Vous êtes sur la page 1sur 12

Improved Seismic Performance of Gusset Plate Connections

Dawn E. Lehman1; Charles w. Roeder2; David Herman3; Shawn Johnson4; and Brandon Kotulka5

Abstract: Current design practices use a strength-based design approach to design gusset plate connections in special concentrically
braced frames 共SCBFs兲, in which the expected tensile and compressive capacities of the brace are used to design the gusset plate and the
weld used to connect the brace to the frame. To achieve brace end rotation, the gusset plate is sized using a linear offset rule, and large,
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

uneconomical gusset plate connections may result. A research program was undertaken to improve the economy and performance of
gusset plate connections. A new approach is proposed in which the gusset plate design is based on a balanced design approach in which
the yield mechanisms of the brace are balanced with the yield mechanisms of the connection and the failure modes of the system to
achieve a target yielding hierarchy and suppress unwanted failure modes. Full-scale one-story, one-bay frames were designed and tested
to investigate the seismic performance of current and proposed design methods. In the test program, variations in balance factors between
the brace, gusset plate, and weld were considered to study the proposed method, to evaluate possible yield mechanisms and failure modes,
and to obtain the desired yielding hierarchy. Comparison of the observed and measured performance of each specimen is made and
specific design expressions to improve the seismic engineering of SCBF systems are proposed.
DOI: 10.1061/共ASCE兲0733-9445共2008兲134:6共890兲
CE Database subject headings: Earthquake engineering; Frames; Bracing; Plates; Connections.

Introduction Gusset plate connections have been used for many years, and they
are designed using the Whitmore width and the AISC Uniform
Concentrically braced frames 共CBFs兲 are economical, and their Force Methods 共UFM兲 共Whitmore 1952, Thornton 1991; AISC
strength and stiffness assist in achieving serviceability limit states 2005b兲. These methods are not based on seismic design prin-
for performance-based seismic design 共PBSD兲. During severe, in- ciples, which add to the uncertainty in the design method. How-
frequent earthquakes, brace yielding and buckling occur, and this ever, they are backed by significant experimental data 共e.g.,
behavior provides the deformation and energy dissipation capaci- Bjorhovde and Chakrabarti 1985; Hu and Cheng 1987; Brown
ties to satisfy life-safety and collapse-prevention performance ob- 1988; Rabinovich and Cheng 1993; Cheng et al. 1994; Grondin et
jectives. The special concentrically braced frame 共SCBF兲 design al. 2000兲. A recent literature review 共Chambers 2004兲 of gusset
requirements in the AISC Seismic Design Provision 共AISC plate connection research identified more than a hundred refer-
2005a兲 address these objectives. SCBFs have increased in usage ences; however these studies do not accurately simulate the seis-
in recent years, because of these advantages and uncertainty as to mic behavior of gusset plate connections.
the performance of special moment resisting frames after the Most past experiments evaluate gusset plate behavior to failure
1994 Northridge Earthquake 共Roeder 2002兲. but few employ inelastic cyclic deformation demands. The pre-
The seismic behavior of SCBFs is not well understood by vious tests ignore complete frame action and exclude the de-
engineers. The design and performance of gusset plate connec- formation demands resulting from cyclic inelastic axial and
tions, which join the brace to beams and columns as schemati- postbuckling deformation of the brace. Past experimental results
cally shown in Fig. 1, are primary concerns in SCBF design. were compared to analytical and design models for connection
performance 共Roeder et al. 2004, 2005兲. The comparison showed
1
Assistant Professor, Dept. of Civil and Environmental Engineering, great variability in the accuracy of current design models, and
Univ. of Washington, Seattle, WA 98103. E-mail: delehman@ suggests that improvements are needed to accurately predict the
u.washington.edu connection resistance and system behavior. Recent tests show that
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of
the seismic performance of these is different from that postulated
Washington, Seattle, WA 98103.
3
Structural Engineer, Magnusson Klemencic Associates, 1301 Fifth in the SCBF design assumptions, and suggest that a broader, ho-
Ave., Suite 3200, Seattle, WA 98101-2699. listic perspective of gusset plate connection design and behavior
4
Structural Engineer, Magnusson Klemencic Associates, 1301 Fifth is required 共Uriz and Mahin 2004兲.
Ave., Suite 3200, Seattle, WA 98101-2699. This broader perspective was the focus of a recent research
5
Structural Engineer, Magnusson Klemencic Associates, 1301 Fifth study, which was aimed at developing such a design method.
Ave., Suite 3200, Seattle, WA 98101-2699. Experiments were carried out to evaluate and improve the seismic
Note. Associate Editor: James S. Davidson. Discussion open until response of SCBF gusset plate connections. Results from 13
November 1, 2008. Separate discussions must be submitted for individual large-scale tests are presented. The test specimens simulated a
papers. To extend the closing date by one month, a written request must
single-story, single bay of a low-to-mid-rise building. The speci-
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on February 5, 2007; mens were integrated component tests, which included beams,
approved on July 31, 2007. This paper is part of the Journal of Struc- columns, a brace, and two gusset-plate connections, and were
tural Engineering, Vol. 134, No. 6, June 1, 2008. ©ASCE, ISSN 0733- conducted to assure a realistic simulation of the demand on and
9445/2008/6-890–901/$25.00. performance of the gusset-plate connection. The results showed

