FINAL REPORT (3-11)

EMBEDDED COLUMN BASE CONNECTIONS SUBJECTED TO

FLEXURE AND AXIAL LOAD: TESTS AND STRENGTH MODELS

DAVID A. GRILLI, AND AMIT M. KANVINDE

DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING
UNIVERSITY OF CALIFORNIA, DAVIS, CA 95616
JUNE 8, 2015
ACKNOWLEDGMENTS
The authors are grateful to the Charles Pankow Foundation (Project 3-11) and the American
Institute of Steel Construction for providing major funding for this project. Supplementary
funding was provided by graduate fellowships at the University of California, Davis. The authors
also thank Bill Sluis and Daret Kehlet, technical personnel at UC Davis for contributing to this
project. Greg Deierlein of Stanford University provided valuable feedback, as did Geoff Bomba
of Forell Elsesser Engineers, and Ron Klemencic and Rob Chmielowski of Magnuson
Klemencic Associates. Tom Schlafly acting on behalf of the American Institute of Steel
Construction was the Industry Champion. The authors acknowledge their contributions, as well
as those of the Industry Advisory Committee. The findings and opinions presented in this report
are solely those of the authors and not of the sponsors or those acknowledged above.

2
EXECUTIVE SUMMARY

Steel column base connections in mid- to high- rise moment frame buildings are often designed
by embedding the column into the footing. This enables the development of moment resistance
through horizontal bearing stresses against the column flanges. This type of connection, referred
to as an Embedded Column Base (ECB) connection, is used when exposed base plate
connections (which resist axial force and moment through vertical bearing forces and tension
anchors) are unfeasible because they require multiple deeply embedded anchor rods or thick base
plates. Moreover, ECB connections may also be used when a stiff or “fixed” base condition is
desired. Significant experimental, analytical, and computational research has been conducted on
exposed base plate connections, leading to design documents that assist with their design,
including the American Institute of Steel Construction’s Steel Design Guide One Series, and the
Structural Engineers Association of California’s Seismic Design Manual. In contrast, research on
ECB connections is sparse, with no established guidelines or methods for their design. To
address this, this report presents the results of five full scale tests on ECB connections subjected
to cyclic lateral deformations in the presence of an axial (tensile or compressive) load. The main
variables interrogated include the embedment depth, column size, and the axial load. The
experiments demonstrate that ECB connections are able to provide high strength and stiffness,
and transfer axial force and moment through a combination of three mechanisms: (1) horizontal
bearing stresses acting on the column flanges (2) vertical bearing stresses acting on the
embedded base plate at the bottom of the column and (3) panel shear. It was determined that
although designed as rotationally fixed, the specimens have some flexibility which must be
considered in simulation and design. An assessment of rotational stiffness of ECB connections
is provided. A strength model utilizing the observed mechanisms is developed. The model
leverages prior research on composite connections, steel coupling beams, and exposed base
connections. The strength model is developed to balance the following considerations (1)
consistency with physics, and minimal reliance on empirical factors (2) simplicity of application,
and (3) agreement with experimental data. The model is able to characterize the observed failure
modes as well as experimental strengths with good accuracy. Limitations of the study are
discussed along with directions for future work.

3
INTRODUCTION

Column base connections are critical components in Steel Moment Resisting Frames (SMRFs),

transferring axial forces, shear forces and moments from the entire building into the foundation.

For low-rise frames (< 3-4 stories), these connections typically include a base plate, which is

welded to the bottom of the column and anchored to the footing below with anchor rods – these

are typically referred to as “exposed” type connections. As shown in Figure 1a, these

connections resist base moments and forces through the development of tension in the anchor

rods, along with vertical compressive bearing stresses under the plate. For taller frames, it is

often unfeasible to use this mechanism for resisting moments, since it necessitates the use of a

large number of deeply embedded anchor rods along with base plates that are extremely thick

(e.g. thicker than 75mm). Consequently, structural designers often employ an “embedded” type

base connection (see Figure 1b), in which the dominant mechanism of moment resistance is

anticipated to be direct bearing between the column and the concrete footing. Significant

research has been conducted on the former (i.e., exposed) type of connections, including tests

and simulations (Gomez et al., 2010; DeWolf & Sarisley, 1980; Thambiratnam & Paramasivam,

1986; Astaneh et al., 1992; Fahmy et al., 1999; Burda & Itani, 1999; Myers et al., 2009), as well

as field observations from earthquake damage (Tremblay et al., 1995), resulting in methods for

strength and stiffness characterization (Gomez et al., 2010; Kanvinde et al., 2012) and ultimately

design guidelines (AISC 341-10, 2010; AISC Design Guide One – Fisher and Kloiber, 2006;

SEAOC Seismic Design Manual SSDM – Grilli and Kanvinde, 2013). However, research on

Embedded Column Base (ECB) connections is sparse, and limited to finite element parametric

studies (Pertold et al., 2000a, b). As a result, the seismic performance of ECB connections is not

well understood, with two consequences: (1) there are no established approaches to facilitate

4
their design, such that AISC Design Guide One, and the SEAOC SSDM exclusively address

exposed type connections, and (2) there is limited understanding of the stiffness they provide,

from the perspective of accurately representing them in simulation and design of the moment

frame itself.

Applied loads

(a) (b)
Steel
Column Steel
Base Column
Plate

Grout

Tensile forces Horizontal

bearing
in anchor rods
Vertical stresses on
Bearing flanges
stresses
Concrete Anchor
Foundation Rods Vertical bearing stresses
on base plate

Figure 1 – Column base connections and force transfer mechanisms

(a) exposed type, and (b) embedded type

For the design of ECB connections, practitioners use ad hoc methods based on research on other

structural components that show mechanisms similar to those expected in ECB connections.

Specifically (referring to Figure 1b), these mechanisms include (1) bearing of the joint panel

against the concrete, accompanied by shear in the panel zone, and (2) restraint to uplift of the

base plate by the concrete above, accompanied by bearing under the toe of the base plate. The

latter mechanism is similar to that of exposed connections. The former mechanism, i.e., bearing

accompanied by panel zone shear, is well-documented in the ASCE guidelines for composite

beam-column connections (ASCE, 1994), and associated research (Deierlein et al., 1989; Sheikh

5
et al., 1989; Cordova and Deierlein, 2005). Moreover, research on steel coupling beams

embedded in concrete shear walls also reports on the bearing mechanism to resist moments in the

embedded element. This research includes experimental work by Marcakis and Mitchell (1980),

Mattock and Gaafar (1981), Shahrooz et al. (1993), and Motter (2015).

The latter mechanism (i.e., resistance to uplift of the base plate) has been reported by Morino et

al. (2003), Cui et al. (2009), and more recently, by Barnwell (2015) for shallowly embedded

base connections where a floor slab is cast over an exposed type connection. These studies

inform physical intuition for each of the mechanisms postulated to be active in the ECB

connections. However, applying these for the design/analysis of ECB connections is challenging

for three reasons –

1. There are key differences between geometric and statical aspects of the ECB connections,

with respect to the mechanisms and specimens examined in the previous studies. For

example, the studies on composite connections and coupling beams disregard the effect of

column axial force, which is present in ECB column. Moreover, the degree of concrete

confinement (in the area of flange bearing) is lower in the coupling beams and beam-column

connections, as compared to the ECB connections, which are typically present in wide

footings.

2. The ECB connections include interactions of the bearing/panel shear mechanism, along with

restrained uplift of the base plate. These interactions occur due to shared stress paths and

deformation compatibility between the different mechanisms. Since the various studies have

(to a large extent) examined these mechanisms in isolation, understanding their interactions

is important for developing strength models.

6
3. In any of these studies, even for these isolated mechanisms, the stiffness has not been

examined. This particularly affects the ECB connections, whose stiffness is critical from the

perspective of building performance.

As a result, the ad hoc adaptation of these previous studies to ECB connections in the absence of

experimental data is susceptible to inaccuracies. Motivated by this, this report presents results

from 5 full-scale tests on embedded column base connections. The tests represent typical column

base connections for moment-frame buildings, such that their failure is controlled by interactions

of flexure and axial load. The primary objective of this report is to present the results of these

experiments and to develop seminal knowledge about the seismic response of ECB connections.

Specifically, this includes (1) load resistance and failure mechanisms, (2) strength, (3) stiffness,

and (4) deformation capacity and dissipation characteristics. Once established, an understanding

of these behavioral aspects is used to develop and verify strength models and design approaches

for ECB connections. The experiments all feature a wide-flanged steel cantilever column

embedded within a concrete footing. Figure 2 shows a schematic illustration of the type of ECB

connections used in the test program. All five test specimens are subjected to a cyclic lateral (or

flexural) displacement history in the presence of a constant (compressive, tensile, or zero) axial

force. The tests investigate the role of various parameters affecting the base connection response,

including (1) column size (2) embedment depth, and (3) axial load. These experiments offer: (1)

insights into the physics of the connections, including failure modes to inform the development

of strength models, and (2) experimental data for validation of the developed models. These

insights are leveraged to introduce a strength characterization approach suitable for the design of

7
ECB connections. The large-scale tests are complemented by ancillary tests for the concrete to

effectively interpret results based on measured (rather than specified) material properties.

The report begins by summarizing the current state of practice, and the scope of this

investigation. This is followed by an outline of the experimental program, including the test

setup, instrumentation, loading protocol, and ancillary tests. The experimental response is

discussed in detail, along with a discussion of failure modes, and the associated strength and

stiffness. A strength design model is then presented, along with a discussion of limitations of the

study and suggested areas of future research.

Column

Footing
Stiffener plate

Base plate

Figure 2 – Schematic illustration of Embedded Column Base

connection used in current study, subjected to axial and lateral loads

8
BACKGROUND AND SCOPE

To provide context for the experimental program, it is useful to discuss the assumptions and

philosophy that underpin the design of ECB connections. The intended yield mechanism in

SMRFs involves the formation of plastic hinges at the ends of all beams, and at the column bases

(Bruneau et al., 1997). The hinge at the column base may be accommodated in the column itself,

or in the base connection. As per current design practice, the former is preferred, assuming that

column plastic hinges have greater rotation capacity. The implication is that base connections

(including ECB connections) are designed to develop the flexural capacity of the attached

column (Grilli and Kanvinde, 2013). As a result, ECB connections may be assumed force–

controlled components that are expected to remain predominantly elastic during design level

shaking. In the context of this study, this observation motivates an understanding of connection

response as it pertains to the onset of nonlinearity or damage, rather than the ultimate strength

(which is somewhat secondary). This is especially relevant, since the tested specimens show high

deformation capacity and dissipation, much of which is not mobilized within the current design

framework.

It is commonly assumed that designing ECB connections to be stronger than the column

guarantees sufficient rotational fixity, such that they may be assumed (and simulated as) fixed in

frame models and design. Studies by Kanvinde et al. (2012) indicate that this is not the case for

exposed type connections. Zareian and Kanvinde (2013) noted that overestimating the fixity of

base connections (or simulating the base connections as fixed) results in significant detriment to

building performance, including increased interstory drift and collapse probability. As a result,

9
developing an understanding of the fixity provided by the ECB connections is an important

motivation for this study.

Embedded Column Base connections may be situated within various types of foundations,

including mats, individual pedestals, pedestals connected by grade beams, or pile caps. Types of

response and failure modes peculiar to these types of foundations affect the strength and stiffness

of ECB connections. In this study, the focus is on response modes in the immediate vicinity of

the column embedment, to the extent they can be isolated from those of the surrounding

foundation. Finally, the scope of this study is limited to only one connection detail, which is

similar to the schematic shown in Figure 2. This detail is distinguished by a base plate attached at

the bottom for resisting uplift, and a stiffener plate (similar to a continuity plate) attached at the

top of the footing. Other common details feature a large base plate of the top, shear studs

attached to column flanges, or supplemental anchor rods on the lower base plate. These

variations are not examined, since (1) the study is the first known experimental examination of

ECB connections, such that primary consideration is to develop fundamental insights into their

response, and (2) the selected detail is highly prevalent in construction practice.

EXPERIMENTAL PROGRAM

All tests specimens were cantilever columns loaded laterally in deformation control as per a

cyclic loading protocol. This protocol was applied under a constant axial load (either

compressive, tensile, or zero). The major variables interrogated were (1) embedment depth (2)

column size, including flange width, and (3) axial load. Table 1 summarizes the test matrix,

along with key experimental results (discussed later).

10
Test Setup

All specimens reflect current construction practice. Figures 3a and b show annotated

photographs of the test setup for compressive load (Test #1, 2, 4) and tensile axial load (Test #5)

respectively. Test #3, which has no axial load, does not have the fixtures for introduction of axial

load, which are present in the other tests. Figure 4 shows a schematic illustration the base

connection detail, which is qualitatively similar in all the experiments. The salient features of the

test setup and specimens are as follows –

1. All specimens were cantilever columns that extended either z = 2.84m (for Tests #1 and #2

– refer Table 1) or 3.1m (for Tests #3, #4, and #5) above the surface of a concrete pedestal,

which measured 3.65m X 1.83m in plan. The height of the cantilever was the location of

application of lateral force via a hydraulic actuator. This height is consistent with the

inflexion point of a 4.5m first story (assuming it occurs at 2/3rds of the story height),

suggesting that the tests may be considered approximately “full-scale” in this aspect. The

columns were all A992 Grade 50 (345 MPa), and were designed to remain elastic to force

failure in the base connection.

2. As shown in Figures 3a and b, axial load was introduced through a crossbeam and hollow

hydraulic jacks attached to the top of the column. To apply axial compression, the hydraulic

jacks were positioned as shown in Figure 3a, and were connected to threaded rods which

were fixed to concrete blocks fastened to the floor. This loaded the crossbeam downward.

