See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327670013

A DUCTILE FUSE FOR SPECIAL CONCENTRICALLY BRACED FRAMES

Conference Paper · December 2018

DOI: 10.18057/ICASS2018.xxx

CITATION READS

1 659

1 author:

Machel Morrison
University of California, San Diego
11 PUBLICATIONS   74 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Machel Morrison on 15 September 2018.

The user has requested enhancement of the downloaded file.

 Ninth International Conference on Advances in Steel Structures (ICASS’2018)
5-7 December 2018 - Hong Kong, China

A DUCTILE FUSE FOR SPECIAL CONCENTRICALLY BRACED

FRAMES

Machel L. Morrison 1*
1
Research Assistant Professor, Dept. of Civil, Construction and Environmental Engineering, North Carolina
State University, Raleigh NC 27607 USA
E-mail: mlmorris@ncsu.edu

Abstract: In special concentrically braced frames, brace yielding and buckling are the primary mechanisms of seismic
energy dissipation, and mid-length brace fracture is the preferred failure mode. However, as evidenced by several studies, the
connections between the braces and gusset plates are susceptible to premature fracture as are other connections e.g. between
the gusset plate and beam or column. This paper describes the preliminary development of a simple ductile fuse to prevent
such failures. The fuse is created by selectively reducing the brace material strength through a thermal treatment process which
in turn reduces connection force demands. Material testing of ASTM A500 steel from square HSS tubing was conducted to
characterize the mechanical property changes from the thermal treatment and to calibrate 3D finite element (FE) models of
brace members. The experimental data and FE models demonstrate the benefits of the fuse, which include, reduced tensile
force demands at the brace to frame connections, improved fracture resistance of brace material, increased drift capacity,
residual stress relief and improved element compactness.

Keywords: Concentrically braced frames; Heat-treatment; Seismic fuse; Connections

DOI: 10.18057/ICASS2018.xxx

1 INTRODUCTION
Due to their inherently high lateral stiffness, special concentrically braced frames (SCBFs)
generally lead to economical building frames. Over the past two decades several advances have
been made in understanding the seismic performance and failure modes of SCBFs [1]. These
efforts have led to among others, improvements in connection detailing, restrictions on brace
configuration and more stringent brace slenderness and section compactness criteria [2-4].
However, despite these advancements, areas for improvement in the economy and performance
of SCBFs still exist.
Rectangular or round hollow structural sections (HSS) constructed to the ASTM A500 grade
B or C specification are commonly specified for brace members due to their structural
efficiency, availability and ease of construction. However, since braces are sized using design
strengths based on the specified minimum yield strength and connections and other elements
are designed to resist forces based on the expected yield strength, the high ratio of expected to
specified yield strength (Ry) which is ≈ 1.3-1.4 for A500 steel, compromises economy [1].
Stated more directly, connections and other frame elements such as columns are designed to
resist 30 to 40% higher brace forces than otherwise necessary because of brace material over
strength.
Another issue is that cold forming at the corners of rectangular HSS members reduces
material ductility and notch toughness [5], unfortunately, this is also the location of high strain
demands when mid-length brace hinges are formed under large load reversals [1]. These
combined effects likely reduce the deformation capacity of rectangular HSS braces under
seismic loading [6].

Proceedings of the ninth international conference on Advances in Steel Structures

Edited by Siu-Lai Chan, Tak-Ming Chan and Songye Zhu. Copyright © 2018 by The Hong Kong Institute of Steel Construction.
 M.L. Morrison

Finally, HSS braces are typically connected to surrounding elements (column, beam or both)
(a) with a gusset plate. The
(b) construction of this connection
involves slotting the brace,
thereby creating a net section
susceptible to premature failure
(Figure 1a). Typically, the net
section strength is determined
as
Reinforcement of
𝑅𝑛𝑒𝑡 𝑠𝑒𝑐𝑡𝑖𝑜𝑛 = 𝜎𝑢 𝑈𝐴𝑛 (1)
net section

where 𝜎𝑢 is the ultimate stress,

(c) U is the shear lag factor and 𝐴𝑛
is the net section area. While
the yield strength of the
member is