890 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Schematic of SCBF gusset plate connections and design checks 共a兲 welded tube brace; 共b兲 bolted angle connection; 共c兲 gusset plate welds;
and 共d兲 free edge buckling

clear limitations in the current design methods and potential im- based upon the work point of the brace and the frame geometry.
provements needed for seismic design. The results were compiled Failure modes that must then be considered, as suggested in Fig.
and analyzed to develop design expressions that assure that the 1, include tearing or fracture of the brace, net section fracture of
gusset plate rotation resistance and stiffness meet the require- the brace or gusset plate, weld fracture of the gusset plate welds,
ments of the SCBF system. Implementing these improvements in shear fracture of the bolts, block shear, excessive bolt bearing
design practice will result in more compact and economical con- deformation, and buckling of the gusset plate. The primary design
nections with increased inelastic deformation capacity and im- steps for a gusset-plate connection, which take these failure
proved performance. modes into consideration, are as follows.
• The welds or bolts used to attach the brace to the gusset plate
must be designed to provide the expected tensile yield resis-
Current AISC Design Method for SCBF Frames tance of the brace, and the weld length or bolt group must also
be checked using the block shear design expression.
The CBF and UFM design methods 共AISC 2005b兲 assume that
• The yield and buckling strengths of the plate are calculated
the frame acts as an in-plane truss, and neglect effects associated
using the Whitmore width and modified Thornton design ex-
with frame bending and brace buckling. The truss assumption is
pressions and compared with the tensile and compressive
made by designers and is explicitly stated in the current AISC
Specification 共2005b兲. Gusset-plate connections are designed to strengths of the brace, respectively. 共Fig. 1 shows these rel-
assure that the factored resistance, ␾Rn, for each connection de- evant variables for bolted and welded connections, respec-
sign failure mode exceeds the axial demand of the brace, Pu. The tively.兲 The Whitmore width is defined by a 30° projected
axial demand, Pu, is a factored design load for most nonseismic angle from the start to the end of the bolted or welded joint. In
applications. For seismic design 共AISC 2005a兲, Pu is defined as addition, an edge buckling check 共Brown 1988; Astaneh-Asl
the expected tensile yield force 共RyAgFy兲 or expected compres- 1998兲 is often employed, as suggested in Fig. 1共c兲, but com-
sive buckling force 共RyAgFcr兲, respectively. In the products, parison with past test results shows poor correlation between
Ry⫽ratio of the expected yield stress to the minimum specified edge buckling predictions and experimental results 共Roeder
yield stress; Ag⫽gross area of the brace; and Fy⫽minimum speci- et al. 2005兲.
fied yield stress of the steel. • The welds, which are fillet or complete joint penetration 共CJP兲
For connection design, the geometric proportions 关dimensions welds, attaching the gusset plate to the beam and column are
a and b shown in Figs. 1共a and b兲兴 of the gusset plate are initially sized 共the interface welds兲 for design forces determined using

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 891

J. Struct. Eng. 2008.134:890-901.


Fig. 2. Photos of damaged CBF gusset plate connections 共a兲 fracture
of beam adjacent to gusset plate; 共b兲 premature bucking of gusset
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. 共a兲 Yield mechanisms; 共b兲 failure modes for SCBFs

equilibrium methods with the expected tensile force in the However, where the capacity-design method assures that the duc-
brace. tile element yields first, the proposed balanced design method
The AISC seismic design provisions require allowances for the maximizes the system ductility and inelastic deformation capac-
brace end rotation, and a 2t p-linear clearance at the end of and ity, because it controls the sequence of yielding and maximizes
parallel to the axis of the brace is commonly employed 共Fig. 1兲, the inelastic deformation. In most cases, greater ductility is
where t p⫽thickness of the gusset plate. This requirement can re- achieved with the proposed method by assuring that multiple,
sult in very large gusset plates which can lead to uneconomical desirable yield mechanisms are developed prior to fracture or
seismic design and inferior seismic performance. failure of the connection 共Roeder 2002兲.
Several design concerns are apparent when past research is This process satisfies serviceability design limits, as each
considered. First, there are significant inconsistencies between de- member has a resistance greater than a specified seismic design
sign model predictions and past experimental results 共Roeder et requirement. The required capacities of the connection must then
al. 2004, 2005兲. Second, substantial variations in design practice exceed the expected plastic resistances of these members.
are noted, because the requirements of the UFM were not specifi- Collapse-prevention and life-safety design objectives require that
cally developed for seismic design and has several variations in the system sustain a significant ductility demand without exhibit-
application. Third, economy and performance of these connec- ing undesirable failure modes, and this is assured by a sequence
tions is uncertain. These issues are exacerbated by results from of yielding before the target failure mode is permitted and unde-
two recent experimental research studies 共Uriz and Mahin 2004; sirable failure modes are suppressed. Some failure modes have
Chen et al. 2004兲 which have shown premature damage to beams, more severe consequences than others, and the balancing method
columns, and braces adjacent to the gusset plate connections such accounts for this by providing greater separation between the less
as the beam fracture shown in Fig. 2共a兲 and gusset plate buckling, desirable failure modes and the primary yield resistance thereby
which occurred at relatively small deformations, shown in Fig. assuring that these undesirable behaviors will not occur.
2共b兲. These recent test results show that the seismic performance Fig. 3 illustrates the possible yield mechanisms and failure
of SCBF gusset-plate connections is different from those postu- modes for SCBFs, and the following equation expresses the pro-
lated in the design assumptions and that brittle and unexpected posed yield mechanism balance procedure used to establish this
failures can result at moderate story drifts, which are well below yielding hierarchy:
the performance required by SCBF design. The goal of seismic
design must be to assure that adequate strength, stiffness, energy Ryield,mean = RyRyield ⱕ ␤y1Ry1Ryield,1 ⱕ ␤y2Ry2Ryield,2 ⱕ ¯
dissipation or ductility, and inelastic deformation capacities are
achieved. Further, for PBSD, these criteria must be achieved to ⱕ ␤yiRyiRyield,i 共1兲
satisfy the multiple performance states. These goals are the focus The nominal yield resistances, Ryield,i, for the identified desirable
of this research. yield mechanisms are separated by balance factors, ␤yi, to control
the resistance of secondary yield mechanisms and to achieve the
yielding hierarchy. The primary yield mechanism, Ryield,mean, is
Overview of Balanced Design Method to Improve followed by secondary yield mechanism 1, which are followed by
Performance secondary yield mechanism 2, etc. The ␤ values, which define the
separations, are based on the ductility and experimental perfor-
A design methodology based on balancing the yield mechanisms mance of the various yield mechanisms and the separation re-
and preventing undesirable failure modes is proposed to address quired to achieve the performance goals. These ␤ have similar
the inadequacies in current SCBF gusset-plate connection design. characteristics to the resistance factors, ␾, commonly used in
Members initially are designed to meet the design seismic force load and resistance factor design, but they are fundamentally dif-
demands, and specific yielding elements 共e.g., braces in SCBFs兲 ferent, because ␤ values are selected to achieve ductility and
are designed to sustain the inelastic deformation demands. Capac- inelastic deformation capacity rather than strength.
ity design principles are used to avoid premature failure and to Failure modes cause fracture, tearing, reduction of resistance,
achieve a progression of inelastic behaviors directed toward a or deterioration of performance. A single failure mode typically
preferred yield sequence and selected failure mode. These two will not result in collapse or total failure of the connection or the
objectives are achieved by balancing the relative resistances of all system, because multiple failure modes are usually required to
of the yield mechanisms and failure modes. achieve this extreme condition. However, a single failure mode
This approach is an extension of the capacity-design approach. causes significant, irrecoverable damage. The following addi-