To apply axial tension, the cylinders were moved near the middle of the crossbeam and

connected to rods that were fixed to the column specimen as shown in Figure 3b. Columns

replaced the thread rods on the sides to support the crossbeam. From the perspective of

interpretation of the test results, two aspects of the mechanism are relevant: (1) the

11
 mechanism was designed to ensure that the axial load was a “follower force,” such that no

additional base moment was induced due to the eccentricity of the axial force. For this

purpose, clevises (shown in Figures 3a and b) were installed at the elevation of top surface of

the concrete footing, and (2) the axial load was held at a constant level throughout the

duration of the test.

3. The embedment depth dembed is defined as the distance between the top of the concrete

surface, and the top of the embedded base plate. The pedestal dimensions and reinforcement

are also illustrated schematically in Figure 4. Referring to the figure, the pedestals were

designed with minimal longitudinal and transverse reinforcement such that the observed

failure modes and strengths were associated (to the extent possible), with the concrete only.

This purpose of this was to facilitate the application of these test results to situations that

were significantly dissimilar in terms of reinforcement quantity and layout.

4. The W-section columns were embedded fully through the depth of the block, with plywood

cast in between the base of the column and the test floor.

5. Plates were provided at the top and the bottom of the embedment region of the column (see

Figure 4). At the bottom, a base plate similar to the ones used in exposed connections was

welded to the bottom of the column; dimensions of these plates for each specimen are

provided in Table 1. The plate served two purposes. First, it allowed the column to be

supported stably as the concrete was poured around it. This is representative of current

construction practice, where the column is often supported on a temporary slab. Second, a

plate at the bottom is typically prescribed by designers to provide resistance to uplift, i.e.,

tensile forces in the column. The weld detail between the base plate and column comprised a

Partial Joint Penetration (PJP) weld with reinforcing fillet welds. A similar weld detail has

12
 been used previously in base plate tests by Myers et al. (2009) and Gomez et al. (2010),

demonstrating excellent performance. The purpose of these welds was to minimize the

likelihood of weld fracture before connection capacity was obtained. To this aim, all welds in

the specimen were toughness rated, fabricated using E70 electrode (480 MPa) minimum

tensile strength, and the Flux Cored Arc Welding (FCAW) process. At the top of the

connection, plates similar to stiffeners were provided between the flanges of the column. The

plates were fillet welded (using 12mm welds) along the contact areas on the web and inner

flange faces of the column. The main purpose of these plates was to provide resistance to

compression in the column; this too is consistent with standard practice. Sometimes,

additional plates are added on the exterior of the flanges for this purpose. However, these

entail additional fabrication costs, and are often not necessary since axial compression forces

in moment frame columns are often low relative to the moments in them.

6. The concrete pedestals were fastened to the laboratory floor using 6 pre-tensioned threaded

rods, 3 on each end of the footing. The locations of these rods were chosen to minimize their

effect on the stress distribution in the vicinity of the column.

Hydraulic jack (axial) Hydraulic jacks (axial)

Lateral actuator
Tension rod

Reaction
U-Frame

Steel column frame

Clevis Base specimen

(a) (b) Base specimen

Figure 3 – Test setup (a) for Tests #1, #2, #4 with axial compression, and (b) for Test #5 with
axial tension

13
Standard cylinder tests were performed on samples taken from concrete pours of all pedestals,

resulting in an average value of f c' = 29.2 MPa, with a standard deviation of 2.6 MPa.

Compression plate similar to stiffener plate

W14x370 (Test #3,4,5)

840m dembed

(a) #3 (10 mm) stirrups (10 hoops)

positioned near tie downs.
Lower embedded plate Present in all specimens
W14x370 (Test #1)
W18x311 (Test #2)

584 mm dembed

508mm

(b)
#6 (19 mm) rebar (4 total)
#5(16 mm) rebar at 6 tie downs, 460mm
top and bottom from edge

1.83m

(c) 3.65m

Figure 4 – Schematic illustration of experimental specimens (a) 762mm embedment

(b) 508mm embedment (c) plan view; common to both

14
Test Matrix

Table 1 summarizes the test matrix. Referring to the table, the parameters varied include: (1) the

embedment depth, (2) axial load, and (3) the column size. The parameter values for each of these

were selected based on a consideration of similarity to full-scale connections, as well as of

limitations of the test setup; specifically –

1. The columns were sized to ensure failure of the connection. As a result, the sections used in

this study are larger than columns that would be embedded in a similar concrete pedestal or

grade beam in practice. In turn, the embedment depths are comparable to those commonly

used for moment frames.

2. The tensile and compressive axial loads were selected to be approximately 10-20% of the

yield capacity of a column that would (in a practical setting) have an embedment depth in the

range of those used in this study, assuming that the embedment depth is based on ensuring

that the connection is stronger than the column in terms of moment capacity. Note that such a

hypothetical column is smaller as compared to the ones actually used in the study.

3. The test matrix may be considered fractional factorial, such that pairs (or trios) of tests may

be used to examine effects of isolated test variables. For example, Tests #1 and #2 provide a

direct examination of the effect of column flange width, whereas Tests #1 and #4 provide a

similar interrogation of the effect of embedment.

Loading Protocol

The deformation history applied to the specimens is expressed in terms of column drift ratio

(similar to inter-story drift angle), defined as the lateral displacement of the column at the

application of the lateral load divided by the distance between the load and the top of the

15
concrete pedestal. For all the tests, displacement-controlled cyclic lateral loading was applied

according to the SAC loading protocol (Krawinkler et al., 2000) to represent deformation

histories consistent with seismic demands in moment frame buildings. Lateral loading was

applied quasi-statically, with a loading rate less than 1.8% drift per minute for all phases of

testing (i.e. less than 0.018 radian rotation per minute). Referring to Table 1, axial loading was

applied to 4 specimens to represent gravity loading and/or tension due to uplift in the column.

Axial loads were applied prior to lateral loading and held constant throughout the test. The axial

loads shown in Table 1 include a correction for the weight of the axial-load system, as well as

the column above the connection area. For tests #1 and 2, the SAC loading protocol was

implemented to failure. For Test #3 and 4, drift amplitudes increased incrementally until

excessive deformations were achieved. Test #5 was terminated prior to completing the loading

protocol due to the slippage of the test block along the test floor and out of plane, and its original

positioning could not be recovered. However, significant damage and inelastic deformation in

the connection was achieved prior to this slippage.

Instrumentation and measurements

For the purposes of performance assessment and model development, primary streams of data

are: (1) lateral displacement at the top of the column and associated actuator force, (2) axial force

(3) displacement transducers to measure rocking of the block, and (3) embeddable concrete strain

gages in the bearing zone directly ahead of the column flanges, whose purpose was to facilitate

characterization of the bearing stress profile in the concrete. Additional transducers were

installed to detect out of plane motion of the column as well other unanticipated response. The

next section discusses experimental results.

16
Table 1 – Test matrix and results

Column
Base Plate, b base
test *
Test Size, z ∆ test y
M base
# ( [kN] [m] [105 kN- *** (%)
[mm]
[mm] [kN-m] m/rad] ∆ fixed max
M base
[mm])

W14x370 51 762 2579(+) 0.87 3.85

1 3.23 1.21
(419) 445 762 2613(-) 0.69 3.82
508 2.84
W18x311 (C) 51 864 2324(+) 0.709 3.01
2 3.84 1.16
(305) 711 2168(-) 0.66 2.89
3741(+) 0.72 6.97
3 0 3.07 1.30
3444(-) 0.67 7.77
W14x370 445 51 762 4124 (+) 0.66 6.48
4 762 3.10 3.38 1.30
(419) (C) 762 3612(-) 0.81 5.09
667 3800 (+) 0.73 2.72**
5 3.25 1.29
(T) 3464(-) 0.72 2.65**
Mean 1.25 0.72 4.98
CoV 0.05 0.07 0.38
*
Average stiffness of both directions
**
Test terminated due to slip prior to failure (Mean, COV does not include these data points)
***
Notional columns used for Tests #1-5 are W14X145, W14X132, W14X193, W14X211, and W14X193
respectively.

17
EXPERIMENTAL RESULTS

The discussion of experimental results is presented in four parts. First, the damage progression

leading up to failure is described. Second, trends in response with respect to test variables are

identified. Third, based on the observed response, a discussion of connection physics and failure

modes is presented, with the objective of informing development of strength models. Finally, the

stiffness of the connections is critically evaluated, in the context of its anticipated effect on

building performance.

Damage progression

Figures 5a-e show the moment-drift plots for all the specimens, whereas Figures 6 and 7 show

schematic and photographic illustrations of damage and failure. As discussed previously, where

applied, the axial load was introduced at the beginning of the test. The introduction of axial load

did not produce any visually observable response in the specimens. Once lateral loading was

introduced, all experiments followed a qualitatively similar progression of damage, with some

variations. The common aspects of response are described first, before discussing the response

peculiar to specific tests.

Figure 6a schematically illustrates damage observed during initial stages of loading (i.e., when

the applied drift was less than ≈ 1%). Small cracks (approximately 2mm wide and extending up

to 70mm long) began to form near the corners of the column almost immediately after the

introduction of lateral load. However, this did not affect the load-deformation response, such that

linear elastic response was observed until drifts of 0.005 radians (i.e. 0.5%). Subsequent to this,

gradual nonlinearity in the load deformation curve was observed, accompanied by the opening of

18
a small gap adjacent to the tension flange, accompanied by the growth of the diagonal cracks

described previously. This nonlinearity was accompanied by strength degradation as well as

pinching response. The pinching response may be attributed to the gapping shown in Figure 6b

resulting in relatively unrestrained movement of the column within its “socket,” as it moved

through the vertical position. Not surprisingly, the extent (i.e., displacement range) of this

pinching region increased as the size of the gaps grew to widths exceeding 30mm. The

photograph in Figure 7a (taken at the end of Test #3) shows such a gap.

Initial spalling of the concrete ahead of the column flanges was observed at approximately 1%

drift. As the loading history progressed beyond this point, the load displacement curve became

highly nonlinear. In general, it was observed that the Tests #3-5 with deeper embedment (i.e.

d embed = 762mm) showed a more gradual decrease in stiffness, as compared to the more

shallowly embedded ( d embed = 508mm) specimens. This nonlinearity was accompanied by three

types of visible damage (shown in Figures 6b and c for Tests #3-5 at 3-4% drift), which

increased in severity (i.e., longer, wider cracks) along with increasing drift –

1. Cracks radiating diagonally outwards from the corners of the column on the top surface of

the pedestal (Figures 6a and b), accompanied by slight upward bulging of the concrete in the

bearing zone between these cracks. On tension side of the column, significant gapping was

observed. This width of this gap was as large as 40mm prior to failure for all experiments

(except Test #5) – see Figure 7b.

2. Flexural cracks observed on the sides and the top of the pedestal, on the tension side of the

connection (Figure 6b). These cracks were parallel to the flanges of the columns, and were

produced by bending of the entire concrete block. These cracks were most prominent in Test

19
 #5, which had the tensile axial load, and least prominent in Tests #1, 2, and 4 which had

compressive axial load, presumably because the compressive force inhibited the uplift of the

block from the strong floor.

3. Diagonal shear cracks on the sides of the block (see Figure 6b), which appeared as straight

cracks parallel to the flanges on the top surface.

All these cracks opened and closed as the loading direction was reversed, and they grew in width

and length as the applied drift was increased. The damage patterns shown in Figures 6a-c are

consistent with drift in one direction, and hence appear unsymmetric, although they are equally

pronounced in both directions of loading. In all the tests, the peak moment was achieved between

1.5 and 5% drift. After this, the strength at successive cycle peaks began to deteriorate. This may

be attributed to the reduction in the moment resisted by bearing ahead of the flanges as the

concrete spalled at the extremities of the bearing block (i.e. at the top and bottom). This process

continued until failure of the pedestal (defined as drop in load of at least 30% of the peak load)

occurred. In one of the experiments (Test #6), a failure point could not be obtained, owing to

irrecoverable slippage of the testing rig. Nevertheless, significant data, including a possible peak

load was obtained. In the other tests, one of two scenarios occurred. Shown photographically in

Figures 7a and b, these are –

1. In Test #1 and #2, with the shallower embedment (i.e., d embed = 508mm), final failure was

accompanied by sudden uplift of a cone of concrete on the tension side of the connection.

See photograph in Figure 7a, and schematic illustration in Figure 6c. Referring to the

introductory discussion of this report, the base moment is shared by the bearing mechanism,

and the restraint to uplift of the base plate at the bottom. As the bearing mechanism becomes

20
 less effective (due to the spalling of concrete, and the associated gapping), a greater fraction

of the base moment is resisted by restraining the uplift of the base plate. When the uplift

force due to this moment reaches a critical value, failure of the type seen in Figure 7a is

observed.

2. For Test #3 and #4 (with the deeper, 762mm embedment), the failure was more gradual, as

increasing deformations were accompanied by a steady drop in load. Referring to Table 1,

this failure occurs between the drifts of 2.2% and 5% for Tests #3, and 4. The steady drop in

load may be attributed to the gradual crushing of concrete ahead of the compression flange

(which also result in the gap behind the tension flange), which reduces the effective lever arm

of the bearing stress. This results in a pattern of widespread cracking damage (as shown in

Figure 7b), rather than the abrupt failure shown in Figure 7a.

In interpreting the discussion above, it is helpful to recall that the columns in the tests were

artificially strong (to force connection failure), and hence stiffer as compared to realistically

sized columns (which are weaker than the connection). For these hypothetical columns, the

elastic cantilever deflections will be larger, resulting in correspondingly greater drifts for each of

the observed damage states.

