You are on page 1of 13

Column Base Connections for Hollow Steel Sections:

Seismic Performance and Strength Models


A. M. Kanvinde, M.ASCE 1; P. Higgins, M.ASCE 2; R. J. Cooke, S.M.ASCE 3; J. Perez 4; and J. Higgins 5

Abstract: The seismic response of exposed hollow steel section columns to base plate connections is examined through a series of eight
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

experiments. The prototype-scale tests investigate a range of variables including base plate size and thickness, column size, and anchor rod
layout (four rods in two rows, and eight rods in three rows). The specimens were subjected to cyclic flexural loading and instrumented to
provide direct (rather than inferred) measurement of tension forces in the anchor rods. All the specimens showed excellent deformation
capacity, with a stable hysteretic response for base rotations as large as 0.057–0.13 rad. Three specimens failed by fracture of the
weld between the column and the plate, whereas five did not fracture. Evaluation of the test data against the current design approach prevalent
in the United States suggests that (1) the design approach is reasonably conservative but (2) does not address the effect of the third
(i.e., central) row of anchor rods; as a result, it cannot be used to design them. A new design method is presented that explicitly incorporates
the third row of rods. The new approach is evaluated against the test data and it is determined that the new approach reflects the internal
mechanics of the connections in an improved way while providing more accurate estimates of forces in the rods. Recommendations for
the design of the connections are outlined, along with ongoing work that leverages the deformation capacity of these connections for
displacement-based design. The limitations of the study are summarized, especially in terms of challenges to the generalization of its findings.
DOI: 10.1061/(ASCE)ST.1943-541X.0001136. © 2014 American Society of Civil Engineers.
Author keywords: Column base plates; Seismic design; Steel connections; Anchor bolt tension; Metal and composite structures.

Introduction one of the most popular approaches to characterizing the strength


of base connections worldwide. Recent work, including experimen-
In steel structures, column base connections transfer forces from the tal work (Gomez et al. 2010; Kanvinde et al. 2013), examines the
steel column to the concrete footing. For low-rise structures, such as assumptions inherent in the methods provided by Steel Design
moment resisting frames with less than three to four stories, or can- Guide One and has determined that, by and large, the design ap-
tilever column systems used in mezzanines, the base connection proaches are reliable and conservative if applied to common con-
often features a base plate similar to the one schematically illustrated nection configurations. However, these studies also point to
in Fig. 1. As shown in the figure, these connections include various limitations of the experimental/analytical data, as well as their adap-
components (i.e., column, base plate, anchor rods, and concrete/ tation to design approaches. Specifically, the following issues are
grout foundation) that interact under a variety of loading conditions noteworthy:
such as axial tension/compression, flexure, and shear. Informed by 1. The vast majority of experiments feature W-section columns
various experimental (DeWolf and Sarisley 1980; Thambiratnam subjected to major axis bending attached to base plates, and
and Paramasivam 1986; Astaneh et al. 1992; Fahmy et al. 1999; none examine the response of hollow steel section (HSS) or
Burda and Itani 1999) and analytical studies (Salmon et al. box column base plate connections. HSSs are commonly used
1957; Blodgett 1966; Maitra 1978; Ohi et al. 1981; Ermopoulos as corner columns in moment frames or as cantilever column
and Stamatopoulos 1996; Drake and Elkin 1999; Wald et al. systems for mezzanines and storage racks. HSS columns re-
1995), the Steel Design Guide One (Fisher and Kloiber 2006), pub- quire the deposition of an out-of-position curved weld at
lished by AISC, provides guidance for the design of these connec- the corner of the section. Moreover, these welds are single-
tions. The method presented in Steel Design Guide One represents sided welds deposited with no access to the inside of the col-
umn, with the possibility of incomplete fusion. As the column
1
Associate Professor, Dept. of Civil and Environmental Engineering, is subjected to flexure, these welds interact with the yield line
Univ. of California, Davis, CA 95616 (corresponding author). E-mail: in the base plate, which also forms at the corners (or edge) of
kanvinde@ucdavis.edu the section, possibly compromising the strength and deforma-
2 tion capacity (which is important in a seismic context).
Peter S. Higgins and Associates, 30765 Pacific Coast Highway #117,
Malibu, CA 90265. 2. The strength characterization methods presented in prior
3
Engineer, Forell Elsesser Engineers, 160 Pine St., San Francisco, CA research (and most of the previous tests) are applicable to a
94111. connection configuration where four anchor rods are used (like
4
Graduate Research Assistant, Dept. of Civil and Environmental the one shown in Fig. 1). Often, alternate configurations may
Engineering, Univ. of California, Davis, CA 95616. be used with eight anchor rods, like the one shown in Fig. 2, to
5
Engineer, Peter S. Higgins and Associates, 30765 Pacific Coast High-
provide additional resistance. The internal stress distributions
way #117, Malibu, CA 90265.
Note. This manuscript was submitted on February 15, 2014; approved in these connections are the result of nonlinear and indetermi-
on June 10, 2014; published online on August 12, 2014. Discussion period nate interactions between the various components, such as the
open until January 12, 2015; separate discussions must be submitted for plate, rods, and grout (Ermopoulos and Stamatopoulos 1996;
individual papers. This paper is part of the Journal of Structural Engineer- Kanvinde et al. 2013). Consequently, these different config-
ing, © ASCE, ISSN 0733-9445/04014171(13)/$25.00. urations alter the internal stress and force distributions in

© ASCE 04014171-1 J. Struct. Eng.

J. Struct. Eng.
Applied force and
the rods and the concrete and affect the pattern of development
moment
of yield lines in the base plate.
Steel 3. With reference to the previous point, the design approaches and
Column strength characterization methods are based on internal stress
Base
and force distributions that have been examined only indirectly
Plate due to the difficulty of measuring internal stresses in a founda-
tion. More specifically, these methods have been validated
through the overall agreement of specimen strength (as ob-
Grout
tained from tests) with strength calculated with a method that
Tensile forces uses these stress distributions, rather than through direct mea-
Bearing
in anchor rods surements of these stresses and forces inside the connection.
stresses in
The preceding Points 2 and 3 are especially relevant if non-
footing
ductile anchors, such as postinstalled anchors, are used. In these
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

Concrete Anchor situations, accurate characterization of internal forces (rather than


Foundation Rods overall connection strength) is even more critical to prevent sudden
failure of the connection. Motivated by the aforementioned issues,
this paper presents test results and an analysis of a series of eight
Fig. 1. Schematic illustration of typical exposed base plate tests on HSS column to base plate connections. Fig. 3 and Table 1
connection illustrate the test setup and the test matrix (discussed in detail sub-
sequently). In reference to these, the test specimens (in the form of
cantilever columns) were approximately prototype scale, with var-
iations in plate size and thickness and in anchor rod layout. All spec-
imens were subjected to cyclic lateral loading, and the anchor rods
were instrumented with load cells for direct measurement of forces.
The objectives of the test program and subsequent analysis are
to (1) characterize the response, i.e., strength, deformation, and
hysteretic characteristics, of HSS column base connections, (2) ex-
amine the accuracy of existing methods for alternate configurations
and relative to directly measured anchor force data, and (3) refine
design methods and philosophies for the use of these types of con-
nections in structural systems. The paper begins by providing a
brief overview of the design methods and other background regard-
ing the design of column base connections. The testing program is
then described. This is followed by an analysis of the test data,
as well as a refined method for strength characterization, for an
alternate connection configuration, such as that shown in Fig. 2.
Conclusions are then presented along with recommendations for
design, ongoing work (that focuses on the development of a dis-
placement-based design method that leverages the deformation
capacity of these connections), and the limitations of the study.

