the heated compartment started to move downward
with their maximum displacement being reached at around 820°C to 840°C. The columns
which were not directly subjected to fire started to displace downwards at 842°C. The whole
first floor reached its displacement limit when it was in contact with the ground at 927°C. This
is depicted in Figure 6(a). For Case 2, a small upward displacement was observed initially
due to thermal elongation of the steel. When the temperature reached 700°C, the structural
steel lost more than half of its strength and the entire floor started to displace vertically
downward as shown in Figure 6(b) through equal displacement of all columns. Full contact
with the ground is reached at 823°C.
(a) (b)
Fig. 6 - Vertical Displacements of First Floor Columns (a) Case 1 and (b) Case 2
Assessment of the column interaction equations is shown in Figure 7(a) and (b) for Case 1
and Case 2, respectively. As shown in both figures, various columns violate the interaction
equation at as low of a temperature as 200oC. Specifically, for Case 1 one column exceeded
the interaction equation at about 200oC and a total of three columns at about 400oC. For
Case 2, one column exceeded the interaction equation at about 200oC and a total of six
columns at about 400oC. This is well below the temperature of approximately 700oC at which
global collapse was triggered for both frames. This indicates that the ASIC Specifications
pertaining to individual column buckling could be overly conservative in determining the limit
state for an entire system performance.
Failure Line
Failure Line Failure Line
(a) (b)
Fig. 7 - Vertical Displacements of First Floor Columns (a) Case 1 and (b) Case 2
SUMMARY AND CONCLUSION
In this study, detailed FE models of 3D six-story steel building structures were created and
utilized for collapse simulations. This was realized through employing detailed FE simulations
to assess the behavior of individual members such as girders, beams, columns and braces as
well as the whole system response that is controlled by the interaction between the various
elements and components. The 6-story building system was subjected to the ASTM E-119
Standard fire curve. Two different scenarios were evaluated including a first-floor corner
compartment and whole first floor.
The main findings and general conclusions obtained from this study are summarized in the
following key points.
• In first floor corner compartment fire cases, the building models was shown to twist,
lean towards the compartment in which the fire is applied, and laterally sway.
• In the full first floor fire cases, failure of the building progressed straight downwards,
and no visible twist was observed during collapse.
• The interaction equation values for all columns was exceeded and in some cases at
low temperature.
REFERENCES
Agarwal, A, and Varma, A. (2013). Fire Induced Progressive Collapse of Steel Building
Structures: The Role of Interior Gravity Columns. Engineering Structures, 58, 129-140.
British Steel. (1999). The Behavior of Multi-Story Steel Framed Buildings in Fire. A report by
the European Joint Research Program. Retrieved from
http://www.mace.manchester.ac.uk/project/research/structures/strucfire/DataBase/Reference
s/MultistoreySteelFramedBuildings.pdf
Ellingwood, B. and Corotis, R. (1991). Load Combinations for Buildings Exposed to Fires.
Engineering Journal, 28 (1): 37–44.
En, Cen. (2005). Eurocode 3: Design of Steel Structures, Part 1.2: General Rules—Structural
Fire Design. London: British Standards Institution. Vol. 2.
Foutch, D., Roeder, C., and Goel, S. (1986). Preliminary Report on Seismic Testing of A. Full-
Scale Six-Story Steel Building.
Gann, R. (2008). NIST NCSTAR 1A: Final Report on the Collapse of World Trade Center
Building 7, Federal Building and Fire Safety Investigation of the World Trade Center Disaster.
(2008). National Institute of Standards and Technology, Gaithersburg, MD.
Gu, L. and Kodur, V. (2011). Role of Insulation Effectiveness on Fire Resistance of Steel
Structures under Extreme Loading Events. Journal of Performance of Constructed Facilities,
25(4): 277–86.
Huang Z, Burgess I, and Plank, R. (2000). Three-Dimensional Analysis of Composite Steel

Framed Buildings in Fire. Journal of Structural Engineering, 126(3):389–97.
Jiang, J., Guo-Qiang L., and Usmani, A. (2014). Progressive Collapse Mechanisms of Steel
Frames Exposed to Fire. Advances in Structural Engineering, 17(3): 381-398.
Jankowiak, T and Lodygowski, T (2005). Identification of Parameters of Concrete Damage

Plasticity Constitutive Model. Foundations of Civil and Environmental Engineering, 6(1): 53-
69
Johnson, G. and Cook, W. 1985. Fracture Characteristics of Three Metals Subjected to

Various Strains, Strain Rates, Temperatures and Pressures. Engineering Fracture Mechanics
21(1): 31–48.
Memari, M. and Mahmoud, H, 2014. Performance of Steel Moment Resisting Frames with
RBS Connections under Fire Loading. Engineering Structures, 75, 126–138.
Mahmoud, H., Ellingwood, M. and Memari, M. (2015). Challenges and Alternative Approaches
for Simulating the Response of Steel Structures Exposed to Fire”, Second International
Conference on Performance-based and Life-cycle Structural Engineering (PLSE), Brisbane,
Australia, 9-11 December.
Newman, G., Robinson, J., and Bailey, C. (2000). Fire Safe Design: A new Approach to Multi-
Story Steel-Framed Buildings. Steel Construction Institute.
Rex, C. and Easterling, S. (2003). "Behavior and Modeling of a Bolt Bearing on a Single Plate."
Journal of Structural Engineering, 129(6): 729 - 800.
Richard Liew, J. and Ma, K. (2004). Advanced Analysis of 3D Steel Framework Exposed to
Compartment Fire. Fire and Materials, 28:253–267.
Sarraj, M. (2007). The Behaviour of Steel Fin Plate Connections in Fire. PhD thesis, University
of Sheffield.
Simulia v6.14, (2014). Dassault Systems, Providence, RI. (www.simulia.com).
Sun, R., Huang, Z., and Burgess, I. (2012). The Collapse Behaviour of Braced Steel Frames
Exposed to Fire. Journal of Constructional Steel Research, 72: 130–42.
Wang Y. and Moore, D. (1995). Steel Frames in Fire: Analysis. Engineering Structures, 17(6):
462–72.
PROGRESSIVE COLLAPSE:
THE CASE OF COMPOSITE STEEL-CONCRETE FRAMES
Riccardo Zandonini1 riccardo.zandonini@unitn.it, Nadia Baldassino1 nadia.baldassino@unitn.it,

Fabio Freddi2 f.freddi@ucl.ac.uk, Giacomo Roverso1 giacomo.roverso@unitn.it
1Department of Civil, Environmental and Mechanical Engineering, University of Trento

Trento, Italy
2Department of Civil, Environmental and Geomatic Engineering, University College of London

London, UK
ABSTRACT
Residual strength and alternate load paths are two fundamental design strategies to ensure
adequate resistance against progressive collapse of structures. This paper presents an
experimental study carried out on two full-scale steel and concrete composite frames to
investigate their structural behaviour in case of a column collapse. The study focusses on the
redundancy of the structure as provided by the beam-slab floor system as well as by the ductile
beam-to-column joints. The specimens were ground floor sub-frames ‘extracted’ from two
reference buildings designed in accordance to the Eurocodes. The frames have the same
overall dimensions, but a different, symmetric and asymmetric, configuration of the column
layout. In both tests, the collapse of an internal column was simulated. The paper presents the
main features of the frames and the principal outcomes of the test on the symmetric frame.
INTRODUCTION
In the last decades, important and numerous studies have been conducted about the collapse
of structures caused by accidental loss of columns. The interest in studying the effects of
extreme loading conditions has being triggered by a few catastrophic events occurred in recent
years since the Ronan Point Building case (UK, 1968). These tragic events lead many countries
to include, in their design codes [1-6], integrity and structural robustness requirements in order
to design a robust structure. As to design principles, the Eurocode 0 [7] prescribes that “a
structure shall be designed and executed in such a way that it will not be damaged by events
such as explosion, impact and the consequences of human errors, to an extent disproportionate
to the original cause”. As to design practice, the Eurocode 1-7 [4] provides several strategies to
design structures against accidental events. In particular, strategies for identified accidental
actions and strategies for limiting the extent of localised failure are specified.
Column loss in a building is one of the most common and effective damage scenario
recommended for progressive collapse investigations. The collapse of vertical members causes
dynamic effects, large deformations of the floor system and high rotation in the beam-to-column
connections. The collapse of an extensive part of the structure can be avoided if the damaged
part is able to redistribute loads to the undamaged parts so that a new stable equilibrium
configuration is achieved. In the context of alternate load paths, the ductility offered by the joints
and the 3D performance capabilities of the floor system represent essential factors for a robust
structural response. In order to improve the knowledge about progressive collapse and to study
the effects that a column loss causes to a structure, 3D full-scale experimental tests are the
most comprehensive approach. On the other hand, these experiments are very complex and
expensive, and, to date, very few experimental data are available.
Recently, the ‘RobustImpact’ research project [8] studied the robustness of composite steel-
concrete frames affected by accidental actions. This European research project, aimed at
developing a new robust design approach against impact loading, is based on the residual
strength and the alternate load path method. Analytical, numerical and experimental activities
were planned and accomplished (Figure 1) to get insight into different issues of robustness.
DEMANDS ON COMPONENT JOINT

JOINTS DUE TO RESPONSES AT HIGH
IMPACT SPEED LOAD
STEEL JOINTS AND

BASE PLATES
RESPONSE UNDER
IMPACT LOADING
DYNAMIC
BEHAVIOUR OF
COLUMNS
JOINT
BEHAVIOUR
3D-SYSTEM
RESPONSE 2D-FRAMES
RESPONSE
Fig. 1 – Issues investigated in the Robustimpact project (after [8])
Within this project, the University of Trento activity focused on the contribution provided by the
concrete slab and by the beam-to-column joints. For a better insight into the mechanisms of
force redistribution in the structures, two experimental tests were conducted on 3D full-scale
steel and concrete composite structures ‘affected’ by the loss of an internal column. The
specimens had the same geometric and structural properties, but two different column layouts.
This paper presents the main features of the experimental study, with particular reference to the
main outcomes of the first ‘symmetric’ test.
THE REFERENCE STRUCTURES
Two ‘typical’ five-storey steel and concrete office buildings were selected as reference
structures. In particular, two different columns layouts were investigated: the first configuration is
symmetric with respect to both the plan X and Y axes (Figures 2-3), while the second one is
symmetric only with respect to the Y axis. The structures were named as “Symmetric” and
“Asymmetric”, respectively.
Steel Braces
34.2 m
12.8 m
A
Y
5.7 m
11.4 m
B
X
5.7 m
5.7 m 5.7 m 5.7 m 5.7 m 5.7 m 5.7 m

1 2 3 4 5 6 7
Fig. 2 – The plan view of the symmetric reference building
34.2 m 11.4 m
5.7 m 5.7 m 5.7 m 5.7 m 5.7 m 5.7 m 5.7 m 5.7 m
3.6 m 3.6 m 3.6 m 3.6 m 3.6 m
3.6 m 3.6 m 3.6 m 3.6 m 3.6 m

18.0 m
18.0 m
a) Y-direction b) X-direction
Fig. 3 – Elevation views
The design was based on the Eurocodes rules [9-13]. The aim was to study frame systems,
where no detail was affected by seismic design rules. Bracings were hence designed under
wind loads only. Besides, structural design aimed at getting, for both structures, the same
section of the structural elements. In particular, columns were HEB220, steel beams IPE240
and concrete slab thickness 150 mm. In absence of previous studies on progressive collapse of
composite frames, it was decided to have fully composite beams. Beam-to-column joints (Figure
4), the same in both structures, were bolted flush-endplates designed according to the
component method as in the Eurocode [12]. Structural steel grade S355, rebars grade B450C,
bolts class 10.9 and concrete C30/37 were the materials selected for the structural elements
and joints.
A Section A-A Section B-B
HEB 220 HEB 220
B B
IPE 240 IPE 240 IPE 240 IPE 240
Bolts M20 10.9
Bolts M20 10.9
A
Fig. 4 – Beam-to-column internal joint
THE SUB-STRUCTURES
The tests were performed on full-scale 2x2 bays sub-frames ‘extracted’ from the first floor of the
corresponding reference building, as illustrated by the dotted area in Figure 2. The circle
individuates the column that was removed during the test. Plan view and cross section of the
symmetric specimen are reported in Figure 5.
Finite element analysis of the full-frames and of the sub-frames provided the background to the
design of the experimental test and, in particular, the lateral restraining system that connects the
specimen to the counter-walls of the laboratory. The goal of the analyses was to mimic in the
test the presence of the bracing system and of the remaining part of the reference structure.
They pointed out the need for extending the columns up to approximately the point of contra-
flexure between the first and second floor. A set of trusses, pinned to the top of the sub-frame
columns (called ‘crowning beams’) were also required to adequately account for the influence of
the upper stories of the reference structure. As to the sub-frame restraining system, three
different options, as illustrated in Figure 6, were considered in the analysis. In Option 1 and 3,
only the steel beams are restrained while the slab is not connected to the reaction wall. The
presence of the bracings in the full-frame prevents from any significant longitudinal
displacement and hence, the relevant d.o.f. U1 is fully restrained in the sub-frame. This d.o.f. is
left free at the central beam (B in Figure 6) where the vertical and lateral displacements (U2 and
U3) are restrained. Besides, in Option 1, the end rotations of beams A and C about both
principal axes (R2 and R3) are restrained, and the central beam’s end is restrained against
rotations R2 and R3. In the Option 2, in addition to the restraints of the Option 1, also the parts
of the slab adjacent to the lateral beams are connected, for a width of 0.5 m, to the reaction
wall, restraining all the translational degrees of freedom. Option 3 is similar to Option 1 but all
the rotations are released. Comparing the results in terms of deformations and internal forces
with the ones obtained from the analysis of the corresponding full-frame, option 3 was then
selected, where all the lateral restraints are made up by truss elements connected to the steel
frame. More details about the design of the lateral restraining system can be found in [14].
B Section A-A
600 4910 1000 4910 600 5700 5700
800 490
2035
3600
5700
12510
A A
Section B-B
5700 5700 1290
5700
2035
310
3600
310 5700 5700 310

12020
B
Fig. 5 – Plan view and sections of the symmetric sub-frame (measures in mm)
a)Option
Option1 1 b)Option
Option2 2 c)Option
Option3 3
A A A
Y, 2
X, 1
B B B
C C C
Z, 3 A A A
X, 1
B B B
Fig. 6 – Lateral restraints options
The specimens were built inside the Laboratory of Materials and Structures Testing of the
University of Trento. The construction of the frame started with the erection of the steel skeleton
and the formwork installation (Figure 7a-b). The reinforcement bars were then positioned and
the concrete poured (Figure 7c). Figure 8a-b shows the symmetric specimen completed, while
Figure 8c shows the ‘central column’ that was replaced by a hydraulic jack in order to simulate
the collapse. During the constructional phases, the central beams were held in position by using
a provisional propping system, which was removed when the hydraulic jack was activated.
a) b) c)
Fig. 7 – The constructional phases
a) b) c)
Fig. 8 – The specimen
THE MATERIALS
Compression and splitting tests on concrete samples were performed according to the criteria of
[15] and [16] respectively. During the casting phase, 18 cubes (150 mm side) and 15 cylinders
(150 mm diameter and 300 mm height) were prepared. In order to appraise the evolution of the
concrete compression resistance, tests on cubes were performed at age, from casting, of 7, 28
and 102 days (e.g. the time of the full-scale test), while splitting tests on cylinders were
conducted at ages of 28 and 102 days. Table 1 reports the measured concrete properties.
Table 1: Concrete properties

Concrete’s age Average cube compressive Average tensile splitting
n. of tests
(days) strength (MPa) strength (MPa)
7 3 43.83 -
Cubes 28 9 56.47 -
102 6 65.74 -
28 6 - 3.81
Cylinders
102 9 - 4.25
As to steel, Table 2 and 3 report the yield stress, the ultimate tensile strength and the elongation
at failure of rebars and structural steel, respectively.
Table 2: Rebars steel properties
Rebar’s diameter (mm) Yield stress (MPa) Ultimate tensile strength (MPa) Agt (%)*
10 496 586 10.5
16 523 631 9.4
*Agt total extension at maximum force
Table 3: Structural steel properties

Yield stress Average yield Ultimate tensile Average ultimate A (%)*
Component
(MPa) stress (MPa) stress (MPa) tensile stress (MPa)
300 441 34.9
Column HEB220 306 303.3 442 440.3 34.5
304 439 36.1
383 537 28.2
Beam IPE240 391 409.3 541 540.7 27.0
454 544 33.3
*A permanent deformation of the gage length after fracture
THE MEASUREMENT SET-UP

The specimen was extensively instrumented to maximise the information gained from the
experiment. Due to the complexity of the frame response, the most important parameters to be
measured during the test were carefully identified. The attention was focused on the response
of the columns, beams and joints. In particular, strain gauges were installed to measure the
strain state at the columns base, at the mid-span of the internal and external beams and near
the central column in the internal beams. From the readings of the strain gauges, it was possible
to obtain some parameters such as the average axial strain and the curvature of the section,
and consequently, assuming the material in the elastic range, the axial force and the bending
moment. Strain gauges to measure the axial strain were also installed in the truss restraining
elements, in the crowning beams and in some reinforcement bars in the vicinity of the central
column. Displacement transducers and inclinometers were installed in correspondence of the
beam-to-column connections to measure the joints’ rotation. Further transducers enabled to
monitor the torsional rotation of the external beams and the rotation of the external columns in
correspondence of the beams’ joints. At the central node of the frame, a wire transducer
measured the vertical displacement and a load cell the force acting on the central column.
Furthermore, the vertical displacements at the centre of the slab panels were monitored.
Instruments’ signals were logged at a frequency of 2 Hz. Figure 9 provides the layout of the
instrumentation set-up, and Table 4 lists the instruments together with the related parameters.
In Figure 9, the letters identify the columns (column A, column B,…), the beams between
columns (beam AB, beam BC, ...), the lateral restraints (br. A, br. G, br. I, br. HG and br. HI).
Furthermore, P1, P2, P3 and P4 are the points where the transducers measuring the vertical
displacements of the slab panels were located.
THE TESTING PROCEDURE