21
 4500 4500
Test 1 Test 2
3000 3000
Mbase (kN.m)

Mbase (kN.m)
1500 1500

0 0

-1500 -1500

-3000 -3000
(a) (b)
-4500 -4500
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Column drift (%) Column drift (%)
4500 4500
Test 3 Test 4
3000 Mbase (kN.m) 3000
Mbase (kN.m)

1500 1500

0 0

-1500 -1500

-3000 -3000
(c) (d)
-4500 -4500
-8 -6 -4 -2 0 2 4 6 8 -8 -6 -4 -2 0 2 4 6 8
Column drift (%) Column drift (%)
4500
δ V
∆=δ/z 3000
Test 5
Mbase (kN.m)

1500
z
0
Mbase = V × z
-1500

-3000
(e)
-4500
-4 -3 -2 -1 0 1 2 3 4
Column drift (%)

Figure 5 – Moment drift plots for all experiments, and schematic illustration of
plotted quantities

22
 Applied lateral load (axial load not
shown)

Reactions from tie-downs

Reactions from
(a) strong floor

Diagonal cracks on
tension and
compression side
Crushing ahead
of compression
flange
Gap between tension
flange and concrete

Shear cracks in
Flexural cracks
concrete panel
on tension side
(b)

Uplift of shaded region

accompanied by
horizontal crack

(c)

Figure 6 – Typical damage evolution (a) below 1-1.5% drift (b) above 1.5% drift (c) sudden
failure mode observed in Tests #1 and #2

23
 40mm gap Crushing

Damaged footing,
gradual loss of strength

Horizontal crack and

uplift (see Figure 5c)
(a) (b)

Figure 7 – Observed failure modes (a) Test #2; representative of Tests #1,2 (b) Test
#3; representative of Tests #3,4.

Effect of test variables on damage progression

Table 1 summarizes key data measured in the experiments. For each specimen, two strength

values are recovered, one corresponding to the maximum moment measured in each direction of
max + max −
loading. These are denoted as M base and M base , where the positive sign denotes the direction

max
of application of the first deformation cycle. The symbol M base without the sign is generically

used to represent the average of these two values. Specifically, the following observations

emerge upon a closer inspection of Table 1, and Figures 5, 6, and 7 –

1. Referring to Table 1, it is immediately apparent that the specimens with deeper embedment

have greater strength. Moreover, the ultimate failure mode is different for specimens with the

two embedment depths. Specifically, Tests #1 and #2 show a concrete failure cone due to the

uplift of the base plate at the bottom of the connection. Tests #3 and #4, have a greater depth

of concrete above the base plate, and cannot mobilize this failure mode. Instead, they show

24
 gradual strength deterioration as the effective bearing depth of the column decreases due to

spalling/crushing of the concrete at the extremities of the embedded region. Note that even in

case of Tests #1 and #2, the strength of the connection is achieved prior to the mobilization

of the ultimate failure mode, and is unaffected by it.

2. Most of the experiments are stronger in the positive direction, such that an average value of
max + max −
M base / M base is 1.08, with a standard deviation of 0.051. This suggests that damage caused

by loading in the positive (forward) direction affects the strength in the negative (reverse)

direction which is loaded subsequently for each cycle.

3. Comparing Tests #1 and #2 (which are similar in terms of embedment and axial load, but

differ in terms of column section and base plate geometry), the 12% higher strength of the

specimen in Test #1 may be attributed to the wider flange of the W14 X 370 (420mm), as

compared to the flange of the W18X311 (305mm), because the width of the bearing area is

proportional to this. Moreover, the moment resisted through the bearing mechanism (relative

to the moment resisted through the restraint of base plate uplift) is dependent on the stiffness

of the embedded column (a highly flexible column will transfer less moment to the base), and

this may be considered another factor in the difference between the observed strengths of

Tests #1 and #2.

4. A comparison between Tests #3, #4 and #5 provides a direct assessment of the effect of axial

force. Interestingly, application of tensile force (Test #5) appears to have only a modest

effect on the flexural strength. On the other hand, the application of a compressive force

increases the capacity to a greater degree (comparing Tests #3 and #4).

max
5. While the peak moment M base is an important quantity, it is attained after significant

inelastic deformation and damage have already occurred. As a result, it may not be suitable

25
 for design of the base connection, especially if it is expected to remain elastic in a “strong-

base-weak-column” design framework, which reflects current practice. On the other hand, it

is challenging to quantify a “yield” moment, since (1) nonlinear response, albeit modest, is

observed even at low moments, and (2) the slope of the backbone curve changes

continuously, rather than suddenly, with respect to increasing deformations. Consequently,

y
the term M base is determined objectively through least-squares fitting of a bilinear curve to

the backbone curve obtained from the test, with the condition that the bilinear curve is

anchored to the origin, as well as the ( ∆max max

base , M base ) point on the backbone curve (which can

be determined uniquely), where ∆max

base represents the drift corresponding to maximum

y
moment. This results in two free parameters, i.e. the “yield” moment M base and

corresponding deformation ∆ ybase . These quantities are summarized in Table 1 along with

y
other test results. Referring to the values of M base in the Table, the average value

y max
M base / M base is 0.72, suggesting a fraction of the ultimate strength that may be suitable for

design. For purposes of convenience, this may be taken as 0.70.

6. All experiments, with the exception of Test #5, achieve deformation capacities in excess of

3% drift (although some strength loss is observed before this deformation). Note that the

column for these experiments was disproportionately strong (to induce connection failure); as

a result, the elastic deformations of the column were lower than those anticipated for a

column in a similar base connection. For reference, the footnotes of Table 1 summarize a

notional column that would be consistent with the connection strengths observed in the

experiments. In any case, the implication is that the deformation and hysteretic characteristics

26
 of these connections may be considered excellent when evaluated relative to deformation

demands in design-level shaking, which are in the range of 2-3% interstory drift.

Connection physics and failure modes

Based on the preceding discussion, and examination of strain gage data (including the embedded

strain gages), Figures 8a and b schematically indicate the postulated internal force distribution,

and failure modes expected in the type of ECB connections tested in this study. It is assumed that

compressive axial force is carried by the top stiffener plate, skin friction along the column, and

the bottom base plate. Tensile axial force is carried by skin friction, and downward bearing on

the bottom base plate. Referring to Figure 8, the base moment is resisted through a combination

of horizontal bearing stresses against the flanges of the column (Figure 8a) and vertical bearing

stresses against the lower base plate (Figure 8b). The horizontal bearing stresses are

accompanied by shear in the panel zone (similar to panel shear in composite connections –

Cordova and Deierlein, 2005). The panel zone consists of the steel web, a compression strut

between the flanges of the column, and a compression field (and complementary tension field) in

the concrete panel outside the flanges. The tension field is responsible for the diagonal shear

cracks observed in Figures 6b and 7b. The column shear is also carried by the horizontal bearing

stresses. These physical mechanisms of response may be used to develop strength models for

ECB connections. However, for the strength characterization method to be general (such that it

may be applied to connections different from the ones tested), consideration of the following

issues is critical –

1. Quantification of failure strength of sub-mechanisms: Each of the three response

mechanisms discussed above (i.e., horizontal bearing, vertical bearing, and joint shear) is

27
 associated with local failure modes (or sub-mechanisms). For example, the vertical bearing

mechanism may result in uplift of the concrete (as observed in Tests #1 and #2), or yielding

of the lower base plate, or even breakout of the concrete under the compression toe of the

base plate if the supporting layer of concrete overlaying the soil is thin. The two latter

mechanisms were not observed in the experiments, but they are possible in connections that

are sized or designed differently. Similarly, joint shear is controlled by shear yielding of the

web, as well as compression strut, and compression field action in the concrete (Cordova and

Deierlein, 2005; Sheikh et al., 1989). For the horizontal bearing mechanism, concrete

crushing is the likely failure mode. Strength checks for each of these mechanisms are

necessary in any strength model. Previous research (e.g., Mattock and Gaafar, 1981;

Marcakis and Mitchell, 1980 – for coupling beams; Cui et al., 2009 – for restrained uplift of

base connections; Cordova and Deierlein, 2005) may be suitably used to inform some of

these strength checks.

2. Quantification of interaction between mechanisms: An understanding of interaction between

the various mechanisms is required to develop a successful strength model. Broadly, the

interaction is active in two ways. First, the distribution of applied forces (i.e., moment)

between the three mechanisms is complex. For example, it may be postulated that the

horizontal bearing mechanism is “in series” with joint shear (similar to composite

connections – Cordova and Deierlein, 2005), i.e., the forces between these mechanism

equilibrate. On the other hand, the horizontal bearing and joint shear mechanisms are “in

parallel,” such that they are constrained through deformation compatibility, resulting in an

indeterminate distribution between the two. A quantitative understanding of these force

distributions is necessary to relate strength checks for sub-mechanisms to the overall

28
 connection strength. Second, some failure modes may interact due to common stress paths.

For example, the concrete that resists uplift of the base plate is also subject to horizontal

bearing stresses in the lower region of the column embedment. Thus, it is anticipated that the

capacities of the sub-mechanisms may be influenced through interaction with other modes of

response.

(a) (b)
Vertical bearing
stresses on
Shear strut in stiffener plate
Skin
panel zone friction

Horizontal Vertical bearing

bearing stresses stresses on base
plate

Figure 8 – Load resisting mechanisms (a) horizontal bearing and

panel shear accompanied by (b) vertical bearing and skin friction

In summary, a successful strength characterization approach will include: (1) strength checks for

failure sub-mechanisms, considering interactions between them, and (2) quantification of force

distributions between the response modes, to enable the determination of overall connection

strength. Finally, the strength model would address the issue that while the failure modes are

associated with ultimate strengths, designing the ECB connection based on ultimate strength will

imply significant inelastic action in the connection itself. This is inconsistent with the intent of

current design practice (AISC, 2010), in which ECB connections are designed to remain elastic

forcing the plastic hinge into the attached column.

29
Assessment of connection stiffness

Referring to the discussion above, ECB connections designed as per current practice are

expected to remain elastic. However, since they are designed based on strength checks, their

rotational fixity is not explicitly determined or considered in design of the frame.

Characterization of the rotational stiffness of ECB connections is critical from the perspective of

ensuring acceptable performance of moment frames. Through parametric simulation, Zareian and

Kanvinde (2013) demonstrated that base flexibility has a major effect on various aspects of

seismic response, including interstory drift, as well as the shaking intensity associated with

collapse. Specifically, simulating the bases as rotationally fixed results in response that is less

adverse (i.e., lower interstory drift, high intensity required to trigger collapse) as compared to

simulating the bases as flexible. Within this context, this section examines the degree to which

the ECB connections tested in this study provide rotational fixity.

An objective evaluation of the rotational fixity is challenging for three reasons. First, referring to

Figures 5-e, the base connection response is nonlinear, even in the early stages of loading.

Consequently, the initial tangent stiffness is not a suitable representation of the effective stiffness

during a design level event. Second, the rotational stiffness of the base interacts with the entire

frame to result in response such as interstory drift. As a result, examining the stiffness of the base

in isolation is not informative, unless this interaction is considered. Third, since the idealized

situation (fixed base) has infinite base stiffness, it is not meaningful to use this as a normalizing

(or benchmarking) value for the measured stiffness. To overcome these issues, the following

process is established for a convenient, yet meaningful, evaluation of base fixity –

30
1. For all tests, the moment-drift plots shown in Figures 5a-e are converted to moment-rotation

plots for the base connection. The interstory drift angle is converted to the base rotation using

the following relationship –

M base × z
θ base = θ drift − − θ footing (1)
3 × Esteel × I column

In the above equation, the second and third terms on the right hand side subtract the

contributions of (1) the elastic rotation due to flexibility of the column, and (2) rotation due

to rocking/uplift of the footing. The latter (determined to be negligible) was measured by

vertical displacement transducers attached to the top surface of the footing block, a

significant distance away from the zones of damage near the column. Recall that the columns

were oversized to force failure in the base, with the implication that the interstory drifts

measured in the tests were larger (for it was stiffer than a column that would be sized in

practice, which would be weaker than the base). Isolation of the base rotation in this manner

enables interpretation of the results in general way, even for columns that are sized

differently. Figure 9 illustrates the moment vs base rotation plot generated (for Test #1,

positive loading direction) in the manner outlined above. Similar plots are generated for all

the experiments.

2. Figure 9 also illustrates the determination of the secant stiffness. As indicated on the figure,
y
the secant stiffness is determined at a base moment corresponding to M base = 0.7 × M base
max
.

Referring to prior discussion, this value is consistent with the expected moment in the base,

assuming it is designed to remain undamaged as the column develops a plastic hinge.

y
Consequently, the secant stiffness at M base is an appropriate parameter for evaluation within

a seismic setting. The use of the secant stiffness (at the design level) for connection rotation

31
 is not without precedent; notably used in the context of partially restrained moment

connections (Bjorhovde, 1988; ASCE Task Committee, 1998). The secant stiffness,

calculated in this manner is denoted b base

test
, and summarized for all specimens in Table 1.

