Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Constructional Steel Research 65 (2009) 1366–1373

Contents lists available at ScienceDirect

Journal of Constructional Steel Research


journal homepage: www.elsevier.com/locate/jcsr

Effect of weld details on the ductility of steel column baseplate connections


A.T. Myers a , A.M. Kanvinde b,∗ , G.G. Deierlein a , B.V. Fell b
a
Department of Civil and Environmental Engineering, Stanford University, Stanford, CA, United States
b
Department of Civil and Environmental Engineering, University of California, Davis, CA, United States

article info a b s t r a c t

Article history: Results from six two-thirds scale tests on moment-resisting steel column base plates are presented.
Received 13 June 2008 The test specimens incorporate Complete Joint Penetration (CJP) and Partial Joint Penetration (PJP) weld
Accepted 19 August 2008 details between the column and the base plate. The test data indicate that both details are resilient to
fracture and sustain inelastic column hinging to story drift ratios of 6%–9%, which exceeds the typical
Keywords: acceptance criteria of 4% drift ratio for seismically detailed special moment frames. In five of the six
Column base plates
tests, fractures initiated in the Heat Affected Zone (HAZ) at the fusion line between the weld and column
Seismic design
Fracture
flange. In the sixth test, fractures initiated at the inside face of the column flange at the upper edge of the
weld access hole of the CJP weld detail. Contrary to the initial expectations, the specimens with the PJP
welds exhibited higher displacement ductility than those with CJP welds. This is attributed to the fillet
reinforcing that strengthens the PJP welds and enables them to sustain stresses and strains necessary to
fully develop yielding in the column flanges. The test data further support the adequacy of the FEMA 350
provisions for determining the required strength of the welds based on the probable moment demand
with the material overstrength factor of Ry = 1.1 and strain hardening factor of Cpr = 1.2.
© 2008 Elsevier Ltd. All rights reserved.

1. Introduction Moment Resisting Frames (IMRFs and SMRFs) with fixed column
bases. Based on an informal survey of the current practice for
Column base connections are critical components in earthquake designing these moment frames in high seismic regions, the most
resistant structural systems, since they must maintain their common welds between the column flanges and the base plate
ability to transfer axial forces, shear forces and moments to the are Complete Joint Penetration (CJP) groove welds or Partial Joint
foundation while sustaining large inelastic deformations. Severe Penetration (PJP) groove welds with fillet reinforcing. The CJP
damage to column base plate components has been observed in details are considered to exhibit superior behavior but are typically
previous earthquakes, such as the 1994 Northridge [1] and 1995 more costly to fabricate. On the other hand, the PJP details are
Kobe earthquakes [2]. Laboratory testing has demonstrated the generally considered more susceptible to failure by fracture, due
susceptibility of column base plate connections to various failure to the built-in flaw formed by the un-fused gap between the
modes, including weld fracture (e.g. [3–5]), base plate yielding, flange and base plate, but are less costly to fabricate. Although
anchor rod fracture and concrete crushing [6]. Recent integrative weld fractures were reported in prior experimental studies [3,8],
studies by Grauvilardell et al. [7] further illustrate the sensitivity of detailed data on the fracture resistance of typical weld connection
these failure modes to the configuration of structural systems and type are scarce [4,5]. Thus, current design guidelines, including
base plate details. These include, but are not limited to, the type the AISC Seismic Provisions [9] and the AISC Design Guide 1 [10],
of building frame, base plate size and thickness, concrete footing provide only general guidance regarding practical column base
strength, anchor rod strength and configuration, and weld details. weld details that relies on engineering judgment and documented
The focus of this paper is the fracture resistance of two weld behavior of other similar components (e.g. end plate moment
connection types between the steel columns and base plates connections, [11]).
in moment-resisting frames. Since the findings of the paper Motivated by the scarcity of data comparing the relative
emphasize the deformation capacities of the details in general, fracture resistance of CJP and PJP weld details and the popularity
they may be suitably interpreted in the context of different Lateral of these details in high seismic regions, this paper presents
Force Resisting Systems (LFRS), including Intermediate and Special experimental investigations on six two-thirds scale column base
specimens tested as a part of a Network for Earthquake Engineering
Simulation and Research (NEESR) project. In addition to their
∗ Corresponding author. Tel.: +1 530 752 2605; fax: +1 530 752 7872. practical relevance to seismic design, these experiments provide
E-mail address: kanvinde@ucdavis.edu (A.M. Kanvinde). data to validate a new type of micromechanics-based model,
0143-974X/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2008.08.004
A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373 1367