The test plan comprises the following steps:
1. activation of the hydraulic jack and removal of the propping system;
2. application of the vertical loads. At this aim, bags filled with sand were placed on the
slab reproducing a uniform distributed load of 8.80 kN/m2 to approximate the factored
design load, including finishes, partitions and live loads (Figure 10a). This load is not the
Code recommendation for checks against progressive collapse, but it makes possible
studying the floor capability of transfer loads in nonlinear range (catenary action,
membrane action, joint ductility). The bags were placed on two layers: the first one is
distributed uniformly on the slab, the second one is placed on a reduced area as shown
in Figure 10b;
3. simulation of the column removal by reducing the pressure of the hydraulic jack down to
zero (velocity=1.4 mm/min);
4. stabilization of the specimen;
5. application, by means of the actuator, of a tensile force in displacement control mode
(velocity = 1.50 mm/min) incremented up to ‘collapse’.
br.G
br.I
br.HG br.HI
GH HI
G H I
P2 P3
DG
EH
FI
DE EF
D E F
P1 P4
AD
CF
BE
br.A AB BC
A B C
displacement transducer (x2) strain gauge inclinometer

wire transducer instrumented column (with strain gauges)
Fig. 9 – The measurement set-up
THE MAIN RESULTS

The deflection of the central node is plotted in Figure 11 with respect to the load measured by
the load cell (in the graph positive values of the load mean compression). The central column
was completely lost at a central node displacement of about 165 mm, which corresponds to
1/33 of the beam span. The load carried by the central column (E) has redistributed in the other
columns as shown in Table 5. As a result, columns B, D and F carried about twice the axial
force acting before the column removal. The axial force increase in column H was the greatest
due to the nearby restraining truss system. At the contrary, the corner column C unloaded due
to the effect of the concrete slab action. The test was ended when, under the applied tension
force, the connection between the central column E and the beam EH ‘failed’ (Figure 12). The
failure was associated with the fracture of the two bolts of the bottom row of the joint c (see
Figures 13a and 14a). This occurred for a vertical displacement of approximately 300 mm (1/18
of the beam span) and a corresponding applied force of 300.85 kN. At this stage, the column H
was the most stressed, while columns B, D and F are stressed almost at the same level (see
Table 5). The slab effect ‘maintains’ the corner column C axial force basically constant.
Table 4: Instruments and parameters measured

Structural element Instrument Parameter measured Parameter deduced
Average axial strain Axial force
Strain gauges at the base Curvature (strong-weak
Bending moments
Columns axis)
Displacement transducers
Rotation
at the beam level
Strain gauges at mid-span Average axial strain Axial force
Internal beams and near the central
column Curvature (strong axis) Bending moment
Strain gauges at mid-span Axial strain Axial force
External beams Displacement transducers
Torsional rotation
at mid-span
Crowning beams Strain gauges at mid-span Axial strain Axial force
Lateral restraints Strain gauges at mid-span Axial strain Axial force
Strain gauges near the
Reinforcement bars Axial strain Axial force
central column
Displacement transducers Rotation -
Joints
Inclinometers Rotation -
Slab panels Wire transducer Deflection -
Hydraulic jack (central Load cell Axial load -
column) Wire transducer Vertical displacement -
2520
1120
2600 2600
1460 1460
2650
3730 2600
1540
a) b)
Fig. 10 – Vertical loads on the slab (measures in mm)
Test phases:
A: Application of the vertical load
B: Column removal
C: Application of the tensile force
Fig. 11 – Load-deflection relation of the central node
Fig. 12 – The specimen at the end of the test
Table 5: Axial force in the columns (kN)

Column End of the loading phase End of the column removal End of the test
E 669.89 2.64 -300.85
B 143.82 319.08 373.08
C 58.01 49.49 63.01
D 148.99 285.07 330.83
F 151.63 308.04 365.02
H 212.07 546.84 635.54
Both central and external beam-to-column joints experienced important deformations at the end
of the test. In particular, the following phenomena occurred: at the central node, significant
plastic deformations of the endplate of the beams BE and EH (Figure 14b-c); in the vicinity of
the external column H (Figure 15a), instability of the bottom flange of the beam EH; in the web
panel of column B, shear deformation in correspondence of the connection with the beam BE
(Figure 15b); at external columns D and F (Figure 15c), horizontal cracks in the slab. Figure 16
reports the connections rotation with respect to the load applied on the hydraulic jack (Joint f
rotation is not reported due to an instrument malfunction). The substantial rotational demand is
apparent.
Focusing on the internal beams, Figure 17 reports the axial force (normalised on the yield force
of the steel section) near the central node and at the beam mid-span. As a first appraisal of the
response, near the central node (Figure 17a) the axial force evolved from negative to positive
due to the change of the bending moment sign during the column removal. In correspondence
with the beam mid-span (Figure 17b), the steel section remains in tension during all the duration
of the test. The process of passing from a negative to a positive bending moment at the central
node was pointed out as well by the evaluation of the axial strains in the instrumented
reinforcement bars near the central column E. During the application of the tensile force by the
actuator, the rebars registered a significant axial force increase, and, at the end of the test,
some bars reached the yielding force.
f c d
+ + +
b a h
Rotation Rotation Rotation
a) Joints position b) Central joints c) External joint

Fig. 13 – Joints position and positive rotation assumption
a) Central node b) Connection a c) Connection c

Fig. 14 – Central connections at collapse (see Figure 13a)
a) Connection g b) Connection e c) Joint f

Fig. 15 – External connections at collapse (see Figure 13a)
CONCLUSIONS
This paper presents the main results of an experimental test performed at the University of
Trento on a full-scale steel and concrete composite structure subjected to a column loss. It
enables investigating the importance of the joints’ ductility and the role of the concrete slab for
allowing the forces redistribution in the structure associated to the alternate loading path
activated after the column loss.
a) Internal connections b) External connections

Fig. 16 – Connection rotation
a) Near the central node b) At the beam mid-span

Fig. 17 – Axial force in the central beams
The full-scale specimen was ‘extracted’ from the first floor of a reference building designed in
accordance with the Eurocodes. The central column collapse was simulated by replacing the
column with a hydraulic jack that was kept inactive before the beginning of the loading. The test
was carried out with the following sequence: the hydraulic jack was first activated and the
propping system removed, the vertical loads were then applied on the slab and the column
removal was simulated reducing the pressure of the jack down to zero. Finally, with the aim of
appraising the structural residual strength, an incremental tension force was applied at the
central node up to the ‘frame collapse’ associated with the failure of a central joint.
The results reported in this paper pointed out that the joints, designed as ductile, enable
achievement of very high rotations. Joint ductility is provided by plastic deformations, mainly of
the end-plate and the column web in shear. Local beam buckling and horizontal cracks of the
outstanding slab at external joints should be considered in design in order to ensure adequate
ductility.
Despite the collapse was associated with joints’ failure, the key role of the slab for ensuring
loads and forces redistribution from the damaged to the undamaged parts of the structure is
apparent. The cracking pattern on the top side of the concrete slab revealed the formation of
‘compressive rings’ typical in slabs where the membrane forces are activated.
ACKNOWLEDGMENTS
The test was made possible by the grant from the Research Fund for Coal and Steel of the
European Community (Contract number No. RFSR-CT-2012-00029). The Authors also
gratefully acknowledge the work by the technicians of the Structural Testing Laboratory of the
University of Trento: Stefano Girardi, Marco Graziadei and Alessandro Banterla.
REFERENCES
[1] GSA, Progressive collapse analysis and design guidelines for new federal office
buildings and major modernization projects, General Service Administration, USA, 2003.
[2] DoD, Unified Facilities Criteria (UFC): Design of building to resist progressive collapse,
Unified Facilities Criteria (UFC) 4-023-03, Department of Defense, USA, 2009.
[3] Office of the Deputy Prime Minister. The building regulations 2000, part A, schedule
1:A3, disproportionate collapse, London, UK, 2004.
[4] EN 1991-1-7, Eurocode 1: Actions on structures – part 1-7: General actions – Accidental
actions, European Committee for Standardization, Brussels, 2006.
[5] BS, BS5950: Structural use of steelwork in buildings, part 1: code of practice for design –
rolled and welded sections, British Standard Institute, London, UK, 2001.
[6] BS, BS6399: Loading for buildings, part 1: code of practice for dead and imposed loads,
British Standard Institute, London, UK, 1996.
[7] EN 1990, Eurocode 0: Basis of structural design, European Committee for
Standardization, Brussels, 2004.
[8] European Commission - Research Programme of the Research Fund for Coal and Steel.
Robust impact design of steel and composite building structures – ‘RobustImpact’. Grant
Agreement number RFSR-CT-2012-00029. Document in press.
[9] EN 1991-1-1, Eurocode 1: Actions on structures – part 1-1: General actions, densities,
self-weight, imposed loads for buildings, European Committee for Standardization,
Brussels, 2002.
[10] EN 1992-1-1, Eurocode 2: Design of concrete structures – part 1-1: General rules and
rules for buildings, European Committee for Standardization, Brussels, 2004.
[11] EN 1993-1-1, Eurocode 3: Design of steel structures – part 1-1: General rules and rules
for buildings, European Committee for Standardization, Brussels, 2005.
[12] EN 1993-1-8, Eurocode 3: Design of steel structures – part 1-8: Design of joints,
European Committee for Standardization, Brussels, 2005.
[13] EN 1994-1-1, Eurocode 4: Design of composite steel and concrete structures – part 1-1:
General rules and rules for buildings, European Committee for Standardization,
Brussels, 2004.
[14] R. Zandonini, N. Baldassino, F. Freddi. Robustness of steel-concrete flooring systems -
An experimental assessment, Stahlbau, 83 (9), 608-613, 2014.
[15] EN 12390-3:2009, Testing hardened concrete – Part 3: Compressive strength of test
specimens.
[16] EN 12390-6:2009, Testing hardened concrete – Part 6: Tensile splitting strength of test
specimens.
Concrete Filled Steel Tubes for Bridge Applications
Charles W. Roeder
Professor, Department of Civil Engineering, University of Washington
Seattle, WA 98115-2900 USA
croeder@uw.edu
Dawn E. Lehman
Professor, Department of Civil Engineering, University of Washington
Seattle, WA 98115-2900 USA
delehman@uw.edu
ABSTRACT
Concrete filled steel tubes (CFST) are composite members that are used in many countries, but
with limited use in the US. Bridge construction uses steel tubes filled with reinforced concrete
(RC) for pile and drilled shaft foundations, but they are designed as RC members where
composite behavior is neglected. Steel tubes have been used for jacketing upgrades or stay-in-
place formwork for RC bridge piers. Recent research at the University of Washington funded by
the CALTRANS, WSDOT, and others focused on using the composite behavior of circular CFST
for accelerated seismic construction of deep foundations and bridge piers. Research has
evaluated the behavior of large diameter circular CFST members and their connections.
Connections with good seismic performance and suitable for economical and rapid construction
were developed. Deep foundations encounter high shear and moment demands due to lateral
spreading and liquefaction of soils, and this has also been investigated. This research is
summarized.
INTRODUCTION
CFST is widely used for bridge construction in Asia, but its use is limited in the US. US bridge
engineers sometimes have serious preconceptions and misconceptions that have limited the use
of CFST. Bridge construction in the US is inherently different than in Asia because of differences
in construction costs and practices. As a result, design and construction practices that are
practical and economical in other countries may not be viable for US construction. Nevertheless,
CFST offers potential advantages for US bridge construction, because of its high strength,
stiffness and ductility, and because of the accelerated construction that can be provided by CFST.
This paper reviews the current use of CFST in US bridge construction, recent research into the
use of CFST in bridges and potential future applications of these composite members.
CFST IN BRIDGE CONSTRUCTION
Composite CFST construction has had limited use in US bridge construction, steel tubes filled
with RC with no composite behavior with the steel tube have been used more frequently. Three
seismic applications of steel tubes with RC infill are noted in US bridge construction. First, older
RC bridge piers suffer severe damage during earthquake load as illustrated in Figure 1, and
jacketing for seismic retrofit of reinforced concrete bridge piers with inadequate shear
reinforcement, confinement, or rebar splice and development lengths have been done to ensure
improved seismic performance. Extensive research has been performed on this issue (Priestley
et al. 1994), because increased pier resistance from composite action increases the seismic
demands on foundations and other parts of the structure. Jacketing confines the concrete but
avoids shear stress transfer (and the resulting composite resistance) between the RC column and
the steel tube. In addition, the steel tube is stopped short of connecting elements to limit force
and moment transfer.
a) Inserting the steel tube b) Inserting the reinforcing

cage
Fig. 1. Seismic damage to
an RC bridge pier Fig. 2. Deep drilled shaft foundations
Bridges in the US employ a range of foundation types. Spread footings are often the most
economical, but bridges designed for seismic loadings may require deep foundations due to large
seismic design forces and soft, weak soil deposits. Pile foundations may use precast concrete
piles, steel H-piles, or steel tube piles, and these piles are a second major application of steel
tubes filled with RC fill. Pile-to-pile cap connections in bridge design often are simple embedded
shear resisting connections, which may develop limited flexural resistance that is neglected in
design. Steel tube piles offer significant advantages, and they are commonly used today. The
tubes can be driven or vibrated into place, cleaned of soil for a portion of their depth. A reinforcing
cage may be inserted into the pile to provide a calculable flexural resistance and increased
stiffness and end bearing capacity of the pile, but the composite behavior of CFST usually is
neglected in design. The reinforcing cage may also extend into the pile cap to provide shear and
moment transfer at the pile-to-pile cap connection.
Drilled shaft foundations are a third application of steel tubes filled with RC for bridge design. The
shafts may be 2 to 4m in diameter and more than 80 to 90m in depth in seismic regions. A steel
tube frequently is inserted into the shaft to hold back loose soil and facilitate construction (see
Figure 2a), and a steel reinforcing cage is inserted into the tube as shown in Figure 2b before the
tube is filled with concrete. The resistance of the shaft is evaluated as a RC member with no
consideration of the composite action or the steel tube.
Bridge engineers often confuse these 3 noncomposite applications with CFST, and US design
specifications probably contribute to this misinterpretation. Composite CFST design provisions
have historically evolved in 3 major US design specifications. The AISC and ACI specifications
are focused on steel and concrete buildings, respectively, (AISC 2010, ACI 2011) and they
consider the composite behavior of CFST. The AASHTO specifications are focused on bridge
design (AASHTO 2016), and these provisions historically lie between the AISC and ACI
provisions. However, there is substantial differences between CFST provisions in ACI and AISC,
and a brief comparison is justified.
Fig. 3. Evaluation models for CFST; a) AISC plastic stress distribution method, b) AISC strain
compatibility, c) ACI strain compatibility
The AISC and ACI specifications provide specific methods for predicting the flexural and axial
resistance of circular CFST. AISC (2010) permits both the plastic stress distribution method
(PSDM) (see Figure 3a) and the strain-compatibility (SC) (see Figure 3b) methods. The ACI
provisions recognize only the SC method as illustrated in Figure 3c. For the PSDM, the section
develops the full yield stress of the steel in tension and compression and a uniform compressive
stress of 0.95f’c in the concrete of circular CFST in compression where f’c is the specified
compressive strength of the concrete. Equilibrium is applied over the cross section for each
possible neutral axis location to develop axial load-bending moment interaction curves, which are
affected by the slenderness, D/t, of the tube. Smaller D/t values result in larger resistance due to
larger area of steel. Larger D/t values significantly increase bending moment for modest axial
loads due to the increased contribution of the concrete fill in compression. However, the total
resistance is smaller, since the concrete contributes more to the resistance. Bishop (2009)
evaluated these alternate methods, compared design predictions to past experimental results,
and showed that the PSDM is conservative and consistently provides more accurate predictions
of circular CFST resistance. He showed that the experimental moment capacity of circular CFST
was 1.24 times the measured moment capacity in prior experiments with a standard deviation of
0.18. The SC method has considerable appeal, since it is the basis of reinforced concrete design,
and it is also provides a measure of curvature and deformation of the member at various load
levels. However, Bishop showed that it is not a particularly reliable model for CFST, because it
requires a strain or deformation limit to establish the resistance. He showed that the experimental
moment capacity of CFST was 1.64 time the ACI SC prediction, because the 0.003 mm/mm strain
limit was employed, and the standard deviation was 1.17. This great scatter was caused because
the .003 strain limit for concrete was proposed to control spalling of concrete members, and
spalling cannot occur in CFST due to the confinement provided by the steel tube.
The AISC SC method employs a linear strain distribution, a bilinear steel material curve, a
parabolic unconfined concrete material behavior, and equilibrium to determine axial and flexural
capacity as shown in Figure 3b. The maximum strain in the extreme compression fiber of the
concrete (which is within the tube) is still 0.003 mm/mm, and so this method also provides a much
more conservative estimate of behavior. Bishop explored other methods of limiting strain,
curvature or other deformation with the SC method. Other limits such as a strain limit in the steel
or curvature of the member resulted in greater accuracy and smaller standard deviation, but they
also produced a significant number of unconservative predictions of CFST resistance.
RECENT EXPERIMENTAL RESEARCH ON CFST FOR BRIDGE CONSTRUCTION