892 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


N⫽number of plate thicknesses, t p, that the elliptical line is to be
offset from the intersection of the beam and the gusset plate.
To study the impact of the gusset plate design on the frame
performance, the study parameters that were investigated in-
cluded: 共1兲 Plate thickness, 共2兲 brace clearance requirement 共stan-
dard or elliptical兲, 共3兲 weld type and dimensions, 共4兲 taper of the
gusset plate, and 共5兲 strength and stiffness of the beams. The last
parameter was included to test the hypothesis that the stiffness of
the framing element also influences the cyclic response 共Yoo
2006兲.
Thirteen specimens were tested and designated as HSS 共indi-
cating that the brace was a hollow steel section兲. Fig. 5 shows a
typical test specimen. Table 1 lists salient details of all specimens.
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Nominally identical sections were used for the beams 共W16


⫻ 45兲, columns 共W12⫻ 72兲, and brace 共HSS 5 ⫻ 5 ⫻ 3 / 8兲, as in-
dicated in Fig. 5. Specimen HSS-11 was the exception since a
larger beam section was used, as indicated in Table 1. The frame
centerline dimensions measured 3.66 m 共12 ft兲 high by 3.66 m
共12 ft兲 wide. The gusset plate was attached to the beam and col-
umn using a fillet weld on both sides 共F兲 or CJP weld. The brace
was connected to the gusset plate using a fillet weld. All welds
were E71T-8 flux core arc welds satisfying the toughness require-
Fig. 4. Proposed elliptical-clearance requirement and associated ments for AISC Seismic Provisions demand critical welds 共AISC
geometry 2005a兲. The length of the interface weld between the gusset and
beam or column were approximately 30– 40 mm less than the
variables a and b, which indicate the length between the gusset
tional balancing expressions are used to separate and balance plate and the flange of the beam or column, respectively. 共This
critical failure mode resistances, and to assure that multiple yield difference results from the clipped corner of the gusset plate at the
mechanisms occur. bam and column flange intersections.兲 The thickness of the fillet
weld used for each specimen is specified in Table 1.
Ryield,mean = RyRyield ⱕ ␤fail,1Rfail,1 ⱕ ␤fail,2Rfail,2 ⱕ ¯ ⱕ ␤yield Each specimen or subseries of specimens were designed to
study a specific parameter. Specimen HSS-1 was designed to re-
⬍ ␤fail 共2a兲 flect current seismic design specifications and practice and served
or if multiple yield mechanisms are desired as a reference; HSS-12 was nominally identical to HSS-1 but
detailed with a CJP weld. These were the only specimens de-
Ry,1Ryield,1 ⱕ ␤fail,1Rfail,1 ⱕ ␤fail,2Rfail,2 ⱕ ¯ 共2b兲 signed using the 2t p linear offset. The gusset plates of the remain-
ing specimens were designed using the proposed elliptical
Eq. 1共a兲 assures that the resistances of all failure modes, Rfail, clearance with additional variations in the elliptical offset dis-
exceed the primary yield resistance, Ryield,mean. Combining Eqs. tance, the interface weld, and the degree of taper, as defined in
共1兲 and 共2b兲 assures that the resistances of all of the failure modes Table 2. The gusset plate of HSS-2 was modified to meet a 6t p
exceed the primary yield resistance and the first secondary yield elliptical clearance requirement and the interface weld used to
mechanism. The magnitude of ␤fail,i is determined such that less connect the plate to the beams and columns was sized to meet the
favorable failure modes have greater separation 共i.e., smaller ␤ plastic capacity of the plate. The thickness of each gusset plate
values兲 than more favorable failure behaviors and smaller prob- used for Specimens HSS-2 and HSS-7 was varied. The offset
ability of occurrence. An earlier paper 共Roeder et al. 2005兲 de- distance used for the elliptical clearance was varied from 3t p
scribes this balance procedure and the rationale used to establish 共HSS-8兲 to 9t p 共HSS-4兲. Taper was studied by using nominally
the ␤ actors. identical plates for Specimens HSS-2 and HSS-10 with a 0 or 15°
taper, respectively. To study the influence of the weld type, the
following pairs of companion, nominally identical specimens
Test Program were detailed with fillet and CJP welds, respectively: HSS-1/
HSS-12, HSS-2/HSS-9, and HSS-10/HSS-13.
Previous tests that have focused on the cyclic behavior of braces The test setup was designed to simulate, to the extent possible,
or connections have neglected the influence of the coupled behav- the boundary and loading conditions present on a one-story, one-
ior of the components of a braced frame and nonlinearity in the bay frame in a SCBF building. Fig. 6 shows the setup and indi-
framing elements. In this test program, the test specimens and cates the components of the test rig. The frame was placed
setup were designed incorporate these important response mecha- horizontally 共parallel to the laboratory floor兲. A horizontal,
nisms. The primary focus of the test program was the influence 1,468 kN 共330 kip兲 actuator applied the cyclic lateral load to the
of the gusset plate design on the overall cyclic lateral response of specimen through a load beam; the load beam extended over ap-
SCBFs. To improve both the constructability and performance of proximately one-third of the specimen width to ensure a uniform
the connections, the 2t p linear clearance distance discussed earlier transfer of the load without excessive restraint of the beam rota-
was replaced using an elliptical clearance requirement, shown in tion. The specimen reactions were distributed to the reaction wall
Fig. 4, for most tests. As shown in the figure, the ellipse is offset through a pair of channels which was securely attached to the
from the beam and column faces a distance of Ntp, where wall. The specimen was supported by direct bearing on the chan-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 893