thicknesses [tbase = 6.35, 12.7, or 19.1 mm] and the axial load
Notation [30% and 100% of the allowable load capacity (nominal design
strength) of the base plate]. All the experiments featured a W6 × 25
bf /2tf Width to thickness ratio of flange (Metric — W150 × 37.1) column which has a width–thickness
Cpr Factor to account for strain hardening (taken as 1.2) ratio bf /2tf = 6.7 (important in the context of this paper since
εn,m Measured strain at the initiation of necking local buckling may delay weld fracture). The base plates were
θinitiation Maximum drift ratio sustained prior to crack supported by a reinforced concrete footing with four 19.1 mm
initiation diameter anchor bolts. All six specimens exhibited reasonably
θfailure Maximum drift ratio sustained prior brittle failure ductile behavior, reaching drift ratios of at least 2% prior to fracture
θy Drift ratio at first yield or loss of strength. A drift ratio of 2% is approximately equal to the
Fy Nominal yield strength of steel, MPa expected story drift demands in moment frame buildings under
Fy,m Measured yield strength of steel, MPa design level ground motions (i.e. ground motions with spectral
Fu,m Measured ultimate strength of steel, MPa acceleration values which may be exceeded with a probability of
My Moment at first yield, kN m 10% in 50 years). The two specimens with thin base plates (tbase =
M2% Moment at 2% drift, kN m (Fig. 4) 6.35 mm) exhibited the most ductile behavior, due to formation
Mmax Maximum moment, kN m (Fig. 4) of stable yield lines in the base plate prior to fracture of the welds.
Mpr Expected moment calculated as Mpr = Cpr Ry Fy Zx , kN In contrast, the thicker base plates showed less ductile behavior
m and failed through a combination of grout crushing and anchor bolt
Py Axial load at first yield of column, N fracture. Another interesting observation was that the tests with
Ry Factor to account for material overstrength, taken as higher axial loads exhibited a decrease in flexural strength and an
1.1 for Grade 50 (345 MPa) steel increase in ductility of the column base plate connection.
tbase Base plate thickness, mm Fahmy et al. [4] conducted two tests on column base
Zx Plastic section modulus for major axis bend- connections under cyclic lateral loads. Both the tests featured
ing, mm3 a W10 × 77 (W250 × 115) column (bf /2tf = 5.9) and a
69.9 mm thick base plate connected to the reinforced concrete
footing by four 50.8 mm diameter anchor bolts. The test specimens
previously developed by the authors [12] to predict earthquake- were identical except for the weld filler material used in each,
induced Ultra Low Cycle Fatigue (ULCF) in steel structures. The thus providing a direct examination of the effect of weld
validation of this model for weld metals and Heat Affected Zones metal toughness classification (toughness versus non toughness
(HAZ) is a part of a broader set of objectives of this investigation. A rated) on the response of the column base connection details.
detailed description of the micromechanical modeling is beyond Both specimens had PJP groove welds and reinforcing fillet
the scope of this paper; however, it is relevant to mention here welds similar to those used in the present study (see Fig. 1
that the micromechanical models, when validated by the test data, and later discussion). Weld toughness was observed to have a
provide insights into the local effects that trigger fracture initiation. strong influence on ductility. The specimen fabricated with the
These insights may be used to extend the specific results from these toughness-rated weld metal sustained about twice the drift before
six tests to more general recommendations on the design of similar fracture (approximately 5% interstory drift ratio) as compared to
weld details. This is part of an ongoing study. the specimen with the non-toughness rated weld metal, which
fractured at a drift ratio of approximately 2.5%.
Burda and Itani [5] performed six tests similar to those of
2. Literature review
Astaneh and Bergsma [3], where cyclic lateral loads were applied
along with a constant column axial load (15% of Py , the yield
Grauvilardell et al. [7] assembled an extensive literature
strength of the column). All tests included a W8 × 48 (W200 ×
review of previous research on column base connections. Drawing
71) column (bf /2tf = 5.9) with base plates thickness of 19.1,
from Grauvilardell et al., previous studies germane to this 25.4, or 31.8 mm that were connected to the reinforced concrete
study are briefly summarized herein. Early research on column footing with four 38.1 mm diameter anchor rods. Three specimens
bases examined axially loaded connections [13–15] and issues featured a detail with fillet welds around the entire perimeter of
pertaining to the base plate bearing on the concrete footing. the W8×48 (W200×71) cross section to connect the column to the
Subsequent studies that examined the seismic response of base base plate, and three others had CJP groove welds, similar to those
connections have led to the development of the AISC Steel Design used in the present study (Fig. 1). All the specimens failed through
Guide 1 [16,10]. These considered tests [6,17] that investigated weld fracture at drift levels ranging from 2% to 13%. Similar to the
base plate yielding, concrete crushing and anchor rod fracture as tests by Astaneh and Bergsma [3], the connection ductility was
controlling failure mechanisms in base connections. Based in part higher for the specimens with thinner base plates. The specimens
on these studies, design procedures have been developed [18] that with the CJP weld details showed greater ductility than those with
are commonly used in practice for sizing various components of the the fillet weld details, where the latter failed at drift ratios between
column base connection, including base plate size and thickness, 2% and 9% and the former failed at drift ratios between 7% and 13%.
and anchor rod dimensions and configuration. Lee and Goel [8] performed four tests on column base
Few studies have specifically investigated failure of the connections. Each featured a W12 × 96 (W310 × 143) column
welds connecting columns and base plates. However, weld (bf /2tf = 6.8) with cyclic lateral load applied in the direction of
fractures have been observed in tests designed to obtain general the column minor axis bending and a 57.2 mm thick base plate.
response characteristics of column base plate connections. Most The key variables included the number and size of anchor rods
previous studies emphasized the catastrophic strength loss due [six 50.4 mm diameter or four 31.8 mm diameter anchor rods]
to connection failure, and do not always report detailed data on and the weld filler material. All four experiments showed fracture
the precise instant and location of weld fracture initiation and initiation in the welds between column and base plate at relatively
subsequent growth prior to unstable propagation. low drift ratios (1.0%–1.5%), with significant strength degradation
Astaneh and Bergsma [3] conducted six tests on column base and complete failure at drifts ranging from 1.0% to nearly 4.0%.
plate details under constant axial and cyclic lateral loads with Of the previous research, five test specimens had weld details
increasing amplitude. The key variables include the base plate that are representative of those currently used in seismic design
1368 A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373