Some recent developments have changed the prospects for using CFST in bridge design. First,
bridge construction has recently focused on accelerated and more economical construction
without resulting in reduced structural performance. CFST clearly offers benefits in this realm
because they provide good structural performance with reduced cost and field construction time
by eliminating formwork, shoring, and internal reinforcement. Further, CFST elements offer
greater resistance with significantly less weight and material than RC members, because the steel
tube provides greater confinement to the concrete and reinforces the concrete at the optimal
location. As a result, significant savings in material may result.
Research clearly shows that circular CFST provides better performance than rectangular CFST
because circular tubes better confine the concrete fill and increase the shear stress transfer (bond
stress) between the tube and the concrete fill. However, US engineers often use rectangular
CFST, because they believe that connections are easier due to the flat steel surfaces provided
by the tube. Recently, significant research has been performed on circular CFST connections with
large diameter tubes suitable for bridge construction. Two CFST column-to-footing or CFST pile-
to-pile cap connections have been developed as shown in Figure 4 (Lehman and Roeder 2012).
Both connections employ an annular ring (see Figure 5) at the end of the tube to develop
connection resistance. The annular ring has the same thickness and yield stress as the wall of
the tube and extends a distance 8 times the thickness into the fill and the encasing concrete.
Figure 4a shows a cast-in-place embedded connection where the annular ring is temporarily
supported on a lift or pedestal and then cast into the footing or pile cap. Construction labor issues
in US may result in a preference for the recessed grouted foundation connection illustrated in Fig.
4b. The footing is cast with a recess formed by a light gauge corrugated steel tube. The CFST
tube with its annular ring is placed into the recess and grouted with high strength reinforced grout.
The tube is then filled with concrete, and construction proceeds as with other connections. In
addition, three CFST column-to-precast cap beam connections illustrated in Fig. 6 have been
developed to facilitate the use of precast pier caps with CFST pier columns and to further enhance
accelerated bridge construction (Stephens et al. 2014).
a) Embedded connection b) Grouted connection Fig. 5 Annular Ring

Fig. 4. CFST column to footing connection
a) Embedded connection b) Welded dowel connection c) RC connection

Fig. 6. CFST column to precast cap beam connections
Eighteen foundation connection specimens were built and tested under cyclic lateral load in the
setup illustrated in Fig. 7 (Lehman and Roeder 2012). The connection performance is strongly
influenced by footing thickness and embedment depth, since they relate to punching shear and
cone pullout of the tube in flexure, and these issues were a focus of the research. Specimens with
shallow embedment depths cause significant damage to the reinforced concrete footing or pile
cap as shown in the lateral force-story drift plot and photo of Figures 8a and 8b. This shallow
embedment depth provides enough resistance to yield the tube and nearly develop the full plastic
capacity of the composite section, but the connection loses much of this moment capacity result
in virtually no damage to the footing as shown in Figure 8e, and they develop and maintain the
full plastic capacity of the composite member through large inelastic deformations with no
deterioration in resistance as shown in Figure 8a. Local buckling of the tube occurs after large
inelastic deformations as shown in Figure 8d, and after multiple cycles of large inelastic
deformation. The cyclic plastic strains due to straightening and rebuckling of walls of the tube
ultimately lead to ductile tearing. The concrete inside the tube is well confined, and no damage
to the concrete fill occurred until after severe local buckling of the tube. After severe local
buckling, crushing of the concrete in this local area of the buckle occurs, and the concrete in this
zone is large pulverized and may pour out of the tear. The required embedment depth and footing
thickness required for punching shear were established from this research.
Figure 7. Test Setup.
a) Force – displacement with shallow b) Photo of footing damage with shallow embedment
embedment
c)Force – displacement with adequate c) Local buckling of tube e) Photo of footing and connection at
embedment end of test
Fig. 8 Typical test results for CFST foundation connections
Calculations show that CFST bridge piers can be constructed with 30% to 40% less material and
weight than an RC pier of equivalent strength and stiffness. Further, CFST piers provide better
seismic performance than the heavier RC member. The CFST member is rapidly constructed
because the steel tube is fabricated off site, is quickly erected, and supports construction loads
during erection and subsequent placement and curing of the concrete fill. Creep and shrinkage of
the concrete is minimized with CFST construction compared to RC construction (Lehman et al.
2013). Environmentally friendly construction practices such as cement replacement can be
employed, because the steel tube supports construction loads during the long period required for
the concrete to gain its full strength with these alternate materials. Reinforcement or shear
connectors are usually not required within the steel tube, and no shoring or formwork is required.
Fabrication is quite simple since it consists of forming and cutting the tube and welding an annular
ring to the end of the tube. CFST pier columns can be constructed in days rather than weeks or
months required for RC bridge piers. The lighter weight of the steel tube eliminates the need for
heavy cranes required with other accelerated construction methods.
The CFST pier column-to-precast cap beam connections are divided into three broad categories:
(1) the embedded connection (see Figure 6a), (2) the welded dowel connection (see Figure 6b),
and (3) the RC connection (see Figure 6c). Eight CFST pier column-to-precast cap beam
connections were tested under cyclic inelastic deformation in either the longitudinal or transverse
directions of the bridge pier. The longitudinal and transverse directions are different, because the
bending moment is either parallel or normal to the axis of the axis cap beam, respectively.
The embedded connection (Figure 6a) is a logical extension of the grouted-recessed foundation
connection (see Figure 4b). A recess is formed in the precast cap beam. The recess of the
precast cap beam is placed over the CFST column and its annular ring, and is grouted into the
recess with reinforced grout. This connection develops the full resistance of the CFST as
illustrated in lateral force-story drift plot of Fig. 9a, and it supports all construction loads without
temporary shoring. As with the foundation connection, the embedment depth is important,
because this relates to potential cone pullout and damage to pier cap. The required width of the
cap beam is important to the economy, esthetics, and performance of the connection, and it was
also investigated. The ultimate failure of this connection is again associated with plastic
deformation of the composite member, local buckling of the steel. and tearing of the steel in the
local buckled region after numerous cycles of large inelastic deformation as shown in Fig. 10a.
The embedded connection develops the full strength of the CFST column, and it has high stiffness
and excellent ductility and inelastic performance.
a) Embedded connection b) Welded dowel connection c) RC connection

Fig. 9. Force-deflection behavior of precast pier cap connections
a) Tear in tube after cyclic local buckling of b) Footing damage cause by welded dowel
embedded connection connection with no debonding
c) Welded dowel connection with debonded d) Fracture of reinforcing bars and spiral in
bars RC connection
Fig. 10. Photos of typical precast pier cap connection behavior
The welded dowel and RC connections shown in Figures 6b and 6c were developed because
bridge engineers may prefer a weaker connection to limit the seismic demands on other structural
elements. Their strength depends upon the size and spacing of the rebar, which are limited by
the inside perimeter of the tube. The welded dowel connection will normally be significantly
stronger than a RC connection, because the reinforcing bar is entirely within the steel tube for the
welded dowel connection and concrete cover is not required. The steel tube does not extend into
the pier cap with either the welded dowel or the RC connection. Figure 6b, shows a welded dowel
connection with debonded reinforcing bars, since debonded bars provide better inelastic
performance (see Figure 9b) with little damage to the pier cap (see Figure 10c) increase the
inelastic deformation capacity and reduce the damage to the pier cap. Dowel bar connections
with bonded bars were tested and significant damage to the pier cap as shown in the photo of
Figure 10b. The actual behavior of unbonded RC and welded dowel connections is a rocking
behavior due to the large axial deformations of the unbonded bars, and significant separation
between the tube and the cap beam were noted as illustrated in Figures 10c and 10d. The
resistance of the welded dowel connection in Fig. 9b is fairly close to that of embedded
connection, because the internal reinforcement had 87% of steel area of the tube but the yield
stress of the rebar was 45% larger than that of the tube. While the resistance of this welded dowel
connection is greater than that possible with a RC column (see Figure 9c) due to the larger
effective diameter of the rebar and the better confinement of the steel, the resistance of welded
dowel connections is smaller than that of a CFST embedded connection, because it is difficult to
weld adequate rebar into the tube to develop the resistance of the tube. The RC connection also
had large inelastic deformation that was limited to yielding in the longitudinal rebars and grout in
the soffit region. The RC connection was weaker and more flexible than the embedded CFST
connection. Its stiffness was 69% of the stiffness of the embedded CFST of identical size and
properties. Significant damage to internal longitudinal bars and spiral reinforcing were noted in
these specimens. Buckling and fracture of the bars were also noted (see Figure 10d).
The placement and design of the welds joining the reinforcing bars to the inside of the tube was
a critical issue for the welded dowel connection. A flare bevel weld was designed to develop the
tensile strength and inelastic deformation capacity of the reinforcing bar, and an experimental
study of this weld was completed prior to testing of the full connections to assure that the
performance goals were achieved. The primary test parameters included rebar size, weld
strength, bonding or debonding the dowels from the concrete, and the embedment depth of the
bar into the steel tube. Weld strengths ranging from 80% to 120% of the dowels strength and
embedment depths of 16db and 24db were evaluated. The bars were pulled to failure using a
hydraulic ram placed into a self-reacting system. The failure mode in all tests was characterized
by dowel yielding followed by strain hardening and fracture of the steel bar. Debonding the
reinforcing decreased the visible concrete damage, and no damage to the tubes in the weld region
was observed in any of the tests. Design and placement of the weld was completed in accordance
with American Welding Society (AWS) Bridge Welding Code (AASHTO/AWS 2010).
The welded dowel and RC connection have substantial penalties with regard to accelerated
construction. The embedded CFST connection easily supports the weight of the cap beam,
girders, and construction equipment during erection, and so shoring is not required. However,
welded dowel connections cannot support the weight of the cap beam, and therefore shoring or
other support system is required to support the pier cap while the cap is grouted into place and
while the grout cures. It is possible that clamp supports may be applied to the CFST column to
support the pier cap with the RC and welded dowel connections. Additional data and information
these tests may be found elsewere (Stephens, Max T. et al. 2016 and Stephens, Max. et al. 2016).
The prior research focused on work done to develop economical and practical connections for
CFST piles, drilled shafts, bridge piers and columns. The shear force demands on foundation and
substructure elements are quite large for piles and drilled shafts subjected to earthquake loading.
Since bridges are commonly built on deep soil deposits, foundation design is often concerned
with liquefiable layers of soil that may introduce lateral spreading or movement during large
earthquakes. Movements of this type may cause large shear forces in piles and drilled shafts,
and there is great interest in establishing the shear capacity of CFST under different applications
and load conditions.
The Washington State Department of Transportation funded a study to address this issue. A few
prior shear tests had been performed (Xu et al. 2009, Xiao et al 2012, and Nakahara and Tokuda
2012), but these prior tests were on moderately small diameter tubes under idealized conditions.
Actual piles and drilled shafts are of much larger size and the actual applications may have dirt
or contamination inside the tube, inaccurate geometric placement and other concerns. This
research study evaluated larger tubes and tubes reflecting current field conditions. Figure 11
shows the test setup and deformation of one of the test specimens failing in shear.
It should be emphasized that shear and flexure are related, since shear is the derivative of the
moment diagram. Hence, shear force can be limited by flexural resistance, by failure in composite
action due to inadequate shear stress transfer between the steel tube and the concrete fill induce
failure, or by a true shear strength failure of the composite CFST member. This distinction does
not appear to be recognized some of the prior research studies, and this distinction was a major
focus of this research. The experimental study considered tubes that were 500 mm in diameter,
with wall thickness of 12.5 or 16 mm, and yield stress of 350 to 490 MPa. The shear span, a (see
Fig. 11a), had to be quite small to obtain a shear failure (usually a< 0.375D). Further, shear stress
transfer is required to develop the shear resistance of the composite CFST member. Hence, the
tail length in the specimens (see Figures 11a and 11b) were important. Specimens without
adequate tail length or with poor shear stress transfer between the steel tube and concrete fill
may only develop the shear resistance of the steel tube acting alone (ie V≈ 0.6Fy 0.5As).
a) Test setup b) Photo of test specimen
c) Photo of test specimen after significant d) Typical shear force-deformation behavior

shear deformation
e) Deformation and ultimate tearing of shear f) Concrete cracking of fill inside a shear
controlled CFST controlled CFST
Fig 11. Shear tests of CFST members
Twenty two large scale specimens were tested, and all three behaviors were obtained. Figure
11d is the force deflection behavior of a CFST specimen truly failing in shear. The shear
resistance of these specimens is significantly larger than 2 times the shear resistance of the bare
steel tube. Specimens failing through flexure or through inadequate tail or development length
have significantly smaller shear force. Figure 11e shows the severe shear strains in the steel
tube with this true shear failure mode. The deflections were extremely large compared to the
short shear span, and the ultimate failure was ductile tearing of the steel tube where the tear
initiated from the shear span of the specimen. The concrete fill clearly contributed to this large
shear resistance. Much of the contribution was provided by the concrete fill retaining the shape
and cross section of the tube. However, large shear strains and severe shear cracking occurred
in the shear span as shown in Figure 11f. Very limited cracking occurred in the concrete located
in the central flexural span of the specimens as shown in the figure.