J. Struct. Eng. 2008.134:890-901.


Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Typical test specimen

nels in compression and a series of bolts provided shear transfer Instrumentation was placed to monitor loads, global displace-
from the lower beam to the channels. A pair of rods was used to ments, and local strains and deformations in the brace, framing
apply a 1,557 kN 共350 kip兲 axial load to each column of the elements, and the gusset plate. Load cells monitored the applied
specimen. This axial load simulated gravity loads from upper sto- lateral and axial loads. Potentiometers monitored the movement
ries and limited column uplift during the test. A series of steel of the specimen and test rig. Additional potentiometers monitored
assemblies fitted with lubricated nylon sliding on polished metal local rotations and translations of the beams, columns, and gusset
surfaces provided out-of-plane restraint with minimal frictional plates as well as elongation of the brace. Strain gauges were at-
resistance. tached to beams, columns, and the brace in regions where limited

Table 1. Test Specimen Properties


Yield and ultimate strengths 共MPa兲 Gusset plate geometry 共mm兲
Brace
Specimen Beam Column Brace Gusset length Weld t2
designation 共A992兲 共A992兲 共A500兲 共A572兲 tp a b Offset Lb共m兲 type 共mm兲
HSS-1 410 516 414 505 483 529 821 876 12.8 863.6 762.0 2t 3.42 F 4.76
HSS-2 405 388 388 516 483 529 454 552 12.4 635.0 533.4 E-6t 4.01 F 12.70
HSS-3 376 503 412 532 483 529 354 503 12.5 635.0 304.8 E-6t 4.01 F 11.11
HSS-4a 396 508 380 514 505 549 354 503 12.5 647.7 543.1 E-9t 3.90 F 11.11
HSS-5 395 501 409 522 505 549 447 602 9.5 635.0 533.4 E-8t 4.01 F 7.94
HSS-6 376 503 412 532 446 481 446 599 9.6 635.0 304.8 E-8t 4.01 F 7.94
HSS-7 395 501 409 522 446 481 407 531 22.0 723.9 622.3 E-6t 3.79 F 19.05
HSS-8 419 532 396 525 446 481 509 621 9.6 596.9 495.3 E-3t 4.12 F 11.11
HSS-9 421 531 392 519 446 481 473 603 12.9 635.0 533.4 E-6t 4.01 CJP —
HSS-10b 412 524 408 524 454 490 473 672 12.9 476.3 415.9 E-7t 3.99 F 7.94
HSS-11c 396 512 404 516 454 490 435 547 22.6 723.9 622.3 E-6t 3.79 CJP —
HSS-12 414 510 414 524 455 490 414 552 12.8 863.6 762.0 2t 3.42 CJP —
HSS-13b 414 510 414 524 455 490 414 552 12.7 476.3 415.9 E-7t 3.79 CJP —
a
Brace splice length was reduced to 349 mm 共from 374 mm兲.
b
Gusset plate detailed with 15° taper for HSS-10 and HSS-13.
c
W16⫻ 89 section used for beams in Specimen HSS-11.