3. In itself, the value b base

test
is not meaningful, unless interpreted in the context of its influence

on structural response (such as first story drift). However, structural response is a result of

indeterminate interactions between the base stiffness and the frame stiffness. To simplify this

issue while retaining the key aspects of structural behavior, substructures (such as the one

shown in Figure 10) are generated as counterparts to each of the experiments. Each

substructure has the following key features –

a. The column in the substructure is notional, and assigned a stiffness that is consistent

with that of a realistically sized column for a given base connection. To achieve this

for each test, a W14 section which has moment capacity M p ≈ M base
y
is selected, and

the corresponding moment of inertia is used in analysis of the substructure. In this

way, the biasing effect of the disproportionately strong and stiff column is mitigated.

b. The rotational spring at the top represents the effective rotational stiffness β frame at

the top of the column due to adjacent framing members. This value is determined as

per the following equation –

1 6 × E steel × I beam
b frame = × 2 × (2)
2 Lbeam

In the above equation, I beam and Lbeam are properties of notional beams framing into

the top of the column. The beam sizes are selected based on strong-column weak

girder checks with respect to the notional column. The factor “6” on right hand side

assumes double curvature bending of the beams, whereas the factor “2” accounts for

32
 the presence of two beams. The factor “1/2” accounts for the fraction of beam

stiffness that contributes to the restraint of the lower column, assuming the other half

of beam stiffness restrains the upper column. It is acknowledged that these

assumptions are approximate, and that true response of the frame may be significantly

different from that implied in the above approach for the following reasons: (1)

member sizes may deviate from the ones assumed in this study, and (2) frame

response may deviate from the idealized response (which assumes beams in double

curvature and first mode response). However, the primary objective of this analysis is

to provide a general basis for interpreting the measured stiffness of the ECB

connections, rather than to precisely estimate story deformations. In this context, the

subjectivity introduced by the approximations is acceptable. Sizes of the notional

beams and columns (that are consistent with realistic design, given the base

connection strength) are summarized in the footnotes of Table 1.

c. If the stiffness β frame is suitably assigned, then the lateral deformation (at the top of

the column) for a unit lateral load (i.e., the first story flexibility) may be determined

through elastic structural analysis for any value of the base stiffness b base . Two such

values are calculated for each experiment. One, termed ∆ fixed represents the story

drift (or story flexibility) for a fixed base; this is determined by setting b base = ∞ . The

second, termed ∆ test , is determined by setting b base = b base

test
. The index ∆ test ∆ fixed

reflects the increase in interstory drift due to flexibility of the base, relative to the

fixed base. A value ∆ test ∆ fixed = 1 , indicates that the base flexibility does not affect

33
 interstory drift (relative to the fixed base assumption), whereas larger values indicate

greater influence. Table 1 summarizes ∆ test ∆ fixed values for all the experiments.

Referring to the values of ∆ test ∆ fixed , an average value of 1.25 is determined suggesting that the

ECB connections are somewhat flexible, such that the first story drift is (on average) 25% larger

than that determined through elastic analysis of a fixed base condition. The specimens with the

deeper (762mm) embedment have a greater average value of ∆ test ∆ fixed (1.30 for Tests #3, 4, 5)

as compared to Tests #1, 2 with the shallower 508mm embedment, such that average ∆ test ∆ fixed

is 1.19. This is somewhat surprising, since a deeper embedment may be expected to provide

higher rotational stiffness. However, this trend may be explained by considering that the

embedded base plate also provides significant restraint to rotation since it is restrained by

concrete. The observed trend suggests that this stiffness provided by the base plate is large

enough, such that reducing the embedment has the effect of lowering the effective bending

length of the column beneath the top of the concrete surface. On the other hand, in the specimens

with deeper embedments, the additional restraint provided by bearing in the concrete does not

offset the increased flexibility due to the greater bending length. In either case, the main

implication of the above observations is that ECB connections may not provide a fixed base

condition, even when designed to be stronger than the column. However, the 20-30% increase in

interstory drifts is relatively modest compared to that generated by the flexibility of exposed type

connections (Zareian and Kanvinde, 2013).

34
 3000

Mbase (kN.m)
2000

1000

0
0 0.01 0.02 0.03 0.04
θbase (radians)

Figure 9 – Schematic illustration of base stiffness calculation, shown for Test #1

Unit load

Idealization of
ECB subassemblage
connection

Figure 10 – Subassemblage analysis for benchmarking base stiffness

35
MODEL FOR STRENGTH CHARACTERIZATION

Based on the observations of the experimental program, this section describes the development

of a strength model to facilitate the design of ECB connections. The model is based on two

competing considerations (1) to reflect critical aspects of physics and internal force transfer, such

that it may be generalized to ECB connections that have not been tested, and (2) to limit

cumbersome or complex mathematical expressions or analysis procedures, such that it may be

applied conveniently within a practical setting. The model is presented in four parts. First, the

problem is rigorously defined to eliminate ambiguity in interpretation of the model. Second, an

idealized representation of internal force transfer is presented; this forms the physical basis of the

model. Third, strengths for each failure mode (or submechanism) within the connection are

discussed, leveraging previous research on similar components. Finally, a scheme is presented

for combining these various submechanism strengths into the overall connection strength.

Problem definition

The strength characterization method aims to characterize base moments associated with various

limit states (and modes of response), given the ECB geometry and applied axial force (whether

axial or tensile), and shear to moment ratio. As a result, the parameters defining the problem may

be listed as the column embedment depth d embed , the column section, axial force P , base plate

dimensions t plate , B, N , and the shear–to–moment ratio, such that for a given base moment M base ,

the corresponding column shear Vcolumn may be determined as a fraction of it. In addition, the

dimensions of the footing as well as material properties of the steel of the base plate Fyplate and

the concrete, i.e., f c' are inputs to the problem.

36
The problem statement may be defined as follows: (1) given the parameters above, determine the

base moments M base , at which each of the possible limit states is attained, and (2) based on these,

determine the ultimate (or design) moment that can be sustained by the ECB connection. Note

that in the above problem statement (and hence the model formulation), the axial force is

considered a given constant quantity, and the limit states are evaluated with respect to base

moments applied in the presence of this axial force. This implies that the base plate footprint and

thickness, as well as the embedment depth have been selected a priori such that failure does not

occur under (tensile or compressive) axial force by itself. This ensures that a finite moment

capacity is available, and recognizes that a design procedure will likely involve preliminary

sizing for one type of force (e.g. axial force), and then iterations to achieve the desired moment

capacity. Methods for sizing the base plate and embedment depth for pure axial load are readily

available in Fisher and Kloiber (2006). Second, this approach reflects the manner of testing,

where the axial force is introduced prior to lateral load. Consequently, the method is based on the

implicit assumption that the strength is insensitive to the order of application of loads.

Idealization of internal force transfer

Figures 11 and 12 schematically illustrate the idealized internal forces associated with force and

moment transfer from the column into the footing. Specifically, Figure 11 shows compressive

axial load transfer, and Figures 12a and b show moment transfer. A corresponding figure for

tensile axial force transfer is not shown, since this is more straightforward to visualize as

compared to that for compressive axial force, because all the tensile force is resisted through

downward bearing stresses on the lower base plate, since the upper stiffener plate is inactive. The

complete force transfer in the connection may be visualized as a superposition of Figures 11,

37
12a, and 12b. Some of the notation and quantities indicated on the figures will be introduced as

the strength method is described in subsequent discussion. The assumed patterns of force

transfer, and contributing joint components are based on a combination of (1) experimental

observations of failure and strain gage data, (2) quantitative agreement of resulting strength

estimates with test data, (3) examination of previous studies on similar components, and (4)

continuum finite element simulations that provide insight into internal force transfer. A detailed

discussion of these is presented in Grilli (2015). It is acknowledged that these assumed patterns

are idealizations of more complicated stress patterns within the connection. Nevertheless, they

are necessary from the standpoint of developing an analytical strength model.

Some assumptions inherent to the force distributions shown in Figures 11 and 12, and

subsequently used for model development are now summarized –

1. If the axial load is compressive, it is resisted through both the top stiffener plate as well as

the bottom base plate. This mechanism is shown in Figure 11. The axial load resisted through

the top stiffener plate is distributed through a compression into the footing in a manner that is

dependent on the geometry and boundary conditions of the footing. For the test specimens,

(and similar configuration) it is assumed that the footprint of this compression field at the

elevation of the lower base plate corresponds to the plan dimension of the lower base plate.

This is a conservative assumption, since it increases the estimated stresses on the lower base

plate. In reality, the compression field (depending on the depth of the footing, and boundary

conditions) will be distributed over a larger footprint. Nevertheless it is expedient from the

perspective of design development because (1) it provides a convenient way to idealize the

vertical stress distribution, especially on the lower plate which resists the applied moment

38
 and axial load through vertical bearing stresses, and (2) the marginal benefit of assuming a

more complex stress distribution is low, given that the net axial load in common moment

frame columns is low relative to the moment.

2. If the axial force is tensile, it is assumed that it is resisted entirely through downward bearing

stresses on the lower base plate, since the upper stiffener plate is not active. In both cases,

i.e., tensile or compressive axial load, the contribution of skin friction is ignored because

lateral loading results in loss of contact between the column flanges and the concrete, at even

low levels of deformation. This mechanism is not illustrated graphically.

3. Referring to Figure 12b which shows an exploded view of the moment transfer, the applied

base moment M base is resisted through the column flange forces at the top of the connection,

top
such that F flange = M base / h . A portion of this moment, termed M HB is resisted through

horizontal bearing stresses against the joint panel, whereas the remainder (termed M VB ) is

resisted through vertical bearing stresses on the lower base plate. This implies that the flange

bottom
forces at the bottom of the flanges are F flange = M VB / h , and the vertical shear force on the

joint panel is V j = F flange

top
− F flange
bottom
. A compression strut induced in the panel zone resists this

shear.

4. It is assumed that the effective width of the joint panel b j = (b f + B) / 2 , where b f is the

column flange width, and B is the width of the lower base plate. This reflects the

development of bearing stresses over a width greater than the column flange, since a portion

of the concrete panel outside the flange is mobilized through the development of a

compression field. The assumed width of the panel zone affects the joint shear strength, as

well as the width over which the horizontal bearing stresses are distributed. In composite

39
 beam column connections, the panel zone width is assumed to consist of an “inner joint,”

mobilized by face bearing plates in the steel beam (analogous to the top stiffener plate), and

an “outer joint,” mobilized through compression field action outside the width of the column

flanges or face bearing plates. (Deierlein et al., 1989). Unlike composite beam column

connections, the ECB connections tested in this study have plates of different width at the top

(stiffener plate), and the bottom (lower base plate). The effective joint width is assumed to be

the average of these two.

Once the internal force transfer has been established as described above, two additional steps are

required before the connection strength can determined. First, the capacities corresponding to the

failure modes of force transfer mechanisms illustrated in Figures 12a and b are characterized.

capacity
Second, these capacities are combined into a net connection moment capacity M base , with

consideration of (1) the nature of the failure modes (i.e. ductile or brittle), and (2) the evolution

of the relative contribution of the different force transfer mechanisms.

Axial force
conservatively assumed
to transfer to lower plate
elevation through
compression field, and
panel zone compression
force

Figure 11 – Idealized force transfer for compressive axial load

40
 (a)

(b)
Resultant of
bearing block
acting at
Effective depth from
Resultant of of steel web
bearing block panel in shear top
acting at
from
bottom

b × b1 × b j
× f c' × d L Concrete compression strut

Figure 12 – Idealization of moment transfer (a) overall equilibrium (b) exploded detail

41
Failure modes and associated moment capacities

Referring to prior discussion, the base moment is resisted due to horizontal and vertical bearing

stresses, such that M base = M HB + M VB . The resistance due to each of these components is now

discussed.

Moment resistance due to horizontal bearing stresses

To characterize the failure modes corresponding to moment transfer through horizontal bearing

stresses, it is useful to consider the free body diagram of the panel zone region (see Figure 12b).

Referring to this free body diagram, a portion of the applied moment as shown in Equation (3)

below is transferred to the panel zone though the joint shear –

(
M HB = M base − M VB = F flange
top
− F flange
bottom
)× h = Vj × h (3)

This moment is counteracted by the horizontal bearing stresses in the stress blocks also shown in

capacity
Figure 12b. A limiting value of M HB , defined as M HB may be determined based on one of

two failure modes, i.e., horizontal bearing failure, or joint shear failure.

bearing
First, the moment corresponding to bearing failure M HB (which is one possible value of

capacity bearing
M HB ) is determined. For this, a trial value of M HB is substituted in force and moment

equilibrium equations for the panel zone, which are constructed from the assumed stress blocks

( dU and d L are the depths of the upper and lower stress blocks respectively, as shown in Figure

12) –

Vcolumn = b × b 1 × f c' × (d U − d L ) × b j (4)

and,

42
 bearing  (
d 2 + d U2 
= b × b 1 × f c' × b j × d L × d effective − L
)
M HB  (5)
 2 

The above equations rely on three assumptions. First, the bearing stress is assumed equal to

β × β 1 × f c' , in which β = 2.0 simulates the effect of confinement, and the factor β1 = 0.85 is

used to establish equivalence between a rectangular stress block (implied in the equations

above), and the idealized parabolic stress distribution which has a peak stress of 2 × f c' . These

factors are consistent with the values used by Mattock and Gaafar (1981), Sheikh et al. (1987)

and Deierlein et al. (1989); and subsequently in the ASCE Guidelines for composite connections

(ASCE, 1994). Second, as discussed previously, the width of the joint b j = (b f + B) / 2 , to

account for the development of the concrete panel outside the flange through the development of

a compression field (which is influenced by the reinforcement in this region, and the base and

stiffener plates). Third, for the purposes of transferring horizontal forces, the embedment is

assumed to be effective only to a limiting value defined as d effective . The following expressions

(Equations 6 and 7), based on analytical derivations by Hetenyi (1946) are proposed for

calculating d effective –

C0
d effective = d ref ≤ d embed , where d ρef = (6)
ρ0

bf × l
14
 
ρ 0 =  
 (7)
 4 × E steel × I column 

In the above equations, C 0 is a constant, whereas the other symbols reflect the stiffness of the

steel column, and the stiffness of concrete surrounding the column. Specifically, the term b f × λ

is the resisting stiffness of the concrete per unit length (of the column), such that λ is the spring

43
stiffness of the concrete per unit area in the horizontal direction. The derivations by Hetenyi

(1946) upon which these equations are based derive bearing stresses (as a function of depth) for

an end-loaded beam embedded in a semi-infinite elastic medium. These derivations, as well as

experimental and computational studies on columns embedded in elastic media such as soil or

concrete (Pertold et al., 2000a, b, Hutchinson et al., 2005) indicate that the a large fraction of the

moment is resisted by bearing stresses in the vicinity of the free surface, and the bearing stresses

attenuate away from this surface. The degree of this attenuation is dependent on the flexural

stiffness of the embedded beam relative to the elastic stiffness of the surrounding elastic

medium. Hetenyi (1946) demonstrated that the characteristic distance for this attenuation d ref

takes the form shown in Equations 6 and 7. As per these equations, a stiff column (relative to the

surrounding medium) results in a low value of ρ 0 , with the consequence of increasing d ref . On

the other hand a flexible column results in a low d ref , indicating that the stresses attenuate

rapidly within the stiffer surrounding medium. Referring to equation 6, the depth d effective is equal

to d ref , with a maximum possible value d embed , the physical depth of embedment. The classical

(Hetenyi, 1946) derivation assumes a canonical semi-infinite half plane within which the column

is embedded, without the influence of boundary conditions. The ECB connection is of a finite

size in the vertical as well as horizontal directions. Solutions for this situation are not readily

available in literature, albeit derivations by Becker and Bevis (2004) suggest a strong influence

of boundary conditions in similar problems. As per these derivations, the spring stiffness λ (per

unit area) in the horizontal direction is inversely proportional to the bearing width (since it

defines an effective gage length for the compression zone) ahead of the flange. In addition, the

44
stiffness is directly proportional to the modulus of elasticity of the concrete. Accordingly,

Equation 7 may be rewritten as follows –

14
 C1 × E concrete 
r 0 =   (8)
 4 × E steel × I column 

Simplifying further, the term d ref may be expressed as –

14
C  E concrete 
d ρef = , where r =   (9)
ρ  4 × E steel × I column 

In the above equation, the factor C , which may be interpreted as a composite factor including

the other assumed constants such that C = C 0 / C1

14
is selected as a calibrated coefficient. This

preserves the convenience of the method, while incorporating basic elements of the column-

concrete interaction. A value of C = 1.77 provides best match with experimental data.