This has been a brief overview of the use of CFST in US bridge construction. It is clear that
the use of CFST in bridge construction is less advanced in the US than in many other countries.
Nevertheless, significant advances have been made in recent years, and it is expected that the
use of CFST in US bridge construction will increase significantly in the future. The primary usage
of CFST has been and will likely continue to be for piles, drilled shafts and bridge pier columns.
Connections suitable for seismic design of bridge pier columns have been discussed. These
connections are practical and economical and they facilitate accelerated bridge construction. A
recent study on the shear resistance of CFST was summarize. It is shown that the shear
resistance of CFST is very large, and that the shear failure is a very ductile failure mode.
ACKNOWLEDGEMENTS
Research was sponsored by the Army Research Laboratory (Cooperative Agreement Number
DAAD19-03-2-0036 through Advanced Technology Institute), by the California Department of
Transportation (CALTRANS) (Agreements 59A0641 and 65A0446) and the Washington State
Department of Transportation (WSDOT) (Agreement T1461). The views and conclusions
contained in this document are those of the authors and should not be interpreted as representing
official policies, either expressed or implied, of the Army Research Laboratory, the US.
Government, CALTRANS or WSDOT. The U.S. Government is authorized to reproduce and
distribute reprints for Government purposed notwithstanding any copyright notation heron. The
authors gratefully acknowledge the financial support of these organizations. In addition, the advice
and assistance provided by Ron Bromenschenkel, Michael Cullen, and Peter Lee of the
CALTRANS, Bijan Khalighi of WSDOT, and Jon Tirpak of the Advanced Technology Institute and
the Vanadium Technology Partnership.
REFERENCES
American Association of State Highway and Transportation Officials (AASHTO) (2012) AASHTO
LRFD Bridge Design Specifications, Washington, D.C.
American Association of State Highway and Transportation Officials (AASHTO) (2014) AASHTO
LRFD Bridge Construction Specifications, 3rd Edition with 2010 through 2014 Interims,
Washington, D.C.
American Concrete Institute (ACI) (2011) Building Code Requirements for Structural Concrete (ACI
318-11), Farmington Hills, MI
American Institute of Steel Construction (AISC) (2010). Specification for Structural Steel Buildings,
AISC, Chicago, Illinois.
American Welding Society (AWS) (2010) “Bridge Welding Code,” AASHTO/AWS D1.5:2010
Standard, Miama, FL.
Bishop, E. (2009) Evaluation of the Flexural Resistance and Stiffness Models for Circular Concrete-
Filled Steel Tube Members Subjected to Combined Axial-Flexural Loading, a thesis submitted in
partial fulfillment of the Master of Science in Civil Engineering Degree, University of Washington,
Seattle, December 2009.
Lehman, D.E. and Roeder, C.W. (2012) "Foundation Connection for Circular Concrete Filled Tubes,"
Journal of Constructional Steel Research, Vol. 78, November 2012, pgs. 212-25, Elsevier.
Lehman,D.E., Kuder, K., Gunnarsson, A.K., Roeder, C.W., and Berman, J.W. (2013) "Circular
Concrete Filled Tubes for Improved Sustainability and Seismically Resilience", Journal of
Structural Engineering, ASCE, Reston, VA, ISSN 0733-9445/B4014008(12).
Moon, J., Lehman, D.E., Roeder, C.W., and Lee, H-E (2013) "Evaluation of Embedded Concrete
Filled Tube (CFT) Column-to-Foundation Connections," Engineering Structures, Vol. 56, pgs 22-
35, Elsevier.
Moon, J., Lehman, D.E., and Roeder, C.W. (2012) "Strength of Circular Concrete Filled Tubes (CFT)
with and without Internal Reinforcement Under Combined Loading," Journal of Structural
Engineering, ASCE, Reston, VA, DOI:10.1061(ASCE)ST1943-541X.000078.
Moon, J., Roeder, C.W., Lehman, D.E., and Lee, H-E (2012) "Analytical Modeling of Bending of
Circular Concrete-Filled Tubes," submitted for publication review, Engineering Structures,
Elsevier.
Nakahara, H., and Tokuda, S., (2012). “Shearing Behavior of Circular CFT Short Columns,”
Proceedings of 10th International Conference on Steel Concrete Composite and Hybrid
Structures, Singapore, pgs 362-369.
Priestley, M.J., Seible, F., and Xiao, Y. (1994) “Steel jacket retrofitting of reinforced concrete bridge
columns for enhanced shear strength; test results and comparison with theory,” ACI Structural
Journal, Vol 91, pp 537-551.
Roeder, C.W., and Lehman, D.E. (2012) “Initial Investigation of Reinforced Concrete Filled Tubes
for use in Bridge Foundations,” Washington State Dept. of Transportation Report WA-RD 776.1,
Olympia, WA.
Roeder, C.W., Lehman, D.E., Stephens, M. (2014) "Concrete Filled Steel Tubes for Accelerated
Bridge Construction," Transportation Research Record No. 2406, Vol. 1, Washington, DC, pgs
49-58.
Roeder, C.W., Lehman, D.E., and Bishop, E. (2010) “Strength and Stiffness of Circular Concrete
Filled Tubes," ASCE, Journal of Structural Engineering, Vol 136, No. 12, pgs 1545-53, Reston,
VA.
Stephens, M. T., Berg, L. M, Lehman, D.E. and Roeder, C.W. (2014) “Circular concrete filled tube
bridge pier connections for accelerated bridge construction,” Proceedings: 2014 National
Accelerated Bridge Construction Conference, Miami, FL, pgs 354-363.
Stephens, Max. T., Lehman, Dawn E., and Roeder, Charles W. (2016) “Design of CFST column-to-
foundation/cap beam connections for moderate and high seismic regions,” Engineering
Structures, Vol. 122, pgs 323-337.
Stephens, Max, Berg, Lisa, Lehman, Dawn E. and Roeder, Charles W. (2016) "Seismic CFST
Column-to-Precast Cap Beam Connections for Accelerated Bridge Construction," Journal of
Structural Engineering, ASCE, Reston, VA, DOI: 10.1061?(ASCE)ST.1943-541X.0001505.
Xiao, C., Cai, S., Chen, T., and Xu, C., (2012). “Experimental study on shear capacity of circular
concrete filled steel tubes,” Steel and Composite Structures, Vol 13, No 5, pgs 437-449, Techno
Press.
Xu, C., Haixiao, L., and Chengkui, H, (2009). “Experimental study on shear resistance of self-
stressing concrete filled circular steel tubes,” Journal of Constructional Steel Research, Vol 65,
pgs 801-807, Elsevier.
HOT-DIP GALVANIZING IN STEEL AND COMPOSITE BRIDGE
CONSTRUCTION
Dieter Ungermann
TU Dortmund University, Institute of Steel Construction
Dortmund, Germany
dieter.ungermann@tu-dortmund.de
Svenja Holtkamp
TU Dortmund University, Institute of Steel Construction
Dortmund, Germany
svenja.holtkamp@tu-dortmund.de
Dennis Rademacher
ArcelorMittal Europe – Long Products
Esch-sur-Alzette, Luxemburg
dennis.rademacher@arcelormittal.com
ABSTRACT
This paper summarizes the most important results of the finished FOSTA research project P835
on the use of hot-dip galvanizing for steel and composite bridge construction, where the corrosion
protection duration and the influence of hot-dip galvanizing on the fatigue behavior of steel were
examined. Based on these results the range of application of the hot-dip galvanizing shall be
transferred to the innovative shear connector of composite dowel strips in composite bridges.
After some basic information about the composite dowel strip, the first results and planned studies
of the current FOSTA research project P1042 will be presented, which shall allow the safe
application of hot-dip galvanizing to composite dowel strips with cyclic loads. Finally, the first hot-
dip galvanized “PreCoBeam” bridge is presented, which combines the economic advantages of
hot-dip galvanizing and composite dowel strips.
INTRODUCTION
Steel components in steel and composite bridge construction are usually protected against
corrosion by organic coatings. Comprehensive research during the last years showed that hot-
dip galvanizing of steel members in bridge structures allows significant economic and ecological
advantages. It could be demonstrated that the first corrosion protection can last over a bridge’s
lifetime of 100 years, even in an environment with a high corrosivity, so that no major maintenance
measures of the corrosion protection are necessary. Several studies demonstrate that almost the
same initial costs for the first corrosion protection can be expected in comparison to the costs of
usual organic coating systems.
Bridges, however, are subject to cyclical loading and require a proof against material fatigue in
accordance with (EN 1993-1-9, 2010). Thus, an application of hot-dip galvanizing was not
possible until recently, as the rules for fatigue design are only valid on non-galvanized details and
the influence of the hot-dip galvanizing on the fatigue strength of steel components had previously
not been studied sufficiently. With the results of many fatigue tests with hot-dip galvanized
specimens in the frame of the recently finished research project P835 (Ungermann, et al., 2014),
now these economic and ecological advantages can be applied for steel and composite bridge
constructions and significant cost savings can be obtained. The results enable the design of
bridges in accordance with the regulations of (EN 1993-1-9, 2010) for frequently appearing details
in bridge construction.
Parallel to this research another innovative solution to increase the efficiency of composite bridges
has been developed. Starting with several research projects and first application in 2003,
composite dowel strips offer a very economical solution for the shear connection between steel
girder and concrete slab in composite bridge constructions. Continuous improvement and further
development finally led to a general technical approval (Allgemeine bauaufsichtliche Zulassung
Z-26.4-56, 2013) by DIBt in Germany. Many bridges containing this new technology are built
meanwhile, not only in Germany.
To combine the economic cost savings of this new technology of shear connection with the
economic and ecological advantages of hot-dip galvanizing during the lifecycle, a research project
was launched with the aim to analyze the effects of the hot-dip galvanizing on the fatigue
resistance of flame cut composite dowel strips and furthermore to enable a simplified evidence of
absence of cracks according to DASt-Guideline 022 (DASt-Richtlinie 022, 2009, revised 2016).
This paper presents the first results as well as the experiences of the first bridge project with hot-
dip galvanized composite dowel strips in Germany.
HOT-DIP GALVANIZING
CORROSION PROTECTION DURABILITY
The durability of corrosion protection of zinc coatings in the atmosphere depends on its layer
thickness as well as the corrosivity of the environment. A classification of the environmental
conditions based on the atmospheric loading is given in so-called corrosivity categories (C1 low
to C5 high) with corresponding zinc abrasion rates in the first year in (ISO 12944-2, 1998). An
annual abrasion rate for long-term behavior of the zinc coatings in the atmosphere can be
calculated according to (ISO 9224, 2012) and thereby enables to determine the durability of the
corrosion protection of a structure or the required zinc layer thickness for a desired protection
period. To assess the current corrosiveness of different atmospheric environments exposure tests
of standard samples were carried out on six bridges in Germany by the Institut für
Korrosionsschutz Dresden GmbH, beginning in May 2011 (Ungermann, et al., 2014). Based on a
comparison with test data of corrosion measurements at four of these bridges in 1983, partly
significant improvements of corrosivity due to lower atmospheric pollution were identified.
Individual local effects, e.g. caused by de-icing salt, can lead to higher removal rates and should
be taken into account. The identified, location-specific corrosivity enables to determine the
theoretically possible corrosion protection duration of hot-dip galvanized steel and composite
structures under current conditions. These studies (Ungermann, et al., 2014) showed that a
theoretically protection duration of 80 to even 100 years can be achieved with a thickness of zinc
coating > 200 μm at corrosion category C4. This is nearly equal to the target service life of a steel
or composite bridge construction. However, the exact zinc layer thickness is difficult to plan, since
it depends on several factors. Thus, the silicon content of the steel has a great influence on the
zinc layer thickness, which also increases with a longer dipping time. In bridge construction
usually a steel with a high silicon content (>0.2%) is used. In combination with large component
thicknesses and thereby necessary longer dipping times in the zinc bath (maximum dipping time
according to DASt-Guideline 022 (DASt-Richtlinie 022, 2009, revised 2016)), zinc coating
thicknesses of more than 200 μm are common in bridge construction, but should be checked prior
to the construction of the bridges by using work samples. Hence, significant cost savings during
the lifetime can be achieved, as 2 - 3 renewals would be required with an organic coating in the
same period.
FATIGUE RESISTANCE
In addition to the static design, bridges are exposed to cyclical traffic loads and therefore require
a proof against material fatigue according to the European design standards (EN 1993-1-9, 2010),
(EN 1993-2, 2010) and (EN 1994-2, 2010). For this purpose a classification of the construction
into fatigue detail categories is necessary, which are assigned to normalized fatigue strength
curves (S/N curves) in (EN 1993-1-9, 2010). Galvanized components are not covered by the
standard, though not explicitly excluded. However, since the notch types were derived from
experimental investigations on construction details generally carried out on non-galvanized test
specimens, they could not be safely applied to hot-dip galvanized components. Accordingly, the
proof against material fatigue for hot-dip galvanized constructions based on current standards
could not be provided until recently.
In order to close these gaps and enable the use of hot-dip galvanizing in bridge construction, the
research project FOSTA P835/IGF-No. 351/ZBG (Ungermann, et al., 2014) investigated the
fatigue behavior of typically used notch details. For the basic proof of the applicability of hot-dip
galvanizing in steel and composite bridge construction, comparative large and small-scale tests
of non-galvanized and hot-dip galvanized components corresponding to the most important notch
details according to EN 1993 for bridge sections with small and medium spans were carried out
at TU Dortmund University and MPA Darmstadt. The results of the small scale tests were
confirmed by corresponding and complementary large-scale tests to consider the actual fatigue
effects of the details. Thereby, an investigation of the isolated influence of the zinc coating on the
cyclic load bearing capacity of structural steel was possible, additionally to the determination of
hot-dip galvanized detail categories based on (EN 1993-1-9, 2010).
The results of comparative fatigue tests of small scale samples with the main intention to identify
the influence of hot-dip galvanizing on fatigue strength and classified in non-galvanized state into
detail category 125 according to (EN 1993-1-9, 2010), are exemplary shown in Figure 1. Both,
different steel grades as well as various cutting processes (water jet cutting, flame cutting, milling)
are included in the evaluation of the test results according to the background document of EN
1993-1-9 (Sedlacek, et al., 2007). The evaluation proofed the detail category 125 for the non-
galvanized specimens, represented as black dots, but the hot-dip galvanized samples, shown as
gray triangles, had to be classified one detail category lower at category 112.
Fig. 1. - Comparison of non-galvanized (detail category 125) to hot-dip galvanized small scale
specimens (Ungermann, et al., 2014)
In a direct comparison of non-galvanized to hot-dip galvanized samples, a reduced fatigue limit

was observed for all examined details, compare Figure 1. The average reduction was around
20% to 30%. Shrinkage cracks which are already existing in the unloaded initial state in the 1-
phase of the iron-zinc alloy layer were found responsible for the decreasing fatigue strength. As
seen in Figure 2 these micro-cracks grow into the base material under cyclic load due to stress
peaks causing premature fatigue crack initiation (Ungermann, et al., 2014).
Fig. 2. - Metallographic section of micro-cracks and fatigue cracks (Ungermann, et al., 2014)
All conducted fatigue tests, including component-like, batch galvanized test samples with
geometrical and structural imperfections as manufacturing tolerances or residual welding
stresses, were statistically evaluated according to the Commentary to Eurocode 3 (Sedlacek, et
al., 2007). The results clearly show that the detail categories defined in (EN 1993-1-9, 2010) for
non-galvanized steel components cannot be transferred directly to hot-dip galvanized steel
components. In fact, some of them must be reduced by a maximum of one category to account
for the influence of hot-dip galvanizing. Some normative detail categories could although be
confirmed, especially in already low classifications due to geometrical notching effects. But this
reduction of detail category does not inevitably lead to the need for an enlargement of the cross
sections, because the fatigue check is often not decisive for the dimensioning.
On the basis of all test results a table of detail categories for hot-dip galvanized steel
complementary to the standard of (EN 1993-1-9, 2010) have been compiled, which is shown in
Table 1. The fatigue check according to the well-known procedure can thus also be applied for
galvanized components and hence the advantages of the hot-dip galvanizing can be used for
steel and composite bridges easily.
Table 1 - Detail categories of tested hot-dip galvanized construction details (Ungermann, et al.,
2014) (Ungermann, et al., 2015)
Detail Construction detail Description *
category
Plates, flats and rolled sections with rolled or milled edges
140 NOTE: The fatigue strength curve associated with category
140 is the highest. No detail can reach a better fatigue strength
at any number of cycles
Material with machine gas cut or water jet cut edges with
112 shallow and regular drag lines
Machine gas cut with cut quality according to EN 1090
Manual longitudinal fillet weld
100
Manual longitudinal fillet weld over manual transverse butt

80 weld
Size effect for Transverse butt weld in plates and flats

100 t > 25 mm:
ks=(25/t)0,2
Size effect for Transverse butt weld in rolled sections with height of weld
t > 25 mm: convexity not greater than 10% of weld width
80 ks=(25/t)0,2
Size effect for Transverse butt weld in plates, flats, rolled sections and
80 t > 25 mm: welded plate girders with height of weld convexity not greater
than 20% of weld width
ks=(25/t)0,2
ℓ ≤ 50 mm Vertical stiffeners welded to a rolled section or plate girder

80
80 Welds under shear loading:

(m=8) Stud shear connectors for composite application
* Requirements for execution in accordance with identical detail in EN 1993-1-9
The general feasibility was proved with the first application of hot-dip galvanizing for a composite
road bridge in Germany, where the theoretical research results were subjected to an in-situ test.
This bridge was completed in summer 2016.
COMPOSITE DOWEL STRIPS

GENERAL BASICS
Composite dowel strips are very efficient composite connectors, which are used mainly for
composite bridges of small and medium spans. Composite dowel strips have been continuously
developed by various research projects over the last few years. Recently, in 2013, a general
technical approval by DIBt in Germany was issued for non-galvanized composite dowel strips with
clothoid and puzzle shape. This general technical approval Z-26.4-56 (Allgemeine
bauaufsichtliche Zulassung Z-26.4-56, 2013) regulates the scope, production and design basis
as well as the determination of load-bearing capacity and the fatigue check.
The composite dowel strip is produced by autogenous flame cutting or a method which leads to
equivalent properties in terms of load bearing capacity and fatigue resistant. Generally, hot rolled
sections are cut in the middle of its web, producing two beams with composite dowels necessary
for transferring the longitudinal shear forces. By the elimination of the upper steel flange and thus
the reduction of steel consumption in areas which contribute only slightly to stiffness and load
bearing capacity (Figure 3b, Figure 3c), composite dowel strips offer economic and product-
technical advantages compared to other connectors as shear studs respectively conventional
prefabricated composite beams with composite dowels strips welded on the upper flange (Figure
3a). Further economic applications are T-beam cross-sections with external dowel strips
functioning as external reinforcement (see Figure 3c).
Fig. 3 - Examples of application of composite beams with composite dowel strips in bridge
construction (Allgemeine bauaufsichtliche Zulassung Z-26.4-56, 2013)
The calculation of composite beams with composite dowel strips is carried out according to the
limit state design of (EN 1992), (EN 1993) and (EN 1994) supplemented by the German general
technical approval (Allgemeine bauaufsichtliche Zulassung Z-26.4-56, 2013). Under cyclic loads,
steel and concrete fatigue has to be verified as well as a rigid shear connection has to be ensured,
whereby the individual components influence each other. (Gündel, et al., 2014)
The procedure for the fatigue check of the steel strip is given in the approval (Allgemeine
bauaufsichtliche Zulassung Z-26.4-56, 2013) in accordance with the local stress concept of (EN
1993-1-9, 2010). The local stress (L) resulting from the transmission of the longitudinal shear
force of the considered dowel and the global stress (G) from the bending action of the composite
beam determine the stress amplitude at the maximum spot. These stresses are calculated by
multiplying the nominal stresses (taking account of both the local and global stress effect) with
stress concentration factors from equation (1).
V  S y  N M 
(1)   k f , L   k f ,G     zD 
I y  tw  A Iy 
 
where k f , L ,CL  7,3 ; k f ,G ,CL  1,5 (Clothoid)
k f , L , PZ  8,6 ; k f ,G , PZ  1,9 (Puzzle)
These stress concentration factors were derived from finite element calculations and verified by
strain measurements in push-out and beam tests (Feldmann, et al., 2012). Depending on the
cutting quality of the flame cut surfaces, detail category 125 or 140 according to (EN 1993-1-9,
2010) applies on the resistance side.
HOT-DIP GALVANIZING OF COMPOSITE DOWEL STRIPS