Background
Fig. 2. Hollow steel section column base plate with eight anchor
rods The main objective of this paper is to examine and refine existing
strength characterization methods for HSS column bases and those

1500 X 900 X 460 mm footing

Loading
Actuator

3200 mm

(b)
460 mm
(a)
Three sets of reusable holes
Set 1 – Four holes at corners of 230 mm square (Test 1--4)
Set 2 – Eight holes at corners and sides of 380 mmsquare (Tests 5 and 6)
Load cells Set 3 – Eight holes at corners and sides of 530 mmsquare (Tests 7 and 8) (c)

Reaction floor

Fig. 3. Test setup: (a) schematic elevation; (b) plan view of concrete footing; (c) test in progress (Test 1 shown at approximately 12% drift)

© ASCE 04014171-2 J. Struct. Eng.

J. Struct. Eng.
Table 1. Test Matrix and Key Results
Column Base plate
material material
Plate Peak Peak
properties properties
size base drift
Test Fy Fu M column
p ¼ Z · Fy B, N tp Fy Fu Rod rotation Δmax M test
max M test
max = M max =
test

(mm) (MPa) (MPa) layout θmax DG−1 M new


number Column size (mm) (MPa) (MPa) (kN · m) (mm) base (rad) (rad) (kN · m) M

1 HSS 127 × 127 × 9.5 457 538 78.8 305 × 305 19 337 487 4 rods 0.130 0.166 59 1.58 NA
2 0.071 0.111 57 1.52
3 HSS 152 × 152 × 9.5 444 524 114.0 305 × 305 19 4 rods 0.086 0.112 77 2.03
4 0.082 0.110 83 2.19
5 HSS 203 × 203 × 9.5 414 514 197.9 457 × 457 19 8 rods 0.092a 0.111a 115 1.60b 1.50
6 0.110 0.129 145 2.02b 1.89
7 HSS 254 × 254 × 12.7 414 518 408.6 609 × 609 31.75 354 518 8 rods 0.057a 0.070a 239 1.26b 1.05
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

8 0.058a 0.070a 234 1.24b 1.03


a
Fracture observed at weld.
b
Steel Design Guide One method applied disregarding inner row of rods.

with alternate anchor rod layouts. To provide context for this, this
section presents an overview of these strength characterization
approaches, the assumptions that they are based on, and current
design guidelines. As discussed previously, the response of
base plates is controlled by nonlinear interactions between the
plate, rods, and supporting concrete. Explicitly incorporating
these effects into strength models is cumbersome and unsuitable
for application within a design setting (e.g., Ermopoulos and
Stamatopoulos 1996). As a result, the prevalent approaches to
designing base connections rely on a predetermined form of stress
distribution (or stress block), such as linear (triangular) or constant
(rectangular) distribution on the bearing side of the connection, and
then determine forces in the various components by establishing
equilibrium on the entire connection; these are by far the most (a)
popular types of methods in use today; in fact, they (specifically
those in Steel Design Guide One) are the sole approach used in
the design of base plate connections in the United States.
As per the approach presented in the Steel Design Guide One,
the base connection resists the applied axial compression and
moment through the mechanism illustrated in Fig. 4. If the
moment is low relative to the axial load, then the load combination
may be resisted exclusively through the development of the bearing
stresses (i.e., the plate does not tend to uplift)—the so-called low-
eccentricity condition—Fig. 4(a). On the other hand, if the moment
is large relative to the axial load, then tension must develop in the
rods to prevent uplift—the high-eccentricity condition—Fig. 4(b).
The two conditions are separated by a critical value of load eccen-
tricity ecrit ¼ M=P, which may be calculated using the following
formula given in Steel Design Guide One:
(b)
N P
ecrit ¼ − ð1Þ
2 2 · B · f DG−1
max Fig. 4. Assumed stress distributions for base connection design:
(a) low eccentricity; (b) high-eccentricity condition; third row of rods
The preceding equation assumes the development of a rectan- shown in (b)
DG−1
gular stress block (with a constant magnitude = f max ) on the bear-
ing side, i.e., the compression side of the connection. The symbols
P, B, and N represent respectively the applied axial load, length,
incorporates the effect of confinement on the bearing stresses. For
and width of base plate. The stress magnitude f DG−1 is the lesser
max
the low-eccentricity condition, i.e., e < ecrit , the only possible limit
of the bearing strength of the concrete or the grout (if used). In
DG−1 state is the flexural yielding of the base plate on the bearing side of
general, fmax may be determined using Eq. (2) as follows:
the connection due to the upward bearing stresses. The moment in
sffiffiffiffiffi the base plate may be calculated in two steps, which are outlined in
DG−1 0 A2
f max ¼ 0.85 × f c × ≤ 1.7 × fc0 ð2Þ Steel Design Guide One. Briefly, these are, first, to consider the
A1 overall force and moment equilibrium of the connection to establish
the bearing stress f, which (for e < ecrit ) is bounded by f DG−1
max , and
The symbols in Eq. (2) are defined in the notation list. the stress block length Y DG−1 , and, second, to calculate the plate
The term in the equation corresponding to the ratio of areas A2 =A1 bending moment based on the assumption of a cantilever bending

© ASCE 04014171-3 J. Struct. Eng.

J. Struct. Eng.
of the flap that extends outward from the compression flange of the present the test results, analysis and discussion of the results,
column, i.e., assuming that the yield failure is parallel to the column and, finally, the implications for design.
compression flange.
On the other hand, if the anchor rods are engaged (almost
always the case in seismic loading), then the high-eccentricity con- Test Setup and Instrumentation
dition is active, i.e., e ≥ ecrit . In this situation, the forces in the Figs. 3(a and b) (introduced previously) show schematic illustra-
anchor rods T DG−1 and the bearing length Y DG−1 may be calculated tions of the test setup (in elevation and plan view, respectively),
using Eqs. (3) and (4) in what follows, also presented in Steel whereas Fig. 3(c) shows a photograph of a specimen being tested
Design Guide One: [with reference to Table 1, i.e., the test matrix, the specimen shown
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi in Fig. 3(c) corresponds to Test 1]. As shown in the figure, salient
2 · ½M þ P · ðN2 − gÞ aspects of the test setup are as follows:
Y DG−1 ¼ ðN − gÞ − ðN − gÞ2 − ð3Þ
f DG−1
max · B 1. All specimens were cantilever columns of length 3,200 mm
with cyclic lateral loading applied at the top. The length
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