Once d effective is determined as above, the equations 4 and 5 may be solved simultaneously to

determine the bearing dimensions dU and d L . Following canonical literature on analysis of

concrete sections (e.g., McGregor and Wight, 2011), these dimensions are indicators of the strain

at the extremity (i.e. top) of the effective embedded zone. As a result, the maximum moment

bearing
M HB associated with bearing failure corresponds to the bearing dimension dU and d L

attaining a critical value. Based on experimental data, this critical value is determined to be 60%

of the distance between the extremity of the embedment (i.e. the free surface for dU and the

distance d effective for d L ) and the “neutral axis,” which is the line that bisects the stress-free

region between the two bearing blocks. Figure 13 illustrates this schematically. This approach

45
for bearing strength (using the 60% value) is similar to that adopted for composite connection

design by the ASCE guidelines for composite connections (ASCE, 1994).

Figure 13 – Critical condition for horizontal bearing failure; in the

case shown (as in the experiments)

capacity
Recall that M HB may be controlled by bearing failure or shear failure of the joint panel. The

latter may be simply calculated by invoking the relationship between the moment and shear

strength, such that –

M HB = (Vsteel × d w ) + (Vstrut ,inner × h) + (1.25 × Vstrut ,outer × h)

shear
(10)

shear
In the above equation, M HB may be calculated as the sum of the moment strengths associated

with various components in the panel zone, i.e., the steel web, the concrete strut, and the concrete

outer joint. As per the ASCE guidelines for joint shear strength, updated by Cordova and

Deierlein (2005), these shear strengths may be determined as –

 d + dL 
Vsteel = 0.6 × Fycolumn × t w ×  d effective − U  (11)
 2 

46
  d + dL 
= 1.7 f c' × b f ×  d effective − U
Vstrut ,inner  ≤ 0.5 × f c × b f × h ,
'
(12)
 2 

 d + dL 
= 1.7 f c' × bo ×  d effective − U
Vstrut ,outer  (13)
 2 

capacity
The moment capacity due to the horizontal bearing mechanism M HB is then determined as

shear bearing
the minimum of M HB and M HB . This estimate is retained for use along with the moment

resisted by the entire connection, which also includes the moment resisted due to the vertical

bearing stresses. This is the topic of the next subsection.

Moment resistance due to vertical bearing stresses

Referring to Figure 12, the base plate at the bottom is subjected to bearing stresses on the lower

as well as the upper surfaces, resisting the moment transferred to the base through the column

flanges, as well as the net axial force transferred to the base plate. If the axial force on the

column is compressive, it has two components (1) the portion transferred through the column,

and (2) the remaining axial force transferred into the footing at the top stiffener plate, which is

ultimately transferred to the lower base plate through the compression field in the footing. If the

axial force is tensile, all of it is resisted by downward bearing stresses on the lower base plate. In

either case, the lower base plate is assumed to resist the total axial force (through upward bearing

stresses in the case of compressive axial force, and downward bearing stresses in case of tensile

axial load) in addition to the moment not carried by the horizontal bearing mechanism discussed

in the previous subsection. Referring to Figure 12a, b (which show the moment transfer), this

moment may be expressed as M VB = F flange

bottom
×h.

47
The bearing stress distribution due to the axial load and moment is idealized, recognizing that the

true stress distribution is a result of a complex interplay between the plate and column flexibility,

coupled with the stiffness of the surrounding concrete. The main simplifying assumption is that

the bearing stresses may be decomposed into those resisting the moment, and those due to the

axial force. For the moment, bearing zones on either end of the base plate may be idealized as

rectangular stress blocks with an equal dimension, which is determined (based on agreement

with test data, and consistency with the approach used for horizontal bearing) as d V = 0.3 × N

where N is the length of the lower base plate. Once this is established, the vertical bearing stress

in the blocks may be calculated from moment equilibrium as –

M VB
f VBM = (14)
0.21 × N × B

The stress due to the axial force f VBP = P /( B × N ) considered uniform over the footprint of the

base plate may be added to (or subtracted from) the stress blocks determined as per Equation

(14) above, resulting in a stepped stress distribution with three zones, as shown in Figure 14. The

central zone of width 0.4 × N carries only f VBP = P /( B × N ) , whereas the two outer zones carry

the stresses f VB = f VBP ± f VBM , since the stresses due to moment may add or subtract from those

due to the axial load, depending on the sign of the axial load, and the end zone being considered.
capacity
Once these stresses are determined, the moment capacity M VB may be controlled by one of

four failure modes (which are schematically illustrated in Figures 15a-d)–

b
1. Bearing failure of the concrete above and/or below the base plate – M VB

c
2. Concrete breakout under the compression toe of the base plate – M VB

t
3. Concrete breakout above the tension side flap of the base plate – M VB

48
 y
4. Yielding of the base plate on the tension or compression side – M VB

Bearing failure of concrete will occur when the stresses in the stress block reach the bearing

strength of concrete (assumed as f VBb = 1.7 × f c' to account for confinement – see prior discussion

b
for horizontal bearing as well as Fisher and Kloiber, 2006). As a result, M VB can be conveniently

defined as the moment that results in the bearing stress f VBb = 1.7 × f c' (calculated as per the

process above) in the compressive end zone.

Figure 14 – Stepped bearing distribution due to moment and

axial load at the base plate

As shown in Figures 15b, c the footing in the concrete may also fail due to breakout. This is

expected when the embedment is shallow (for tension side breakout – Figure 15c), or if the lower

base plate rests on a thin layer of the footing (e.g., a thin slab provided to facilitate erection)

supported by the underlying soil (Figure 15b). Recall that the former type of breakout was noted

in Tests #1 and #2 of the experimental program. The breakout strength is controlled by the total
c t
force in the bearing block, rather than the bearing stress. As a result, the terms M VB and M VB are

determined by performing a strength check on the total force in the stepped bearing stress

49
distribution. For tension side breakout, the total bearing force is the force contained in the block

(or blocks) that represent the downward stress distribution. For compression side breakout, the

total bearing force is the force contained in the block (or blocks) that represent the upward

bearing stresses. In either case, the total force may be compared against the capacity for breakout

as determined from the equation below –

40 1
t,c
Fbreakout =η × × × f c' × A35 (15)
9 d cov er

The equation above for concrete breakout capacity is based on the Concrete Capacity Design

(CCD) method proposed by Fuchs et al., (1995) which has been demonstrated (Gomez et al.,

2009) to successfully characterize concrete breakout capacity for various fastenings and concrete

embedments. The above equation is especially attractive because it incorporates the “size-effect”

in concrete, through the introduction of the square root term d cov er , which accounts for the

phenomenon (Bažant, 1984) that larger embedments are weaker on a unit basis as compared to

smaller embedments. The term A35 represents the projected area of a failure cone emanating

from the edges of the stress blocks being considered, such that the angle between the cone

surface and the horizontal is 35 degrees. Note that depending on the direction of axial force, one

or two stress blocks may be active for tension or compression side breakout. In the above

equation, d cov er is the thickness of the material that must but ruptured for breakout. Thus, for

tension side breakout, d cov er = d embed . The factor η is taken as 1.5 to correct for a discrepancy

between the CCD method which assumes that the embedments loaded with a uniform stress. In

contrast, the stepped stress distribution on the lower base plate represents a stress gradient, such

that the plate rotates (rather than translates) to cause concrete breakout. For the same magnitude

of total force, this situation is less intense, and results in a higher capacity. The factor η = 1.5 is

50
calibrated to match test data. Alternative models for tension side concrete breakout (e.g., Cui et

al. 2009) were examined, but they did not produce reasonable agreement with the experiments.

t c
Based on the above process, the moments M VB and M VB may be determined as moments

(applied to the lower base plate) that result in either tension or compression side breakout.

Finally, the bearing stress corresponding to base plate yielding may be determined assuming

cantilever bending of the base plate flaps on either side of the column, with yield lines forming

parallel to the column flanges. In the following expression (Equation 16), it is implicit that the

(1) dimension of the bearing zone is greater than the length of the flap, and (2) the full plastic

moment capacity of the base plate is developed in bending. An analogous expression may be

derived if this is not the case –

2
t plate × Fyplate
fVBy = (16)
2 × n2

In the above equation, n is the length of the flap of the base plate. Once fVBy is determined in this

manner, it may be substituted in Equation (14) to calculate the corresponding moment capacity
y y b c t capacity
M VB . The minimum of M VB , M VB , M VB , and M VB is then taken as M VB .

capacity capacity
Once the moment capacities M HB and M VB have been determined as per the processes

outlined in this subsection, they must be combined to characterize the net connection strength.

This is a non-trivial issue owing to several factors, which are discussed in the next section.

51
 (a) (b)

(c) (d)

Figure 15 – Failure modes corresponding to vertical bearing moment transfer (a)

bearing failure above and below base plate (b) concrete breakout under compression
toe (c) concrete breakout above tension flap, and (d) base plate yielding

Combination of moment strengths due to horizontal and vertical bearing

capacity capacity
The combination of M HB and M VB to determine a net (or overall) moment capacity for the

capacity
ECB connection M base is based on a consideration of (1) the evolution of response, and

distribution of moments between the horizontal and vertical bearing mechanisms through the

loading process, and (2) the interaction between the failure modes. Referring to prior discussion,

mechanisms that rely on vertical and horizontal moment transfer are constrained by deformation

compatibility, with the implication that they may be considered “in parallel,” such that their

52
moment contributions are additive. However, this also introduces indeterminacy with respect to

the relative contributions of these moments. To resolve this, it is assumed that before the

moments, M HB or M VB reach their capacities (as determined from the previous section), the net

moment M base is distributed in a constant proportion between the two mechanisms, such that –

M VB = a × M base (17)

and,

M HB = (1 − a )× M base (18)

In the above equations, the ratio α controls the relative contribution of the two mechanisms. The

contribution of the vertical bearing stresses will diminish as the embedment depth increases,

since most of the moment will be carried by the horizontal bearing stresses – refer discussion

regarding the effective depth of the embedment. Based on the same discussion, it is implied that

the vertical stresses will not resist any moment if the depth exceeds the reference depth of d ref

(Equation 6). To reflect this, the following expression is proposed for the ratio α –

α = 1 − (d embed d ref ) ≥ 0 (19)

Given this, various scenarios may be invoked to determine the net moment carried by the

connection. The following discussion summarizes each of these scenarios, methods for

calculating the associated moment capacity, and the corresponding rationale. Broadly, two

situations may be considered (1) failure due to vertical bearing occurs before failure due to

horizontal bearing, and (2) failure due to horizontal bearing occurs before failure due to vertical

bearing.

53
Scenario 1 – Failure due to vertical bearing occurs before failure due to horizontal bearing

This scenario occurs if the following inequality is valid –

cαpαcity cαpαcity
M VB M HB
≤ (20)
α 1−α

In this scenario, ultimate strength of the connection may be determined based on the precise

mode of failure within the vertical bearing mechanism. As discussed previously, three modes are

possible –

1. Concrete breakout on the tension side: This type of failure results in a failure plane that

interferes with the horizontal bearing mechanism as well, since it is angled upwards towards

the free surface. In fact, in two of the specimens tested, i.e. Tests #1 and 2, this type of failure

resulted in a sudden loss of capacity of the specimen. As a result, if concrete breakout on the

tension side is the controlling mechanism of failure given that vertical bearing failure occurs

first, the net moment capacity of the connection may be simply determined as the moment

carried by the connection at the instant when the vertical bearing mechanism reaches its

failure mode, i.e., -

cαpαcity
M bαse = M VB
t
α (21)

2. Concrete breakout on the compression side: Failure of this type results in the loss of moment

carrying capacity due to vertical bearing, but because the failure plane is below the lower

base plate (and therefore does not interfere with the horizontal bearing mechanism), the

horizontal bearing capacity is still retained. Accordingly, if the critical value of M VB is

controlled by breakout on the compression side, the moment capacity of the base may be

characterized as –
capacity
M base = M HB
capacity
(22)

54
3. Concrete crushing or base plate yielding: Both of these are ductile failure modes and may be

assumed continue to carry moment until the horizontal bearing capacity is reached (although

this has not been verified in the experiments – which showed other failure modes). As a

result, if any of these control M VB , the connection strength may be estimated as per Equation

(23) below –
capacity
M base = M HB
capacity
+ M VB
b capacity
or M base = M HB
capacity
+ M VB
t
(23)

Scenario 2 – Failure due to horizontal bearing occurs before failure due to vertical bearing

This occurs if the following inequality is valid –

cαpαcity cαpαcity
M VB M HB
> (24)
α 1−α

capacity
As discussed previously, M HB may be controlled by bearing failure or joint shear failure.