𝑅𝑦𝑖𝑒𝑙𝑑 = 𝑅𝑦 𝜎𝑦 𝐴𝑔 (2)
Heat-treated
sections
where σy is the yield stress and
Ag is the section area. Typically,
the shear lag factor, U, is
approximately 0.85, and the
ratio between the yield and the
Figure 1: (a) Rupture of brace to gusset plate connection [1] (b)
sketch of connection between brace and gusset plate (c) sketch
ultimate stress is also a similar
of inverted V-braced frame with heat-treated braces. value. Thus, even in absence of
material over strength (i.e. Ry =
1) the inequality σuUAn > RyσyAg , cannot be satisfied unless the net section is reinforced with
plates as shown in Figure 1b [7].

2 NOVEL FUSE TECHNIQUE

To overcome these challenges, a technique is borrowed from previous work of the author
and co-workers in moment resisting frames [8]. As shown in Figure 1c, the proposed technique
involves heat-treating the mid-section of the brace (up to one brace depth away from the edge
of gusset plate) by subjecting this region to high temperatures followed by controlled cooling.
Temperatures during this heat-treatment are high enough to relieve residual stresses from prior
cold work and to cause metallurgical changes to the material. This technique has several
attractive features. In Figure 2a and 2b the tensile stress-strain responses of A500 steel and heat-
treated (HT) A500 steel are shown. Coupons were subjected to various peak temperatures (700,
800 and 900°C), held at this temperature for 30 minutes (to allow for thermal equilibrium),
cooled at 6°C/min (Figure 2a) or 0.6°C/min (Figure 2b) to 550°C and then removed from the
furnace to cool in air. From the figure, it is observed that with increasing temperatures and
slower cooling rates the strength of steel is reduced while the elongation is increased. Also note
that the ability of the steel to work harden is restored. Temperature has a far greater effect than
cooling rate, as such, for the remainder of this paper, discussion will be limited to the thermal
history shown in Figure 2a (i.e. the faster cooling rate).

2
 M.L. Morrison

600 600
A500 (a) A500
700°C 700°C (b)
500 800°C 500 800°C
*900°C *900°C
400 400
σ (MPa)

σ (MPa)
300 300
T °C t @ Tmax = 30min. T °C t @ Tmax = 30min.
200 Tmax Heating rate: 10°C/min. 200 Tmax
Heating rate: 10°C/min.
Cooling rate: 6°C/min. Cooling rate: 0.6°C/min.
550 550
*Sample cooled from 900°C *Sample cooled from 900°C
100 Air cool to 850°C at 10°C/min.
100 Air cool to 850°C at 10°C/min.

t (min.) t (min.)
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
ε ε

A500 (c) HT A500 @900°C (d)

A500 (e) (f) HT A500 @900°C

Curling of wall
from release of
No distortion due
residual stresses
to residual stress
relief

(g)

A500 HT A500 @900°C

Figure 2: (a) and (b) Tensile stress-strain reponses of A500 and HT A500 steel (c) and (d) optical
micrographs of A500 and HT A500 steel (e) and (f) photographs of saw cutting HSS before and
after HT (g) photographs of fractured Charpy V-notch bars before and after heat treatment.

The reduction in strength and increase in ductility are due to microstructural transformations
(recovery and partial or full recrystallization) which take place during the annealing cycle. A
detailed description of these metallurgical processes is beyond the scope of this paper, but a
brief overview is provided here. Recrystallization of a cold worked metal reduces the