T DG−1 ¼ f DG−1
max · B corresponds to a typical cantilever column system (such as a
 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi mezzanine), in which flexure dominates response. No axial
2 · ½M þ P · ðN2 − gÞ load was applied;
· ðN − gÞ − ðN − gÞ2 − −P
fDG−1
max · B
2. All welds were toughness-rated complete joint penetration
(CJP) welds deposited from the outside of the column;
ð4Þ 3. The base of the column was attached to a reusable concrete
footing supported by spacer beams, also shown in Fig. 3(a).
The preceding equations are derived by establishing vertical and
The spacer beams allowed access to the underside of the con-
rotational equilibrium on the connection. The symbols used in the
crete footing for the installation of load cells, which were in-
equations are defined in the notation list at the end of this paper. In
stalled at the lower end of all anchor rods between a nut–
the context of this study, it is useful to note two points. First, T DG−1
washer assembly and the underside of the concrete footing.
represents the total tension force in all anchor rods on the tension
In this way, the tension in the rods exerted compression on
side; if only two are present, then the sharing of load between them
the load cells. The top end of the anchor rods featured a similar
may be assumed to be equal, not necessarily the case if alternate rod
nut–washer arrangement, such that the anchor rods could only
layouts (e.g., three in a row) are used. Second, the methods outlined
be engaged in tension during uplift of the plate;
in Steel Design Guide One do not address the effect of an additional
4. Fig. 3(b) shows a plan view of the test setup. As shown in this
set of anchor rods [e.g., as shown in Fig. 2 or by the dashed lines in
view, the footing (designed to be reusable) had 24 holes (lined
Fig. 4(b)]. As a result, the force T DG−1 is an estimate of T outer
with PVC pipe) through the depth to allow for passage of
shown in the figure. Note that the addition of the inner rods intro-
anchor rods. The 24 holes were designed to accommodate
duces an indeterminacy in the system such that the internal forces
the three different base plate footprints and corresponding
may no longer be inferred based only on the force and moment
rod layouts;
equilibrium of the connection. As discussed subsequently, the
5. Major instrumentation included load and deformation at the
present study addresses this issue.
top of the column, i.e., at the actuator location; tension forces
In any case, once T DG−1 and Y DG−1 are determined as per
in all the anchor rods, measured by the load cells described
the foregoing process, the bending moments in the plate may be
previously; various displacement transducers to measure the
determined assuming an upward cantilever bending of the flap
uplift/rocking of the entire foundation; and strain gages on
on the bearing side of the connection or the downward forces in
the top surface of the base plate to monitor local strains.
the anchor rods on the tension side. These may then be used to
The base plates were not entirely flat at the time of specimen
appropriately detail the plate (i.e., thickness) or the rods (size,
installation, presumably because of warping induced by the column
grade, and embedment, based on consideration of several failure
welding process. To ensure contact between the base plate and
modes; e.g., Cimellaro and Reinhorn 2011). Note that (1) this
concrete (and to restore vertical alignment of the column), the nuts
method has not been developed for HSS members or connections
on the anchor rods were tightened, inducing a prestress in each of
with alternate rod layouts; in fact, consideration of additional rows
the anchor rods; this prestress ranged from 4–9 kN (for the thinner
of rods requires fundamental changes to the method to address the
plate, i.e., 19 mm) to roughly 25–30 kN for the thicker
indeterminacy produced by the additional rods; (2) even for con-
(i.e., 31.75 mm) plate. The results of the experiments are inter-
nections similar to those implied in the design methods, i.e., those
preted (in a subsequent section) with appropriate consideration
with two rows of rods, the distribution of forces among these rods is
of this prestress. In heavier columns (for multistory frames), col-
not addressed by the current methods; and (3) the internal stress and
umn alignment and base plate contact are often ensured using shim
force distributions are not based on independent measurements of
stacks or leveling nuts followed by grouting (e.g., Gomez et al.
rod force but rather the overall connection response. The following
2010). However, for low-rise frames, especially featuring postin-
section describes the testing program conducted to respond to these
stalled anchors, this type of leveling/grouting procedure is unfea-
issues.
sible. As a result, the prestressing method used in this study is
similar to construction practice (where the column is first plumbed
Test Setup, Instrumentation, Ancillary Testing, and and the nuts are then differentially turned without disturbing the
Test Matrix plumbing).

A series of eight experiments was conducted on HSS column base


Ancillary Testing and Material Properties
connections. These tests were complemented by material testing of
the base plate, anchor rod, and concrete used in the fabrication of Table 1 summarizes the material properties of the various materials
the test specimens. This section describes the test setup and instru- in the connection. The 28-day strength of the concrete in the
mentation, test matrix, and applied loading. Subsequent sections footing (which was common for all the tests) was measured to

© ASCE 04014171-4 J. Struct. Eng.

J. Struct. Eng.
be f c0 ¼ 28.1 MPa based on standard cylinder tests. All rods were column. The symbols in the equation are defined in the notation list
19 mm diameter ASTM F1554 Grade 105 (i.e., Fu ¼ 723 MPa); at the end of the paper. A qualitative assessment of the experimental
by design, they did not yield in any of the experiments. The steel response for the various tests is now presented to facilitate the in-
properties of the column and base plates were obtained from mill terpretation of quantitative data, which is discussed subsequently.
certificates. The columns were nominally A500 Grade B (Fy ¼ All the tests share some features of the qualitative response. These
317 MPa), whereas the plates were nominally A36 (i.e., Fy ¼ are first described, and then responses peculiar to specific speci-
248 MPa). The table shows that the column and plate yield (and mens are presented. For all specimens, the initial elastic response
ultimate strengths) are greater (by approximately 35%) as com- was observed until a base rotation of approximately 0.0075–
pared to the specified values. The difference between the specified 0.01 rad (0.75–1%); this corresponds to a column drift of approx-
and expected strengths is common (Liu et al. 2005) and typically imately 1–1.5% [which is consistent with the expected yield drift
incorporated into capacity design by way of the Ry or Rt factors in moment frame systems (Krawinkler et al. 2000)]. This was fol-
(AISC 2010). The columns were sized to remain elastic even as lowed by yielding of the base plate on the compression side (due to
the connection developed its full capacity. As shown in the table, the upward bearing stress) and then yielding of the base plate on the
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

the flexural capacity of the column (based on the measured yield tension side due to the tension in the anchor rods. The reversed
strength) was greater compared to the maximum moment M test max for cyclic yielding of the base plate produced a fairly stable hysteretic
all the tests, suggesting that there was no yielding in the column and response, as shown in Figs. 5(a–h). A slight pinching was also ob-
that the measured peak moments correspond to the connection served for the higher amplitude cycles (more than 2–3% drift) as
strength. All subsequent analyses (of connection strength/response) the base plate (which at this point in the loading had permanent
are based on the measured, rather than the specified, material deformations) engaged and disengaged (i.e., contacted and sepa-
properties. rated from) from the supporting concrete and the anchor rods. Note
that in some tests, this so-called pinching plateau is observed at
Test Matrix and Loading nonzero loads due to the differential prestress in the rods. This hys-
teretic response continued in a stable cyclic manner until either the
Table 1 also summarizes the test matrix. The table shows that the test was concluded (due to actuator limitations) or fracture of the
key variables were the plate dimensions (i.e., footprint and thick- weld between the column and the base plate was observed (in Tests
ness) and the anchor rod layouts, i.e., four or eight anchor rods. 5, 7, and 8).
Tests 1–4 (with 305 mm square base plates) featured the four- Fig. 6 photographically shows typical posttest deformations for
rod pattern, whereas Test 5–8 (with 457 or 610 mm square base two of these tests, i.e., Tests 2 and 5. Tests 1, 3, and 4, which feature
plates) featured the eight-rod pattern. Thus, three plate sizes the four-rod layout, show qualitatively similar deformations as Test
(i.e., footprints) were investigated, along with two plate thick- 2. Tests 6–8 with the eight-rod layout, show a deformation pattern
nesses, i.e., 19 and 31.75 mm. The anchor rods were placed in similar to that of Test 5, with the exception that Test 6 did not
a square pattern with an edge distance of g ¼ 38 mm for all the exhibit fracture, whereas Tests 5, 7, and 8 showed weld fracture.
experiments. Table 1 shows that all the experiments featured square Other than plate yielding (which is clearly visible for the specimens
HSS columns. The dimensions and layouts of the test specimens shown in Fig. 6), no visible distress was observed in any of the
are representative of prototypical configurations commonly used specimens until fracture (or conclusion of the test). In the early
in current construction practice. For each of the parameter sets, stages, plate yielding was evidenced by flaking of the paint in
two replicate tests were conducted. the region of the yield lines, whereas in the later stages of loading
All the specimens were subjected to the ATC-SAC loading pro- (drifts greater than 3–4%) it was observable in the form of perma-
tocol (Krawinkler et al. 2000) to represent deformation histories nent deformations in the plate. On the bearing side, slight (and
consistent with seismic demands in moment frame buildings. highly localized) crushing of the concrete was observed under the
The deformation history applied to the specimens (at the top) as corners of the compression side of the base plate. The anchor rods
per this protocol is expressed in terms of column drift ratio defined did not yield in any of the tests.
as the lateral displacement of the column at the application of the The main qualitative difference between the response of the tests
lateral load divided by the distance between the load and the top of was in the mode of deformation of the base plate. In this regard, the
the base plate (3,200 mm for all tests). In all except three of the response may be grouped into two categories corresponding to the
eight tests (i.e., Tests 5, 7, and 8) failure did not occur during two different types of anchor rod layouts. For Tests 1–4, the yield
the loading history. Consequently, a monotonic excursion was lines in the base plate on the tension side of the connection were
appended to the end of the history, in which the magnitude of inclined with respect to the loading direction because only two an-
the excursion was controlled by the stroke limit of the actuator chor rods were present in the corners. This is indicated schemati-
(consistent with a column drift of approximately 13%). cally in Fig. 6(a). This is similar to the type of response observed in
other testing programs (e.g., Gomez et al. 2010). Although Steel
Design Guide One does not explicitly address this type of response
Test Results
(and methods for characterizing associated strengths), it is ad-
Figs. 5(a–h) indicate the moment-rotation response for all the ex- dressed in research by Mazzei (2012) and AISC’s Steel Design
periments. Figs. 6(a and b) show photographs of two specimens at Guide Ten (Fisher and West 2003) in the context of the construction
the conclusion of testing. The moment-rotation response shown in of bracing of low-rise frames. Plate yielding in Tests 5–8 was con-
Fig. 5 plots the base moment (calculated as M base ¼ V top × Lcolumn ) strained by the presence of additional anchor rods. Consequently,
against the base rotation θbase , which is determined using Eq. (5) yield lines formed parallel to the HSS column edges, and the mode
of plate bending was different compared to the tests with the four-
Δtop V top × L2column rod layout. This is indicated schematically in Fig. 6(b). Also, as
θbase ¼ − ð5Þ
Lcolumn 3ðEIÞcolumn shown in this figure, the presence of additional rods also constrains
the yielding of the base plate. Presumably, this constrained mode of
This provides a direct assessment of the deformation response deformation led to fracture (in the weld connecting the HSS column
(and capacity) of the connection, discounting the flexibility of the to the base plate) in three of the four tests featuring the eight-rod

