Documente Academic
Documente Profesional
Documente Cultură
211
two beams. Due to an eccentricity created by the lateral movement of the compression
flanges, a torque is applied to the column. Deep columns tend to have thinner flanges and a
web than a shallower column, resulting in a reduced torsional resistance. Consequently, there
have been concerns that the use of an
RBS connection to a deep column in a
(a)
SMRF can lead to inferior seismic
Erection bolt
performance because of the connection
being susceptibility to torsional loading
from the beams.
Doubler Plate
(b)
e1
FF1
1
e2
F2 2
TEST MATRIX
The test matrix for the experimental program is given below in Table 1, where some of the
details of the six full-scale RBS beam-to-deep column connection specimens are
summarized. All specimens represented an interior RBS connection in a perimeter SMRF
with a composite floor slab, with the exception of SPEC-6 which did not have a composite
floor slab. The parameters investigated in the experimental program included: (1) column
size; (2) beam size; (3) the floor slab; and (4) supplemental lateral brace at the end of the
RBS.
The beam and column section sizes for each specimen were selected on the basis of
introducing different degrees of torsional effects, predicted by the recommended design
procedure of Chi and Uang (2), while also satisfying the weak beam-strong column criteria in
the ASIC Seismic Provisions (4). The design procedure by Chi and Uang considers the total
normal stress in the column at 4% story drift due to axial load, flexure load, and torsion. The
predicted total normal stress in the column flange is shown plotted in Figure 2 for various
column sections, including those of the test specimens. Figure 2 indicates that SPEC-2,
SPEC-4, and SPEC-5 are predicted to develop column flange yielding. The columns for all
specimens and the beams for SPEC-3 through SPEC-6 were fabricated from A992 steel. The
beams for SPEC-1 and SPEC-2 were fabricated from A572 Gr. 50 steel. Both A992 and A572
Gr. 50 have a nominal yield strength of 345 MPa.
212
Column
size
Beam
Size
Doubler
Plate
Supp.
Floor
Lat.
Slab Brace @
RBS
Yield Stress
Flange/Web
(MPa)
Beam
W36x230 W36x150
W27x194 W36x150
W27x194 W36x150
W36x150 W36x150
W27x146 W30x108
W24x131 W30x108
6x800x
1067
13x610x
1067
13x610x
1067
10x8160x
1067
10x610x
914
13x533x
914
Tensile Stress
Flange/Web
(MPa)
Col
Beam
Col
Yes
No
343/378 356/393
478/492
496/514
Yes
No
343/378 372/392
478/492
520/502
Yes
Yes
365/396 356/403
508/506
497/521
Yes
No
365/396 365/396
508/506
508/506
Yes
No
344/353 363/399
471/469
499/513
No
Yes
344/353 334/359
471/469
499/493
Shear Plate
SPEC-4
W36x150
SPEC-2
W27x194
300
Shear tab
6"
SPEC-6 W24x131
SPEC-3 W27x194
100
6"
E71T-8
(supplemental bracing)
200
27"
(supplemental bracing)
9"
200
400
600
800
1000
1200
1400
E71T-8
CJP
No Runoff Tabs
E70T-1
RBS
Flange Cut
Beam
E70T-6
R CJP(TC-U4a-GF)
E70T-1
0
0
E71T-8
E70T-6
CJP(TC-U4a-GF)
erection bolts
SPEC-1
W36x230
400
E70T-1
6"
SPEC-5
W27x146
500
10"
600
Continuity plate
Doubler plate
E70T-1
(TYP)
6"
700
Column
E71T-8
E70T-1
213
the CJP groove welds, and the weld at the edges of the beam flanges ground to a smooth
transition. The backing bar of the top flange weld was left in place and a reinforcement fillet
weld was provided between the bottom surface of the backing bar and the column flange
using the E71T-8 electrode. The beam bottom flange backing bar was removed using the airarc process, back gouged, and reinforced with a fillet weld using an E71T-8 electrode. No run
off tabs were used for the vertical beam web CJP groove welds. All CJP groove welds were
inspected using the ultrasonic test procedure in order to evaluate whether they complied with
the criteria in AWS D1.1 (7) for weld quality.