3
 M.L. Morrison

dislocation density in the crystal lattice and produces a new set of “strain-free” grains [9]. The
process requires mass transport (i.e. diffusion), making it especially sensitive to temperature
(typically following an Arrhenius type equation). Below the phase transformation temperature
e.g. @ 700°C (Ac1≈717°C for this steel), the thermodynamic driving force for recrystallization
is the free energy stored in dislocation networks and the process allows the material to achieve
a lower internal energy state by reducing the strain energy in the lattice. At higher temperatures,
recrystallization is due to the ferrite to austenite phase transformation, which reverses upon
cooling. Because of recrystallization, the effects of prior cold work are removed either partially
or completely [9]. After full-recrystallization (e.g. HT @900°C) a uniform equiaxed grain
structure is produced (compare Figure 3c and 3d). Note that there is no significant grain growth,
this is likely due to the solute drag and Zener pining effects from alloying elements such as
Niobium and Aluminum (see table 1 for the chemical composition of this steel).
As a consequence of these microstructural changes, notch toughness is remarkably improved
and residual stresses in the HSS section are relieved (Fig. 2 e-g and table 2).These mechanical
property changes have several desirable effects for braced frames. First and foremost, material
overstrength can be reduced significantly, for example as shown in Figure 2a and table 2, heat-
treating at 800°C reduces the yield strength to 345 MPa which is the specified minimum yield
strength of the material. Further, material strength can be controlled to allow the designer more
flexibility in proportioning members. This will improve structural efficiency and likely overall
economy of the structure. Second, the combination of improved ductility, notch toughness and
work hardening ability should reduce the susceptibility of braces to strain localization and
fracture, allowing them to sustain larger deformations, more cycles and, better energy
dissipation. Third, reinforcement of the connection between the brace and gusset plate (see
Figure 1a) is no longer needed since the inequality σuUAn > RyσyAg can be readily satisfied for
HT braces. Finally, residual stress relief should lead to improved compressive strength.
It is noted that recently the ASTM A1085 specification for cold formed HSS has been
introduced and has been given a lower overstrength factor (Ry =1.25) than A500 steel [4]. While
improved performance is anticipated for the A1085 grade, the heat-treatment proposed in this
study will further reduce material overstrength and will provide better ductility and notch
toughness than A1085. In addition, it solves the problem of connection reinforcement, which is
not addressed by the A1085 specification.
In this study, the proposed fuse technique is studied through detailed finite element (FE)
simulations and an uncoupled ductile fracture prediction model. The study focuses on the effect
of tensile properties and b/t ratio on brace strength and deformation limit states such as the onset
of local buckling and ductile fracture initiation. Details of the modelling strategy and results are
presented in the following.

3 FE MODELLING
The modelling strategy employed in this study follows closely with that of Fell [10] with a
few exceptions. Three-dimensional nonlinear FE models were developed for brace members
using the commercial software ANSYS Mechanical ADPL. Geometric and material
nonlinearities were incorporated in the FE models. An example of the FE mesh and boundary
conditions is shown in Figure 3a. The brace and gusset plates were modeled with continuum 20
noded solid hexahedral elements (SOLID186). FE models accounted for material nonlinearity
through rate-independent metal plasticity theory based on the von Mises yield criterion, additive
strain decomposition and associated flow rule. A multilinear kinematic hardening model
calibrated from tensile material tests, strain-controlled cyclic material tests and smooth notch
tensile (SNT) tests was used to capture material strain hardening (Figure 2a and Figures 3a-c).
The strain-controlled cyclic material tests allowed for model calibration at low strain ranges

4
 M.L. Morrison

while the tensile and SNT tests allowed for simulation of behavior at large strains. This
combination was essential for capturing the response of the brace throughout the load history
(up to fracture initiation).
6
4
Far-field loading
(a)
2
Drift (%)

0
-2
-4 𝒖𝒙 = 𝒖𝒚 = 𝟎
-6
Y
DUR = 11.7 mm
X DNR =6.4 mm
R =1.6 mm
Z
DUR
-21453.4

-4711.04

25 mm 12031.4

28773.8

gauge R 45516.2

62258.6

DNR
79000.9

95743.3

112486

𝒖𝒙 = 𝒖𝒚 = 𝒖𝒛 = 𝟎 129228

800 (c)
(b) Fracture initiation
30
Experiment-A500
400
Simulation-A500
Experiment HT@900°C
Force (kN)
σ (MPa)

20 Simulation HT@900°C
0

Experiment-A500
-400 Simulation-A500 10
Experiment-HT@900°C Fracture
Simulation-HT@900°C initiation
-800 0
-0.02 -0.01 0 0.01 0.02 0.03 0 0.5 1 1.5 2 2.5
ε
Elongation (mm)

Figure 3: (a) Far field loading history, FE mesh and boundary conditions, photographs of
cyclic and SNT tests, contour plot of hydrostatic stress distribution in notch (b) experimental
and simulated stress-strain response of A500 and HT A500 steels (c) experimental and
simulated SNT response of A500 and HT A500 steels.