© ASCE 04014171-5 J. Struct. Eng.

J. Struct. Eng.
90 90
Test 1 Test 2
60 60

(kN. m)
30 30

0 0

-30 -30

-60 -60
(a) (b)
-90 -90
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 -0.15 -0.1 -0.05 0 0.05 0.1 0.15
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

90 90
Test 3 Test 4
60 60
(kN. m)

30 30

0 0

-30 -30

-60 -60
(c) (d)
-90 -90
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 -0.15 -0.1 -0.05 0 0.05 0.1 0.15

180 180
Weld fracture
Test 5 Test 6
120 120
(kN. m)

60 60

0 0

-60 -60

-120 -120
(e) (f)
-180 -180
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 -0.15 -0.1 -0.05 0 0.05 0.1 0.15

300 300
Test 7 Test 8
200 200

Weld Weld
(kN. m)

100 100
fracture fracture
0 0

-100 -100

-200 -200
(g) (h)
-300 -300
-0.15 -0.1 -0.05 0 0.05 0.1 0.15 -0.15 -0.1 -0.05 0 0.05 0.1 0.15
(rad) (rad)

Fig. 5. Moment rotation curves for all eight specimens; dashed lines indicate estimated capacities (Table 1)

layout. In these tests, sudden fracture (not preceded by any rotation θmax
base , and the maximum drift sustained by a column,
observable ductile tearing) propagated through the tension flange i.e., Δmax . In addition, the table also shows a comparison of the
(or edge of the HSS) and nearly all the way through the webs maximum base moment observed in the experiment M test max to the
(i.e., perpendicular edges of the HSS). Fig. 6(b) indicates this frac- estimates of connection strength. Two such estimates are used in
ture for Test 5. Table 1, introduced previously, indicates the defor- the table. One of these, denoted by MDG−1, is the estimated moment
mations corresponding to fracture for Tests 5, 7, and 8. based on Steel Design Guide One. This estimated base moment
Table 1 (introduced previously) summarizes key quantitative corresponds to the initiation of flexural yielding on the tension side
data measured from the tests. This includes the maximum moment of the base plate. For Tests 1–4, this is based on the assumption of
sustained by a connection during a test (i.e., M max
test ), the peak base inclined yield lines as defined by Mazzei (2012) and the measured

© ASCE 04014171-6 J. Struct. Eng.

J. Struct. Eng.
(a) (b)

Column Weld fracture


Column
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

Footing Inclined yield Footing Straight yield


lines in plate line in plate

Fig. 6. Posttest photographs indicating deformation modes and damage: (a) Test 2; (b) Test 5

material properties summarized in Table 1. Note that the method stresses that are even higher Fu than (since Fu corresponds
outlined in the Steel Design Guide One also considers the limit state to necking and localization of a tension coupon; such necking
of plate yielding on the compression side of the connection. How- is absent during plate bending); (3) as the plate is lifted up,
ever, as demonstrated by Gomez et al. (2010), this limit state does membrane action is mobilized, further increasing the load car-
not affect the overall connection strength, which requires the mo- rying capacity; finally, (4) referring to the prior discussion, the
bilization of other yielding modes (such as yielding of the base rods carried some prestress at the beginning of the test; this
plate on the tension side or yielding of the anchor rods). Moreover, elevates the base moment required for lift-off of the plate
Steel Design Guide One does not address three rows of anchor rods and subsequent yielding of the base plate. For Tests 5–8,
(such as are present in the eight-rod layout, i.e., Tests 5–8). Thus, the Steel Design Guide One is inherently conservative (as are
for Tests 5–8, M DG−1 is calculated by disregarding the middle row the previously considered factors) because it disregards the
of the anchor rods to examine the efficacy of the Steel Design Guide effect of the third (middle) row of anchor rods. However, it
One in characterizing the connection strength under such condi- is interesting to note that the degree of conservatism for
DG−1
tions. Three observations may be made based on Table 1 and Tests 1–4 is larger (average M test
max =M ¼ 1.83) compared
DG−1
the moment-rotation curves shown in Fig. 5 to that for Tests 5–8 (average M test
max =M ¼ 1.53). This may
1. In all the experiments, an excellent deformation capacity be attributed to fracture in three of these (latter) tests, which
and stable hysteretic response were observed. As shown in diminished the deformation capacity relative to Tests 1–4,
Table 1, the maximum observed rotations were in the range thereby reducing the influence of material hardening and
of 0.057 rad (Test 7) to 0.13 rad (Test 1). The corresponding the associated increase in the base moment. It appears that this
column drifts were in the range of 0.07–0.16 rad (i.e., 16% effect offsets the additional strength provided by the third row
drift). This observation is even more noteworthy when viewed of anchor rods.
in the context of prequalification standards for beam–column A large difference between test and predicted moment capacities
connections (e.g., FEMA 2000), which require a stable re- has been observed in the context of base connections (Gomez et al.
sponse up to drift angles of 0.06, i.e., 6%. On the other hand, 2010) and in the context of other connections (end plate moment
it is not entirely unexpected, given that previous tests on connections) that have similar plate deformation modes producing
exposed column bases (Gomez et al. 2010) demonstrated a high strain hardening and membrane action (Adegoke and Kemp
similar deformation response. 2003). In general, when a connection is designed to remain elastic,
2. The rotation capacity for specimens with an eight-rod layout this overstrength is not a problem (although it does indicate that
was lower compared with those with a four-rod layout. An in- plates may be somewhat oversized). However, it may become a
spection of Fig. 6(b) suggests the underlying cause: the addi- relevant issue if either the connections are designed as a ductile
tional anchor rods constrain the yielding of the base plate, (or fuse) element or the anchors are designed to develop plate yield-
thereby promoting the initiation of fracture. ing. In these cases, more accurate methods to characterize plate
DG−1
3. The ratio M test
max =M (based on measured, rather than spe- strength must be developed. Given that the strength is controlled
cified, material properties) are in a range of 1.24 (Test 8) to by strain hardening and membrane action, this type of strength es-
2.03 (Test 3). This indicates that the method presented in Steel timate will likely be sensitive to the expected level of deformation.
Design Guide One may be conservative. This observation is As discussed previously, all the anchor rods were instrumented
also consistent with previous research (Gomez et al. 2010) with load cells for direct measurement of tension forces.
and may be attributed to four factors: (1) the method uses Figs. 7(a–c) examine these measured forces (T test outer ) relative to those
the yield stress Fy of the base plate and disregards strain hard- estimated as per Steel Design Guide One (T DG−1 ). More specifi-
ening, which is evidently mobilized in the base plate given cally, the figures plot the evolution of the test to predicted ratio
T test DG−1 versus the deformation. Before discussing the trends
the extent of inelastic deformation; note that the measured outer =T
Fu (Table 1) is in the range of 480–520 MPa, roughly 50% observed in the figures, it would be useful to note a few points
higher than Fy used for the calculation of M DG−1 ; (2) the yield about the figures themselves. First, the T test outer values reflect the
lines in the plate are subjected to extremely large strains at sum of all the rods in the outer row of anchor rods minus the
high deformations; these strains may generate so-called true prestress recorded at the beginning of the test, thereby providing