Each of these cases is now considered separately. If bearing failure controls, then

capacity
M HB = M HB
bearing
, and the moment carried by the horizontal bearing stresses degrades gradually

with increasing deformations. This is because spalling of the concrete at the extremities of the

bearing zone reduces the effective depth of bearing. This type of behavior is observed in all tests

(note the post-peak response in Figure 5). This type of response cannot be described as sudden or

brittle, and moreover it does not appear to reduce the effectiveness of the vertical bearing

mechanism. However, since horizontal bearing failure decreases the moment M HB , it is expected

that (1) the ratio α defining the moment distribution between the two mechanisms is no longer

valid, and (2) the vertical bearing mechanism will now carry a larger proportion of the moment.

The implication is that as deformations increase beyond those required to initiate horizontal

bearing failure, the moment carried by horizontal bearing will reduce, accompanied by a

55
corresponding increase in the moment carried by vertical bearing. Ultimate failure will occur

when one of the failure modes in vertical bearing is triggered. This type of response is shown by

Tests #1 and #2 where the tension-side breakout (a vertical bearing failure mode) is triggered at

large deformations after significant reduction in the moment carried by horizontal bearing. To

reflect this response, the maximum moment carried by the connection may be determined as –

 cαpαcity
cαpαcity M HB 
M cαpαcity

= mαx κ × M HB
cαpαcity
+ M VB ,  (25)
bαse
1−α 
 

The right hand side of the above equation contains two candidates for the moment capacity

associated with this failure mode, to reflect two possible scenarios, which are –

capacity
1. The decrease in M HB (after M HB is reached) is greater than the corresponding increase in

M VB before failure is achieved in the vertical bearing mode. In this case, the peak moment

capacity
sustained by the connection is simply M HB . This is the case as observed in Tests # 1 and

#2.

capacity
2. The decrease in M HB (after M HB is reached) is lower relative to the increase in M VB ,

before it reaches failure. This is possible, for example, in shallowly embedded bases (not

tested in the current study) where horizontal bearing provides only modest moment resistance

(which is reached quickly), followed by a period during which the vertical bearing stresses

capacity
are mobilized, to achieve a moment strength higher than M HB . In this case, the peak

capacity
moment sustained by the connection is the sum of M VB and the reduced moment

κ × M HB
capacity
carried by the horizontal bearing mechanism at the instant of failure in the

vertical bearing mechanism.

56
The factor κ (which can have a maximum value of 1.0) reflects the reduction of M HB , and is

capacity
physically dependent on the magnitude of deformations applied after M HB has been achieved.

Characterization of κ based on deformations is challenging for several reasons, primarily

because the strength method is based on assumed stress distributions, rather than deformation

fields. As a result, the following relationship is proposed to determine the factor κ -

 M cαpαcity × (1 − α ) 
−2

κ =  VB cαpαcity  (26)
 M HB × α 

cαpαcity
M VB × (1 − α )
In the above equation, the quantity reflects a normalized index of the residual
M HB × α
cαpαcity

(or available) capacity in the vertical bearing mechanism at the instant when the horizontal

cαpαcity
M VB × (1 − α )
bearing strength is reached. For example, if = 1 , then the implication is that the
M HB × α
cαpαcity

vertical and horizontal bearing failures are reached simultaneously. As a result, κ = 1 , because

capacity
no additional deformations can be applied to the connection after M HB capacity has been

capacity
reached (since M VB is also simultaneously reached). On the other hand, if

cαpαcity
M VB × (1 − α )
> 1 , then the implication is that additional deformations may be applied before
M HB × α
cαpαcity

capacity
M VB is reached, and these deformations will be accompanied by a reduction in M HB . The

relationship in Equation 26 above reflects this assumed response, and is calibrated to match test

data, especially the breakout failure strengths observed for Tests # 1 and #2.

capacity
If shear failure controls, in which case M HB = M HB
shear
, then the failure is brittle, with no

possibility of additional moment being sustained in any of the failure modes. In this case, the

moment capacity of the connection may be calculated through Equation (27) below –

57
 cαpαcity
M bαse = M HB
sheαr
/(1 − α ) (27)

capacity
In the above equation, M base is the moment carried in the connection when the horizontal

bearing mechanism reaches its capacity. The above equation is similar to Equation 21 presented

previously for the brittle failure mode (i.e., tension breakout) in the vertical bearing mechanism.

The flowchart shown in Figure 16 schematically illustrates the process described above to
capacity
determine base connection capacity M base . Referring to the Figure, it is evident that the

algorithm is suited for automated implementation using a computer program, given its

complexity. To design the base (i.e., to solve the inverse problem), the computer program may be

executed with various trial parameters (such as embedment, plate thickness etc.) until an

acceptable solution is determined within the design constraints. A note of caution here is that the

strength determined as per the method above is consistent with the ultimate strength of the

connection, with the implication that if designed based on this strength, the connection will

sustain inelastic deformations during design level shaking. As discussed previously, the average

y max
ratio M base / M base = 0.70 may be considered a suitable fraction of the ultimate strength for

design purposes, since it limits inelastic deformations. Given the nonlinearity of the load-

deformation curve (and the processes responsible for it, e.g., progressive bearing failure ahead of

the column flange) is gradual, it is challenging to identify a precise physical event and a

corresponding first yield moment for the connection. Consequently, designing the connection for

a fraction of the ultimate strength is an expedient strategy. The next section evaluates the

efficacy of the proposed method (i.e., the determined strengths as well as the implied response)

against the test data.

58
 Given quantities: Determine and using
• , , , Equations 6, 9 and 19
• Column size
• Base plate dimensions
Iteration loop

Increment and V
Compute , , , using
Equations 14-16. Set
Compute , , ,
, using Equations 4, 5, 10-13

Determine failure scenario (Eq 20, 24)

Scenario 1, Scenario 2, No
Eq 20 is true Eq 24 is true

?
Yes
Yes No
No Yes

No Yes

Figure 16 – Flowchart illustrating strength characterization process; dark borders

indicate final estimates.

59
ASSESSMENT OF STRENGTH MODEL RELATIVE TO TEST DATA
capacity
Figure 17a-e show the strength estimate M base , determined as per the method presented in the

previous section, overlaid on the moment-drift curves for all the five specimens. The figure

graphically indicates that the agreement between the predicted and observed moment strength is

excellent for all the five experiments. Table 2 summarizes these observations quantitatively. The

key observations from the Table and Figures 17a-e are –

1. The Table shows test-predicted ratios for both directions of loading for each test. On

average, the test-predicted ratio is 1.01 (with a standard deviation of 0.06), which may be

interpreted to indicate that the model appears to characterize the strength of the base

connections with accuracy.

2. The Table includes the maximum experimental moments M max determined from both

directions of loading in the cyclic tests. The value for the direction loaded first (denoted

forward) is listed first is followed by the reverse direction. The strength in the forward

direction is on average 1.08 times stronger as compared to the strength in the reverse

direction. This trend is also reflected in the test-predicted ratios such that the average test-

predicted ratio for the forward direction is 1.03, whereas for the reverse direction, it is 0.95.

The model cannot account for strength degradation due to loading in the opposite direction,

which is the reason for this bias.

3. In three of the tests, (i.e., Test #3, #4 and #5 with d embed = 762mm), the peak strength (as

determined by the method) was controlled by horizontal bearing, such that

cαpαcity
M bαse = M HB
cαpαcity
/(1 − α ) . The observed physical response was consistent with this, such

that a loss in moment capacity was observed after reaching a peak strength, which was

60
 accompanied by spalling of concrete in the bearing region directly ahead of the column

flange.

4. The capacity of Test #1 (with d embed = 508mm) was controlled by bearing as well. However,

after reaching the peak strength (accompanied by concrete spalling as in Tests #3, #4, and

#5), the moment capacity gradually diminished until sudden failure occurred due to concrete

breakout on the tension side of the connection. The moment corresponding to this breakout

event (which occurs only on one side of the specimen, after which the test is terminated)

represents the base moment κ × M HB

capacity
+ M VB
t
(refer Equation 25). However, since it is

cαpαcity
lower than M bαse = M HB
cαpαcity
/(1 − α ) , it does not control the strength of the connection.

Nevertheless, the agreement of the measured value of this moment M base = 2177 kN.m (as

shown in Figure 17a) relative to its analytical counterpart κ × M HB

capacity
+ M VB
t
= 2337 kN.m

provides additional physical validation of the proposed method.

5. The strength of Test #2 is also ultimately controlled by Equation 25 such that

cαpαcity
M bαse = M HB
cαpαcity
/(1 − α ) . However, as with Test #1, strength degradation and breakout

capacity
occurred subsequent to reaching M base . The measured value of this moment was M base =

1807 kN.m, and its analytical counterpart is κ × M HB

capacity
+ M VB
t
= 2083 kN.m, with κ = 0.59.

In summary, the connection strengths determined by the method are generally in good agreement

with the experimental values. The test-to-predicted ratios do not appear to show significant bias

with respect to any of the test variables. Moreover, the failure modes predicted by the method are

also consistent with those observed experimentally, indicating that the method is able to reflect

key physical aspects of connection response. The latter is especially important from the

61
perspective of generalizing the approach to connections configurations that are different from the

ones tested.

4500 4500
Test 1 Test 2
3000 3000
Mbase (kN.m)

Mbase (kN.m)
1500 1500

0 0

-1500 -1500

-3000 -3000
(a) (b)
-4500 -4500
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Column drift (%) Column drift (%)

4500 4500
Test 3 Test 4
3000 3000
Mbase (kN.m)

Mbase (kN.m)

1500 1500

0 0

-1500 -1500
(c) (d)
-3000 -3000

-4500 -4500
-8 -6 -4 -2 0 2 4 6 8 -8 -6 -4 -2 0 2 4 6 8
Column drift (%) Column drift (%)
4500

capacity Governing failure Test 5

M HB 3000
capacity
M base = case for all tests,
1−a
Mbase (kN.m)

shown by solid 1500

li 0
capacity Failure mode
κ × M HB
capacity
+ M VB -1500
observed for (e)
Tests #1 and #2, -3000
shown by dashed
li -4500
-4 -3 -2 -1 0 1 2 3 4
Column drift (%)

Figure 17 – Load displacement curves with moment strengths estimated as per proposed
method

62
Table 2 – Test results and comparison to strength model

Column Base Plate, max

d embed z max
M base d ref * M base
Test Size, P tp ×N ×B
[m]
α **
# ( b f [mm]) [kN] [mm] [kN.m] [mm] M capacity
[mm] base

W14x370 2579(+) 1.05

1 51 762 762 920 0.45
(419) 2613(-) 1.06
445 (C) 508 2.84
W18x311 2324(+) 1.04
2 51 864 864 978 0.48
(305) 2168(-) 0.97
3741(+) 1.00
3 0 920 0.17
3444(-) 0.92
W14x370 4124(+) 1.10
4 445 (C) 762 51 762 762 3.10 920 0.17
(419) 3612(-) 0.97
3800(+) 1.02
5 667 (T) 920 0.17
3464(-) 0.93
Mean 1.01
COV 0.06
*In all cases d ref > d embed , such that d effective = d embed .
**In all cases
cαpαcity
M bαse = M HB
cαpαcity
/(1 − α ) .

DESIGN CONSIDERATIONS

It is anticipated that the method will be used iteratively in a design setting until the trial

configuration (which is characterized by embedment depth, lower base plate dimensions)

capacity
provides a strength M base that is sufficient to resist the applied moment (typically the moment

capacity of the embedded column). The method was tested by applying it to several column sizes

in the range of W14 to W33. In all cases, reasonable designs (embedment depths between 350-

1000mm – depending on the moment capacity of the column, along with a base plate thickness

of 50mm) were determined as per the method. An interesting observation was that for some

column sections (heavy W14s, which had a high moment capacity but low depth), panel zone

63
 cαpαcity
shear was the critical mode of failure, i.e., M bαse = M HB
sheαr
/(1 − α ) . If these cases are

shear
encountered in practice, it may be expedient to increase M HB by providing reinforcement in the

compression field region in the joint panel outside the column flanges. This was not considered

in the experiments. In a majority of the cases (where joint shear did not govern), the dominant

modes of failure were horizontal bearing (Equation 25), with or without breakout. Details of

these design trials are provided in Grilli (2015).

SUMMARY, CONCLUSIONS, AND LIMITATIONS

This report presents findings from five tests representative of embedded column base (ECB)

connections in mid- to high-rise steel moment frames, along with a strength characterization

method based on experimental data and observed behavior. While these connections are common

in current construction, their design is based on adaptations of guidelines for composite beam-

column connections, or for coupling beams embedded in shear walls. Moreover, while they are

assumed to be fixed connections, little evidence substantiates this assumption. Both of these

knowledge gaps may be attributed to the lack of experimental data on ECB connections. The

experiments described in this report, being the first of their kind, provide a direct assessment of

the response of these connections, and the opportunity to develop design methods that reflect

physical response peculiar to them. The main objective of these experiments is to examine the

seismic response of these connections in three contexts: (1) to understand overall hysteretic

response including failure modes, and deformation characteristics, (2) to support a strength

characterization/design method, and (3) to examine whether ECB connections provide adequate

rotational stiffness.