Geometric nonlinearities were accounted for via a large deformation formulation which
accompanied by small eccentricities/imperfections in the geometry allowed for the simulation
of global and local brace buckling. Initial imperfections were obtained by first conducting an
eigenvalue buckling analysis of the perfect structure and then prescribing superimposed scaled
values of the global buckling mode (first eigenmode) and local buckling mode (third
eigenmode) displacement fields as the initial configuration of the structure. The maximum
value of geometric imperfection for the global mode was L/1000, which is half of the
permissible variation in straightness allowed by ASTM A500. The maximum value of
geometric imperfection for the local mode was 0.07mm based on measurements by Fell 2008
[10]. Global and local imperfections were essential to simulating buckling deformations

5
 M.L. Morrison

observed in experimental testing of braces [10]. It is noted, that residual stresses were not
included in the FE analysis leaving the effect of this variable for future study.
Stress and strain indices from the FE model were used as inputs to a micromechanics based
cyclic void growth model (CVGM) proposed by Kanvinde and Deierlein [11] to predict the
initiation of ductile macroscopic cracks that were reported during laboratory tests [10]. The
CVGM predicts the initiation of a ductile macroscopic crack when the following condition is
satisfied
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝑉𝐺𝐼𝑐𝑦𝑐𝑙𝑖𝑐 > 𝑉𝐺𝐼𝑐𝑦𝑐𝑙𝑖𝑐 𝑓𝑜𝑟 𝑙 ≥ 𝑙 ∗ (3)

𝑉𝐺𝐼𝑐𝑦𝑐𝑙𝑖𝑐 = ∑ ∫ 𝑒 1.5𝑇 𝑑𝜀𝑃 − ∑ ∫ 𝑒 1.5𝑇 𝑑𝜀𝑝 (4)

𝑇𝑒𝑛𝑠𝑖𝑙𝑒 𝐶𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑣𝑒
𝐶𝑦𝑐𝑙𝑒𝑠 𝐶𝑦𝑐𝑙𝑒𝑠

𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑎𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑒𝑑

𝑉𝐺𝐼𝑐𝑦𝑐𝑙𝑖𝑐 = 𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 𝑒 −𝜆𝜀𝑝 (5)
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
where, 𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 is obtained from the following failure criterion establish for monotonic
loading:
𝜀𝑝
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 = ∫ 𝑒 1.5𝑇 𝑑𝜀𝑝 > 𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 (6)
0

where 𝑇 = 𝜎𝑚 ⁄𝜎𝑒 (ratio of mean stress to effective stress, also called stress triaxiality) and
𝑑𝜀𝑝 is the increment of equivalent plastic strain. λ is a material dependent parameter which
represents the degradation of material resistance to ductile fracture due to cyclic loads and
𝜀𝑝𝑎𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑒𝑑 is the equivalent plastic strain that has accumulated up to the beginning of each
tensile excursion of loading. The CVGM fracture initiation criterion (eq. 3) should be satisfied
over a length scale representative of the physical events leading to ductile fracture (𝑙 ∗ ). This
model feature typically requires mesh sizes on the order of 𝑙 ∗ . The element length in the
region of interest in this study is 0.3 cm which is sufficient to capture strain and stress
gradients but is much larger than 𝑙 ∗ . However, other studies demonstrate reasonable results
using similar element sizes [10, 12].
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 and λ are calibrated by experimental testing of notched bar specimens
subjected to monotonic and cyclic loading histories respectively. In this study, SNT bars
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
subjected to monotonic loading were used to calibrate 𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 (see Figure 3c and Table 2)
but due to limited resources, cyclic notched bar testing was not carried out. Therefore, a value
of λ=0.17 was used based on the work of Fell [10]. While this value was found to be reasonably
accurate for A500 steel it may be conservative for HT A500 steel.
3.1 Finite Element Model Validation
For FE model validation, specimen HSS 1-1 and HSS 1-2 tested by Fell [10] were modelled.
Both specimens consisted of square HSS102x102x6.35 members. The overall length of the
specimens (i.e. the distance between the outer edges of the gusset plates) was 3124.2 mm. The
gusset plates were welded to thick endplates and loaded axially. The reader is referred to that
study for details of the experimental setup, specimen dimensions, loading history etc. Specimen
HSS 1-1 was subjected to a far-field loading history (see inset of Figure 3a) akin to that
commonly used in the testing of moment frames, while specimen HSS1-2 was subjected to a
near-fault loading history with asymmetric compression loading cycles. Story drift angle
expressed in radians (θ ) is determine from the assumed kinematics of the braced frame during
seismic loading [10] and is calculated as