The specimen composite floor slab had a total thickness of 133 mm, and consisted of 27.6
MPa nominal compressive strength concrete cast on a 20-gage zinc coated metal deck. A
W4x4 welded wire mesh with wire 152 mm on center was placed in the floor slab prior to
pouring the concrete. The width of the floor slab was 1220 mm to one side, with a 305 mm
overhang on the other side to simulate the conditions for a perimeter SMRF. The ribs of the
decking ran parallel to the main beam (i.e., the beams with the RBS connections) of each test
specimen. To develop the composite action, 19 mm diameter shear studs were placed
outside the RBS region at 305 mm spacing along the beams to attach the deck to the main
beams as well as transverse W14x22 floor beams. These transverse beams were placed at a
spacing of 3048 mm to provide lateral bracing to the main beams and column, where the
distance of 3048 mm satisfied the AISC Seismic Provisions (4).
SPEC-6, which had no composite floor slab, had a supplemental lateral brace at the end of
the RBS in addition to the other lateral bracing noted above for the beams. The lateral
bracing was attached to a W36x150 section that was placed parallel to the beams of the test
specimen to simulate a parallel beam in the prototype building. This parallel beam in the test
setup was allowed to move horizontally with the test specimen, but restrained from out-ofplane movement. The corresponding stiffness of the lateral bracing setup satisfied the AISC
LRFD Specification (9). SPEC-3 also had supplemental lateral braces, but these were
anchored in the floor slab.
TEST SETUP, LOADING PROTOCOL, INSTRUMENTATION
The test setup is shown in Figure 4 (a), with the lateral bracing detail given in Figure 4(b) for
the main beams. The ends of the members in the test setup had pin-connected boundary
conditions, using cylindrical bearings to simulate inflection points at the beam midspan and
column midheight in the prototype frame. The ends of each beam away from the column were
supported by instrumented rigid links, which simulated a roller boundary condition and
enabled horizontal movement of the end of each beam. The lateral bracing detail shown in
Figure 4(b) was used to prevent out-of-plane movement of the beams and column (the
diagonal double angles were not used at the column), and designed for strength and stiffness
in accordance with the AISC LRFD Specification (9). The top of the column was braced
against torsion, while at the base of the column a clevis was used to create the pin boundary
condition. The beams were also braced at the rigid links in order to stabilize the test setup.
The torsional bracing provided at both ends of the column in the test setup was evaluated
using a nonlinear finite element model (3) to examine whether the stiffness would be
representative of the torsional restraint at the column inflection points in the prototype
structure. It was found to be satisfactory and not influence the test results by over-restraining
the ends of the column from twisting.
The specimens were tested by imposing a cyclic story drift history based on the loading
sequence defined in Appendix S of the AISC Seismic Provisions (4). The loading protocol
consisted of initial elastic cycles of story drift, followed by cycles of increased amplitude to
cause inelastic response. A test was terminated when either a fracture occurred, resulting in
214
a significant loss of specimen capacity, or after reaching a story drift of 6%. Each specimen
was instrumented to enable measurement of the applied loads, reactions at the rigid links,
specimen story drift; strains in the beam, column, panel zone, and continuity plates; in
addition to panel zone deformation, plastic beam rotation, twisting of the column, and lateral
displacement of the beam at the center of the RBS.
SYM
CL
10'
1312"
Load Cell
Column
Floor Beam
(North Side Only)
Floor Slab
Actuator
Shear Stud
12"
A325
diam.
(TYP)
W36x150
3
Beam (East)
Beam (West)
14'-9"
(b)
Load Cell
1312"
6'-6"
6'-6"
21"
10'
29'-6"
48"
Floor Slab
51 4"
(a)
4"
Double
Angle
2 L2x2x5 16
W14x22
14'-9"
Figure 4. (a) Test setup and (b) beam lateral bracing detail for specimens with a
composite floor slab (Note: 1 inch = 25.4 mm).