The regulations of the German general technical approval for composite dowel strips (Allgemeine
bauaufsichtliche Zulassung Z-26.4-56, 2013) do not include the use of hot-dip galvanizing for
cyclic loaded components. In principle, the fatigue check is conceivable with the table of detail
categories for hot-dip galvanized steel on the basis of (EN 1993-1-9, 2010), which has been
derived within the research project P835 (Ungermann, et al., 2014), see Table 1. However, the
specific influences from complex geometry, production-induced residual stresses and the load-
bearing behavior of the dowel strip were not taken into account in the underlying tests. In order to
reliably exploit the economic advantages of hot-dip galvanizing for composite bridges with
composite dowel strips and to provide a fatigue check by means of extended scientific
certifications, the fatigue behavior of the hot-dip galvanized composite dowel strips is investigated
within a current research project FOSTA P1042/IGF-No. 18624N (Feldmann, Kühne, Ungermann,
& Holtkamp, exp. 2019). While maintaining the design concept for fatigue resistance of the
general technical approval (Allgemeine bauaufsichtliche Zulassung Z-26.4-56, 2013) as
described in the previous section, a new detail category for flame-cut, hot-dip galvanized
composite dowel strips is to be determined for the fatigue check according to (EN 1993-1-9, 2010)
or the classification according to Table 1 is to be confirmed.
Initially, numerous extensive numerical studies at the RWTH Aachen University investigated the
stresses of the composite dowel strips occurring during the hot-dip galvanizing. For this, the
immersion of large steel girders with clothoid geometry and the most unfavorable dipping
parameters according to the limits of DASt-Guideline 022 (DASt-Richtlinie 022, 2009, revised
2016) as a dipping angle of 0° and a dipping speed of 0.8m/min (both not used in practice) was
simulated by applying a temperature of 450°C, corresponding to the usual temperature of a zinc
bath. The numerical model was validated by measurements of the temperature distribution inside
the web and measurements of strains at one dowel during the galvanizing process of four 7m
long beams with dowel shape, see (Feldmann, Kühne, Ungermann, & Holtkamp, exp. 2019). As
small scale samples allow a better and simpler experimental procedure for the planned fatigue
tests than only doing full scale component tests, the numerically determined stresses of beams
with typical bridge profiles occurring during the galvanizing process had to be transferred
subsequently to small scale samples of heavy plates. Numerous investigations have shown that
the reduction to a dowel strip of heavy plate and thus the removal of the flange and the lower web
height, which has an influence on the magnitude of the resulting stresses, can be compensated
by dipping the dowel strips upside-down. With a web height of 270mm, this procedure offers
similar strains in the critical point of the dowel’s ground as girders with unfavorable profiles for
bridge constructions.
To analyze the influence of hot-dip galvanizing on the fatigue strength of composite dowel strips
a large number of experimental investigations will be carried out at TU Dortmund University.
Initially, comparative fatigue tests of small scale samples with clothoid dowel shape and steel
grades S355 and S460 will be performed in non-galvanized finish and galvanized finish realized
by dipping upside-down. At this stage the samples are tested as dowel strips without concrete
slab. The experimental set-up is designed in such a way that the load results in stresses similar
to those from real loading. The maximum load from the "global" stress occurs at a similar position
to the maximum stress during hot-dip galvanizing. The numerical investigation of the experimental
set-up and the investigations to determine the point of failure of the small-scale specimen will be
extensively presented in the final report (Feldmann, Kühne, Ungermann, & Holtkamp, exp. 2019).
In order to take into account the interaction between the steel dowel and the concrete slab and to
investigate a possible interference or influence of the zinc layer to the composite effect, additional
cyclic push-out tests, similar to the standard test set-up according to (EN 1994-1-1, 2010),
appendix B, will be carried out. Thereby the specimens’ stresses correspond to the “local” bearing
behavior of the composite beams. For the final verification of a preliminary S/N curve, determined
by small scale sample experiments, four-point bending tests on large composite beams with
component-like dimensions and with welded composite dowel strips and halved rolled sections
(see Figure 3a, Figure 3b) will be carried out. This combines both, the global and local bearing
behavior. Thereby previously excluded effects can be taken into consideration. While maintaining
the described design concept based on the general technical approval of composite dowel strips
(Allgemeine bauaufsichtliche Zulassung Z-26.4-56, 2013), the possibly new S/N curve and the
subsequent defined detail category of the flame cut, hot-dip galvanized composite dowel strips
for the fatigue check according to (EN 1993-1-9, 2010) will permit a scientifically verified
application of hot-dip galvanizing for composite dowel strips in future composite bridge
constructions.
Additionally to the fatigue studies, the research project examines a classification of the composite
dowel strips into a detail class according to DASt-Guideline 022 “Hot-dip galvanization of load-
bearing steel structure components” (DASt-Richtlinie 022, 2009, revised 2016), which has to be
applied in Germany for batch galvanizing of structural steel components. This guideline was
developed by extensive research since 2005, as an accumulation of heavy crack-damage cases
during or after galvanizing of large supporting steel components occurred in the period from 2000
to 2007 in Germany. During the dipping process of steel components the high temperature
changes and restraint effects due to the component geometries result in potentially large transient
stresses and local plastic deformations (Pinger, 2009). Due to design and production high residual
stresses occur additionally in many bridge details such as flame cut composite dowel strips. In
unfavorable conditions, an overlapping of these stresses can lead to small micro cracks or even
to the failure of whole components. Therefore, the DASt-Guideline 022 (DASt-Richtlinie 022,
2009, revised 2016) defines requirements for material and process parameters to reduce these
effects and avoid crack initiation. A simplified procedure with classification into a construction
class, a detail class and therefrom resulting confidence zone allows determining the necessary
tests for the galvanized construction to ensure absence of cracks. Since no classification of the
composite dowel strip is possible in an existing detail class, the influence of various unfavorable
process and geometrical parameters on the stresses of the composite dowel strips occurring
during the hot-dip galvanizing are analyzed with the developed numerical models at RTWH
Aachen University. Thus, the general applicability of the hot-dip galvanizing for composite dowel
strips shall be ensured.
FIRST APPLICATION - A GALVANIZED PRECOBEAM BRIDGE IN HALLE-OSENDORF

The Elster-Bridge Osendorf (Germany) is the first “PreCoBeam” bridge with hot-dip galvanized
steel structure. This bridge combines the economic and ecological advantages of prefabricated
composite beams (PreCoBeams) with composite dowel strips with those of hot-dip galvanizing. It
therefore results in an extremely economic construction, both in production and in service life
(Seidl, Danders, Gunkel, Rademacher, & Pinger, 2017). The bridge is located in Halle-Osendorf
and transfers a rural road across the Weisse Elster to a nature reserve. The single-span frame
bridge with twin-girder cross-section and a span length of 21 m (see Figure 4) replaces an old
three-span structure that was damaged by a flood.
Fig. 4 - Cross-section with two PreCoBeams and in-situ concrete (© SSF Ingenieure AG) (Seidl,
Danders, Gunkel, Rademacher, & Pinger, 2017)
The rolled sections (HD320x300, steel grade S355ML), cut into halves with composite dowel
shape, are used as an external reinforcement of the T-beams (see Figure 3c). After the flame
cutting the edges and surfaces of the steel teeth were ground. The halved beams were
subsequently cambered by cold forming with a pitch of 1.08 m, which could still be measured
without changing after galvanization at 450°C. The construction height is 0.70 m in mid-span and
1.40 m at the abutments of the bridge, resulting in a slenderness of l/30 respectively l/15.
For transport reasons, the approximately 20.4 m long halved beams were divided into 2 beams
each. With total lengths of approximately 10.2 m they complied with the maximum zinc bath
lengths and could be galvanized without problems. The galvanizing was carried out according to
DASt-Guideline 022 (DASt-Richtlinie 022, 2009, revised 2016) close to the construction site in
Landsberg/Halle. The necessary length of immersion time was derived from preliminary tests, so
that finally mean layer thicknesses of at least 350 μm could be measured on the flange bottoms
of the four rolled profiles (see Figure 5). These thicknesses ensure a theoretical corrosion
protection period of >100 years, see (Ungermann, et al., 2014). The cut surfaces of the dowel
geometry gained significantly lower layer thicknesses. As they are finally completely enclosed in
concrete, this has no adverse effects in terms of durability (corrosion), but has a positive effect on
the fatigue behavior.
Fig. 5 - Cooling of galvanized beams (Seidl, Danders, Gunkel, Rademacher, & Pinger, 2017)
After hot-dip galvanizing each two 10.2 m long pieces had to be connected again to one beam by
welding. To avoid high efforts to remove zinc around the intended site joints, the areas close to
the weld preparation were protected by a special lacquer before galvanizing. Thus zinc deposit
could be prevented. According to the procedure (Ungermann, et al., 2014), which was developed
for site joints of galvanized bridge components, at first residues of the lacquer at the beam ends
were removed, subsequently the joints were welded and the weld seams ground flush. The areas
around the joint were blasted, spray-galvanized and sealed.
After pouring the concrete of the PreCoBeam into the prepared formwork and hardening of the
concrete, the prefabricated beams with external reinforcement were lifted in place. Finally the
deck slab was completed by in-situ concrete (Figure 6).
Fig. 6 - PreCoBeam in final position before adding in-situ concrete for the slab (© SSF
Ingenieure AG) (Seidl, Danders, Gunkel, Rademacher, & Pinger, 2017)
In addition to the very long, maintenance-free service life of the steel components due to the
galvanizing, the high robustness of the zinc coating enables to resist the usual mechanical
stresses occurring during transport and installation without damage. Thus, the construction
process does not have to be interrupted or delayed for necessary repair work on the corrosion
protection system. The use of hot-dip galvanized steel components therefore supports the
essential advantages of the composite construction with regard to short construction times.
SUMMARY
The finished research project P835 (Ungermann, et al., 2014) revealed that hot-dip galvanizing
has great advantages with regards to life cycle costs and durability of steel and composite bridges
and can easily be applied with moderate modifications, amongst others due to the reduction of
fatigue strength for hot-dip galvanized details. Their cause was detected in micro-cracks in the
zinc layer emerging during the hot-dip galvanizing and growing to the base material under cyclic
loading.
Furthermore, several investigations in recent years have contributed to an innovative shear
connector, the composite dowel strip, which is regulated in a general technical approval in
Germany (Allgemeine bauaufsichtliche Zulassung Z-26.4-56, 2013).
To combine and take advantage of both developments by enabling the safe and scientifically
proven application of hot-dip galvanizing for composite dowel strips, the presented research is
carried out in the current FOSTA research project P1042 (Feldmann, Kühne, Ungermann, &
Holtkamp, exp. 2019), with particular consideration of the fatigue behavior. The fatigue resistance
will be examined by small-scale tests of not embedded dowel strips and push-out tests.
Experiments on large composite beams shall finally verify the new derived S/N curve of flame cut,
hot-dip galvanized composite dowels for the fatigue check according to (EN 1993-1-9, 2010).
Furthermore, numerous numerical investigations examine the stresses on the composite dowel
strip occurring during the hot-dip galvanizing process, in order to determine the geometry of the
small scale samples and, on the other hand, to define a detailed class according to DASt-
Guideline 022 (DASt-Richtlinie 022, 2009, revised 2016) by means of parameter studies.
The first-time use of hot-dip galvanizing at a prefabricated composite bridge with dowel strips
(PreCoBeam) has shown that the application is basically possible and can be realized without
any problems, considering certain design and process parameters.
NOTE
This report was written for Composite Construction CVIII in 2017. The research project P1042
has now been completed. The final research results can be found in the final report, which is
expected to be published in 2019.
ACKNOWLEDGEMENTS
Both the research project P835 (IGF-No. 351/ZBG) and the research project P1042 (IGF-No.
18624N) were carried out with the financial support of the Arbeitsgemeinschaft industrieller
Forschungsvereinigung „Otto von Guericke“ e.V. (AiF) Cologne, Germany and funding from the
German Federal Ministry for Economy. Many thanks to these supporting committees. Also many
thanks to Forschungsvereinigung Stahlanwendung e.V. (FOSTA) and Gemeinschaftsausschuss
Verzinken e.V. (GAV) for their supervision and support. Furthermore, many thanks to the project
partners, the involved industry companies and the project support committee of both projects for
their support.
The first bridge with hot-dip galvanized precast composite beams with dowel strips (PreCoBeam)
was initiated by SSF Ingenieure AG and the municipality of Halle (Germany). The realization of
the hot-dip galvanized steel construction was carried out by ArcelorMittal Europe - Long Products
and Voigt & Schweitzer on behalf of OST Bau - Osterburger Strasse-, Tief- und Hochbau GmbH.
REFERENCES
Allgemeine bauaufsichtliche Zulassung Z-26.4-56. (2013). Verbunddübelleisten (coposite dowel
strips). Berlin: Deutsches Institut für Bautechnik (DIBt).
DASt-Richtlinie 022. (2009, revised 2016). Guideline for hot-dip-zinc-coating of prefabricated
loadbearing steel components. Deutscher Ausschuß für Stahlbau DASt.
EN 1992. (n.d.). Eurocode 2: Design of concrete structures.
EN 1993. (kein Datum). Eurocode 3: Design of steel structures.
EN 1993-1-9. (2010). Eurocode 3: Design of steel structures - Part 1-9: Fatigue. Brussels:
European Committee for Standardization (CEN).
EN 1993-2. (2010). Eurocode 3: Design of steel structures - Part 2: Steel bridges. Brussels:
European Committee for Standardization (CEN).
EN 1994. (kein Datum). Eurocode 4: Design of composite steel and concrete structures.
EN 1994-1-1. (2010). Eurocode 4: Design of composite steel and concrete structures - Part 1-1:
General rules and rules for buildings. Brussels: European Committee for Standardization
(CEN).
EN 1994-2. (2010). Eurocode 4: Design of composite steel and concrete structures – Part 2:
General rules and rules for bridges. Brussels: European Committee for Standardization
(CEN).
Feldmann, M., Gündel, M., Kopp, M., Hegger, J., Gallwoszus, J., Heinemeyer, S., . . . Hoyer, O.
(2012). Neue Systeme für Stahlverbundbrücken - Verbundfertigteilträger aus hochfesten
Werkstoffen und innovativen Verbundmitteln, research report FOSTA P804. Düsseldorf.
Feldmann, M., Kühne, R., Ungermann, D., & Holtkamp, S. (exp. 2019). Ermüdungsfestigkeit
feuerverzinkter Verbunddübelleisten im Verbundbrückenbau, IGF-No. 18624N, FOSTA
P1042.
Gündel, M., Kopp, M., Feldmann, M., Gallwoszus, J., Hegger, J., & Seidl, G. (2014). Die
Bemessung von Verbunddübelleisten nach neuer Allgemeiner bauaufsichtlicher
Zulassung. Stahlbau 83(2), 112-121.
ISO 12944-2. (1998). Paints and varnishes - Corrosion protection of steel structures by protective
paint systems - Part 2: Classification of environments. Beuth Verlag.
ISO 9224. (2012). Corrosion of metals and alloys – Corrosivity of atmospheres – Guiding values
for the corrosivity categories. Berlin: DIN Deutsches Institut für Normung e. V.
Pinger, T. (2009). Zur Vermeidung der Rissbildung an Stahlkonstruktionen beim Feuerverzinken
unter besonderer Berücksichtigung der flüssigmetallinduzierten Spannungsrisskorrosion,
dissertation. RWTH Aachen University.
Sedlacek, G., Hobbacher, A., Nussbaumer, A., Müller, C., Stötzel, J., & Schäfer, D. (2007).
Commentary to Eurocode 3 - EN 1993-Part 1-9 - Fatigue. unpublished: RWTH Aachen
University.
Seidl, G., Danders, A., Gunkel, F., Rademacher, D., & Pinger, T. (2017). Elster bridge Osendorf
– a hot-dip galvanized composite bridge with external reinforcement. Stahlbau 86(2).
Ungermann, D., Rademacher, D., Oechsner, M., Landgrebe, P., Simonsen, F., Friedrich, S., &
Lebelt, P. (2014). Design of hot-dip galvanized bridges, Proceedings of 7th European
conference on steel and composite structures. Eurosteel 2014, Napoli, 741-742.
Ungermann, D., Rademacher, D., Oechsner, M., Landgrebe, R., Adelmann, J., Simonsen, F., . .
. Lebelt, P. (2014). Feuerverzinken im Stahl- und Verbundbrückenbau, IGF-No. 351/ZBG,
research report FOSTA P835. Düsseldorf.
Ungermann, D., Rademacher, D., Oechsner, M., Simonsen, F., Friedrich, S., & Lebelt, P. (2015).
Feuerverzinken im Brückenbau – Teil 1: Zum Einsatz feuerverzinkten Baustahls bei
zyklisch beanspruchten Konstruktionen. Stahlbau 84(1), pp. 2-9.
Operational testing of composite railway bridges with innovative
composite dowels
Daniel Pak
Universität Siegen, Chair of Steel and Steel-Composite Construction
Siegen, Germany
pak@bau.uni-siegen.de
Maik Kopp
RWTH Aachen University, Division 10.2 - Construction Management
Aachen, Germany
maik.kopp@zhv.rwth-aachen.de
Günter Seidl
Fachhochschule Potsdam, Chair of Steel and Steel-Composite Construction
Potsdam, Germany
seidl@fh-potsdam.de
ABSTRACT
Composite dowels are increasingly used as shear connectors, especially for prefabricated
composite bridges. Advantages compared to headed studs are in particular an increased
strength, a sufficient deformation capacity even in high strength concrete and a simple
application in steel sections without upper flange. In Germany composite dowels are
designed according to a national technical approval available for any design office and
construction company. In 2012, the VFT Rail® system, adopting these composite dowels,
was approved in accordance with CEN and railway standards by the German Federal
Railway Office (EBA) for operational testing. Due to positive experiences gained during
operation, two additional bridges were designed by SSF Ingenieure AG and brought into
operation in 2014 by the German railway Deutsche Bahn (DB). To survey those bridges, a
monitoring campaign was set up, investigating (among others) the stress distribution
between concrete and composite dowels. The results are presented in this paper.
1. Composite dowels with clothoid shape as shear connectors for railway bridges
Composite dowels are shear connectors for composite beams, which consist of openings in
steel plates, cast with concrete. They are either made of steel plates welded on the upper
flange of steel beams or fabricated directly out of the web of steel beams. Up to 2013, they
were used for road and railway bridges in Germany with approvals in the individual case. In
2013, the national technical approval Z-26.4-56 (Deutsches Institut für Bautechnik DIBt,
2013) was issued, allowing for a design of composite dowels in the regular case. In parallel,
the prefabricated composite bridge type VFT Rail® with composite dowels in clothoid shape
was approved by the German Federal Railway Office (EBA) for operational testing, based on
the same design principles. The first composite bridge structure using external reinforcement
was put into operation by the German Railway Company DB Netz AG on the railway link
Bingen-Saarbrücken, crossing the river Simmerbach (Seidl, Mensinger, Koch & Hugle,
2012). Two years later, two additional railway bridges with the same cross section were
designed by SSF Ingenieure AG and brought into operation in Upper Bavaria (Germany).
2. VFT railway bridges
2.1. Design requirements
In the case of railway bridge replacements, the clearance below the bridge is often widened
and the span increased. Therefore, the construction height of the new superstructure is of
central importance. To address this problem, new bridge types have been developed within
the framework of a project initiated by Germany’s railway infrastructure provider DB Netz
AG: these bridges are designed to meet the requirements for replacement in existing
networks with regard to a slender design, manufacturing process, traffic flow and
maintenance. As a result, two bridge types with external reinforcement (composite dowels),
either with direct rail fastening or with a ballast bed, were developed by SSF Ingenieure AG
(Germany) (Seidl, Mensinger, Koch & Hugle, 2012), whereas the one with direct rail
fastening is discussed in this paper.
2.2. Design principles and construction