Fig. 1. Drawings illustrating (a) CJP weld detail and (b) PJP weld detail.

practice (two PJP from [4] and the three CJP specimens from [5]).
However, given the other differences in the experiments (column
size, the presence of axial load, base plate thickness, number and
size of anchor bolts, and loading history), the impact of weld
detail on performance cannot be accurately assessed. The tests
presented in this paper complement the existing data by providing
a controlled comparison between base plate connections with CJP
versus PJP weld details. In addition, the tests evaluate the effect of
loading history (general cyclic versus near-fault) on the fracture
response. With careful attention paid to documenting fracture
behavior, the tests provide data to validate micromechanical
fracture models, which can be used for future parametric studies
considering various design variables and loading effects.

3. Experimental investigation of column base connections

As shown in Fig. 2, the test specimen consists of a W8 × 67


(W200 × 100, A992 Gr. 50–345 MPa) cantilever column and a
457 × 457 × 57.2 mm (A572 Gr. 50–345 MPa) base plate that is
supported on a rigid steel foundation plate and loaded transversely
by a hydraulic actuator. Because of constraints on the allowable
forces in the testing lab, the specimen dimensions were chosen
to be roughly two-thirds that of a more realistically sized first
floor column. This specimen scale produced the largest forces
that the testing facility could safely withstand. The specimen was
loaded in the direction of the column major axis bending with
load applied 1.75 m above the base to approximately represent
the point of inflection in the bottom story of a fixed-base moment
frame (roughly 2/3rd of the story height). For clarity, the restraint
system provided to prevent lateral (out of plane) deformations is
not shown in Fig. 2.
The column cross-section has a flange width–thickness ratio of Fig. 2. Schematic illustrating column base plate testing setup: (a) elevation view
bf /2tf = 4.4, which is significantly smaller than the limiting value and (b) plan view of base plate.
of bf /2tf = 52/ Fy = 7.35 (for Fy = 345 MPa steel) permitted
p
by the Seismic Provisions [9]. This highly compact section was crushing or base plate yielding) and thereby isolate weld fracture
selected to allow the formation of a plastic hinge region in the as the governing failure mode. Thus, the intent of the tests is
column without local buckling, thereby maintaining high moment to provide an upper bound evaluation of the stress and strain
demand on the welds throughout the tests. This stands in contrast demands imposed by flexural hinging of the column base.
to columns with less compact sections, where local buckling tends The test variables considered and selected test results are
to limit the moment demand on the welds. The base plate also summarized in Table 1. The primary test variable is the type of weld
was proportioned to develop the full column required moment detail, CJP or PJP, as shown in Fig. 1. As discussed previously, the
strength, per FEMA 350 [11], with limited deformations in the details were selected based on a survey of current design practice
base plate. The column base plate was attached to a 102 mm thick for SMRF systems. In the CJP detail (Fig. 1a), the column flanges
steel foundation plate with pre-stressed A325 high strength bolts, were welded to the base plate with two-sided groove welds that
resulting in a fixed-base boundary condition. The thick foundation necessitate a weld access hole in the column web. The column
plate was used to preclude other failure modes (such as concrete web was fillet welded to the base plate. The flange welds were
A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373 1369