894 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


Table 2. Global Experimental Results
Drift ratio 共%兲 Load 共kN兲
Specimen
Specimen tp nt p tw / t p description ⌬max ⌬min ⌬comp,tear Range Max Min
HSS-01 12.78 n/a 0.4 Ref. 2t w / 1 / 2 in. pl 1.3 −1.6 −1.0 2.9 1,483 −860
HSS-02 12.42 6 1.0 6t w / 1 / 2 in. pl 1.5 −2.1 −2.0 3.6 1,514 −757
HSS-03 12.50 6 0.9 6t w / 3 / 8 in. pl 1.5 −3.0 −2.2 4.5 1,635 −757
HSS-04 12.50 9 0.9 9t w / 3 / 8 in. pl 1.5 −2.9 −2.5 4.4 1,483 −766
HSS-05 9.53 8 0.8 HSS-03 w / 5 / 16 in. weld 1.7 −3.1 −1.7 4.8 1,586 −721
HSS-06 9.55 8 0.8 HSS-05 w/reinf. weld 1.7 −3.0 −1.9 4.8 1,577 −712
HSS-07 22.02 6 0.9 6t w / 7 / 8 in. pl 1.3 −2.8 −2.7 4.0 1,595 −927
HSS-08 9.63 3 1.2 3t w / 3 / 8 in. pl 2.3 −2.6 −1.7 4.9 1,519 −771
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

HSS-09 12.85 6 cjp 6t w / 1 / 2 in. pl & cjp weld 1.3 −2.4 −1.8 3.7 1,631 −762
HSS-10 12.85 7 0.6 HSS-02 w/15° taper 1.9 −2.5 −2.2 4.5 1,564 −708
HSS-11 22.58 6 cjp HSS-07 w/large beam 1.1 −1.5 −1.5 2.6 1,841 −896
HSS-12 22.78 n/a cjp Ref. 共HSS-01兲w/CJP 1.4 −2.1 n/a 3.5 1,671 −806
HSS-13 12.70 7 cjp Tapered w/CJP 1.9 −2.1 n/a 4.0 1,564 −788

yielding was expected to monitor axial load, shears, and mo- testing protocol 共ATC 1992兲. Fig. 7 shows the target history. The
ments. For some specimens, biaxial strain gauges were used to magnitude of each cycle was a multiple of the yield drift, ␪y, but
monitor local yielding of the gusset plate. Additional details of the different drifts corresponding to buckling and tensile yielding of
test setup may be found elsewhere 共Johnson 2005; Herman 2006兲. the brace are expected. As a result, the number of cycles applied
in these tests was larger than normally required by the ATC pro-
tocol in order to fully capture the performance of yielding and
Experimental Response
buckling in both directions of loading. Differences in the applied
The specimens were tested to observe the cyclic, nonlinear re- drift history and that actually sustained by the frame resulted from
sponse with increasing amplitude drift cycles based on the ATC slip, column uplift, and local bearing deformations. These effects

Fig. 6. Test configuration

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 895

J. Struct. Eng. 2008.134:890-901.


Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Normalized displacement history

were measured and the data was corrected to obtain the actual
drift history of the frame.
One of the primary objectives of the test program was to de-
velop PBSD tools for SCBFs. Therefore, the primary and second-
ary yield mechanisms and all failure modes, which are indicated
in Fig. 3, were carefully observed and the frame drifts and local
deformations at which they occurred were noted. Similar mecha-
nisms and modes were observed for most test specimens.
In all specimens, the initial yield mechanism was brace buck-
ling, as shown in Fig. 8共a兲, and it occurred at drift ratios ranging
from −0.2 to −0.5%. Tensile yielding of the brace occurred under
load reversal at drift ratios ranging from 0.4 to 1%. With addi-
tional drift demand, the brace buckling increased to a moderate
Fig. 8. Brace buckling progression 共a兲 initial; 共b兲 moderate; and 共c兲 level 关Fig. 8共b兲兴. In most test specimens, severe brace buckling
pinching occurred in which local pinching and cupping of the tube was
observed towards the center of the brace 关Figs. 8共c兲 and 9兴. The
drift at which severe buckling occurred varied depending on
the connection design. This response was followed by tearing of
the brace, as shown in Fig. 9共b兲.
Although a brace is typically idealized as a pin-ended 共truss兲
member analytically, the test results show that this may not be
accurate. Out-of-plane movement of the brace caused rotation of
the gusset plate, which in some cases resulted in gusset plate
yielding, as shown in Fig. 10. With cycling, yielding of the plate
in tension occurred in the specimens with thinner or tapered gus-
Fig. 9. Brace failure progression set plates. In some specimens, the large rotation demands on the

Fig. 10. Progression of yielding in gusset plate 共a兲 formation of yield line; 共b兲 significant yielding

896 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


Fig. 11. Progression of weld damage 共a兲 initial tearing; 共b兲 separation; and 共c兲 complete fracture
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