6
 M.L. Morrison

2∆𝑎
𝜃= ⁄𝐿 (7)
𝐵

where Δa is the axial deformation of the brace and LB is the distance between the fold lines of
the gusset plate (2984.5 mm).
The predicted brace force-drift (%) response is plotted against the experimental response in
Figure 4a and 4b. The model prediction is found to be in close agreement with the experimental
response for both loading histories. The evolution of the void growth indices (for the CVGM)
for specimen HSS 1-1 is shown in Figure 4c. The location considered is the brace centerline at
the corner of the HSS as shown in Figure 4d. Note that fracture initiation is predicted to occur
at the end of the second excursion of loading at 2.8% drift while fracture was observed slightly
before the end of the second excursion at 2.8% drift (Figure 4a).
1.5 Fracture initiation 1.5
predicted (a) Experiment (b)
1 1 Simulation
Experiment

Brace Force (MN)

Brace Force (MN)

Simulation
0.5 0.5

0 0
Fracture in test
-0.5 -0.5

-1 -1
-4 -2 0 2 4 -8 -4 0 4
Drift (%) Drift (%)
2.5 (c)
𝑽𝑮𝑰𝒄𝒓𝒊𝒕𝒊𝒄𝒂𝒍
𝒄𝒚𝒄𝒍𝒊𝒄
𝒑
𝜺𝒆𝒒=0.74 (d) (e)
2 Fracture initiation
(Void growth resistance) T =0.5
predicted at corner Ductile tearing of
of HSS HSS corner
1.5 CVGM failure criteria
satisfied during 2nd
1 loading cycle to 2.7% drift

0.5 𝑽𝑮𝑰𝒄𝒚𝒄𝒍𝒊𝒄
(Void growth demand)
0
0 200 400 600 800 1000

Figure 4: Finite element model validation (a) comparison of experiment and simulated force-drift
responses of specimen HSS 1-1-far field loading [10] (b) comparison of experiment and simulated
force-drift responses of specimen HSS 1-2 near-fault compresion loading[10] (c) CVGM fracture
prediction for specimen HSS 1-1 (d) FE prediction of accumulated equivalent plastic contours of
specimen HSS 1-1 (e) photograph showing fracture initation of HSS 1 [10].

4 FE EVALUATION OF PROPOSED TECHNIQUE

The proposed fuse technique is evaluated using the previously validated FE model. See
Table 3 for details of the specimens considered. The length of the brace and boundary
conditions are the same as those used for FE model validation. Also, for sake of simplicity, the
far field load history used in the validation is applied to all specimens. This evaluation considers
the effect of the maximum temperature of heat treatment (i.e. the change in mechanical
properties from heat-treating the brace to 700, 800 and 900°C respectively). In addition, the
evaluation considers two different HSS wall thicknesses to evaluate the effect of section
compactness i.e. b/t ratio on brace response. Note that measured properties (from table 2) are
used for calculation of all applicable quantities reported in table 3. The yield strength from heat-
treating at 700, 800 and 900°C resulted in three different ratios of σuUAn / σyAg (capacity to

7
 M.L. Morrison

demand) all of which are greater than unity. This implies that net section failure would be
circumvented for all HT braces considered. The prediction of fracture initiation at the net
section is outside the scope of the current study, but needs future attention.
Several important insights are drawn from the results of FE analysis presented in Table 3.
The normalized tensile (𝑃𝑇𝑚𝑎𝑥 ⁄𝑃𝑦 ) and compressive strengths (𝑃𝐶𝑚𝑎𝑥 ⁄𝑃𝑦 ) of the brace are
increased from the heat treatment. The increase in normalized compressive strength is far
greater than the increase in tensile strength. The mild increase (3-9%) in normalized tensile
strength when HT braces are compared to their unconditioned counterparts is likely due to
material strain hardening (see Figure 2a and Table 2).