The superstructure of the VFT railway bridge (VFT Rail®) with direct rail fastening is
designed as a simply supported composite beam. Composite action between concrete and
steel girders (which act as external reinforcement) is realized by composite dowels
(geometry MCL250/115 (Seidl, Mensinger, Koch & Hugle, 2012), Figure 1, right). Shear
reinforcement is placed in cut-outs between the steel dowels according to (Deutsches Institut
für Bautechnik DIBt, 2013).
Fig. 1 – transverse cross section (left) VFT-Rail® with rail support points directly fastened to
the superstructure and composite dowel and longitudinal cross section (right, depicting the
clothoidal shape of the dowels)
Each single rail is taken by a recessed concrete channel to increase the effective height of
the cross section (Figure 1, left). The steel-reinforced track channels are furthermore acting
as derailment guard. As a result, the concrete pressure zone is reduced by 40%; the pure
concrete cross-section is no longer able to take the compressive stresses even if high-
strength concrete is used. Therefore the compression zone is strengthened by external
reinforcement as well, which furthermore results in an increase of stiffness of the
superstructure. The superstructure is completely prefabricated; the prefabricated
accompanying paths are subsequently fixed to the structure to reduce the installation weight.
Due to prefabrication, the quality of the structure is high. Quality deficiencies can (if
necessary) be remedied in the factory already and are therefore decoupled from the time
pressure of the track possession period (Seidl, Mensinger, Koch & Hugle, 2012).
Due to concrete shrinkage, state II (cracked concrete) builds up over the concrete’s cross
section. Therefore, for design purposes shear forces are assigned to the concrete and
bending forces to the structural steel.
Fig. 2 – Structure of adjustable rail support point system Krupp ECF
The elastic, adjustable rail support system Krupp ECF (ThyssenKrupp GfT Gleistechnik,
2006) is used to fasten the rails directly to the composite girder (Figure 1, left and Figure 2).
Each rail support point is anchored by means of two bolts that are braced to pressure
sleeves embedded in the precast concrete. Lining plates are used to compensate vertical
tolerances.
Fig. 3 – Pattern of rail fasteners at the end of superstructure
The rail support points are placed in a regular pattern of 60 cm. Next to the transition
between superstructure and backfilling, an additional point is placed to reduce the
compressive forces in this region (Figure 3). The regular distance of 60 cm is continued on
the ballast bed in front of the bridge. To reduce differential settlements in that region, wide
B90 concrete sleepers are placed in front and behind the bridge structure. The rail support
points are adjustable in position and height to compensate for superstructure tolerances.
A possible maintenance replacement of the first four rail support points needs to be ensured
according to the approval for operational testing. Therefore, the first four rail support points
are fixed by means of stainless steel bolts, fixed to the superstructure. The type of fixation
depends on the specific type of rail support fixation. One solution requires the steel bolts to
be screwed into nuts that are fixed at the end of pressure sleeves, embedded in the
superstructure. Another solution calls for screwing threaded rods into friction welded
stainless steel sleeves, anchored in the VFT concrete girder.
Fig. 4 – VFT Rail® bridges constructed in Germany (left: crossing the river Simmerbach,
middle / right: Upper Bavaria)
The first VFT Rail® bridge was built in 2012 in the course of railway track 3511 between
Bingen and Saarbrücken (Germany) (Figure 4, left) (Feldmann et al., 2012). It replaced two
existing steel troughs with ballast bed, which had reached their live span. After successful
operation of this bridge, the approval for an extended operational testing of the VFT-Rail®
system was granted by the German Federal Railway Office (EBA) in 2013
(Eisenbahnbundesamt, 2013). Three years later, two additional bridges of the same type
came into operation in Upper Bavaria (Germany) (Figure 4, middle and right). The start of
operation of all three bridges was accompanied by measurement campaigns, conducted by
RWTH Aachen University; the results of the strain measurements performed at the steel
dowels of the two Bavarian bridges are discussed within the scope of this paper. The
measured values are compared to values calculated according to the general technical
approval Z-26.4-56, which is summarized shortly in the next section.
3. Design of composite dowels according to German general technical approval

Within the scope of the national technical approval Z-26.4-56 (Deutsches Institut für
Bautechnik DIBt, 2013), the design of composite dowels in clothoid (CL) and puzzle (PZ)
shape to be used under sagging and hogging moment is regulated (Kathage et al., 2016).
The geometry can be scaled in dependence of the distance of the openings between 150
and 500mm with a plate thickness between 6 and 40mm. Structural steel grades S235, S355
and S460 are covered. The steel dowels have to be fabricated by gas cutting or a cutting-
process which is similar in terms of strength and fatigue. For fatigue loading, the quality has
to meet the requirements for detail category 125 or detail category 140, respectively,
according to EN 1993-1-9.
The verification of the longitudinal shear connection has to be proofed in accordance with
EN 1994-1-1, whereas an equal spacing is also allowed for steel sections without upper
flange, taking into account additional design constraints.
Three possible failure modes under static loading are governed by the technical approval.
Particularly for small openings and large steel plate thicknesses, the dominating failure mode
is double shearing of the concrete dowel. For composite members with low distances
between concrete dowel and concrete surface, the ductile failure mode “concrete pry-out”
governs, which is similar to the concrete pry-out of anchors subjected to shear forces. For
small plate thicknesses and low steel strength, ductile steel failure can occur. In beam-type
sections with thin webs and composite dowels as external reinforcement, another non-
ductile failure mode can occur. In that case splitting tensile forces can exceed the concrete
tensile stress, which results in a horizontal crack at the height of the composite dowel.
The fatigue design concept comprises steel fatigue, concrete fatigue and securing of a rigid
shear joint.
Regarding concrete fatigue design, it is distinguished between the loss of bearing capacity
due to trickling of crushed concrete out of the composite joint and cyclic concrete pry-out of
composite dowels with insufficient concrete cover subjected to high upper loads. To
guarantee for a rigid shear connection between the steel and the concrete section, the strain
shift between steel and concrete section is used as criterion for a relevant degradation of the
composite joint.
The steel fatigue design is based on the geometric stress approach. On the resistance side
the fatigue strength is described by the fatigue strength curve of detail category 125
(machine gas cut edges having shallow and regular draglines) or of detail category 140
(machine gas cut edges with subsequent dressing) in accordance with EN 1993-1-9.
Fig. 5 – Calculated stress distribution (schematic, left: “L”ocal loading due to shear force
acting at steel dowel, right: “G”lobal loading due to axial deformation of the external
reinforcement due to bending of the composite section)
On the load side, the geometric stress amplitude is calculated. Therefore, it has to be
distinguished between two possible maximum stress distributions. On the one hand, a stress
maximum occurs in the upper region of the clothoid due to (“L”ocal) shear forces acting at
the steel dowel (Figure 5, left). On the other hand, a stress maximum occurs close to the
base of the dowel, caused by (“G”lobal) axial deformation of the external reinforcement due
to global bending (Figure 5, right).
However, a simple prediction of the actual position of the relevant hot-spot and the resulting
geometric stress amplitude is not possible, as both effects interfere, based on the position of
the steel dowel as well as the position of the loading. Therefore, the geometric stress
amplitude is calculated conservatively as the sum of those stresses, even though they do not
occur at the same position (Feldmann et al., 2014) (Mensinger, 2010).
The geometric stresses are not calculated by the designer directly; in fact they are
determined by an amplification of nominal stresses, defined as longitudinal shear stresses
(“L”ocal) and normal stresses (“G”lobal) at the dowel base (eq. 1).
Δ𝑉∙𝑆y Δ𝑁 Δ𝑀
Δσ = |𝑘f,L ∙ 𝐼y ∙𝑡w
| + |𝑘f,G ∙ ( 𝐴 + 𝐼y
∙ 𝑧D )| (1)
The stress amplification factors 𝑘f,L and 𝑘f,G were calculated and verified experimentally for
both dowel shapes using different numerical models:
 for clothoidal shape “CL” of the steel dowel: 𝑘f,L,CL = 7.3 and 𝑘f,G,CL = 1.5
 for puzzle “PZ” shape of the steel dowel: 𝑘f,L,PZ = 8.6 and 𝑘f,G,PZ = 1.9
On the safe side for the determination of stresses due to global bearing behavior the
influence of tension stiffening can be neglected (cracked section). For stresses due to
longitudinal shear transfer, the more unfavorable value from calculations with a cracked and
an un-cracked section should be used.
To exclude low cycle fatigue the geometric stress amplitude is limited to 2∙fy and the upper
geometric stress is limited to 1.3∙fy.
In compliance with the approval for operational testing, the rail operator made the long-term
measurement of these strains in the steel girders of the three VFT Rail® bridges conditional.
The measurements allowed for a confirmation of considerations underlying the design of the
composite dowel and for a confirmation of correct application of rules based on stress
concentration factors respectively. In the following, the monitoring concept for the bridges
constructed in Upper Bavaria is described, followed by a discussion of measurement results.
4. Monitoring concept
For measurement of strains in the composite dowels of all bridges (Figure 1, right), strain
gauges were applied and connected by wire during prefabrication of the composite elements
in the workshop.
The measurements were performed as short-term measurements, allowing for a high-
frequency (1000 Hz) recording of short-term loading caused by single train crossings. These
temporary measurements were performed three times during a period of half a year.
Fig. 6 – “Stop-run” of single locomotive
For the first measurements a single locomotive of known weight (load per axle = 216 kN)
came to stop every single meter on the VFT-Rail® girder for at least one minute (Figure 6).
The data recorded during this “stop-run” was used to check precisely the monitoring analysis
tools as well as the correct implementation of data into the evaluation routines. All following
measurements were performed during regular train crossings.
5. Local stresses in external reinforcement

5.1. Determination of design values
The internal forces needed for the determination of design stresses were determined by
means of a framework model. The section stiffness was calculated according to elasticity
theory for state I (uncracked concrete, Figure 7, left) as well as for state II (cracked concrete
in tension zone, Figure 7, right). In the case of state II, tension stiffening was disregarded
(cracked concrete not contributing to stiffness, Young’s modulus of cracked concrete set to
Ec = 0). In the region of compression, concrete action was considered with non-reduced
Young’s modulus. For the determination of the tension zone’s height, neutral axis zi,total =
37.9cm from state I was considered without iteration. In both cases, next to external
reinforcement, conventional internal reinforcement (32 x Ø32) was taken into consideration
as well. Especially in state II, the influence of this so-called redundancy reinforcement on the
total stiffness of the cross section could not be disregarded.
Fig. 7 – Cross section for state I (left) and state II (right), rail support point drawn in for
illustration purposes only
Internal bending moment M as well as internal shear force V were determined in the
monitored sections for different load positions. Therefore, the loading of a single axle was
applied step-wise to the framework model with a step-size of 0.1m. Based on these internal
forces, nominal normal stresses were calculated at the inner side of the upper and lower
flanges, at the web as well as at the base of the steel dowel (close to the stress hot-spot).
The stress calculation in the flanges as well as in the web was based on the global bending
moment only. An amplification due to local stress concentration had not to be considered in
this region; the stress amplification factor 𝑘f,G was set to 1.0. The calculation of stresses in
the region of the steel dowel was performed in compliance with the procedure given in the
general technical approval (Deutsches Institut für Bautechnik DIBt, 2013) as described in
chapter 3, based on the global bending moment as well as on internal shear forces. The
stress amplification due to structural effects as well as local bending of the steel dowel
(caused by internal shear forces) was considered by means of the stress amplification
factors 𝑘f,G,CL and 𝑘f,L,CL (eq. 1).
5.2. Field measurements

5.2.1. Test set-up and measurement
The strain gauges were applied at one pair of outer external reinforcement (lower steel
girder G7 and associated upper steel girder G3) of the VFT-Rail® girder (Figure 8).
Fig. 8 – Measurement sections at superstructure of the VFT rail bridge
In axial direction, two measurement sections were defined. On the one hand, the position of
the section close to the bearings (section “support”) had to be chosen in a way to allow for
measurement of high shear forces in the composite section. On the other hand, local effects
due to the load transfer into the bearings had to be minimized. Therefore, the section
“support” was defined at a distance of about 0.875m to the bearing axis around the 5th and
6th steel dowel. The section “field” was defined in mid-field around steel dowel 23 and steel
dowel 24, as in this section the highest bending moments and deformations occurred (Figure
8).
Fig. 9 – Position of strain gauges
In the two measurement sections (Figure 8) strain gauges were applied at the flanges and
the web (Figure 9). The strain gauges placed at the flanges and at the web allowed for the
determination of the linear strain distribution over the cross section. Additional strain gauges
were applied close to the hot-spots (Figure 5, Figure 9). Based on experiences gained in
former research projects (Feldmann et al., 2014) (Feldmann, Hegger, Hechler & Rauscher,
2007) (Heinemeyer, 2011) the position of these strain gauges was chosen at an angle of 24°
and 41° to measure the maximum strains occurring in that region.
Fig. 10 – Strain gauges in hot-spot region (steel dowel 6) (left), protected strain gauges,
covered with 2-component adhesive (PS-adhesive) and permanent elastic cement (AK22)
However, it was not possible to measure the maximum strain at the actual hot-spot directly,
as it always occurs directly at the cutting line. Therefore, the strain gauges were applied in a
distance of 5mm to the cutting line, which still covers the predicted hot-spot region (Figure
10, left). As the stress concentration factors as given in the general technical approval do not
reflect exactly the stresses measured there, they were adopted to the actual measurement
position.
The strain gauges were applied prior to casting in the workshop. In total 32 conventional
steel strain gauges (R = 100 Ohm) were applied to the steel girders of the bridge (Figure 9).
As all strain gauges were covered by concrete, special care needed to be taken prior to
casting regarding the protection of the gauges. Therefore, all strain gauges were covered by
a 2-component adhesive (PS-adhesive) after application of the signal wires. After hardening,
this adhesive formed a stiff protective cover. Finally, a layer of permanent elastic cement
(AK22) was applied, which minimized the risk of the gauges to be pulled off due to concrete
pressure (Figure 10, right). Furthermore, all signal wires were placed inside elastic ducts,
placed in the radius of the steel profiles and leading to one of the abutments of the bridge
structure (axis 20, Figure 8).
5.2.2. Measurement results
Within the scope of this paper, selected results of the strain measurements of the second
bridge constructed in Upper Bavaria (Germany) are discussed. Therefore, the strains
measured are transformed into stresses. At first, results are presented gained by
measurement of strains at flanges, web and within the hotspot region during the stop-runs of
the diesel electric locomotive. These results are compared with the theoretic characteristic
values determined according to the national technical approval (whereas the stress
concentration factors are adopted to the actual measurement position). Afterwards the strain
values recorded during regular train crossings are checked to identify time-dependant
effects.
Undisturbed region: flange of lower external reinforcement (stop-run)

In Figure 11 the strains on the inner side of the flange (S1, S8, S9, S16, see Figure 9 for
exact location) are compared to the theoretic characteristic values. The stresses are plotted
over position of the first axis of the train on the bridge. The single measurement sections
(section “field”: green and blue, section “support”: yellow and orange) can be clearly
identified. The highest strains in each region are measured when the axis is placed at the
position of the particular section, which complies with the bending moment influence line.
Fig. 11 – Measured strains / stresses on the inner side of the flange (lower external steel
girder, points) compared to calculated stresses for state I (dashed line) and state II (solid
line)
A comparison of the measured values to the theoretical values of state I (uncracked

concrete, dashed lines) shows, that the measured strains / stresses exceed the
characteristic design values slightly in mid-field (Figure 11). This leads to the conclusion that
cracking of the concrete occurred in the section’s tension zone, redistributing the stresses to
the structural steel. However, all measured values are situated clearly below the
characteristic design values for state II (cracked concrete, solid lines, Figure 11).
Undisturbed region: web of lower external reinforcement (stop-run)

The strains measured in the web (S2, S7, S10, S15, see Figure 9 for exact location) show
the same behaviour during train crossing as the strains measured at the flange (Figure 12).
The stresses in the web are by 35% lower than the corresponding stresses in the flange,
which can be attributed (amongst others) to the smaller lever arm to the neutral axis. All
values are increased compared to the theoretic values of state I (uncracked concrete), which
again shows that state II (cracked concrete) governs and needs to be considered during
design.
Fig. 12 – Measured strains / stresses at web of lower external reinforcement, compared to
calculated values for state I (dashed line) and state II (solid line)
Hot-spot region: steel dowel of lower external reinforcement (stop-run)