© ASCE 04014171-7 J. Struct. Eng.

J. Struct. Eng.
2
Test 1 authors (Kanvinde et al. 2013). This also suggests that for
Test 2 the bolt edge distances and plate thicknesses used in this study,
prying forces (induced by the corners of the plate bearing on
the concrete) are not a significant factor.
3. For Test 5–6, the test-predicted ratios approach values lower
1 than 1.0. This is not unexpected since the estimates do not
consider the contribution of the middle row of anchor rods.
However, it is interesting to note that the values are not sig-
nificantly lower than 1.0 but in the range of 0.8–1.0, suggest-
ing that Steel Design Guide One may be used conservatively to
0 size these anchors.
(a) 0 0.05 0.1
In summary, the method presented in Steel Design Guide One
2 provides reasonably accurate estimates of the total force in the outer
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

Test 3
Test 4 row of the anchor rods. It also provides conservative estimates of
the connection strength. However, in the context of the connections
with an eight-rod layout, it has some limitations. First, since it dis-
regards the inner row of rods [Fig. 4(b)], it cannot be used to de-
1 termine forces in them or to design them. Second, the method
calculates the forces in a row of rods, rather than in individual rods.
For a four-rod layout, only two rods are present in each row; the
distribution of the force between these is trivial (i.e., the force dis-
tributes equally). However, as discussed in the next section, the dis-
(b)
0 tribution is more complex for the situation where three rods are
0 0.05 0.1 present in a row. Third, though the method provides reasonable es-
2 timates of overall strength and anchor rod forces (in the outer row)
Test 5
for Tests 5–8 in this study, it cannot be generalized to other plate
Test 6
Test 7
sizes or thicknesses since it disregards the inner row of rods. In fact,
Test 8
if the contribution of the inner row is disregarded in the design
method, it is difficult to justify the use of these rods, which entail
1 additional cost. The next section proposes an approach to character-
izing connection responses in these situations.

Proposed Method for Characterizing Response of


0 (radians) Connections with Eight-Rod Layout
0 0.05 0.1
(c) (rad) As discussed previously, the response of the base connections is the
result of nonlinear interactions whose explicit characterization is
Fig. 7. Comparison of estimated rod forces (as per Steel Design challenging even for the four-rod layout. Thus, previous ap-
Guide One) with measured forces for (a) Tests 1, 2; (b) Tests 3, 4; proaches (including Steel Design Guide One) rely on assumed
(c) Tests 5, 6, 7, 8 stress distributions under the plate to facilitate design. The
eight-rod configuration is even more challenging to characterize,
primarily because of the increased degree of static indeterminacy
a basis for consistent comparison with corresponding estimates within the connection. The method presented in this section has the
T DG−1 , calculated as per Eq. (4). Second, the values are plotted following features: (1) it does not characterize the nonlinear inter-
against the absolute value of the peak base rotation (i.e., jθbase j) actions (discussed previously) in an explicit way but rather seeks to
for each positive and negative cycle of loading. Third, since the load- provide a straightforward method that may be applied in a practical
ing is reversed-cyclic, the T test setting, and (2) it augments the Steel Design Guide One method
outer values reflect the forces in the rods
that are in tension for a particular cycle; in this way, data from both with a relatively simple extension, such that it degenerates to
positive and negative cycles may be plotted conveniently. Finally, the Steel Design Guide One method if the middle row of anchor
since the objective of Fig. 7 is to assess the efficacy of the Steel rods is absent.
Design Guide One, only the forces in the outer rows of rods are Referring to Fig. 4 shown previously, the applied force and
shown. With reference to Figs. 7(a–c), three points may be made: moment may be resisted through the development of a bearing
1. For all tests, initially (i.e., for drifts less than 0.25%) the test- stress block on the compression side and forces in the anchor
predicted ratios oscillate as the rods begin to engage; note that rods on the tension side of the connection. For the low-eccentricity
while the values oscillate, the magnitude of the forces in the condition, i.e., when there is no plate uplift, the four- and eight-rod
rods is fairly low (i.e., less than 5–10% of the value at peak configurations may be assumed to respond in an identical manner,
moment). However, as loading progresses (deformation in- i.e., the eccentric force is resisted entirely by the development of the
creases), the test-predicted ratios stabilize. stress block in the concrete. However, when e ≥ ecrit [Eq. (2)], the
2. For Tests 1–4, the values approach a value of 1.0, indicating anchor rods are engaged. As shown in Fig. 4(b), the following re-
that Steel Design Guide One provides a fairly accurate esti- lationships may be established based on equilibrium of the entire
mate of the rod forces. For the range of plate thickness inves- connection:
tigated in this study, this is consistent with similar observations new × f DG−1
from finite element simulations conducted previously by the P þ T new
outer þ T inner ¼ B × Y
new
max ð6Þ

© ASCE 04014171-8 J. Struct. Eng.

J. Struct. Eng.
    2
N N Y new Test 5 - Outer
T new
outer × − g þ B × Y new × f DG−1
max × − ¼ M ð7Þ
2 2 2 Test 6 - Outer
Test 5 - Inner
Test 6 - Inner
Eq. (6) establishes the vertical force equilibrium, whereas
Eq. (7) establishes the moment equilibrium. The two preceding
equations contain three unknowns (i.e., T new outer , T inner , and Y
new new ) 1
and thus cannot be solved unless an additional condition is intro-
duced. It is proposed that the tension in the inner rods should be
considered as being directly proportional to the tension in the outer
rods, i.e., T new
inner ¼ k × T outer . This assumption is based on a general
new

kinematic consideration by which the deformations in the two rows 0


of rods are constrained by the plate. However, it is also recognized (a) 0 0.05 0.1
that the plate is flexible and that the plate or rods will likely yield as
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

2
loading progresses, such that this direct proportion may no longer Test 7 - Outer
be valid. However, an explicit consideration of these effects will Test 8 - Outer
Test 7 - Inner
necessitate not only further complexity (an obstacle to implemen- Test 8 - Inner
tation) but, perhaps more importantly, the introduction of additional
parameters that cannot be robustly calibrated with the available
data. For the tests in this study, P ¼ 0; substituting this into Eqs. (6) 1
and (7) and solving simultaneously results in
M
T new
outer ¼ ð8Þ
N − 2g ð1 þ kÞ
0
T new
outer × ð1 þ kÞ
Y new ¼ DG−1
ð9Þ 0 0.05 0.1
B × f max (b)