1.4 1.4
(a) HSS 102x102x9.5 (A500)
Fracture initiation
(b) HSS 102x102x9.5 (HT@700°C)
predicted

0.8 0.8

P/Py
P/Py

0.2 0.2
Strength loss
predicted due
to necking
-0.4 -0.4
Local buckling
predicted
-1 -1
-7 -5 -3 -1 1 3 5 7 -7 -5 -3 -1 1 3 5 7
Drift (%) Drift (%)

HSS 102x102x9.5 (A500) HSS 102x102x9.5 (HT@700°C)

(c) (d)
1st cycle @ 2.7% Drift 1st cycle @ 2.7% Drift

Plastic strains
widely
Plastic strains localize
distributed

(e) HSS 102x102x9.5 (A500) (f) HSS 102x102x9.5 (HT@700°C)

1st cycle @ 4% Drift 1st cycle @ 4% Drift

Plastic strains
Plastic strains localize widely
distributed

Figure 5: Finite element analysis of braces (a) force-drift response of HSS102x102x9.5 (A500) (b)
force-drift response of HSS102x102x9.5 (HT@ 700°C) (c) equivalent plastic strain contours of
HSS102x102x9.5 (A500) at 2.7% drift (d) equivalent plastic contours of HSS102x102x9.5 (HT@
700°C) at 2.7% drift) (e) equivalent plastic strain contours of HSS102x102x9.5 (A500) at 4% drift
(f) equivalent plastic contours of HSS102x102x9.5 (HT@ 700°C) at 4% drift

8
 M.L. Morrison

The b/t ratios for the HSS members have been normalized to the limiting values specified
in the AISC Seismic Provisions [4] and tabulated in Table 3. Note that the heat treatment
reduces the normalized b/t ratios implying improved compactness and greater deformation
capacity. Indeed, as shown in table 3, HT braces showed delayed onset of local buckling and
fracture initiation. In fact, for HSS102x102x9.5 braces HT @700, 800 and 900°C no local
buckling or fracture initiation was predicted before the analysis was terminated after loading
up to 6% drift.
The improved deformation capacity of HT braces is attributed to the improved work
hardening behavior and the increased fracture resistance of the material. For example,
consider the responses of HSS102x102x9.5-A500 and HSS102x102x9.5-HT@700°C braces
shown in Figure 5. Due to the low work hardening rate of A500 steel, plastic strains localize
in the center of the brace during cycles at 2.7% drift. Note the relatively large strain gradients
at the center of the A500 brace shown in Figure 5c relative to those of the HT brace shown in
Figure 5d. This localization of plastic strains eventually leads to local buckling during cycles
at 4% drift and fracture soon after (Figure 5a and 5e). This localization of strain is
circumvented in the HT brace due to the improved material work hardening (Figure 5b and
5f).

4.1 Practical Issues

The proposed heat-treatment will require a special furnace or induction heating system
adjustable to the desired HT length of the brace. Note that heat-treatment of the brace can be
performed before the brace is cut to its final length and before fabrication steps such as hole or
slot cutting. Therefore, it will not interrupt the fabrication process. Also, based on the findings
of this study, the process should take approximately two hours. This process time can be
reduced with faster heat-up rates and further efficiency can be gained by heat-treating several
braces simultaneously. Non-destructive test methods such as hardness testing can be calibrated
to verify desired strength is achieved by the proposed heat-treatment or alternatively, small
pieces of steel from the same heat as the brace members can be HT and destructively tested to
ensure strength requirements are satisfied.