plate led to ductile tearing 共drift levels for initial weld tearing are drift ranges 共⌬tens,max − ⌬comp,max兲 are also provided for each speci-
noted in Table 2兲, or complete fracture of the weld 共HSS-1 only兲, men, because this range represents a better measure of total
as shown in Figs. 11共a and b兲, respectively. ductility.
In some specimens, large frame deflections led to yielding or Only Specimen HSS-1 failed as a result of complete fracture
local buckling of the beams and columns adjacent to the gusset of the gusset-plate interface weld. Specimen HSS-1 was designed
plate because of bending moments in the beams or columns in- for weld forces obtained using the UFM and based on the ex-
duced by frame action, as depicted in Fig. 12. pected brace yield force. Failure of the other specimens was a
result of brace fracture, which was the more ductile and desirable
failure mode. The maximum drift in compression was achieved
Influence of Study Parameters by HSS-5 and HSS-6, which were detailed with thin 共9.5 mm兲
plates using an 8t p elliptical offset clearance requirement. The
Fig. 13 shows the measured normalized force-drift response of six maximum drift 共brace in tension兲 was sustained by Specimen
specimens, specifically: HSS-8, which was detailed with a thin plate and a 3t p elliptical
• HSS-1—reference specimen designed using current AISC offset. Specimens HSS-5, HSS-6, and HSS-8 achieved the maxi-
specification; mum drift ranges. In contrast, the minimum drift capacities in
• HSS-2—elliptical, 6t p clearance requirement with t p tension and compression, and the smallest drift range were
= 12.7 mm 共1 / 2 in.兲; achieved by Specimen HSS-11, which was designed using a large
• HSS-5—elliptical, 8t p clearance requirement with t p = 9.5 mm beam and thick gusset plate. These results show the importance of
共3 / 8 in.兲; the connection flexibility on the global frame response.
• HSS-7—elliptical, 6t p clearance requirement with t p A more detailed evaluation of specific behaviors permits
= 22.2 mm 共7 / 8 in.兲; evaluation of several gusset plate design parameters and issues
• HSS-10—elliptical, 7t p clearance requirement tapered to meet including plate flexibility, use of and variation in the elliptical
Whitmore width; and clearance requirement, weld geometry, and impact of frame
• HSS-11—nominally identical to HSS-7 with larger 共W16 stiffness. The following subsections evaluate each of these
⫻ 89兲 beams parameters.
In the plots, the measured shear force was normalized by the
shear force corresponding to the design yield strength of the brace
共RyFyAg兲. Curves of the remaining specimens are similar and Offset Distance Using Elliptical Clearance
may be found elsewhere 共Johnson 2005; Herman 2006兲. Table 2
Figs. 13共a and b兲 show Specimen HSS-1, the reference specimen,
provides maximum drifts achieved for the brace in compression
and Specimen HSS-2, designed using the elliptical clearance with
⌬comp,max 共given as a negative percentage兲, and the brace in ten-
a 6t p offset. The comparison suggests that the elliptical clearance
sion ⌬tens,max, as well as the corresponding story-shear forces. The
provides improved system drift capacity, but HSS-1 failed prema-
turely because of the inadequate capacity of the gusset plate weld.
Specimen HSS-12 was nominally identical to HSS-1 expect that a
CJP weld was used, which prevented all weld cracking. Table 2
shows that HSS-12 had somewhat comparable performance to
HSS-2 共same plate thickness with different offset criteria兲, but its
performance was inferior to HSS-5 and HSS-8, which had
slightly larger elliptical offsets. These results clearly show that the
elliptical clearance model provides equal or better performance
than the 2t p linear clearance model if the proper design param-
eters are employed.
Selection of the proper elliptical offset distance 共Ntp兲 depends
on the performance objectives of interest. Comparison of HSS-04
共9t p兲, HSS-05 共8t p兲, and HSS-08 共3t p兲 permits evaluation of the
Fig. 12. Local yield mechanisms in beams and columns 共a兲 extensive influence of the offset distance on the brace and system perfor-
yielding; 共b兲 severe local buckling mance. Table 2 presents measured global drift and load data as

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 897

J. Struct. Eng. 2008.134:890-901.


Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Normalized force-deflection responses for selected specimens

well as the compressive drift demand corresponding to initial achieved, simultaneously delaying weld tearing and nonlinear
weld tearing, ⌬comp,tear. Each specimen had the same specified analyses support these experimental findings 共Yoo 2006兲. Thus,
gusset-plate thickness of 9.7 mm 共3 / 8 in.兲. The “compressive” 8t p is the recommended clearance limit.
drift capacity 共brace in compression兲 is smallest for Specimen
HSS-8 共3t p兲. Conversely, the “tensile” drift capacity 共brace in Plate and Beam Geometries
tension兲 was smallest for Specimen HSS-4 共9t p兲. Specimen Fig. 14共b兲 presents measured drift ranges for three approximate
HSS-5 had the maximum drift range 共8t p兲 before fracture of the gusset plate thicknesses, 9.5, 12.5, and 22 mm. Figs. 13共c and d兲
brace. provide a comparison of specimens with 22 and 9.7 mm thick
Compiling all of the experimental results shows a similar gusset plates. The test data indicate that initial buckling capacity
trend. Fig. 14共a兲 indicates the compressive drift at the onset of was largest for Specimen HSS-7, because the thick gusset plate
weld tearing, which was lowest for the 3t p offset specimen resulted in additional end rotation restraint. However, the speci-
共Specimen HSS-08兲. Fig. 14共b兲 plots the maximum drift range as mens with thinner plates achieved larger tension drifts and drift
a function of the elliptical clearance. These results indicate that ranges. This improved response is likely a result of the increased
using an offset of 8t p permitted the maximum drift range to be flexibility and secondary yielding of the gusset plate.

898 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. Impact of elliptical offset distance on 共a兲 compressive drift at weld tearing, 共b兲 total drift range