Once T new
outer is determined, the force in the inner row of rods may Fig. 8. Comparison of estimated rod forces (as per proposed method)
be determined as T new inner ¼ k × T outer . Note that for this study, direct
new
with measured forces for (a) Tests 5, 6; (b) Tests 7, 8
measurements of T outer and T inner are available. Based on these
new new

measurements, a value of k is determined (through trial and error)


to achieve the best overall agreement with force measurements
from all tests. This optimal value is determined to be k ¼ 0.5. distributions that produce them) are sensitive to the relative
Figs. 8(a and b) present the evolution of force in the anchor flexibility of the various interacting components such as the
rods (in Tests 5–8) in a manner similar to Figs. 7(a–c). Unlike base plate and supporting concrete.
Figs. 7(a–c), Figs. 8(a and b) illustrate the results of the new method 4. As shown by Fig. 8, the test-predicted ratios for the inner rods
and, thus, also include estimated forces for the inner row of anchor approach unity only as the deformations become fairly large
rods. With reference to Figs. 8(a and b) (and Table 1 introduced (i.e., 3–4% drift). Thus, it may be argued that their influence
previously) some observations can now be made: can be neglected for lower drifts that are more consistent with
1. For all four tests, the test-predicted ratios of the forces in the design-level deformations. Nonetheless, the use of the refined
outer rods are either close to 1.0 or slightly lower, indicating method is still desirable for two reasons. First, for the outer
that the method is reasonable (and perhaps slightly conserva- row of rods, the refined method (with k ¼ 0.5) provides less
tive) for estimating the forces in these rods. For all four of conservative and more accurate estimates of forces (relative to
these tests the estimated forces in the outer row of rods is be- the Steel Design Guide One method) even for smaller defor-
tween 5 and 10% lower than the corresponding estimates from mations (compare test-predicted ratios for the outer rods in
the current, i.e., Steel Design Guide One, method because the Figs. 7 and 8). Second, and perhaps more importantly, the pro-
contribution of the inner row is considered. posed method provides a way to design the inner row, whereas
2. The test-predicted ratios for the inner rods evolve from a value the Steel Design Guide One method provides no such method.
of zero and approach (or even slightly exceed) unity as the test In this context, too, the validity of k ¼ 0.5 may be questioned,
progresses. This is because initially, only the outer row of rods given that this characterizes rod forces well only at large dis-
is engaged (such that no forces are measured in the inner row). placements. It is important to consider the alternatives, which
Upon subsequent loading, and deformation of the plate, the include the use of a nonlinear evolution relationship for k,
inner rods begin to develop force. which would be cumbersome to generalize and apply in a
3. The preceding Point 2 suggests that the relationship between design setting, or the use of k < 0.5, which would be noncon-
the forces in the inner and outer rows is nonlinear, such that the servative if deformations greater than 3–4% drift were indeed
inner rods carry little (≈0) load at the beginning of the test, but encountered. Thus, the use of the refined approach (with
as deformations increase, the accuracy of the method (which k ¼ 0.5) appears to have little downside while providing a
assumes a constant ratio between the forces in the inner and rational approach to designing the connection, especially
outer rows) increases. It is also interesting to note that the test- the inner row of rods.
predicted ratios for all the rods are lower for Tests 7 and 8 5. As shown by Table 1 (the last column, i.e., M test
max =M
new ), the

compared to Tests 5 and 6. Note that Tests 7 and 8 have a test-predicted ratios for the new method are slightly lower
larger and thicker base plate (610 × 610 × 31.75 mm) and a (i.e., less conservative) than those determined as per the Steel
larger column (HSS 254 × 254 × 12.7) compared to Test 5 Design Guide One method. The moment M new was calculated
and 6. This underscores that the rod forces (and the stress in a manner identical to MDG−1 , with the exception that T newouter

© ASCE 04014171-9 J. Struct. Eng.

J. Struct. Eng.
were used instead of T DG−1 to determine the magnitude of
plate bending.
As shown by Table 1 (and the prior discussion regarding
M test DG−1 ), the average test-predicted ratio for the four-rod
max =M
tests from the Steel Design Guide One method was higher
(i.e., 1.83) than the corresponding test-predicted ratio when applied
to the eight-rod tests (1.53). Thus, the premise of the proposed
method that increases the estimates of base moment capacity
(thereby resulting in a lower average M test max =M
new
¼ 1.37) may
be questioned because it suggests that the new method predicts
a greater strength for the subgroup of tests (Tests 5–8), which
on average have a smaller test-predicted ratio as per the Steel
Design Guide One method. Two points may be made in response
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

to this concern. First, the measured M test


max values for both the four-
rod and the eight-rod tests are obtained at base rotations in the range Fig. 9. Distribution of tension forces among three outer rods
of 0.058–0.13 (drifts of 0.07–0.166). These are well in excess of
any expected rotations and, thus, are not representative of the
strengths that may be expected in design-level scenarios. Within deformations increase, the middle rod carries a progressively
this range, the higher test-predicted ratios of the four-rod specimens larger fraction of the load, reaching a peak value of roughly
are an artifact of their greater deformation capacity relative to the two-thirds (≈0.65) the total load in the three rods (for Test
eight-rod specimens rather than a consequence of the internal 5; the ratios are slightly lower for the other three tests). This
stress/force distribution or the mechanics. Second, the measure- is not surprising, given that the upward force is introduced
ments of anchor rod forces indicate that the proposed method pro- into the base plate by the tension flange of the HSS located
vides a better indication of the internal stress distribution. From a in the central portion of the plate width (i.e., near the middle
design standpoint, where the deformation magnitudes observed in rod rather than the corner rods). This may be alternatively
the tests are not expected, it is prudent to use a method that reflects explained based on plate-bending solutions (Timoshenko
the internal mechanics of the connection. Consistency with internal and Woinowsky-Krieger 1959) or by invoking a so-called
mechanics is even more important for the design of nonductile an- tributary-width argument according to which the middle rod
chors since they are sensitive to force estimates in them, not to the carries a larger share of the load. In any case, it is clear that
overall connection strength. Thus, the use of the Steel Design Guide designing the rods to carry equal forces is not a satisfactory
One method for the eight-rod configuration has two potential prob- approach.
lems. First, it has the potential of mischaracterizing the internal 2. At very large rotations (i.e., jθbase j>0.05–0.06 rad), the ratio
stresses, especially the force in the inner row of rods. Second, it again drops to roughly 0.5 (i.e., the middle rod is still carrying
is influenced by the observation of higher overall connection half the total load). This may be attributed to a redistribution of
strengths in the four-rod specimens, which in turn are only an ar- stresses as yield lines form in the plate or to perhaps small
tifact of increased deformation capacity. As an alternative argu- prying forces in the corner rods (due to the downward bearing
ment, consider a hypothetical situation where each of the three of the plate corners).
rows of rods contains two rods (i.e., a six-rod configuration). The next section summarizes the study and presents conclusions
One may imagine that this has a deformation capacity similar to and their implications for improved design.
the four-rod configuration since the plate will deform with less
constraint and with the formation of inclined yield lines. Given
the similar deformation capacity, one may further anticipate that Summary, Design Considerations, and Limitations
the additional (inner) row of rods in the six-rod configuration will
result in additional strength over the four-rod configuration. This This paper presents findings from eight tests on exposed column
hypothetical, but consistent, comparison of the three-row versus base connections to examine their seismic response and the efficacy
two-row configuration (with two rods in each row) supports the of current design methods in ensuring safety and economy. With
use of the proposed method, which reflects the internal stress respect to previous work in the area, the main distinguishing
distribution in an improved way. features of this study were an examination of HSS rather than
The distribution of forces among the three rods in the outer row W-section columns, an examination of alternate anchor rod layouts
also bears scrutiny. This is because when only two rods are present featuring eight rods (i.e., three rows of rods), which is not addressed
per row, the distribution of forces among them is trivial (i.e., forces by current design methods or previous research, and a direct meas-
are shared equally), an assumption that is not self-evident when urement of anchor rod forces that provide refined insights into the
three or more rods are present in a row. Fig. 9 plots the ratio of internal force distribution within the connection for a critical evalu-
the load carried by the middle rod to the total anchor rod forces ation of current design methods.
in all three rods in the outer row, i.e., T middle =T test
outer . The figure All eight test specimens showed excellent deformation capacity,
is similar to Figs. 7 and 8 such that the values are plotted against such that column drift ratios in the range of 7–16% were achieved,
the absolute value of the peak base rotation (i.e., jθbase j) for each accompanied by a stable hysteretic response and only modest deg-
positive and negative cycle of loading. Moreover, the forces plotted radation in strength. The primary deformation mode was inelastic
in Fig. 9 also do not include the prestress in the rods present at the bending of the base plate, with minor concrete crushing at the
beginning of the test. With reference to the figure, the following extremities of the plate. This deformation mode was exceptionally
observations may be made: ductile, resulting in base rotation capacities (excluding column de-
1. At the beginning of the test, the T middle =T test
outer ratio is in the formations) in a range of 0.057–0.13 rad. In fact, five of the eight
range 0.3–0.45 (suggesting that the load sharing is approxi- tests were terminated without failure (owing to limitations of test
mately equal among the three rods). However, as the equipment). Fracture was observed in three of the tests (with the