TEST RESULTS
A summary of test results for each specimen is given in Table 2, where Rv/Vpz, max, Mf/Mpn,
K,col, , flg, and bf are equal to the ratio of panel zone shear capacity-to-panel zone shear
force corresponding to the plastic flexural moment developing in the RBS, specimen drift from
the last cycle prior to any fracture or strength deterioration to below 80% of the specimen
nominal capacity, ratio of maximum measured beam moment developed at the column faceto-nominal beam flexural capacity, column elastic torsional stiffness, specimen column twist
at 4% story drift, lateral displacement of the beam bottom flange at the RBS at 4% story drift,
and beam flange width, respectively. Typical observed behavior during the testing of a
specimen consisted of yielding in the RBS and the panel zone, followed by cyclic local web
and flange buckling in the RBS. Following the development of local bucking in the RBS,
lateral movement of the bottom beam flange began to occur in the RBS of specimens with a
composite floor slab at 2% to 3% story drift. The combined effect of cyclic local buckling and
lateral flange displacement resulted in a gradual deterioration in specimen capacity to occur
during subsequent cycles where the story drift amplitude was increased. This is evident in the
lateral load-story drift hysteretic response of SPEC-4 shown in Figure 5. The lateral
displacement of the bottom beam flange occurred when it was in compression, and caused
some column twist to develop. Figure 7(a) and (b) shows photographs of SPEC-4 at 4% and
6% story drift, where the yielding in the members and panel zone in the connection region
and lateral beam flange movement in the RBS are visible. The maximum column twist among
the specimens with a floor slab at 4% story drift was 0.037 rads. (SPEC-4). 4% story drift is
the drift at which connections are judged for qualification for seismic use by the AISC Seismic
Provisions (4). SPEC-4, like the other specimens, developed a flange fracture in the RBS
where extensive local flange buckling had occurred (see Figure 7(c)). This occurred at a story
drift of 6%, and was caused by local buckling in the beam flange that led to large cyclic
strains, resulting in a low cycle fatigue failure. SPEC-6, which had a supplemental brace and
lateral bracing attached to the beam that is parallel to the test beam, had minimal
deterioration in capacity as well as column twist (0.004 rads. at 4% story drift), see Figure 6.
Connections in Steel Structures V - Amsterdam - June 3-4, 2004
215
2000
1500
1500
1000
1000
Lateral Load (kN)
2000
500
0
-500
500
0
-500
-1000
-1000
-1500
-1500
-2000
-2000
-6
-4
-2
0
2
Story Drift (% rad.)
-6
-4
-2
0
2
Story Drift (% rad.)
fracture
(b) Beam bottom flange lateral movement
at RBS, 4% story drift
216
Rv
Vpz
max
(% rad)
Mf
M pn
K,col
(kN-m/rad)
(1)
(rad)
flg (1)
(mm)
0.2bf
(mm)
1.26
4.0
1.03
3190
0.016
53
61
1.14
4.0
1.13
1404
0.025
34
61
1.28
5.0
1.15
1404
0.006
35
-(2)
1.24
4.0
1.11
947
0.037
38
61
1.21
5.0
1.20
900
0.007
26
53
1.03
4.0
1.00
577
0.004
-(2)
SPEC-4
SPEC-6
Strain (microstrain)
4000
2000
y=1765
0
-2000
-4000
-6000
-150
-100
-50
0
50
100
Distance across column flange (mm)
150
217
representative of typical specimen behavior, and show little evidence of a strain gradient
across the flange that would result from the effects of warping normal stresses due to column
torsion.
DESIGN RECOMMENDATIONS
The strains in the beam bottom flange near the column face were examined to evaluate the
stress distribution across the beam flange that leads to a torque T applied to the column.
Shown below in Figure 9(a) is the distribution of longitudinal stress across the beam bottom
flange at 4% story drift. These stresses are based on measured longitudinal strains in the
specimens. These results correspond to a negative beam moment at the column face (i.e.,
when the bottom flange of the beam is in compression). Similar results for longitudinal
stress across the beam compression flange were obtained from finite element studies (see
Figure 9(b)). The results in Figure 9 show a trend where the stress distribution across the
beam flange has a reduction in stress, which is due to a moment in the plane of the beam
flange caused by the lateral movement of the beam flange at the RBS. This moment is
equivalent to the torque T that is applied by the beam flange to the column. Shown in Figure
10(a) is an idealized uniform longitudinal stress distribution prior to lateral movement of the
beam flange in the RBS (at 2% story drift). The idealized longitudinal stress distribution at 4%
story drift based on the measured and finite element analysis results is given in Figure 10(b).
At 4% drift local buckling and lateral beam flange movement has occurred in the RBS.
Elastic-perfectly stress-strain behavior is assumed in Figure 10, where Fye is the yield stress.
300
Stress (MPa)
200
100
SPEC-1
SPEC-2
SPEC-3
SPEC-4
SPEC-5
0
Longitudinal stress (MPa)
400
100
0
-100
(a)
-200
(b)
-100
-200
-300
-400
-300
-400
-150
-100
-50
0
50
100
Distance across beam flange (mm)
150
-500
-150
-100
-50
0
50
100
Distance across beam flange (mm)
150
Figure 9. Longitudinal stress distribution across beam flange for (a) all test
specimens, and (b) finite element analysis of SPEC-2.