The drift capacities were smallest for Specimen HSS-11, Engineering Recommendations
which had a thick plate and larger beam, because of the increased
stiffness of the gusset plate and beam. This increased end restraint Experimentally the effective length factor was determined using
increased the local deformation demand in the brace for a given the buckling capacity of the brace and the total brace length 共Lb兲.
inelastic drift demand thereby reducing the ductility of the frame. On average, the experimental results indicate that an effective
Fig. 13共f兲 shows the limited inelastic deformation range of this length coefficient, k, of 1 used with this true brace length is ap-
specimen. The drifts at initial yielding and brace fracture are larg- propriate for design. Specific values are presented elsewhere
est for the specimens with the thinnest 共HSS-5兲 and tapered 共Johnson 2005, Herman 2006兲.
共HSS-10兲 gusset plates. This increased capacity results from the The elliptical design criterion was developed to achieve large
flexibility, deformability, and yielding of the plate and reduced frame postbuckling drift capacities without excessive weld tear-
strain concentration in the buckled brace. The use of the flexible ing. The experimental results indicate that using an offset distance
gusset plate clearly permitted larger deformation of the brace, and of approximately eight 共8兲 times the plate thickness 共t p兲 meets this
therefore the frame, prior to brace fracture, a direct result of the performance objective. This elliptical clearance model provided
balanced design method. brace rotational capacity and cyclic, inelastic performance that
exceeded that achieved with the current linear clearance model,
simultaneously providing a thinner, more compact gusset plate.
Weld Size and Type The thinner, more compact gusset plate had added benefits in that
they induced less damage in the welds and faming members ad-
Three pairs of specimens were designed to compare the impact of jacent to the gusset plate. With this model, a clear 共8t p wide兲
using complete joint penetration and fillet welds. These were elliptical band is maintained, as shown in the shaded area of Fig.
HSS-1/HSS-12, HSS-2/HSS-9, and HSS-10/HSS-13. The results 4. The actual dimensions of the elliptical band are easy to visu-
indicated that CJP welds either eliminate 共in the case of HSS-12, alize, and they can be determined graphically from the gusset
reference specimen, and HSS-13, specimen with tapered plate, plate dimensions. Although the offset can not be expressed by a
which did not exhibit weld tearing兲 or reduce 共in the Specimen single equation, an approximate analytical solution may be useful
HSS-9, which sustained only initial weld tearing兲 the extent of to the designer and has been developed using the geometric no-
weld tearing over that noted with specimen design using fillet tation shown in Fig. 4. The additional terms are needed for this
welds sized to achieve the plastic capacity of the plate 共Herman approximate derivation:
2006兲. • ␣ is the inclination angle of the brace from the horizontal;
Fig. 14共a兲 presents the drifts corresponding to weld tearing, • c is maximum distance from the centroidal axis to the extreme
⌬comp,tear, as observed during testing, as a function of the ratio of fiber of the brace; and
the weld size 共tw兲 to the plate thickness 共t p兲. Table 2 provides the • l⬘ is the length from the imaginary corner of the rectangular
elliptical clearance distances 共3t p – 9t p兲, weld-to-plate thickness gusset plate to the end of the square cut of the brace. This
ratios, and degree of plate taper for all specimens. Specimens length must include the length of the weld or bolt group as
with a 3t p offset or a weld-to-plate thickness ratio, tw / t p, less than well as the corner that is typically cut from the gusset.
0.9 sustained weld tearing at less than 2% drift. Specimens with a The dimensions a and b should be selected so that the imagi-
3t p offset or tw / t p ⬍ 0.9, sustained additional weld damage to nary corner of the gusset intersects the centroidal axis of the brace
achieve the maximum drift capacity of the frame, but the tough- as shown in Fig. 4. The radii of the ellipse are established as
ness of the welds prevented dramatic weld fracture and assured
reasonable inelastic deformation capacity of the SCBF system.
The acceptability of weld tearing at a given drift level depends on a⬘ = a − 8t p ; b⬘ = b − 8t p 共3兲
the design criteria, however, it is noted that larger fillet welds and
offset distances delay this damage state. and

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 899

J. Struct. Eng. 2008.134:890-901.


a⬘ Table 3. Comparison of Drift Capacities for Various Design ␤ww Values
␳= 共4兲
b⬘ Specimen Study parameter ␤ww ⌬max共%兲 ⌬range共%兲
HSS-05 HSS-03 w/5/16 in. weld 1.0 1.7 4.8
Then, x⬘ and y ⬘ 共as shown in Fig. 4兲 are established as the exact
HSS-06 HSS-05 w/reinf. wield 1.0 1.7 4.8
centerline dimensions to assure the 8t p elliptical clearance
HSS-03 6t w/ 3/8 in. pl 1.0 1.5 4.5
HSS-10 HSS-02 w/15° taper 0.8 1.9 4.5
y ⬘ = a⬘ sin共arctan共␳ tan共␣兲兲兲 共5a兲
HSS-09 6t w/ 1 / 2 in. pl & cjp wield 0.8 1.3 3.7
and HSS-02 6t w/ 1 / 2 in. pl 0.8 1.5 3.6
HSS-01 Ref. 2t w/ 1 / 2 in. pl 0.8 1.3 2.9

x⬘ = a⬘ 冑 冉 冊
1−
y⬘
b⬘
2
共5b兲
HSS-07
HSS-11
6t w/7/8 in. pl
HSS-07 w/large beam
0.4
0.4
1.3
1.1
4.0
2.6
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

The x⬘ and y ⬘ dimensions 共measured from the imaginary cor-


ner兲 are the exact clearance dimensions for the brace if the brace Conclusions and Recommendations for Future
had no width. However, the brace has finite width, and the goal is Work
to assure that the extreme corners of the brace also clear the
elliptical zone. An approximate solution is required here based The experimental program presented studied the inelastic, seismic
upon the slope of the ellipse at the centroid of the brace performance of SCBF gusset plate connections to improve the
performance relative to the current AISC design provisions, si-