4.2 Future Work

This proposed fuse technique shows promise but needs to be validated through large scale
testing. Also, given the many variables that influence HSS material strength such as
manufacturing history (e.g. direct formed vs. continuously formed), section geometry and steel
chemistry, a detailed metallurgical study is needed to develop appropriate thermal cycles that
efficiently produce the desired mechanical properties. Fortunately, the kinetics and
thermodynamics of recrystallization and phase transformations in low carbon steels are fairly
well established [9]. So too are metallurgically based strength prediction models [13]. For
example, based on the chemical composition (Table 1) and an average grain size of 11.7 μm
for A500 steel HT @900°C (measured from Figure 2d), one such strength model [13] predicts
a yield strength of 279 MPa which is remarkably close to the average measured yield strength
of 276 MPa (Table 2). However, much work is needed to verify these models across a range of
material chemistries and processing histories.

5 CONCLUSION
A novel fuse technique for SCBFs is introduced and analyzed in this study. The technique
involves heat-treating the mid-section of the brace to reduce material strength and improve
work hardening, ductility and notch toughness. The technique has the potential to

9
 M.L. Morrison

simultaneously reduce construction cost and improve seismic performance for new building
construction and may also be used to upgrade older non-ductile brace frames. However, the
results of this study are preliminary and much effort is needed to validate these claims and to
eventually implement this technique in building construction.

ACKNOWLEDGEMENTS
The author is grateful to Steel Fab Inc. for donating the steel for this study. The efforts of
Christopher Stanford in machining test specimens, the assistance provided by Dr. Harvey West
in performing optical microscopy and that of Heramb Mahajan in preparing isometric drawings
are gratefully acknowledged.

REFERENCES
[1] Sabelli, Rafael, Roeder, Charles W., and Hajjar, Jerome F. “Seismic design of steel special
concentrically braced frame systems:A guide for practicing engineers,” NEHRP Seismic Design
Technical Brief No. 8, produced by the NEHRP Consultants Joint Venture,a partnership of the
Applied Technology Council and the Consortium of Universities for Research in Earthquake
Engineering, for the National Institute of Standards and Technology, Gaithersburg, MD, NIST
GCR 13-917-24, 2013.
[2] Lehman, D. E., Roeder, C. W., Herman, D., Johnson, S., and Kotulka, B., “Improved seismic
performance of gusset plate connections.” Journal of Structural Engineering, 134(6), 890–901,
2008.
[3] Roeder, C. W., Lumpkin, E. J., and Lehman, D. E., “A balanced design procedure for special
concentrically braced frame connections.” Journal of Constructional Steel Research, 67(11),
1760–1772, 2011b.
[4] Seismic provisions for structural steel buildings., American Institute of Steel Construction,
Chicago, USA, 2016
[5] Sun M., Packer J.A., “Charpy V-notch impact toughness of cold-formed rectangular hollow
sections”, Journal of Constructional Steel Research, 97, 114–126, 2014.
[6] Maranian, P., Reducing Brittle and Fatigue Failures in Steel Structures, American Society of
Civil Engineers, 2010.
[7] Shaw S.M., Kanvinde A.M., Fell B.V., “Earthquake-induced net section fracture in brace
connections-experiment and simulations”, Journal of Constructional Steel Research, 66, 1492–
1501, 2010.
[8] Morrison, M., Schweizer, D., and Hassan, T. (2015). “An innovative seismic performance
enhancement technique for steel building moment resisting connections.” J. Const. Steel Res.,
109(6), 34–46.
[9] Krauss G., Steels: processing, structure, and performance, ASM International, 2005.
[10] Fell, B.V., Large-Scale Testing and Simulation of Earthquake-Induced Ultra Low Cycle Fatigue
in bracing Members Subjected to Cyclic Inelastic Buckling, Ph.D. Dissertation, University of
California-Davis, 2008.
[11] Kanvinde, A. M., and G. G. Deierlein. "Cyclic void growth model to assess ductile fracture
initiation in structural steels due to ultra-low cycle fatigue." Journal of engineering mechanics
133(6), 701-712, 2007.
[12] Kanvinde, A. and Deierlein, G., “Validation of Cyclic Void Growth Model for Fracture Initiation
in Blunt Notch and Dogbone Steel Specimens”, Journal of Structural Engineering, 134(9), 1528–
1537, 2008.
[13] Totten G. E., Xie L., Funatani K., “Handbook of Mechanical Alloy Design” Marcel Dekker Inc.,
New York, 2004