64
Experimental results

The experimental specimens were designed to represent current construction practice. All the

specimens were cantilever columns loaded with cyclic lateral deformations under a constant

axial load. The major test variables were the embedment depth, axial load (compressive, tensile,

or zero), and the column cross-section. The experiments revealed that the base connections show

hysteretic response with a high deformation capacity. In the initial stages of loading, damage was

concentrated near the column flanges in the form of diagonal cracks. As loading progressed,

crushing of concrete ahead of the compression flange resulted in nonlinearity in load

displacement response. Diagonal shear cracks on the side of the joint were observed along with

flexural cracks due to bending of the footing. Two types of failure were observed, one was

associated with a gradual loss of strength owing to crushing of concrete in the compression zone

ahead of the flange, whereas the other was characterized by uplift of a cone of concrete on the

tension side of the connection. Based on these observations, three mechanisms for load

resistance are postulated: (1) horizontal bearing of the column flanges against the concrete, (2)

vertical bearing stresses on the embedded as well as stiffener plates, and (3) panel zone shear.

Each of these mechanisms is further associated with one or more sub-mechanisms and local

failure modes. An understanding of these mechanisms, sub-mechanisms, and their interactions is

necessary for the development of a strength model. An analysis of the rotational stiffness of the

ECB connections indicates that the connections may be flexible, such that first story drifts may

be (on average) 25% larger than those estimated from a fixed base assumption.

In summary, the study indicates that if designed with consideration of both strength and stiffness,

ECB connections are an attractive option for transferring large base moments and forces from

65
moment frame columns into the foundation. Moreover, they indicate excellent deformation

capacity such that ≈2–3% drifts are reached in the experiments without significant loss of

strength. A similar trend has been noted previously (Gomez et al., 2010; Kanvinde et al., 2013)

for exposed base plate connections. In the context of current design practice, which typically

requires the base connection to be stronger than the attached column (i.e., a capacity design),

these observations suggest that utilizing the deformation and hysteretic characteristics of well-

designed base connections may facilitate more economical design.

The general observations of this study must be interpreted within the context of its limitations,

which are numerous. First, all the experiments examined only one generic detail, i.e., with the

base plate at the lower end and the stiffener plate at the top. Other details may feature (1) a larger

base plate at the top (2) supplemental anchor rods (3) additional stiffeners or haunches between

the column and the base plate. Extrapolation of the method to these situations will require careful

consideration of the physical distinctions between these and the tested configurations. Second,

reinforcement in the footing was sparse by design, and the addition of reinforcement has the

potential to affect the failure modes of horizontal bearing, panel zone shear, as well as vertical

bearing (especially tension-side blowout). Third, extrapolating the observations to

footing/embedment sizes that are significantly different from those tested is prone to error, since

they are empirical and may be sensitive to size. In addition, the observed strengths as well as the

failure modes may be sensitive to the manner of application of loading. Two issues are of

importance here: (1) in the experiments, loading was applied in a non-proportional manner where

the lateral loads were applied in the presence of a constant axial load, and (2) a constant moment-

to-shear ratio was maintained throughout the tests. Seismic loads will, in general, not follow this

66
manner of load application. Finally, as discussed previously, the experiments address only the

failure modes that occur in the immediate vicinity of the embedded connection; the boundary

conditions of the experiments were designed to interrogate only these modes. Failure modes

specific to foundation type (e.g., flexural failure in a grade beam or mat) are not addressed. To a

large extent, the limitations above arise because the main objective of this study was to develop

fundamental insights about the response of these connections. These insights may be refined and

generalized through future experiments (e.g., on more connection types) and analytical research

(e.g., finite element simulations) for more accurate and broader application. In any case, it is

important to recognize that these limitations of the test program are inherited by strength

characterization models that are based on these tests.

Strength characterization method

The strength characterization method developed in this report builds on the experimental

observations that in ECB connections, the applied moment and axial load are resisted three

primary mechanisms: (1) horizontal bearing of the column flanges against the concrete, (2)

vertical bearing stresses on the embedded as well as stiffener plates, and (3) panel zone shear.

These mechanisms are interactive, and each is associated with one more failure modes. The

method characterizes the strengths associated with each of these failure modes, and establishes

the nature of interactions between these mechanisms such that strengths associated with the

individual failure modes may be suitably combined to characterize overall connection strength.

The proposed method seeks to balance four considerations: (1) reliance on published models and

fundamental physics/mechanics to characterize interactions and failure modes, (2) minimization

67
of ad-hoc, calibrated factors, (3) agreement with test data in terms of strengths as well as

qualitative failure modes, (4) convenience of application within a practical or design setting. The

empirical (or calibrate d) aspects of the method include: (1) the relationship for d effective , which

characterizes the attenuation of stresses through the depth of the footing, (2) the approach to

characterize the distribution of moment between the horizontal and vertical bearing mechanisms,

i.e., the α factor, (3) the approach to quantify the degradation of the horizontal bearing strength

after it has achieved its peak, through the κ factor, and (4) the modification factor η to account

for the effect of stress gradient on concrete breakout. While calibrated empirically, these have

basis in classical derivation (in the first case), and physical observations of test response (in the

three latter cases).

The method assumes that all dimensions and material properties of all components of the

connection are known or specified. In addition, the method also assumes that the axial force is

given. It is anticipated that this method will be used iteratively to size various parts of the footing

given the loading and other (dimensional, architectural, planning) constraints. Since the method

identifies the type of failure controlling the moment capacity (and the moment capacities with

other response modes), it explicitly provides guidance for optimal re-sizing and iteration. While

the method can be applied through hand-analysis or a spreadsheet, a computer program is most

convenient for its implementation. The method is tested to design connections for several column

sizes, and in general, is determined to provide reasonable designs (in terms of embedment and

base plate dimensions).

68
The proposed method has several limitations, which must be considered in its interpretation and

application. As mentioned previously, is important to note that the limitations of the

experimental program are inherited by the method as well. While the main purpose of the

method is to enable extrapolation beyond the conditions tested, it is prudent to recognize that the

empirical aspects of the method may not be generally applicable. In fact, extrapolating the

method to footing/embedment sizes that are significantly different from the method is prone to

error, since the empirical aspects of the method may be sensitive to size. Where available,

methods that are based on similarly sized (or detailed) connections may be used. For example,

the methods developed by Cui et al. (2009), or by Barnwell (2015), may be more suitable for

shallowly embedded connections with d embed ≤ 200mm. On the other hand, where such

approaches are not available, aspects of the presented method may be leveraged or modified

where possible. For example, the effect of anchor rods (if provided on the lower base plate) may

be readily incorporated by modifying the vertical bearing limit states by adding the anchor rod
t
strength in the calculation of M VB . Moreover, as discussed previously, the method (like the

experiments) only addresses failure modes that occur in the immediate vicinity of the embedded

connection; and not those triggered by overall foundation failure. These are presumably sensitive

to foundation type. To overcome some of these limitations, future work may involve additional

experiments on varied details, and finite element simulations for more accurate understanding of

internal force transfer. Finally, the method provides deterministic estimates of nominal strength.

Appropriate φ − factors are required to ensure adequate margins of safety against failure; these

may be developed through reliability analysis.

69
NOTATIONS

α Fraction applied moment resisted by vertical bearing mechanism

A35 Projected area of a concrete 35 degree failure cone emanating from the edges of

concrete stress block

B , N t plate Base plate dimensions: Width, length (in direction of load), thickness

β , β1 Factors to account for concrete confinement and effectiveness in bearing

bbase , bbase
test
Rotational stiffness of base connection: generic, experimental

bf , t f Width, thickness of column flange

β frame Rotational stiffness of framing members at top of column

b j , bo Effective joint width, outer joint panel zone width

d , dw Column depth, web depth

d cov er Thickness of concrete that must be ruptured for breakout

d embed , d effective Embedment depth, effective embedment depth

d ref Depth at which horizontal bearing stresses attenuate to zero

dL Depth of lower horizontal concrete bearing stress block

dU Depth of upper horizontal concrete bearing stress block

dv Length of vertical bearing zone on base plate

∆ base
max
Drift corresponding to Mbase
max

y
∆ base
y
Drift corresponding to Mbase

∆ fixed , ∆ test Calculated drift of column due to a unit load applied at the top of the column –

assuming a fixed base, experimental value

70
E steel , E concrete Moduli of elasticity of steel and concrete

t ,c
Fbreakout Force at which breakout occurs, on tension or compression side

fc' Compressive strength of concrete

top bottom
Fflange , Fflange Forces in column flanges at the top and bottom of embedment zone

Fy Minimum specified yield strength of steel

Fycolumn , Fyplate Minimum specified yield stress of column and base plate

fVBb Vertical bearing stress on base plate

fVBM , fVBP Vertical bearing stress on base plate due to moment, axial load

h Distance between the centerline of column flanges

I beam , I column Moment of inertia of notional beam, column

κ Degradation factor for horizontal bearing

Lbeam Length of notional beam framing into top of column

y
M base , Mbase
max
, Mbase Column base moment – generic, maximum in test, yield

capacity
M base Base moment capacity

capacity
M HB , M HB Moment resisted through horizontal bearing stresses

bearing shear
M HB , M HB Moment capacity of provided by horizontal bearing, shear

Mp Plastic moment capacity of column

capacity
M VB , M VB Moment resisted through vertical bearing mechanism, capacity

b
M VB Moment capacity of vertical bearing mechanism due to concrete crushing

71
 c
M VB Moment capacity of vertical bearing mechanism due to concrete breakout under

the compression toe of the base plate

t
M VB Moment capacity of vertical bearing mechanism due to concrete breakout under

the compression toe of the base plate

y
M VB Moment capacity of vertical bearing mechanism due to yielding of base plate on

tension or compression side

n Length of flap of the base plate in the direction of loading

η Factor used in concrete breakout equation accounting for differences in loading

conditions implied by CCD method and ECB connections

P Column axial load

θ drift , θbase , θ footing Column drift angle, base rotation, footing rotation

Vj Vertical shear force in the joint panel

Vcolumn Imposed column shear

Vsteel , Vstrut ,inner , Vstrut ,outer Shear capacity of steel web, concrete strut, outer panel

z Height of load application above concrete surface

72
REFERENCES

AISC. (2010). Seismic Provisions for Structural Steel Buildings (ANSI,AISC 341-10), American

Institute of Steel Construction, Inc., Chicago, IL.

ASCE Guidelines (1994). “Guidelines for Design of Joints Between Steel Beams and Reinforced

Concrete Columns,” Journal of Structural Division, ASCE, Vol. 120(8), pp. 2330-2357.

ASCE Task Committee (1998). “Design Guide for Partially Restrained Composite Connections,”

Journal of Structural Engineering, ASCE, Vol.124 No. 10, pp.1099-1114.

Astaneh, A., Bergsma, G., and Shen J.H. (1992). “Behavior and Design of Base Plates for

Gravity, Wind and Seismic Loads,” Proceedings of the National Steel Construction

Conference, Las Vegas, Nevada, AISC, Chicago, IL.

Barnwell, N. (2015). “Experimental Testing of Shallow Embedded Connections Between Steel

Columns and Concrete Footings,” Masters Thesis, Brigham Young University.

Bažant, Z.P. (1984). “Size Effect in Blunt Fracture: Concrete, Rock, Metal,” Journal of

Engineering Mechanics, ASCE, 110(4), 518–535.

Becker, J., Bevis, M. (2004). “Love’s Problem”, Geophys. J. Int., 156, 171 - 178.

Bjorhovde, R., Brozzetti, J., Colson, A. (1988). Connections in Steel Structures: Behaviour,

Strength, and Design. New York: Elsevier Science Pub. Co. Print.

Bruneau, M., Uang, C., Whittaker, A. (1997). Ductile Design of Steel Structures. New York:

McGraw-Hill Professional; 1 edition (October 1, 1997). Print.

Burda, J.J., and Itani, A.M. (1999). “Studies of Seismic Behavior of Steel Base Plates,” Report

No. CCEER 99-7, Center of Civil Engineering Earthquake Research, Department of Civil

and Environmental Engineering, University of Nevada, NV.

73
Cordova, P.P. and Deierlein, G.G. (2005). “Validation of the Seismic Performance of Composite

RCS Frames: Full-Scale Testing, Analytical Modeling, and Seismic Design,” Technical

Report 154, Blume Earthquake Engineering Center.

Cui, Y., Nagae, T., Nakashima, M. (2009). “Hysteretic Behavior and Strength Capacity of

Shallowly Embedded Steel Column Bases.” Journal of Structural Engineering, ASCE

Vol. 135, No. 10, October 2009, pp. 1231-1238.

Deierlein, G.G., Sheikh, T.M., Yura, J.A., and Jirsa, J.O. (1989). “Beam-Column Moment

Connections for Composite Frames: Part 2,” Journal of Structural Engineering ASCE,

Vol. 115, November 1989, pp.2877-2869.

DeWolf J.T., and Sarisley, E.F. (1980). “Column Base Plates with Axial Loads and Moments,”

Journal of the Structural Division, ASCE, Vol. 106, No. 11, November 1980, pp. 2167-

2184.

Fahmy, M., Stojadinovic, B., and Goel, S.C. (1999). “Analytical and Experimental Studies on

the Seismic Response of Steel Column Bases,” Proceedings, 8th Canadian Conference on

Earthquake Engineering, Vancouver, Canada.

Fisher, J.M. and Kloiber, L.A. (2006), “Base Plate and Anchor Rod Design,” 2nd Ed., Steel

Design Guide Series No. 1, American Institute of Steel Construction, Inc., Chicago, IL.

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. and Deierlein G.G. (2010). “Exposed Column Base Connections

Subjected to Axial Compression and Flexure,” Report Submitted to the American

Institute of Steel Construction (AISC), Chicago, IL.