© ASCE 04014171-10 J. Struct. Eng.

J. Struct. Eng.
eight-rod layout). The fracture initiated at the tension flange weld possible, then the Steel Design Guide One method may be
between the HSS column and the base plate and was sudden with used conservatively to characterize the forces in the outer
no prior observation of initiation. Presumably, the more constrained row of rods, and the total force in the inner row may be de-
form of plate yielding in the eight-rod tests promoted fracture in termined as being half of this value (based on experimental
these tests. observations). Once the force in each row is determined, spe-
The currently used approach for designing exposed column base cial attention must be paid to the outer row if it contains three
connections (e.g., the one in Steel Design Guide One) was evalu- rods because experimental data suggest that the load is not
ated against the test data. The approach was directly applicable to shared equally among them. The central rod carries up to
Tests 1–4 (which had the four-rod configuration, i.e., with two rows two-thirds of the total load in the row and should be designed
of anchor rods). For these, it was determined that the current ap- with that mind.
proach was able to characterize the forces in the anchor rods with While the preceding points pertain to the calculation of design
good accuracy (suggesting that it characterizes internal force/stress forces within the rod and the plate, perhaps the most important find-
distributions in connections in a reasonable manner). Moreover, ing of the study is the excellent deformation capacity (and hyster-
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

with regard to connection strength (i.e., maximum moment capac- etic response) of the base plates. This type of response is not
ity), the approach was significantly conservative; this is attributed anomalous but, rather, consistent with observations in previous
to prestress in the rods and hardening in the plate material, which studies that have similarly shown excellent deformation response
was mobilized by the high degree of inelastic deformation. This in exposed-type base connections. As per current design practice
approach was also applied to Tests 5–8, disregarding the inner for moment frames (e.g., SEAOC 2012), base connections are de-
row of anchor rods, to examine its efficacy in these situations. signed to be stronger than the attached column to force plastic
The relatively similar degrees of conservatism for the two sets yielding into the column rather than the base connection. Thus,
of tests (i.e., with four-rod and eight-rod configurations) also sug- base connections are often designed for amplified seismic loads
gests that perhaps the inner row of rods does not provide signifi- that are based on the factor or the maximum expected column base
cantly improved strength but may reduce deformation capacity. moment. This necessitates detailing (such as thicker base plates or
However, an eight-rod layout may be required to provide adequate deeply embedded anchor rods, or sometimes even embedded col-
strength if the connection is also designed for out-of-plane bending. umns) that are expensive. The assumption underlying this practice
To address the limitations of Steel Design Guide One in these is that base connections are brittle and must be protected against
(eight-rod) situations, a refinement is proposed to the current ap- force-controlled failure, while promoting yielding and plastic hinge
proach that explicitly incorporates the contribution of the inner row formation in the adjacent column. The current study (and previous
of anchor rods. The refined approach seeks to provide an improved research on base connections) shows that this may not be the case
characterization of connection response while also preserving sim- and that base connections may be relied upon for inelastic response
plicity such that the approach may be applied in a design setting. if detailed appropriately (notwithstanding issues such as reparabil-
Consideration of this additional set of rods introduces an indeter- ity). On the other hand, some recent studies indicate that columns
minacy in the derivations, which is resolved by establishing a linear (especially deep columns susceptible to lateral torsional buckling)
relationship between the forces in the outer and inner rows of rods. may not be as ductile in bending as previously thought. Ongoing
Based on test data, the ratio of the forces in the inner rods to the efforts are focused on developing methodologies that effectively
outer rods is calibrated to be 0.5, such that T newinner ¼ 0.5 × T outer .
new incorporate the deformation and energy dissipation characteristics
Application of the refined method to Tests 5–8 provides closer of the connections into a design process.
agreement with test data in terms of overall connection strength Though this study provides new insights into the response of
and anchor rod forces (in the outer row), i.e., the degree of con- base connections and refinements to current approaches, it does
servatism is reduced with respect to the Steel Design Guide One have several limitations that must be considered when interpreting
method. Excepting some minor issues, the method provides a con- and generalizing the results. First, none of the tests featured axial
venient way of calculating design forces. An interesting observa- compressive (or tensile) loading. For cantilever column systems,
tion (from the measured force data) is that the middle rod (in the the axial loads are typically low (less than 10% of the concrete bear-
outer row of rods) carries a disproportionate fraction (up to two- ing capacity); thus, care must be exercised in applying the findings
thirds) of the total force in the row. This observation is important of this study to situations where larger axial loads are present. Other
since current approaches do not address this and implicitly assume loading scenarios, such as the presence of biaxial bending of the
that each row contains only two anchor rods. Based on the test data column, will limit the applicability of the results—computational
and its assessment with respect to the two methods (Steel Design research by Kanvinde et al. (2013) suggests that biaxial bending
Guide One and the proposed refinement), this study has the follow- may compromise the strength of connections—but no experimental
ing implications for the design of these types of column base: data are available to confirm this finding. Second, an obvious issue
1. For base connections with a four-rod layout (and two rows of is that anchor rods were not embedded in concrete but rather passed
rods), the method presented in Steel Design Guide One is rea- through the concrete to enable the attachment of load cells and
sonable and conservative in characterizing the overall connec- allows reusability of the concrete block. Third, though the prestress
tion strength (controlled by base plate yielding) and the rod in the rods was measured and considered in this study, it may not be
forces. These observations are based on the incremental forces representative of prestress in field details; appropriate adjustments
introduced into the rods, not considering the prestress intro- should be made to design forces to account for the best estimates of
duced into them during installation. It is recommended that prestress. Another consideration in the interpretation of results is
estimated or measured prestress forces, which depend on that the evaluation of the design methods is based on limited test
the method of installation, be added to those determined by data, and generalization of these approaches should incorporate ap-
the Steel Design Guide One approach to designing anchors. propriate measures of reliability, e.g., resistance factors (ϕ-factors)
2. For base connections with three rows of anchors, it is recom- developed through the analysis of previous standards, specifica-
mended that the refined method be used to characterize incre- tions, and similar test data. Fourth, in this study, columns were de-
mental (i.e., not including prestress) forces in the anchor rods signed to remain elastic because the focus was on isolating the
and the corresponding base moment capacity. If this is not connection response, whereas in design, yielding may be expected

© ASCE 04014171-11 J. Struct. Eng.