10
 Ninth International Conference on Advances in Steel Structures (ICASS’2018)
5-7 December 2018 - Hong Kong, China

Table 1 Chemical composition of A500 steel

Chemical Composition (wt. %)
Item Fe C Mn P S Al Si Cu Ni Cr Mo V Nb Ti N
A500 Grade B & C
(HSS125x125x12.7)c Bal. 0.07 0.74 0.015 0.001 0.049 0.04 0.02 0.01 0.04 0.01 0.004 0.041 0.002 0.003
c
Square HSS member with 125mm width and 12.7mm wall thickness

Table 2 Mechanical properties of A500 and HT A500 steels

Material Property A500 HT@ 700°C HT@ 800°C HT@ 900°C

𝜎𝑦 a (MPa) 532 434 345 276

𝜎𝑢 (MPa) 565 513 404 368
𝑒𝑢 (% elongation at max. force)b 3.65 12.1 16 22.6
𝑒𝑓 (% elongation at fracture)b 26.9 30 36.2 45.9
Charpy impact energy @ 20°C (J) 20 NM NM 324
𝜎𝑦 ⁄𝜎𝑢 0.94 0.85 0.85 0.75
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝑉𝐺𝐼𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 2.35 3.01 3.63 4.31
a
0.2% offset method used to determine 𝜎𝑦 .
b
Elongation measured over a 50.8 mm gauge.
Average values are reported for all properties.
HT specimens were cooled at 6°C/min to 550°C then air cooled.
NM-Not measured.

Proceedings of the ninth international conference on Advances in Steel Structures

Edited by Siu-Lai Chan, Tak-Ming Chan and Songye Zhu. Copyright © 2018 by The Hong Kong Institute of Steel Construction.
 M.L. Morrison

Table 3 Summary of simulation results

Brace member (Material) 𝜎𝑢 𝑈𝐴𝑛 ⁄𝜎𝑦 𝐴𝑔 𝑃𝑇𝑚𝑎𝑥 ⁄𝑃𝑦 𝑃𝐶𝑚𝑎𝑥 ⁄𝑃𝑦 Normalized θLB θmax (# of cycles)
b/t*
HSS102x102x6.35 (A500) 0.83 1.06 0.65 1.12 1.9% 2.7% (1st)
HSS102x102x9.5 (A500) 0.83 1.05 0.58 0.67 4% 5% (1st)
HSS102x102x6.35 (HT@ 700°C) 1.02 1.11 0.79 1.02 1.9% 4% (1st)
HSS102x102x9.5 (HT@ 700°C) 1.02 1.14 0.71 0.61 NP NP
HSS102x102x6.35 (HT@ 800°C) 1.28 1.10 0.72 0.91 2.7% 4% (2nd)
HSS102x102x9.5 (HT@ 800°C) 1.28 1.10 0.67 0.54 NP NP
HSS102x102x6.35 (HT@ 900°C) 1.60 1.12 0.88 0.80 2.7% 4% (2nd)
HSS102x102x9.5 (HT@ 900°C) 1.60 1.13 0.90 0.48 NP NP
*
Calculated based on 2016 AISC Seismic Provisions [4]; measured yield strength used in calculating limiting b/t ratios.
U (Shear lag factor) taken as 0.85.
𝑃𝑦 = 𝜎𝑦 𝐴𝑔 where 𝐴𝑔 is the gross cross-section area, 𝑃𝑇𝑚𝑎𝑥 and 𝑃𝐶𝑚𝑎𝑥 are the maximum simulated tensile and compressive brace forces
respectively.
𝐴𝑛 is the net cross-section area.
θLB is the story drift amplitude of the cycle during which brace local buckling is predicted.
θmax is the story drift amplitude of the cycle during which fracture initiation is predicted.
NP-Not predicted, analysis terminated after 1 cycle of loading @ 6% drift, and these limit states had not yet occurred.

12

View publication stats