Stresses due to global bending of the composite girder and stresses caused by local
bending of the steel dowel superimpose within the region of the so-called hot-spot at the foot
of the dowel. Furthermore, the stresses are increased due to geometric notch effects. In mid-
field, global bending still governs the stress state (Figure 13), whereas strain gauges S12
and S13 show distinctively higher values as strain gauges S11 and S14. On the one hand,
this can be attributed to the slightly larger lever arm. On the other hand, this confirms the
prediction that the stress state is governed by global effects at mid-field, yielding an increase
of stresses at the position of the lower strain gauge (measuring hot-spot strains / stresses
due to global effects, compare Figure 5).
Fig. 13 – Measured strains / stresses within the region of the hot-spot, lower external
reinforcement, compared to calculated values for state I (dashed line) and state II (solid line)
The characteristic design values exceed the measured values, as for design the stress sums
of global and local stress are taken into account on the safe hand side. However, neglecting
the local effects (bending of the steel dowel due to shear forces) leads to an underestimation
of the stresses measured in the hot-spot region, which shows the necessity of stress
superposition. It is further recalled that for design, the hot spots not only at midfield and at
the bearings have to be investigated to find the critical cross section, but all other hot spots
in between as well.
Time dependency, stresses in flange of lower external reinforcement (regular traffic)

To check the development of the measured values over time, the crossing of scheduled
trains is recorded as well. The first measurement of strains caused by the diesel electric
locomotive during normal operations (Figure 14) shows a very good agreement with the
values obtained during the stop-run (Figure 11). The gliding mean value of all strains /
stresses measured is situated close the curve of state I (uncracked concrete), which
corresponds to the results discussed before. The measurement performed six months later
(Figure 15) does not show a distinctive change of strain / stress magnitude.
Fig. 14 – Stresses in lower flange, 1st measurement (03/11/2014), diesel electric locomotive
and bulk freight wagons (empty)
Fig. 15 – Stresses in lower flange, 3rd measurement (08/05/2015), diesel electric locomotive
and bulk freight wagons (empty)
It can be concluded that there is a good correlation between calculated characteristic

stresses and measured strains / stresses. Together with the evaluation of results in the
region of the hot-spot, this leads to the conclusion that the given design concept and safe-
sided design rules based on stress concentration factors (Deutsches Institut für Bautechnik
DIBt, 2013) is applied correctly.
6. Conclusions and outlook

Since 2013, the national technical approval Z-26.4-56 regulates the design and application
of such composite dowels in Germany, reflecting the consistent continuation of the
standardization development for composite dowels. However, the validity of that approval
expires in May 2023. For the future, provisions are made to include these design rules into
EC4. The successful realization of the VFT Rail® bridges in Simmerbach (Germany)
(Feldmann et al., 2012) and Bavaria (Germany), combined with the confirmation of the
applicability of state of the art calculation methods by monitoring measures proved the
applicability, reliability and readiness for the market of the new bridge type. Furthermore the
results again show that the given design concept and design rules based on stress
concentration factors are significantly safe-sided. For inclusion of these design rules into
EC4, they might be slightly modified to allow for an even more economic design.
REFERENCES
Deutsches Institut für Bautechnik DIBt. (2013). Allgemeine Bauaufsichtliche Zulassung,
Verbunddübelleisten, Z-26.4-56. Berlin.
Eisenbahnbundesamt. (2013). Erweiterung der Zulassung zur Betriebserprobung vom
06.12.2010 der Bauart VFT-Rail der SSF Ingenieure AG zur Verwendung in Strecken
der Eisenbahn des Bundes mit zulässigen Geschwindigkeiten bis 160 km/h. Bonn.
Feldmann, M., Hegger, J., Gündel, M., Kopp, M., Gallwoszus, J., Heinemeyer, S. & Hoyer,
O. (2014). P804 - Neue Systeme für Stahlverbundbrücken – Verbundfertigteilträger
aus hochfesten Werkstoffen und innovativen Verbundmitteln. VERBUND-DÜBEL-
LEISTEN. Düsseldorf: FOSTA - Forschungsvereinigung Stahlanwendung e. V.
Feldmann, M., Hegger, J., Hechler, O. & Rauscher, S. (2007). P621 - Untersuchungen zum
Trag- und Verformungsverhalten von Verbundmitteln unter ruhender und
nichtruhender Belastung bei Verwendung hochfester Werkstoffe,. Düsseldorf:
FOSTA - Forschungsvereinigung Stahlanwendung e. V.
Feldmann, M., Pak, D., Kopp, M., Schillo, N., Wirth, T., Mensinger, M. & Koch, E. (2012).
Eisenbahnüberführung Simmerbach. Stahlbau, 81, pp. 737-747.
doi:10.1002/stab.201201610
Heinemeyer, S. (2011). Zum Trag- und Verformungsverhalten von Verbundträgern mit
Puzzleleisten und ultrahochfestem Beton (doctoral dissertation). Aachen, Germany:
RWTH Aachen University.
Kathage, K., Feldmann, M., Kopp, M., Pak, D., Gündel, M., Hegger, J. & Gallwoszus, J.
(2016). Composite Dowels as Shear Connectors for Composite Beams—Background
to a New German Technical Approval. Composite Construction in Steel and Concrete
VII, (pp. 593-606).
Mensinger, M. (2010). Gutachterliche Stellungnahme G 1008 Low Cost Bridge: Beurteilung
der Konzepte zum Nachweis der statischen Tragfähigkeit und der
Ermüdungssicherheit der Verdübelung der externen Bewehrung des neuartigen
Brückentyps "Low Cost Bridge". München.
Seidl, G., Mensinger, M., Koch, E. & Hugle, F. (2012). Projektbericht Eisenbahnüberführung
Simmerbach - Pilotbrücke in VFT-Rail Bauweise mit externer Bewehrung. Stahlbau,
81, pp. 100-107.
ThyssenKrupp GfT Gleistechnik. (2006). Oberbauhandbuch. Bochum.
Integral vs. Seismic Control in Composite Deck
Bridges
Uwe E. Dorka
University of Kassel
Kassel, Germany
uwe.dorka@uni-kassel.de
Andrea Klarendic
University of Kassel
Kassel, Germany
andrea.klarendic@uni-kassel.de
ABSTRACT
This paper presents a study that compares an integral design for a typical elevated road
consisting of a composite deck and rc piers, with a Hyde System design, where force limits
can be controlled during an earthquake (seismic control concept). There is no question that an
integral bridge can be designed safely for modern code requirements. But the definition of
robustness requires a minimal functionality after an extreme event and a speedy recovery,
especially for elevated roads and railroads. During such events, design ground motions are
exceeded and additional hazards are triggered like tsunami, landslides and fires.
Focusing on an extreme ground motion, this study compares the remaining functionality,
reparability and recovery time to full functionality for both designs of the elevated road
mentioned before. The 2011 Mw 9.0 Tohoku scenario is used. It is the current basis for
seismic risk studies in the Osaka Bay area.
Keywords: seismic control, integral bridges, Hyde System, emergency functionality, Tohoku
event
1. INTRODUCTION
More and more integral bridges are being built today, whose evolution dates back to late
1930s. The first known integral bridge in the United States is the Teens Run Bridge in Ohio,
which was built in 1938. In the 1960s they started to spread around the world. They are
either reinforced or pre-stressed concrete bridges, like the Naibekoshinai River Bridge in
Japan built in 1996, or a composite deck bridges made of concrete slabs and steel
girders, like the A27 Brochampton Road Bridge in the United Kingdom built in 2001, which
is similar to the bridge studied in this paper. The world's longest known integral bridge is
the FAS 50 over Happy Hollow Creek built in 1997 with a total length of 358 m (Burke Jr,
2009). These are all integral bridges where a solid connection is provided between piers
and superstructure and abutments and superstructure, replacing the usual bearings.
On the other hand, bridges with seismic control require a "seismic link" where forces or
force limits can be controlled during an earthquake (Dorka, 2004). In the NATO Science-
for-Peace project "Seismic upgrading of Bridges in South-East Europe by Innovative
Technologies" (ISUbridge) a 3-span model of a deck bridge was placed on the shaking
table at IZIIS, Skopje (“Figure 1”, Dorka & Ristic, 2013). The controllable friction device
UHYDE-fbr (“Figure 2”, Dorka, 1995) was placed between piers and deck. With it the
behavior of various passive devices were simulated (stiff-plastic, bi-linear, base isolation
type, viscous, acceleration dependent, stiff-plastic device failure). The bridge model was
tested under sine- sweep excitations and the records from the Ulcinj-Albatros and EL-
Centro earthquakes. It was shown that forces in the piers were under control: Especially the
stiff-plastic control devices effectively limit the forces and protect the piers, even if the
devices are overloaded and fail (Dorka, 2014). Thus robustness was added, which is
especially important for elevated roads where the failure of one pier eliminates the
functionality of the whole road.
Fig. 1 – Model of a 3-span deck bridge on the shaking table at IZIIS, Skopje (Dorka &
Ristic, 2013)
Fig. 2 – Schematic representation of patented bi-directional semi-active friction device

UHYDE-fbr. The friction force is controlled by pressurized air (Dorka, 1995).
Integral bridges cannot have such links, so seismic control is not applicable. This raises
the question: Which concept is superior in earthquake prone regions?
2. STRUCTURAL CONTROL
In this paper, structural control is understood as the control of selected response variables,
such as displacements or their time derivatives (velocities, accelerations) and/or forces, of a
structure subjected to dynamic loading (Dorka, 2004). Such control of a structure can be
achieved, if the structural system is properly selected. Rigid-body motion control systems
appear to be the ones most suitable for seismic actions: They minimize the kinetic and
potential energies (and therefore displacements and forces) and allow for the control of
forces in the links between the rigid bodies. Four such systems have been suggested so far
for buildings (“Figure 3”).
Fig. 3 – Structural concepts suitable for seismic control: Base Isolation, Hysteretic
Device System, Tendon System, Pagoda System (Dorka, 2004)
It is evident that a deck bridge has a structural concept that naturally suits a Hysteretic
Device Systems, where deck and piers are rigid bodies. The link between them is ideal
for providing seismic control (control of force limits) by placing suitable devices there.
In general control devices can be passive, active or semi-active.
Passive control devices do not require any external energy source and are activated by
the motion of the structure. They should not be confused with “dampers” or dissipators” as
described in (Housner et al., 1997) and (Symans & Constantinou, 1999), because the
conversion of mechanical energy into (harmless) heat is not their primary purpose: It is the
control of forces, especially force limits! Therefore, friction or metal-yielding based devices
provide better controllability than viscous or base-isolation devices, which do not have a
force limiting hysteresis.
Active control devices can provide full controllability of forces throughout the earthquake, but
are expensive, use large amounts of external energy, require a sophisticated electronic
sensing and control system and need extensive maintenance. This makes them unsuitable for
almost all earthquake related applications where their action may be only required a
couple of times in several decades.
Halfway between passive and active control lies semi-active control in which a certain key
parameter of the control device is controlled. Semi-active control devices require an
external energy source to operate, but the energy needed is significantly smaller than for
active devices, which makes them less expensive and more attractive. They can improve
controllability in comparison to passive devices and have the potential to achieve the
performance of a fully active system for certain applications. Such devices include variable-
orifice fluid dampers, controllable friction devices (like the UHYDE-fbr, “ Figure 2”), variable-
stiffness devices, controllable tuned liquid devices, electro-rheological and magneto-
rheological devices and controllable impact devices (Housner et al., 1997; Symans &
Constantinou, 1999).
3. COMPOSITE BRIDGE SELECTED FOR THIS STUDY

A composite bridge typical for a 4-lane elevated road has been selected for this
comparative study. Only two spans (70 m each) were modeled in order to study the
interaction of one pier with the deck, which is the critical detail for both, the integral and the
seismic controlled bridge.
The superstructure is made of a typical concrete slab and 2 steel I-shaped girders, as
shown in "Figure 4". The deck slab has a variable height: 24 cm at cantilever ends, 46 cm
at joints with steel girders and 34 cm in the middle. Two girders at axial distance of 10.8
m are 3.2 m in height in total. Flange width is 1.0 m at the bottom and 0.9 m at the
joints with the slab, with thickness of 10 cm.
Fig. 4 – Selected typical cross section of the composite bridge
According to CSI Knowledge Base, there are several ways of modeling composite behavior
(Computers and Structures, 2010). Different approaches are shown in "Figure 5".
Composite action in these bridges is modeled using frame insertion points as in 'Model 4',
where frames and shells are drawn at the elevation of deck centroid.
This ´Model´ showed better full composite behavior (full connection through the span length)
than the ones with body constraints or fixed links, which is the reason why is it chosen.
Fig. 5 – Approaches to modeling composite behavior (Computers and Structures, 2010).
Model 4 is selected in this study
The superstructure is supported by piers in the middle and abutments on both ends. Piers
are formed out of two round columns, each 120 cm in diameter, spaced at 10.8 m just
below the steel girders. Height of the piers is a typical 6 m. Abutments are designed as
walls at the ends of the bridge, 2.5 m wide and 4 m high.
All the steel elements were designed with steel grade S355. Piers, abutments and cross
beam were designed with concrete grade C35/45, slab and transverse girder were designed
with concrete grade C45/55.
Both (integral and seismic control) bridges have been analyzed using SAP2000 v15.
4. THE INTEGRAL BRIDGE

To achieve integral behavior in this bridge, the piers are fixed at the top to the transverse
girder, which connects and stiffens the longitudinal steel girders. The width of the
transverse girder is 1.5 m. Abutments are extended to the top end and connected with the
concrete slab. "Figure 6" shows the model as defined SAP2000.
Fig. 6 – Integral bridge
5. THE BRIDGE WITH SEISMIC CONTROL (HYDE-SYSTEM)

In this bridge piers have a cross beam 1.5 m in width and 1.2 m in height on top providing a
link to the superstructure where elastomeric bearings are placed underneath the two steel
beams and a passive control device in the middle. Bearings and control device are
connected directly to the steel girders and concrete deck, respectively. "Figure 7" shows
the model as defined in SAP2000.
Bearings are modeled as linear links with the stiffness of 828 kN/mm in vertical direction
and 11.3 kN/mm in both horizontal directions. These values are taken from technical
specifications of elastomeric bearing pads made by Trelleborg Engineered Systems
(Trelleborg, 2009).
The seismic control device is modeled in SAP2000 as Plastic (Wen) link, which is based on
the hysteretic behavior proposed by Wen (Wen, 1976), with nonlinear behavior in both
horizontal directions. The stiffness of this plastic link is 140 kN/mm and the yield force is
200 kN. These values are selected to achieve the desired performance for the design
earthquake, which requires linear behavior throughout the structure (except in the control
device) and no exceedance of the design displacement limit for the control devices.
Fig. 7 – Bridge with passive seismic control
6. EARTHQUAKE DATA
All earthquake data used in analyzing these bridges was obtained from NIED Database via
the Strong-motion Virtual Data Center (National Research Institute for Earth Science and
Disaster Resilience, 2011).
The bridges were subjected to the records from the 2011 Mw 9.0 Tohoku earthquake. The
data was collected from the Oshika MYG011 station, whose hypocentral distance from the
earthquake is 81.3 km. Three directions were recorded as shown in “Figure 8”. Direction N-
S of the earthquake captured the maximal acceleration of 9.21 m/s2, which is applied to the ´x
directions´ of both bridges. In the E-W direction the maximum acceleration was 6.88 m/s2,
which is applied to the ´y directions´. Lastly, the station recorded the maximal acceleration of
2.54 m/s2 in the U-D direction, which is applied in the ´z directions´ of the bridges. The
duration of the recorded earthquake is 300 s, but for the purposes of this study the 20 s of
the second wave (80 – 100 s) were sufficient, where the biggest accelerations occurred.
7. RESULTS
By looking at the stress in both bridges and comparing them, one can see the
differences between the behavior of the integral bridge and the controlled bridge. The
integral bridge shows excessive stresses in the whole slab, especially in the region where
the slab is connected to the transverse girder and abutments (“Figure 9”) indicating serious
damage in these locations.
The bridge with seismic control shows significantly smaller stresses in general (“Figure
10”). Moments in the plate are essentially negligible (±45 kN-m), as oppose to moments in
the plate of an integral bridge.
First simple estimate gives rough values of a yield moment for concrete piers 5940 kN-m,
and of a full plastic capacity 10080 kN-m.
Moments in the piers of both bridges are shown in “Figure 11” and “Figure 12”.
Fig. 8 – Earthquake data, recorded acceleration (National Research Institute for Earth
Science and Disaster Resilience, 2011)
The full moment capacity is not reached, but the integral model with the maximum
moment in the pier around 9040 kN-m is close to reaching the capacity limit, and is well
into the plastic behavior. Moments in the piers of the bridge with seismic control stay
well within the elastic zone, with its maximum moment of 4740 kN-m. This is 2 times
smaller than the moment in the integral bridge.
Fig. 9 – Moments in the integral bridge (N-m)
Fig. 10 – Moments in the controlled bridge (N-m)
Fig. 11 – Moments in the piers (integral bridge, N-m)
Fig. 12 – Moments in the piers (controlled bridge, N-m)
The hysteresis loop of the modeled seismic control device is shown in the “Figure 13”,
which shows the well-controlled limit force in the link (200 kN). Displacement of the deck
ranges between -100 mm to +80 mm.
Fig. 13 – Hysteresis loop of the control device (force in N, deformation in m)
8. POST-QUAKE FUNCTIONALITY
This study has shown that an integral composite bridge is most likely going to experience
severe damage in and around the piers during an extreme earthquake, most likely causing
failure and collapse. This bridge is likely to lose its functionality. Reconstruction is going to
be extensive and time-consuming, in the meantime causing severe problems to the
functioning of the road network. In some familiar cases, like the 1994 Northridge and 1995
Kobe earthquakes, pier failures shut down entire road networks, with reconstruction taking
several years.
For the bridge with seismic control it is highly unlikely that deck and piers will be damaged
to a degree that functionality is severely reduced. The passive seismic device in the link
is clearly capable of saving the piers from collapsing. Under more extreme conditions,
some cracking of the deck may occur, if the displacements in the link become excessive
and the superstructure slips off the bearings. But even under such conditions, the piers are
protected (as the ISU-bridge experiments have shown: Dorka, 2014) and emergency
functionality is unlikely to be lost. This kind of damage is also much easier and faster to
repair than in the integral case, where some piers have likely collapsed.
All in all post-quake functionality of damaged bridges is little understood and needs further
research.
9. CONCLUSIONS
Two composite bridges, an integral and a seismic controlled one, were compared in
order to study their behavior and assess their post-quake functionality under an extreme
earthquake event. The 2011 Tohoku scenario was used for this purpose. Both bridges where
meant to represent a typical elevated road with hundreds of piers, although only one pier
was modeled in order to study the effect of such an extreme event on this vital
connection between superstructure and pier: If this fails, and especially if the pier
collapses, the whole road looses its functionality.
The study shows that the integral bridge is clearly in danger of sustaining heavy damage in
and around this vital detail. This damage may occur in the deck as well as in the piers,
which may even collapse. This can easily result in a total loss of functionality (one collapsed
pier is enough) and extended repair and reconstruction works. The integral bridge is therefore
not a robust structural concept suited for elevated roads in earthquake prone regions.
More robust structural systems must be used for bridges and elevated roads in such
regions, and seismic control concepts (like the Hyde-system studied here) are capable of
providing this for composite bridges within traditional, cost-effective bridge concepts. This
even opens the way for retrofitting such bridges without excessive interference: With the
necessary link already in place, only a suitable control device must be inserted to provide
the necessary robustness.
10. REFERENCES
Burke Jr, M. P. (2009). Integral and Semi-Integral Bridges. Chichester: Wiley-Blackwell.