74
Gomez, I.R., Kanvinde, A.M., Smith, C.M., and Deierlein, G.G., (2009), "Shear Transfer in

Exposed Column Base Plates," Report Submitted to the American Institute of Steel

Construction, Chicago, IL.

Grilli, D.A. (2015). “Seismic response of embedded column base connections and anchorages,”

Doctoral Dissertation, University of California, Davis, CA 95616.

Grilli, D.A., and Kanvinde, A.M. (2013). “Special Moment Frame Base Connection: Design

Example 8,” 2012 IBC SEAOC Structural/Seismic Design Manual, Volume 4, Examples

for Steel-Frame Buildings, 255-280.

Hetenyi, M. (1946). Beams on Elastic Foundation: Theory with Applications in the Fields of

Civil and Mechanical Engineering. University of Michigan Press (August 1, 1946). Print.

Hutchinson, T.C., Chai, Y.H., and Boulanger, R.W. (2005). “Simulation of Full-Scale Cyclic

Lateral Load Tests on Piles,” Journal of Geotechnical and Geoenvironmental

Engineering, ASCE, Vol. 131, No. 9, pp. 1172-1175.

Kanvinde, A.M., Jordan, S.J., and Cooke, R.J. (2013). “Exposed Column Baseplate Connections

in Moment Frames – Simulations and Behavioral Insights,” Journal of Constructional

Steel Research, Elsevier, 84, 82-93.

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.

Krawinkler, H., Gupta, A., Medina, R., and Luco, N. (2000), “Loading Histories for

Seismic Performance Testing of SMRF Components and Assemblies,” SAC Joint

Venture, Report no. SAC/BD-00/10. Richmond, CA.

75
Krawinkler, H., Parisi, F., Ibarra, L., Ayoub, A., and Medina, R., (2000). “Development of a

Testing Protocol for Woodframe Structures,” CUREE Publication No. W-02.

Marcakis, K. and Mitchell, D. (1980) “Precast Concrete Connections with Embedded Steel

Members,” Prestressed Concrete Institute Journal, V. 25, No. 4, July/Aug. 1980, pp. 88-

116.

Mattock A. H. and Gaafar G. H. (1981). “Strength of Embedded Steel Sections as Brackets,”

ACI Journal, Vol. 79, No. 2.

Morino, S., Kawaguchi, J., Tsuji, A., and Kadoya, H. (2003). “Strength and Stiffness of CFT

Semi-embedded Type Column Base,” Proceedings of ASSCCA 2003 Conference,

Sydney, Australia, A. A. Balkema, Sydney, Australia.

Motter, C.J., (2015). Large-scale Testing of Steel-reinforced Concrete (SRC) Coupling Beams

Embedded into Reinforced Concrete Structural Walls. Dissertation, University of

California, Los Angeles. Ann Arbor: ProQuest/UMI, 2014. (Publication No. 3629336).

Myers, A.T., Kanvinde, A.M., Deierlein, G.G., and Fell B.V. (2009). “Effect of Weld Details on

the Ductility of Steel Column Baseplate Connections,” Journal of Constructional Steel

Research, Elsevier, Vol. 65, No. 6, June 2009, pp. 1366-1373.

Pertold J., Xiao, R. Y., and Wald, F. (2000a). “Embedded Steel Column Bases – I. Experiments

and Numerical Simulation,” Journal of Constructional Steel Research, Vol. 56, pp. 253-

270.

Pertold, J., Xiao, R. Y., and Wald, F. (2000b). “Embedded Steel Column Bases – II. Design

Model Proposal,” Journal of Constructional Steel Research, Vol. 56, pp. 271-286.

Shahrooz, B. M., Remetter, M. E., and Qin, F. (1993), “Seismic Design and Performance of

Composite Coupled Walls,” Journal of Structural Engineering, ASCE, Vol. 119, No. 11,

76
 pp. 3291–3309, Reston, VA.

Sheikh, T.M., Deierlein, G.G., Yura, J.A., and Jirsa, J.O. (1989). “Beam-Column Moment

Connections for Composite Frames: Part 1,” Journal of Structural Engineering ASCE,

Vol. 115, November 1989, pp. 2858-2876.

Thambiratnam, D.P., and Paramasivam, P. (1986). “Base Plates Under Axial Loads and

Moments,” Journal of Structural Engineering, ASCE, Vol. 112, No. 5, May 1986,

pp. 1166-1181.

Tremblay, R., Timler, P., Bruneau, M., and Filiatrault, A., (1995). “Performance of Steel

Structures During the 1994 Northridge Earthquake,” Canadian Journal of Civil

Engineering, Vol. 22, pp. 338-360.

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.

77
 APPENDIX

LISTING OF PROGRAM TO CALCULATE MOMENT STRENGTH OF EMBEDDED CONNECTIONS

AS PER THE PROPOSED METHOD

78
function[Mcap] = Mbase_predict(P,dembed,section,z,N,B,tp,W,fc,Fy)

%READ ME===================================================================
%Function implements a strength prediction method to calculate the
%maximum base moment for an embedded column base connection

%INPUTS:
%axial load, embedment depth, column dimensions, base plate dimensions,
%and steel and concrete material

%definitions
% P: axial load, compressive (+), tensile (-) [kips]
% dembed: embedment depth [in]
% section: [string] Embedded column section (must use section.m, or recode
to
% define column dimensions manually
% z: height of horizontal load application (moment/shear ratio) [in]
% N: length of base plate (in loading direction) [in]
% B: width of base plate [in]
% tp: thickness of base plate [in]
% W: width of foundation (perpindicular to loading direction) [in]
% fc: compressive strength of concrete [ksi]
% Fy: yield strength of base plate [ksi]

%OUTPUTS:
% Base moment capacity [k-ft]

%definitions
% Mt_vb: moment due to vertical bearing at breakout on tension side
% Mb_vb: moment due to vertical bearing at bearing failure
% My_vb: moment due to vertical bearing at base plate yield
% Mbearing_b: moment due to horizontal bearing at bearing failure
% Mh_ps: moment due to horizontal bearing at panel shear failure
% Mcap_h: maximum moment due to horizontal bearing
% Mcap_v: maximum moment due to vertical bearing
% Mcap: maximum moment

%supplimentary information printed to command window

%==========================================================================

format shortG
close all;

numpts = 1.0e5;
M = linspace(1.0e2,1.0e6,numpts); %moment increments

%column section properties

[~,d,tw,bf,tf,Iz,~,~,~,~,~,~,~,~,~] = wsection(section);

%constants
V = M./z; %constant moment-to-shear ratio
Mpa = 6.89475908677537; %ksi to Mpa conversion
ki = 1.7; %inner panel strength factor set to 1.7
ko = 1.25; %outer panel strength factor set to 1.25

79
bi = bf; %inner joint width
bo = (B - bf)/2; %outer joint width
bj = bi + bo; %effective joint width
dj = d - tf;
Es = 29000; %elastic modulus steel
Ec = 3605; %elastic modulus concrete

%concrete bearing factor

beta1 = 0.85;
if fc >= 4.5; beta1 = 0.80; end;
if fc >= 5.0; beta1 = 0.75; end;
if fc >= 5.5; beta1 = 0.70; end;
if fc >= 6.0; beta1 = 0.65; end; %ACI 318 (R10.2.7)

%parameters
gamma = beta1.*2.0; %concrete bearing capacity enhancement due to
%confinment effects
beta = 0.6; %maximum fraction of neutral axis depth that
%horizontal bearing depth can attain
dv = 0.3.*N; %fraction of N that contributes to vertical bearing
mu = 2; %calibrated parameter to adjust kappa
nu = 1.5; %calibrated parameter to adjust breakout capacity
C = 1.77; %calibrated parameter to adjust dref

%failure flags
hb_flag = 0;
ps_flag = 0;

%calculated constants
n = 1/2*(N - d); %flap length

%determine dref and alpha==================================================

betaref = (Ec/(4*Es*Iz))^(1/4);
dref = C/betaref; %reference depth
if dembed > dref,
dembed = dref;
end

alpha = 1 - dembed/dref; %alpha assumed linear function of dembed

%RESPONSE==================================================================

for j = 1:numpts

%horizontal moment response================================================

%horizontal bearing

Vbar = V(j)./(gamma.*fc.*bj);

dR = (1/2).*(Vbar + dembed - sqrt(dembed.^2 - 2.*dembed.*Vbar -...

Vbar.^2 - 4.*Vbar.*z + 4.*alpha.*Vbar.*z));

80
dL = dR - Vbar;

Mbearing_hb_test = M(j).*(1 - alpha);

Mbearing_max = (-dembed - 2.*z + 2.*alpha.*z +...
1./2.*sqrt(-4*(-2*beta*dembed^2 + beta^2*dembed^2) +...
(2*dembed + 4*z - 4*alpha*z)^2)).*z.*gamma.*fc.*bj.*(1 - alpha);

if hb_flag == 0
if Mbearing_hb_test > Mbearing_max
hb_flag = 1;
Mbearing_hb = M(j).*(1 - alpha);
end
end

%inner panel

if ps_flag == 0
if Mbearing_hb_test > Mbearing_max
Mshear_hb = inf; %if shear failure not achieved when dR = dRmax,
%panel shear capacity set to infinite
ps_flag = 1;
end
end

Vspn = 0.6*Fy*tw*(dembed - (dR + dL)/2); %strength of steel web panel

Vicn = min(ki*sqrt(fc*Mpa)*bi*(dembed ...

- (dR + dL)/2)./Mpa,0.5*fc*bf*dj); %strength of inner strut

Von = ko.*sqrt(fc*Mpa)./Mpa.*bo.*(dembed -...

(dR + dL)/2); %strength of outer strut

M_ps = (Vspn*(d - 2*tf) + Vicn*dj + 1.25*Von*dj);

if ps_flag == 0
if M_ps <= Mbearing_hb_test
ps_flag = 1;
Mshear_hb = M(j)*(1 - alpha);
end
end
%==========================================================================
end

%vertical moment response==================================================

if P >= 0
alphaP = 1.0; %fraction of P transferred to the base, compression
elseif P < 0
alphaP = 1.0; %fraction of P transferred to the base, tension
end

%bearing failure check

Mb_vb = (1.7*fc - alphaP*P/(B*N))*(N - dv)*B*dv;

81
%base plate yield check

My_vb = (tp^2*Fy/(2*n^2) - alphaP*P/(B*N))*(N - dv)*B*dv;

%breakout check

%CCD

Anc = min(W*(dv + 3.*dembed),(dv + 3.*dembed)*(B + 3.*dembed));

%breakout area
%adjusted only if width of block interferes
%with breakout cone

Qc = (1./sqrt(dembed).*(40./9).*sqrt(fc.*1000)./1000.*Anc).*nu;

Mt_vb = (Qc + alphaP*P*dv/N)*(N - dv);

% end

%NOTE: no check for downward bearing in code

%failure mode combinations ================================================

[Mcap_v,indv] = min([Mb_vb,My_vb,Mt_vb]);
[Mcap_h,indh] = min([Mbearing_hb,Mshear_hb]);
kappa = nan;

if Mcap_v/alpha <= Mcap_h/(1 - alpha)

cse = 1;
[Mcap,ind] = min([Mt_vb/alpha, Mcap_h + Mb_vb, Mcap_h + My_vb]);
else
cse = 2;
kappa = ((Mcap_v*(1 - alpha))/(Mcap_h*alpha)).^(-mu);
if Mbearing_hb < Mshear_hb
[Mcap,ind] = max([kappa.*Mcap_h + Mcap_v, Mcap_h./(1 - alpha)]);
else
Mcap = Mcap_h/(1 - alpha);
ind = 3;
end
end

%print out=================================================================

if indv == 1
vcontrol = 'Vertical bearing';
elseif indv == 2
vcontrol = 'Base plate yield';
elseif indv == 3
vcontrol = 'Breakout';
end

if indh == 1
hcontrol = 'Horizontal bearing';
elseif indh == 2

82
 hcontrol = 'Panel shear';
end

if cse == 1
if ind == 1
maxcond = 'Breakout only';
elseif ind == 2
maxcond = 'Vertical bearing, horizontal';
elseif ind == 3
maxcond = 'Base plate yield, horizontal';
end

elseif cse == 2
if ind == 1
maxcond = 'Degraded horizontal bearing, vertical';
elseif ind == 2
maxcond = 'Horizontal bearing at maximum';
elseif ind == 3
maxcond = 'Panel shear only';
end

end

%print information to command window

fprintf('PARAMETERS: %g\n','')
fprintf(' %g\n','')
fprintf('alpha: %g\n',alpha)
fprintf('kappa: %g\n',kappa)
fprintf('gamma: %g\n',gamma)
fprintf('beta: %g\n',beta)
fprintf('dembed: %g\n',dembed)
fprintf('dreference: %g\n',dref)
fprintf(' %g\n','')
fprintf('CONTROL AND MAXIMUM CONDITION: %g\n','')
fprintf(' %g\n','')
fprintf('Case: %g\n',cse)
fprintf('Horizontal control: %s\n',hcontrol)
fprintf('Vertical control: %s\n',vcontrol)
fprintf('Maximum condition: %s\n',maxcond)
fprintf(' %g\n','')
fprintf('VALUES: %g\n','')
fprintf(' %g\n','')
fprintf('Mh_b: %g\n',Mbearing_hb./12)
fprintf('Mh_ps: %g\n',Mshear_hb./12)
fprintf('Mv_bpyield: %g\n',My_vb./12)
fprintf('Mv_tbreak: %g\n',Mt_vb./12)
fprintf('Mh %g\n',Mcap_h./12)
fprintf('Mv %g\n',Mcap_v./12)
fprintf(' %g\n','')
fprintf('Mcap %g\n',Mcap./12)
fprintf(' %g\n','')
Mcap = Mcap./12;

83