Fye
Fye
(a)
(b)
Figure 10. Idealized longitudinal stress distribution across beam bottom flange at
(a) 2% story drift and (b) 4% story drift.
218
For the longitudinal stress distribution shown in Figure 10(b), T can be shown (3) to be equal
to
T=
11
Fyeb 2f t f
150
(1)
where Fye, bf, and tf are equal to the expected beam flange yield stress (1.1Fy), the beam
flange width, and beam flange thickness, respectively.
A design procedure was thus developed in order to determine the total design longitudinal
stress ftotal in the column flange that is attached to an RBS connection. The procedure
involves determining the elastic warping normal stresses fw that develop in the column flange
due to the torque T (10) and superimposing them with the column flange normal stresses
due to bending (fb) and axial loading (fa) to obtain the total normal stress ftotal, where
f w = EWnO "
(2)
In Equation (2) E, WnO, and are equal to the Youngs modulus, normalized warping
function at the column flange tip (10), and the second derivative of the angle of twist in the
column (10), respectively, where is a function of the torque T.
The total stress ftotal is compared to the criteria in the AISC LRFD Specification (9), Equation
(H2-1), where
f total = Fy
(3)
in which and Fy are the resistance factor (0.9) and nominal yield stress of the column
flange, respectively. The above design procedure is similar to that developed by Chi and
Uang (2), except for the method in which the torque is determined.
Table 3. Comparison of column normal flange compression stresses with design procedure.
SPEC
Column
W36x230
W27x194
W27x194
W36x150
W27x146
W24x131
Beam
W36x150
W30x108
Axial
load Bending
stress stress
fa
fb (MPa)
(MPa)
Experimental
Total normal
results, total
stress ftotal (MPa) stress & strain,
4% story drift
Chi
ProProStrain Stress
and
posed
posed
()
(MPa)
Uang
Warping
stress fw
(MPa)
Chi
and
Uang
190
128
66
318
256
1277
255
299
182
101
481
400
2151
372(1)
332
332
332
1797
356(1)
337
321
163
658
500
3296
365(1)
252
180
95
432
347
1598
319
347
347
347
2525
334(1)
219
methods is the normal warping stress fw predicted by the above procedure is based on a
more accurate value of the torque T applied to the column. For specimens with a
supplemental lateral brace it was assumed that the restraint of the supplemental brace
resulted in no torque applied to the column (i.e., the normal warping stress fw is equal to
zero). This results in a lower predicted stress than the measured response.
SUMMARY AND CONCLUSIONS
An experimental program was conducted in order to evaluate the seismic performance of
RBS connections to deep wide flange columns. The study involved testing six full-scale
specimens to evaluate the effects of column depth, beam size, composite floor slab, and a
supplemental lateral brace.
Based on the experimental study, the following main conclusions are noted:
1. A composite floor slab can significantly reduce the lateral displacement of the beam
bottom flange in the RBS and the amount of twist developed in the column. The slab
appears to be effective in reducing the twist in deeper columns attached to an RBS
connection, and enables the cyclic strength of the beam with an RBS connection to
be better sustained.
2. All of the specimens were able to satisfy the criteria in the AISC Seismic Provisions
(4) for qualifying the connection for seismic use.
3. A weaker panel zone in a deep column RBS connection will not develop as much
column twist and strength degradation as a connection with a stronger panel zone.
However, a weaker panel zone can significantly increase the potential for ductile
fracture of the connection (3). It is recommended that connections be designed with a
balanced panel zone strength condition.
4. A supplemental brace at the end of the RBS significantly reduced the transverse
movement of the beam flanges in the RBS and column twist that leads to cyclic
degradation in specimen capacity.
5. Basing the column torque on a transverse movement of the beam flange in the RBS
of 0.2bf for calculating column flange warping stresses appears to be conservative. A
new procedure for estimating the torsional load applied to the column due to the local
and lateral buckling in the RBS shows improvement in predicting the correct column
flange normal stress.
ACKNOWLEDGEMENTS
The research reported herein was supported by a grant from the American Institute of Steel
Construction (Mr. Tom Schlafly program manager) and from the Pennsylvania Department of
Community and Economic Development through the Pennsylvania Infrastructure Technology
Alliance (PITA) program. The following companies donated materials for the experimental
testing conducted in this research project: Arcelor International America of New York, NY
(steel sections); Nucor Vulcraft Group of Chemung, NY (metal decking); and the Lincoln
Electric Company of Cleveland, OH (welding wire). The support provided by the funding
agencies and companies is greatly appreciated.
220
REFERENCES
(1)
221
222