Table 1
Test matrix and results
Test Weld detail Loading history θinitiation (%) θfailure Crack location M2% /Mpr Mmax /Mpr Energy norm Energy ratio
Cyc. # Cyc. (%) Fail (%)

1 CJP General cyclic 3 2 5.0 3.1 Corner-HAZ 0.81 1.04 122 –


2 CJP Near-fault 6 1 6.0 4.5 Corner-HAZ 0.75 1.06 141 0.48
3 CJP General cyclic 4 2 6.0 0.4 Access hole 0.81 1.06 139 –
4 CJP General cyclic 4 2 6.0 0.2 Corner-HAZ 0.78 1.08 130 –
5 PJP General cyclic 5 2 9.0 0.3 Corner-HAZ 0.83 1.19 205 –
6 PJP General cyclic 6 1 8.0 2.7 Corner-HAZ 0.83 1.14 165 –
θinitiation — Maximum drift ratio sustained prior to crack initiation. The 6% reported for test #2 refers to the first tensile pulse of the near fault loading.
θfailure — Drift at the point of brittle failure. For general cyclic loading, results may be interpreted as follows: For test #1, total flange fracture occurred during the second cycle
of the 5.0% drift cycles at a deformation of 3.1% drift. For test #2, total fracture occurred during the appended general cyclic loading.
M2% — Moment at 2% drift (see Fig. 4).
Mmax — Maximum moment (see Fig. 4).
Mpr — Cpr Ry Fy Zx ; Cpr = 1.2; Ry = 1.1; Fy = nominal yield stress.
Energy norm — Total energy dissipated prior to brittle failure normalized by My θy .
Energy ratio — Energy dissipated by the near fault loading protocol divided by the total dissipated energy.

Table 2
Measured material properties for weld and column base metal
Fy,m (MPa) Fu,m (MPa) εn,m (–)
True Eng. True Eng. True Eng.

Column base metal 352 352 614 490 9.1% 9.5%


Weld metal 579 579 710 648 16.0% 17.0%
Fy,m — Measured yield stress.
Fu,m — Measured ultimate stress.
εn,m — Measured strain at necking.

designed to develop full yield strength of the column flange, as


per the Seismic Provisions [9]. Four specimens (Tests #1 through
#4) were fabricated with CJP welds. In the PJP detail (Fig. 1b), a
groove weld was installed from the outside flange face through
80% of the flange thickness, and a reinforcing fillet weld is provided
on the inside of the flange. The PJP weld and the throat of the
reinforcing fillet weld together provide a total throat area 25%
larger than the flange. The web is fillet welded to the base plate,
similar to the CJP detail. Unlike in the CJP detail, no access hole
was provided in the column web. The tests with the PJP welds
were used to establish whether the built-in flaw created by the un-
fused section of the flange is likely to trigger fracture and thereby
reduce the connection ductility. Two specimens (Tests #5 and #6)
were fabricated with the PJP detail. All welds were fabricated with
E70T-1-H8 (480 MPa) filler metal, deposited through a flux core arc
welding (FCAW) process as per the appropriate American Welding
Society welding procedure specification, D1.1. Three Charpy V-
notch tests were conducted on the electrode at −29 ◦ C as per FEMA
350 [11]. The average Charpy V-notch toughness for these tests was
33 J. Table 2 summarizes measured material properties from round
coupon tests, designed per the ASTM standard for tension testing of
metallic materials [19], extracted from both the weld and column Fig. 3. (a) General cyclic loading protocol and (b) Near-fault loading protocol
material. The reported properties are averages of two coupon tests followed by general cyclic protocol.
for each material.
As illustrated in Fig. 3, loads were applied according to two 3.1. Qualitative description of experimental response
alternative displacement histories that are based on ATC-SAC
protocol [20] to represent realistic seismic demands. The general Representative plots of the column base moment versus drift
cyclic loading history (representative of typical far-field ground ratio are shown in Fig. 4a, for general cyclic loading history
motions — Fig. 3a) was applied to five specimens (Test #1, 3, 4, 5, (Test #4), and Fig. 4b, for the near-fault loading history (Test
and 6). The ‘‘near-fault’’ loading protocol (Fig. 3b) was applied to #2). Also indicated are the observed points of fracture initiation
Test #2 and features a large unidirectional pulse, followed by the and complete failure. Qualitatively, all the specimens that were
general cyclic loading history similar to that applied in the other subjected to the general cyclic loading history (Tests #1, 3, 4, 5,
tests. The general cyclic history was appended to the near-fault 6) exhibit a similar response. Yielding in the column, evidenced by
history in the event that fracture does not occur during the near- flaking of mill scale on the column flanges near the base plate, was
fault protocol cycles. The SAC loading protocols are expressed in observed at drift ratios of approximately 1%. The base plate and
terms of story drift ratios and are converted to displacements by foundation plate did not yield and the measured deformations in
multiplying them with the column height (69 in.). The results are both plates were negligible. Initiation of ductile cracking at drift
presented in terms of story drift ratios, expressed in percent drifts, levels ranging from 3% to 6% (Table 1) was observed in the fusion
for convenient interpretation. line between the column flange and the Heat Affected Zone (HAZ)
1370 A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373