Computers and Structures, Inc. (8. January 2010). CSI Knowledge Base. Retrieved
November 1, 2016 from CSI America: https://wiki.csiamerica.com/display/kb/Home
Dorka, U. E. (1995). Patentnr. 5456047. United States.
Dorka, U. E. (2004). Erdbebensicherung durch Structural Control. Stahlbau 73, Heft 9 , 661-
667.
Dorka, U. E. (1994). Hysteretic device systems for earthquake protection of buildings. 5th US
Conference on Earthquake Engineering (S. 775-785). Chicago: EERI.
Dorka, U. E. (2014). Seismic control for elevated roads. Building materials and structures , 9-
20.
Dorka, U. E., & Ristic, D. (2013). Seismic Upgrading of Bridges in South-East Europe by
Innovative Technologies; ISUbridge (SfP Project Number 983828); Final Report. Skopje:
The NATO Science for Peace and Security Programme.
Housner, G., Bergman, L., Caughey, T., A.G., C., Claus, R., Masri, S., et al. (1997).
Structural control: Past, present and future. ASCE Journal of Engineering Mechanics
(123(9)), 897-971.
National Research Institute for Earth Science and Disaster Resilience. (11. March 2011).
NIED: Database. Retrieved December 8, 2016 from NIED: National Research Institute for
Earth Science and Disaster Resilience:
ftp://www.kyoshin.bosai.go.jp/knet/3comp/2011/03/20110311144600/MYG0111103111446.t
a r.gz
Symans, M. D., & Constantinou, M. C. (1999). Semi-active control systems for seismic
protection of structures: a state-of-the-art review. Engineering Structures (21(6)), 469-487.
Trelleborg. (25. November 2009). Engineered Products: Trelleborg. Retrieved November 1,
2016 from Trelleborg AB: http://www.trelleborg.com/en/engineered-products/products--
and-- solutions/bearings
Wen, Y.-K. (1976). Method for Random Vibration of Hysteretic Systems. Journal of the
Engineering Mechanics Division, Vol. 102 (2) , 249-263.
AUTHORS’ INDEX
Abspoel, R.............................239 Gajda, J.................................132 Klemencic, R.............................1

Agarwal, A.............................274 Gajewski, M...........................322 Knobloch, M..........................714
Aggelopoulos, E......................74 Garcia, M...............................381 Kopp, M.........................144, 800
Alzeni, Y.................................643 Garlock, M.............................358 Korpas, G..............................346
Ataei, A..................................597 Geißler, K..............................346 Kostadinova, N......................346
Ayyad, H................................274 Gizejowski, M........................322 Kuhlmann, U....109, 194, 310, 574
Gordon, J.................................13 Kurz, W..........................132, 346
Baldassino, N........................762 Gross, J.................................702
Barcewicz, W.........................322 Guo, F....................................206 Lai, Z.............................392, 631
Bradford, M............................597 Guo, Z...................169, 450, 667 Lam, D...................................181
Broschart, Y...........................132 Landolfo, R............................678
Bruneau, M............................643 Hajjar, J.........................251, 381 Lee, H....................................404
Burgan, B................................74 Han, L....................................462 Lehman, D.............................775
Han, X...................................474 Leon, R. ................................381
Cao, Y....................................474 Hanya, K................................334 Lepourry, C............................157
Chan, S.................................690 Hauf, G..................................194 Li, G.......................................206
Chen, H.................................169 Hee Kang, W...........................86 Li, J........................................550
Chen, Y..........................502, 550 Helwig, T................................217 Li, M.......................................502
Cheng, X...............................655 Hicks, S.......................62, 86, 98 Li, T........................................690
Chhabra, A..............................38 Hiroshima, S..........................334 Liang, J..................................513
Choe, L..................................702 Hjiaj, M...................................416 Liao, Y....................................450
Chuan, G...............................550 Holtkamp, S...........................787 Liew, R.....................................50
Ciutina, A.........................62, 538 Hooper, J...................................1 Lignos, D...............................428
Clayton, P..............................217 Hou, C...................................462 Lin, H.....................................667
Cordova, P.............................486 Hu, H.....................................667 Liu, S.....................................690
Hüttig, L...................................25 Liu, X-J..................................450
Dai, X.....................................181 Liu, X-P. ................................597
D’Aniello, M...........526, 607, 678 Inamasu, H............................428 Liu, Y......................169, 450, 667
Deierlein, G...........................486 Lu, Y.......................................450
Denavit, M.....................369, 381 Jeon, C..................................440
Don, R...................................538 Ji, J........................................474 Mahmoud, H..........................750
Donahue, S...........................217 Ji, X.......................................655 Main, J...................................702
Dorka, U................................812 Jia, X.....................................655 Mathias, N...............................13
Dubina, D......................538, 678 Jian, X...................................502 Mazeika, A...............................38
Jiang, J..................................702 Morgen, B..................................1
Easterling, S..........................286 Judd, J...................................619 Neuenschwander, M..............714
Eatherton, M..........................286
Eggert, F................................310 Kang, W...................................98 Nguyen, Q.............................416
Engelhardt, M........................217 Kanno, R.......................334, 738 Nijgh, M.................................263
Kanvinde, A...................428, 586
Fargier-Gabaldon, L..............486 Keo, P............................157, 416 O’Brien, P...............................286
Feldmann, M.........................144 Kim, D....................................440 Odenbreit, C............................62
Fleischman, R.......................274 Kimura, K...............................738 Ozaki, F.................................738
Fontana, M............................714 Kitaoka, S..............................334
Freddi, F................................762 Klarendic, A...........................812
AUTHORS’ INDEX
Pak, D....................................800 Schorr, J................................194 Wang, L.................................251

Pakwan, N.............................619 Seidl, G..................................800 Wang, P.........................358, 513
Palas, F..................................157 Seif, M...................................702 Wang, W................................502
Pan, J....................................513 Selvarajah, R.........................702 Wang, Y...................................50
Park, H..................................404 Seo, J....................................631 Wang, Z.................................513
Parra-Montesinos, G.............486 Shahrooz, B...........169, 450, 667 Webster, M............................251
Pascual, A.....................109, 726 Sheehan, T............................181 Weigand, J.............................702
Pavlovic, M............................263 Shim, C..................................440 Williamson, E.........................217
Perea, T.................................381 Shimada, Y............................121 Wolters, K..............................144
Somja, H........................157, 416
Qin, C....................................750 Stark, J..................................239 Yamada, S.............................121
Stratan, A...............................678 Yang, B..................................561
Rademacher, D.....................787 Stroetmann, R.........................25 Yang, H......................................1
Raichle, J...............................109 Swanson, J....................526, 607 Yang, Z..................................206
Ranzi, G................................228 Szczerba, R...........................322 Ye, Y.......................................169
Rassati, G......................526, 607 Yong, D....................................50
Ricles, J.................................274 Takada, K..............................334 Yoshida, F..............................334
Roeder, C..............................775 Tartaglia, R............526, 607, 678 You, T.....................................474
Romero, M.............................726
Roverso, G............................762 Uang, C.................................274 Zandonini, R..........................762
Ruiz-Sandoval, M..................381 Ungermann, D.......................787 Zhang, J..................................13
Ruopp, J................................574 Uy, B..................................86, 98 Zhang, X................................206
Zhao, J..................................561
Sadek, F................................702 Varma, A........................392, 631 Zimbru, M..............................678
Sarkisian, M.............................13 Veljkovic, M...........................263
Sause, R................................274 von Arnim, M.........................263
Schäfer, M.............................298 Vulcu, C.................................538
KEYWORD INDEX
ambient vibration...................381 composite structure...............169 encased columns..................381

analytical......... 392, 98, 121, 228, composite walls...........1, 74, 631, engineering model.................144
263, 274, 346, 369, 416, 526, 643, 655 environmental product
607, 619, 702, 714, 750, 812 concrete anchors...................561 declarations.........................25
analytical/experimental............286 �� Eurocode 2............................416
anchor connections...............561 column...............................502 Eurocode 4....................298, 416
anchorage.............................574 ��= Experimental......... 50, 132, 157,
beam test...............................263 tube....................................513 181, 206, 251, 263, 310, 404,
beam-column connection......550 ��= 486, 513, 574, 586, 643, 655,
blast behavior........................762 (CFST)...............................462 714, 787
blind bolts......................502, 726 ��ë ..............775 experimental analysis............762
bracing design.......................334 �� experimental data collection....474
bridge applications.........98, 775, columns.............................726 experimental investigations.....322
787, 800, 812, 358, connected towers....................38 experimental study........550, 667
buckling design formula.........334 connections...................502, 702 experimental tests.................144
buckling restrained bracing...440 constitutive model....................50 fastening technology.............574
��í . ....334 construction schedule........... 631 fatigue resistance..................787
buildings..................................86 corrosion................................462 ��
�� . .......206 coupled systems....................631 (FEA)........450, 597, 334, 678,
ComFlor.........................210, 239 cyclic shear behavior............ 655 726, 206, 462
composite action...................513 database................................392 �� ......169, 538
composite beams.........298, 132, decks and slabs......62, 206, 228, �� ......322
334, 346 274, 286, 607 �� .........................................358
composite beams and deconstructable.....................597 ��. .........702, 714, 750
trusses............13, 62, 74, 121, deep deck..............................239 �� ........................726
157, 181, 194, 251, 263, 310, deep foundation.................... 775 ��.......... 738
574, 702 deformation limits..................474 ��ë ...........................25
composite bridges................ 787 demountability.......................263 force transferring
composite columns.....13, 50, 74, design....................................392 mechanism........................513
369, 381, 392, 404, 416, 440, design for deconstruction......263 frame collapse mechanism....738
450, 513, 714 design formulas.....................655 friction-grip bolts....................597
composite construction....86, 561 design method.......................450 geometrical imperfections.....322
composite dowel strips..........787 details of gusset plate............513 gravity framing connections...217
composite joints.....................597 direct analysis........................690 headed studs.........................574
composite links......................678 distortional instability.............322 headed studs close to
composite slab and metal drilled shafts..........................775 the edge.............................109
deck...................................217 ductility..................................310 high strength concrete...........392
composite slabs.............206, 239 ecological balance.................. 25 high strength steel.........392, 690
composite steel-concrete ��ó .........381 high-rise buildings...................50
beams................................322 ��ë ...........381, 416 hot-dip galvanizing................787
composite structural ��ó ............381 hybrid columns......................416
systems......1, 13, 38, 74, 194, ��ë ......................381 hybrid coupling walls.............667
251, 286, 322, 346, 428, 440, elevated temperatures...........702 impact....................................462
486, 513, 526, 550, 574, 607, embedded column base imperfections.........................690
619, 750, 762, 800 connections.......................428 injection bolt..........................263
KEYWORD INDEX
innovative hybrid members...416 plastic design.........................298 shear strength.......................655

innovative hybrid system.......631 post-welding residual �� ....................538
large-scale testing.................217 stresses.............................322 small-scaled..........................144
��ë ......................631 prediction methods................263 SRC columns................381, 474
lateral strength...................... 631 project....................1, 13, 38, 194 stability of steel columns...... 428
lateral-torsional buckling........334 project descriptions..............586, standards and
life-cycle analysis (LCA)........462 38, 194 �� ..........62, 74, 86,
linked towers...........................38 push-out test..........157, 169, 263 98, 228, 310, 369, 404, 486,
load slip behavior..................310 recommendations for design and 762, 800
longitudinal shear execution...........................263 static shear capacity..............144
resistance..........................239 regressive analysis................474 steel.......................................358
Mexico...................................381 reinforced concrete................416 steel and composite
m-k method...........................239 reliability................................392 constructions.......................25
��...........416 replaceable component.........450 steel columns........................428
moment resistance................298 replaceable details................667 steel coupling beams.............667
moment-resisting replaceable links....................678 steel slit damper....................450
connections.......................538 resins.....................................263 steel-concrete
non-seismic building frames..738 robustness.....................762, 738 connections...............561, 574
numerical simulations............144 ��í ...............416 structural redundancy........... 738
�� seismic...................................217 sub-assemblage
buildings..............................25 seismic applications.......13, 121, experiments.......................334
parametric studies.................206 274, 440, 502, 526, 550, 586, sustainability............................25
partial shear method..............239 619, 643, 655, 775, 812 tall buildings...........................561
performance based design.... 440 seismic behavior....538, 631, 667 ultimate capacity....................206
�� seismic design.......................678 ultra-high strength concrete.....50
design................................738 seismic engineering...............561 U-shaped steel beam............157
performance levels................474 shear buckling.......................358 validation of numerical
piers and superstructure shear connections.................109 model.................................322
connection.........................812 shear connectors............62, 132, welded plate girders..............322
piles.......................................775 144, 157, 169, 181, 194, 206, wind behavior........................631
pin-structures.........................144 263, 310, 787, 800
Smarter. Stronger. Steel.
American Institute of Steel Construction
312.670.2400 | www.aisc.org
P705-20

p705-20w

Uploaded by

Copyright:

Available Formats

p705-20w

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

p705-20w

Uploaded by

Copyright:

Available Formats

COMPOSITE

Edited by Gian Andrea Rassati,

July 29 – August 2, 2017

Wright Runstad & Co.

Edited by Gian Andrea Rassati,

July 29 – August 2, 2017

American Institute of Steel Construction

Printed in the United States of America

These proceedings summarize the state-of-the-art in composite construction worldwide,

Gian Andrea Rassati, Jerome F. Hajjar, and Roberto T. Leon

Ulrike Kuhlmann, University of Stuttgart, Germany

Composite Floor, Slab, and Diaphragm Systems

Composite Columns and Column Bases

Composite Lateral Force Resisting Systems

Fire and Progressive Collapse

Hee Jae Yang

AISC 341-16 includes expansion of Section H Composite Braced-Frame and Shear-Wall

TALL BUILDING APPLICATION

(a) (b) (c)

CF-CPSW CORE WALL DETAILING

Fig. 2 – CF-CPSW Multi-Cell Core Wall

CODE-LEVEL ELASTIC STRENGTH DESIGN

Table 1 – MCE Acceptance Criteria for Deformation-Controlled Actions

Fig. 8 – Representative CF-CPSW Normalized Wall Shear Stresses

Figure 2. Plan of the project at level 6, showing the three buildings

Figure 3. Sections through Tower A

Figure 4. Construction photo of the transfer truss between levels 23 and 25

Figure 5. Coupling truss between level 43 and roof level

Figure 7. Cantilever trusses between levels 25 and 42

Figure 8. Long cantilever beam supporting swimming pool at level 24

Figure 9. Long cantilever truss supporting hotel spa at level 24

Prof. Dr.-Ing. Richard Stroetmann

Dipl.-Ing. Lukas Hüttig

2.1 ECOLOGICAL SUSTAINABILITY

2.2 Economical sustainability

3 DESIGN OF BEARING STRUCTURES

4 DESIGN AND EVALUATION OF REFERENCE BUILDINGS

Multifunctional reference building with three

5 SUSTAINABILITY AND INFLUENCE OF FACADE SYSTEM

Aaron Mazeika, PE SE AIA

CONVENTIONAL COMPOSITE CONSTRUCTION

COMPOSITE BUILDING CORE

Wujiang Greenland Structural System

Figure 02- Architectural Rendering

Mode 1: 7.3s (Translation) Mode 2: 4.8s (Translation) Mode 3: 2.9 (Torsion)

Figure 04 - Dynamic properties of building

Figure 06 - Embedded truss at the split core levels

Resistance to Racking Motion

Figure 08: Architectural rendering

Floor hung from bridge

Outrigger Transfer truss

Figure 11: Enhanced torsional performance with connected bridge trusses

Figure 12: Connecting multiple buildings study

Table 1. Limitation on characteristic strength (N/mm2) in modern design codes

2 NEW TESTS AND COLLECTED TEST DATA

(a) Circular specimen (b) Square specimen

(a) Instrumentaion (b) Circular specimen (c) Square specimen

3 INFLUENCE OF CONCRETE STRENGTH

𝐸𝐸cm = 22[(𝜂𝜂 ⋅ 𝑓𝑓ck + 8)⁄10]0.3 (2)

Table 4. Effective compressive strength and modified secant modulus of HSC