J. Struct. Eng.
in columns. Nominally, this should not affectthe connection Δmax = maximum column drift;
strength, but interactions between the column yielding and connec- Δtop = displacement at top of column;
tion response cannot be ruled out. Finally, the proposed method θbase = base rotation; and
was “validated” against the test data set used to develop it because θmax
base = maximum base rotation.
no other data set is currently available. This limits the generality of
the method, and future data sets may be used to refine the approach
or to propose new methods.
In conclusion, it is emphasized once again that the response of References
these connections is controlled by highly nonlinear and complex
interactions between the various components. As a result, the Adegoke, I. O., and Kemp, A. R. (2003). “Moment-rotation relation-
development of a design (or strength characterization methods) that ships of thin endplate connections in steel beams.” Advances in Struc-
explicitly satisfies the equilibrium, compatibility (including gap- tures: Proc., Int. Conf. on Advances in Structures (ASSCCA’03), Vol. 1,
Swets & Zeitlinger, Lisse, Netherlands, 119–124.
ping and contact), and nonlinear constitutive response of the
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

AISC. (2010). Seismic provisions for structural steel buildings, Chicago.


various components is intractable. Consequently, the methods Astaneh, A., Bergsma, G., and Shen, J. H. (1992). “Behavior and design of
presented in this study (as well as previous methods) are based on base plates for gravity, wind and seismic loads.” Proc., National Steel
simplifying assumptions such as a predetermined stress distribu- Construction Conf., AISC, Chicago.
tion. This implies that caution should be applied in extrapolating Blodgett, O. W. (1966). Design of welded structures, James F. Lincoln Arc
the results of this study to any situation that is significantly dissimi- Welding Foundation, Cleveland, 3.3-1–3.3-32.
lar (in terms of geometry, plate size or thickness, loading, or other Burda, J. J., and Itani, A. M. (1999). Studies of seismic behavior of steel
parameters) from those examined in this study. base plates, Rep. CCEER 99-7, Center of Civil Engineers Earthquake
Research, Dept. of Civil and Environmental Engineering, Univ. of
Nevada, Reno, NV.
Acknowledgments Cimellaro, G. P., and Reinhorn, A. M. (2011). “Multidimensional
performance limit state for hazard fragility functions.” J. Eng. Mech.,
The authors gratefully acknowledge Wildeck for donating the 10.1061/(ASCE)EM.1943-7889.0000201, 47–60.
DeWolf, J. T., and Sarisley, E. F. (1980). “Column base plates with axial
column specimens and providing funding. The assistance of Bill
loads and moments.” J. Struct. Div., 106(11), 2167–2184.
Sluis, Daret Kehlet, Jeffrey Louie, Eliot Wu, and Kimberly Drake, R. M., and Elkin, S. J. (1999). “Beam-column base plate design—
Stillmaker during the testing is also greatly appreciated. The LRFD method.” Eng. J. AISC, 36(1), 29–38.
opinions in this article are those of the authors and do not represent Ermopoulos, J., and Stamatopoulos, G. (1996). “Mathematical modeling
those of the sponsors. of column base plate connections.” J. Constr. Steel Res., 36(2),
79–100.
Fahmy, M., Stojadinovic, B., and Goel, S. C. (1999). “Analytical and
Notation experimental studies on the seismic response of steel column bases.”
Proc., 8th Canadian Conf. on Earthquake Engineering, Canadian
The following symbols are used in this paper: Association for Earthquake Engineering.
A1 = bearing area of plate; FEMA. (2000). NEHRP recommended provisions for seismic regulations
A2 = plan area of foundation; for new buildings and other structures, Building Seismic Safety
B = width of base plate; Council, Washington, DC.
e = load eccentricity; Fisher, J. M., and Kloiber, L. A. (2006). “Steel design guide 1—Base plate
and anchor rod design.” 2nd Ed., AISC 801-06, AISC, Chicago.
ecrit = critical eccentricity;
Fisher, J. M., and West, M. A. (2003). Design guide no. 10, erection brac-
f = maximum bearing stress; ing of low-rise structured steel buildings, 2nd Printing, Steel Design
Fu = ultimate strength of steel; Guide Series, AISC, Chicago.
fc0 = crushing strength of concrete; Gomez, I., Kanvinde, A., and Deierlein, G. (2010). “Exposed column base
DG−1
fmax = maximum bearing stress per Steel Design Guide One; connections subjected to axial compression and flexure.” Final Rep.,
Fy = yield strength of steel; American Institute of Steel Construction.
g = rod edge distance; Kanvinde, A. M., Jordan, S. J., and Cooke, R. J. (2013). “Exposed column
Lcolumn = length of column; baseplate connections in moment frames—Simulations and behavioral
M base = bending moment at base of column; insights.” J. Constr. Steel Res., 84, 82–93.
M DG−1 = moment capacity as per Steel Design Guide One; Krawinkler, H., Gupta, A., Medina, R., and Luco, N. (2000). Loading
histories for seismic performance testing of SMRF components
M new = new method;
and assemblies, Rep. SAC/BD-00/10, SAC Joint Venture, Richmond,
M test
max = test capacity; CA.
N = length of base plate; Liu, J., Sabelli, R., Brockenbrough, R. L., and Fraser, T. P. (2005). “Ex-
P = axial compression in column; pected yield and tensile strength ratios for determination of expected
T DG−1 = tension force in outer row of rods per Steel Design member capacity in the 2005 AISC seismic provisions.” Eng. J. AISC,
Guide One; 44(1), 15–25.
T outer = tension forces in outer rows of anchor rods (generic); Maitra, N. (1978). “Graphical aid for design of base plate subjected to
T inner = tension forces in inner rows of anchor rods (generic); moment.” Eng. J. AISC, 15(2), 50–53.
T test Mazzei, D. (2012). “Designing column base plates for uplift.” Structure
outer = measured incremental forces in outer rows;
T test magazine, National Council of Structural Engineering Associations,
inner = measured incremental forces in inner rows; 27–28.
T new
outer = estimated forces in outer rows as per new method; Ohi, K., Tanaka, H., and Takanashi, K. (1981). “Ultimate strength of steel
T new
inner = estimated forces in inner rows as per new method; column bases.” Bulletin No. 14, Earthquake Resistant Structure
T middle = measured force in middle rod in the outer row of rods; Research Center, Tokyo, Japan.
tp = thickness of base plate; Salmon, C. G., Shaker, L., and Johnston, B. G. (1957). “Moment-
Y DG−1 = bearing width per Steel Design Guide One; rotation characteristics of column anchorages.” Trans. ASCE, 122(1),
Y new = bearing width per new method; 132–154.

© ASCE 04014171-12 J. Struct. Eng.

J. Struct. Eng.
Structural Engineers Association of California. (2012). IBC structural seismic Timoshenko, S. P., and Woinowsky-Krieger, S. (1959). Theory of plates
design manual vol. 4: Examples for steel-framed buildings, San Francisco. and shells, McGraw-Hill, New York.
Thambiratnam, D. P., and Paramasivam, P. (1986). “Base plates under axial Wald, F., Sokol, Z., and Steenhuis, M. (1995). “Proposal of the stiffness
loads and moments.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1986) design model of the column bases.” Proc., 3rd Int. Workshop on
112:5(1166), 1166–1181. Connections in Steel Structures, Elsevier Science.
Downloaded from ascelibrary.org by Queen's University Libraries on 08/22/14. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04014171-13 J. Struct. Eng.

J. Struct. Eng.

You might also like