Fig. 5. (a) Crack initiation at corner of flange in HAZ, Test # 4 (b) Re-entrant corner
at weld access hole triggering fracture initiation and subsequent failure, Test # 3.
Fig. 4. Representative load deformation curve indicating key values and limit states
for (a) Test #4 — general cyclic loading protocol and (b) Test #2 — near-fault (black
line) appended by general cyclic loading protocol (gray line).

at the flange corners (Fig. 5a). The initiated crack was observed to
grow steadily as the loading progressed. The initiation and growth
of the ductile crack did not produce any discernible changes in
the load deformation curve (refer Fig. 4). Stable crack growth
continued for a considerable number of cycles (approximately
10–15) until the growing ductile crack finally triggered a sudden
failure at drift levels in the range of 5%–9%.
Post-test inspection of the fracture surfaces from Tests #1,
4 and 6 indicates that the cracks initiated and grew by ductile
tearing at the flange corners and then transitioned to a brittle
cleavage fracture that completely severed the column flange. In
Test #3, ductile cracks initiated both at the flange corners and in
the web/flange intersection at the top of the weld access hole,
and brittle failure propagated from the crack at the weld access
hole towards the column flange (Fig. 5b). In the CJP specimens,
the brittle crack propagation is arrested by the weld access hole Fig. 6. Test #5 (PJP) at failure, illustrating crack propagation through web.
after severing the flange (Fig. 5b), however, for PJP tests the
crack propagates partly or wholly through the web. Fig. 6 shows flange tips during the subsequent general cyclic loading protocol,
a photograph of a PJP specimen (Test #5) after brittle crack leading to a total flange fracture during its first 6% drift cycle.
propagation.
For Test #2, subjected to the near-fault loading history (Fig. 4b), 3.2. Experimental results
ductile fracture initiation was observed in both corners of the
flange in tension during the initial unidirectional pulse to 6% drift. Table 1 summarizes the test data corresponding to ductile crack
On subsequent cycles of the near fault loading protocol, fracture initiation, θinitiation , and complete fracture propagation, denoted as
initiated at all four corners, although the ductile crack growth θfailure . The test results are also presented graphically in Fig. 7a,
remained somewhat limited. The cracks grew rapidly in all four b, where the square and round markers are superimposed on the
A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373 1371

measurements. This comparison has implications, which are


described below, for capacity design of welds and other critical
components of the connection. In addition to the ultimate strength,
strength levels at a drift of 2% (which are typically lower than the
ultimate strength) are reported (see Fig. 4 for an illustration), to
reflect moment demands at design level earthquakes. The expected
strength is determined according to the provisions of FEMA 350
(see Table 1 footnote) for determining moment demands for the
design of beam–column connections. The FEMA 350 procedures
adjust for the expected steel yield strength (through the Ry factor)
and strain hardening (through the Cpr factor) in determining the
required strength. For Grade 50 (345 MPa) steel, the expected
strength is equal to 1.3 (Cpr × Ry = 1.2 × 1.1) times the nominal
plastic moment of the columns.
Referring to Table 1 and Figs. 4 through 8, the main observations
from the tests are as follows:
• All the specimens sustained drift levels of 5%–8% before
exhibiting any type of deterioration in the load–deformation
response. In all six tests, the onset of deterioration was sudden
and corresponds with a complete connection failure. This stable
behavior up until 5%–8% drift exceeds the expected earthquake
drift demands of approximately 2% for design level earthquake
loads and 4%–5% for maximum considered earthquakes (per
Krawinkler et al. [20]). Moreover, none of the specimens
showed fracture initiation before the 3% drift cycles, indicating
that the weld details considered may prevent fracture initiation
during design level earthquake loads. Previous results by Fahmy
et al. [4] and Burda and Itani [5] indicate similar findings.
• Five of the six specimens (including both PJP specimens) show
Fig. 7. Location of fracture initiation, indicated by square markers, and total flange
fracture, indicated by circle markers for (a) Tests #1, 3, 4, 5, and 6, general cyclic
fracture initiation and failure in the HAZ at the fusion line
loading history and (b) Test #2, near-fault loading history. Text labels indicate between the weld and the column flange, as illustrated in
specimen label and base moment in kN m. Fig. 5a. While expected for the CJP specimens, this behavior was
not expected for the PJP specimens, where fracture initiation
displacement loading history to indicate the occurrence of fracture was expected to occur at the initial flaw formed by the un-
initiation and complete fracture, respectively. The data for fracture fused portion between the column flange and the base plate.
initiation in Table 1 refer to the largest drift sustained prior to However, the test results clearly indicate that the combination
initiation, while the data for complete fracture include both the of the extra strength provided by the fillet weld reinforcement
maximum drift and drift cycle sustained prior to fracture and (with the combined throat area 25% larger than the column
the drift at fracture. For example, for Test #1, fracture initiation flange) flange and the weld toughness were sufficient to resist
occurs during the first cycle to 3% drift and complete fracture fracture at the weld root and to concentrate yielding in the
occurs during the second cycle to 5% drift at a drift ratio of column flange.
3.1%. Another index for fracture endurance is the energy norm, • Overall, specimens with PJP welds performed better than those
which is total energy dissipated by the specimen prior to fracture, with CJP welds, with the PJP details sustaining drifts as large as
normalized by the product of the yield moment and yield drift 8%–9% before failure, as compared to 5%–6% drifts for the tests
(My θy ). Total dissipated energy is calculated through a simple with the CJP detail. This is attributed to the concentration of
trapezoidal numerical integration technique. The last column of stresses and strains in the fusion line and HAZ of the column,
the table includes a breakdown of the energy dissipated during the created by the weld access hole in the CJP specimens (Fig. 5b).
near fault loading to the total energy dissipated for Test #2. The concentration of strains is exacerbated by the fact that
Also included in Table 1 is a comparison between the predicted material in the HAZ is generally less ductile than either the
ultimate strengths of the base connection (associated with column parent base metal or the weld metal [21,22]. It should be noted
hinging, per FEMA 350 [11]) and the corresponding experimental that the large difference in sustained drifts between the PJP and

Fig. 8. Ductile crack growth prior to fracture initiation in Test #4 (a) photo crack at flange tip, (b) fractured surface of flange section.
1372 A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373

the CJP specimens observed in these tests is probably a function loaded in major axis bending. The data complement previous
of the relative column size reduction in the CJP specimens due to studies (Fahmy et al. [4], Burda and Itani [5], Astaneh and
the weld access hole. For these tests with the W8 × 67 (W200 × Bergsma [3], Lee and Goel [8]) by providing additional data to
100), the standard 31.8 mm wide weld access hole of the CJP assess and compare the fracture resistance of base connections
detail reduces the column cross-section area by 10%. However, with CJP and PJP weld details. In both cases, toughness rated filler
for larger columns, the area of the access holes relative to the metal was used.
area of the entire cross-section will generally be less, and hence, Qualitatively, all the experiments follow a similar sequence of
CJP weld details may perform better than observed in these events. Ductile cracks typically initiate at column drift ratios of
tests. 3%–5% in the Heat Affected Zone at the fusion line between column
• Fracture initiation at the top of the access hole on the inside of flange and groove weld. Upon continued cyclic loading, the ductile
the flange in Test #3 (Fig. 5b) is likely caused by the stress/strain cracks grow to about 0.2–0.6 inches in length, until they propagate
concentration introduced by the re-entrant corner of the access unstably by brittle failure at drift ratios of 5%–9%. The initiation
hole, and a lowered local material toughness due to flame- and growth of ductile tearing does not affect the load displacement
cutting of the access hole. Similar cracking at weld access holes response until brittle failure occurs. These results are generally
was previously reported in tests by Krawinkler et al. [23]. consistent with previously published data and meet the minimum
• In all the tests, a significant delay was observed between ductility requirements for seismic design as implied by the AISC
Seismic Provisions (AISC [9]).
the initiation of ductile cracking and complete failure. As
The specimens with fillet-reinforced PJP welds were observed
mentioned earlier, the growth of the ductile crack during this
to be more ductile than those with CJP welds. Whereas the PJP
period did not visibly affect the load–deformation response
welds were expected to fracture from the built-in flaw at the un-
of the specimens. For the five specimens for which fracture
fused section between the column flange and base plate, fracture
initiation occurred in the HAZ at the fusion line between the
was observed at the top of the weld in the HAZ of the column
weld and the column flange, the crack propagated between
flange. This is attributed to the strength and toughness at the weld
0.2 and 0.6 in. along the outside flange surface before brittle root of the reinforced PJP welds, which permitted the development
fracture. Fig. 8a shows the crack growth just prior to the of a fully yielded and strain-hardened column flange. On the other
brittle fracture (Test #4), and Fig. 8b shows a representative hand, the CJP details tended to concentrate plastic straining in
photograph of the corresponding flange fracture surface, the reduced section produced by the weld access hole, thereby
indicating the through-thickness growth of the ductile crack reducing their ductility. Additionally, one of the four CJP specimens
before it transitions to a brittle fracture. exhibited fracture initiation at and propagation from the area of
• The type of loading protocol does not appear to have a stress/strain concentration created at the top of the weld access
significant effect on either the mode of failure or the ductility of hole geometry at the column web/flange interface.
the specimens. Test #2, which featured the near-fault loading In summary, the experimental data from the six tests indicate
protocol appended by the general cyclic loading protocol, that the CJP and PJP column base weld details, commonly used
exhibited fracture ductility (with respect to complete fracture) for design in high seismic regions, can sustain deformations that
similar to that of the other CJP specimens. This is despite the are sufficient for seismic design. This conclusion is based on the
early initiation of fracture in Test #2 during the near-fault tests in which column flange buckling and base plate yielding
portion of the loading protocol. were precluded through the use of stronger columns and base
• Generally, the energy dissipated by the specimens is propor- plates. While these tests are thought to provide lower-bound
tional to the drifts observed at the complete fracture, mainly be- limits on weld performance, the potential impacts of local column
cause there is little degradation in the specimen properties until flange buckling or base plate deformations adjacent to the weld
the point of complete fracture and since the specimen configu- should not be ruled out. The experiments described in this paper
ration and details are similar. For test #2, approximately 50% of provide valuable insights into the response of column base details;
the energy was dissipated during the near fault loading. however, they are limited in size, configuration, and may not
• The low bf /2tf ratio (=4.4) for the W8 × 67 (W200 × represent all relevant effects, material properties, and boundary
100) column inhibited local buckling up to the ultimate drift conditions that arise in design and construction practice. Ongoing
demands of 5%–6%. Test #5, with a PJP weld detail and very large work by the authors to investigate these effects through testing
sustained drifts, showed flange local buckling during the 8% and finite-element fracture analyses of column base plates is
drift cycles. Thus, the tests are successful at providing an upper- underway.
bound of the inelastic column moment demands of column base
plates. However, the effect of local buckling on the response of Acknowledgments
sections with higher bf /2tf ratios is unknown.
• For these test specimens, the FEMA 350 provisions [11] for The research presented herein was supported by the National
calculating the probable column moment underestimates the Science Foundation (NSF) under the grant number CMS 0421492,
maximum moment demand by about 4% to 19% (see column of the George E. Brown Jr. Network for Earthquake Engineering
Mmax /Mpr ratios in Table 1). However, the calculated demands Simulation (NEES), the Department of Defense (DOD) and the
do exceed the measured moments at 2% drift ratios (see column McNair scholars program at the University of California, Davis. The
of M2% /Mpr ratios in Table 1) by about 20%. Thus, the FEMA 350 knowledgeable support of the University of California, Berkeley
NEES lab personnel, and the diligent assistance of Jorge Camacho,
predictions appear reasonable and conservative for estimating
an undergraduate researcher at the University of California, Davis,
moment demands in base connections during design level
is gratefully acknowledged.
events.

References
4. Summary and conclusions
[1] Northridge Reconnaissance Team. Northridge earthquake of January 17, 1994.
This paper presents findings from six two-thirds scale tests of Reconnaissance report. (Suppl. C-2 to Vol. 11). Oakland (CA): EERI; 1996.
column base connection specimens subjected to earthquake-type [2] Midorikawa M, Hasegawa T, Mukai A, Nishiyama I, Fukuta T, Yamanouchi H.
Damage investigation of steel buildings in specific areas observed from the
cyclic loading. The main objective of this paper is to examine the 1995 Hyogoken-Nanbu earthquake. In: Proceedings of US–Japan workshop on
effect of weld details on the response of column base connections steel fracture issues. 1997.
A.T. Myers et al. / Journal of Constructional Steel Research 65 (2009) 1366–1373 1373

[3] Astaneh A, Bergsma G. Cyclic behavior and seismic design of steel base plates. [13] Hawkins HM. The bearing strength of concrete loaded through flexible plates.
Proceedings, Structures Congress, ASCE 1993;409–14. Magazine of Concrete Research 1968;30(62):95–102.
[4] Fahmy M, Stojadinovic B, Goel SC. Analytical and experimental studies on the [14] DeWolf JT. Axially loaded column base plates. Journal of the Structural
seismic response of steel column bases. In: Proceedings of the 8th Canadian Division, ASCE 1978;104(5):781–94.
conference on earthquake engineering. 1999. [15] Murray TM. Design of lightly loaded steel column base plates. Engineering
[5] Burda JJ, Itani AM. Studies of seismic behavior of steel base plates. Report Journal, AISC 1983;20(4):143–52.
no. CCEER 99-7. Reno (NV): Center of Civil Engineering Earthquake Research, [16] DeWolf JT, Ricker D. Column base plates. In: AISC steel design guide 1. 1st ed.
Department of Civil and Environmental Engineering, University of Nevada; Chicago (IL): American Institute of Steel Construction; 1990.
1999. [17] Thambiratnam DP, Paramasivam P. Base plates under axial loads and
[6] DeWolf JT, Sarisley EF. Column base plates with axial loads and moments. moments. Journal of Structural Engineering, ASCE 1986;112(5):1166–81.
Journal of the Structural Division, ASCE 1980;106(11):2167–84. 1980. [18] Drake RM, Elkin SJ. Beam–column base plate design — LRFD method.
[7] Grauvilardell JE, Lee D, Hajjar JF, Dexter RJ. Synthesis of design, testing
Engineering Journal, AISC 1999;36(1):29–38.
and analysis research on steel column base plate connections in high-
[19] ASTM Standard E8. Standard test methods for tension testing of metallic
seismic zones. Structural engineering report no. ST-04-02. Minneapolis (MN):
materials. American Society for Testing and Materials. 2008.
Department of Civil Engineering, University of Minnesota; 2005.
[20] Krawinkler H, Gupta A, Medina R, Luco N. Loading histories for seismic
[8] Lee D, Goel SC. Seismic behavior of column base plate connections bending
about weak axis. Research report UMCEE 01-09. Ann Arbor (MI): Department performance testing of SMRF components and assemblies. SAC Joint Venture,
of Civil and Environmental Engineering, University of Michigan; 2001. Report no. SAC/BD-00/10. Richmond (CA); 2000.
[9] American Institute of Steel Construction, Inc. (AISC). Seismic provisions for [21] Johnson MQ, Ramirez JE. Preliminary evaluation of heat affected zone
structural steel buildings. Chicago, (IL, USA): American Institute of Steel toughness in structural shapes used in the construction of seismic moment
Construction; 2005. frames. SAC Joint Venture, Report no. SAC/BD-00/13. Richmond (CA); 2000.
[10] Fisher JM, Kloiber LA. Base plate and anchor rod design. In: AISC steel design [22] Myers AT, Kanvinde AM, Fell BV, Fu X. Large scale tests and micromechanics-
guide 1. 2nd ed. Chicago (IL): American Institute of Steel Construction; 2006. based models to characterize ultra low cycle fatigue in welded structural
[11] FEMA. Recommended seismic design criteria for new steel moment-frame details. In: Proceedings of ASCE/SEI structures congress. 2007.
buildings. In: FEMA 350. Washington (DC): FEMA; 2000. [23] Krawinkler H, Zohrei M, Lashkari-Irvani B, Cofie N, Hadidi-Tamjed H.
[12] Kanvinde AM, Deierlein GG. A cyclic void growth model to assess ductile Recommendations for experimental studies on the seismic behavior of steel
fracture initiation in structural steels due to ultra low cycle fatigue. Journal components and materials. Technical report 61. Stanford (CA): John A Blume
of Engineering Mechanics, ASCE 2007;133(6):701–12. Earthquake Engineering Center, Stanford University; 1983.

You might also like