␤ = arctan 冉 冑 冊
−2

a ⬘2
x ⬘2
共6a兲
multaneously improving the constructability of SCBF systems.
The study focused on the gusset plate clearance needed to accom-
modate brace end rotation, the size and type of weld used to
connect the gusset plate to the framing elements, and the impact
Corr = c sin共␤兲cos共␣兲 共6b兲 of the gusset plate stiffness, thickness and geometry and the con-
nection and system performance. The study indicates that:
and finally • The elliptical gusset plate clearance model with an 8t p offset
distance provides greater system ductility and deformation ca-
l⬘ = 冑x⬘2 + y ⬘2 + Corr 共7兲 pacity, simultaneously delaying or limiting fracture of the
welds or brace.
This procedure was evaluated for different brace angles and cross • Use of the elliptical clearance model results in more compact,
sections, and gusset plate sizes and dimensions. The results show flexible gusset plates, reduced damage to the welds and beams
that it provides a corner intersection of the brace that approxi- and columns than the current linear offset model.
mates the 8t p band, except in cases where the brace angle be- Design recommendations appropriate for SCBF systems were
comes very flat or very steep. developed and include:
The weld size requirements were studied both experimentally • Use of an effective length 共K兲 coefficient of 1 共one兲 for tubular
and analytically 共Johnson 2005; Yoo 2006兲. The results indicate braces with gusset plate connections to provide a good esti-
that for a matching electrode, the fillet weld size should be equal mate of the brace buckling capacity.
to or greater than the thickness of the gusset plate. Yoo 共2006兲 • An approximate expression for the elliptical 8t p offset.
presents an equivalent expression for a different electrode; how- • Use of interface fillet welds, which connect the gusset plate to
ever the expression has not been verified experimentally. the beams and columns, with a thickness equal to the gusset
The experimental results indicate that yielding and flexibility plate thickness and an electrode that matches the gusset plate
of the gusset plate permits increased SCBF drift capacity, and metal to prevent premature weld cracking. Using a CJP weld at
reduced cracking of the weld and yielding of the beams and col- this location also reduces the potential for weld cracking.
umns adjacent to the connection. To quantify the benefit of bal- • The thickness of a rectangular gusset plate with welded HSS
ancing the tensile strength of the brace and the plate, the balance tube braces should be sized using Eq. 共8兲 which is based on the
factor for the plate strength based on the Whitmore width, ␤ww, Whitmore width for the expected tensile capacity of the plate
was calculated using the expression provided in and the brace with ␤ww = 1 for rectangular plates.
The research results still leave unanswered questions regarding
Ry,bFy,bAb other aspects of the SCBF frame design including alternative
␤ww = 共8兲 brace sections, beam-to-column connection and bolted connec-
Ry,gpFy,gpbwt p
tions. Additional research is needed to address issues regarding
The overstrength factors used were Ry,b = 1.4 and Ry,gp = 1.1 for the system response including brace configuration, geometry, and
the materials specified. The variables Fy,b and Fy,gp indicated the three dimensional effects. Future research by the writers and oth-
specified yield strengths for the brace and gusset plate, respec- ers is planned to explore these topics.
tively 共345 MPa兲, and the variable bw⫽Whitmore width. Table 3
provides the calculated balance factors for all specimens designed
using the 8tp elliptical clearance requirement and the reference Acknowledgments
共HSS-1兲 specimen. Specimens HSS-05, HSS-06, HSS-03, and
HSS-10 achieved the maximum drift range. For the rectangular The research results presented here were the result of studies
specimens, the balance factor, ␤ww, was approximately 1. Only funded by the National Science Foundation, Grant No. CMS-
Specimen HSS-10, which was designed using a tapered plate, 0301792, Performance-Based Design of Concentrically Braced
sustained equivalent drift capacities with a lower ␤ww factor. Frames 共Dr. Steven McCabe, Program Manager兲 and the Ameri-

900 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008

J. Struct. Eng. 2008.134:890-901.


can Institute of Steel Construction 共Mr. Tom Schlafly, Research strength of gusset plate connections.” J. Struct. Eng., 120共2兲, 538–
Director兲. The structural steel shapes were donated by Nucor- 559.
Yamato steel 共coordinated by Mr. Michael Engstrom兲 and the Grondin, G. Y., Nast, T. E., and Cheng, J. J. R. 共2000兲. “Strength and
high-strength steel tubes were donated by Columbia Structural stability of corner gusset plates under cyclic loading.” Proc., Annual
Tubing 共coordinated by Mr. Engstrom兲. The writers gratefully ac- Technical Session and Meeting, Structural Stability Research Council.
Herman, D. 共2006兲. “Further improvements on and understanding of
knowledge this support. SCBF systems.” MS thesis, Univ. of Washington, Seattle.
Hu, S. Z., and Cheng, J. J. R. 共1987兲. “Compressive behavior of gusset
plate connections.” Structural Engineering Rep. No. 153, Univ. of
References Alberta, Alberta, Ont., Canada.
Johnson, S. 共2005兲. “Improved seismic performance of special concentri-
American Institute of Steel Construction 共AISC兲. 共2005a兲. “Seismic pro- cally braced frames.” MS thesis, Univ. of Washington, Seattle.
visions for structural steel buildings.” AISC/ANSI Standard341-05, Rabinovitch, J. S., and Cheng, J. J. R. 共1993兲. “Cyclic behavior of steel
gusset plate connections.” Structural Engineering Rep. No. 191, Univ.
Downloaded from ascelibrary.org by COLUMBIA UNIVERSITY on 02/17/13. Copyright ASCE. For personal use only; all rights reserved.

Chicago.
of Alberta, Alberta, Ont., Canada.
American Institute of Steel Construction 共AISC兲. 共2005b兲. Steel construc-
Roeder, C. W. 共2002兲. “Connection performance for seismic design of
tion manual, 13th Ed., Chicago. steel moment frames.” J. Struct. Eng., 128共4兲, 517–525.
Applied Technology Council 共ATC兲. 共1992兲. “Guidelines for cyclic seis-
Roeder, C. W., Lehman, D. E., and Yoo, J. H. 共2004兲. “Performance
mic testing of components of steel structures.” ATC 24.
based seismic design of braced-frame connections.” 7th Pacific Struc-
Astaneh-Asl, H. 共1998兲. Seismic behavior and design of gusset plates,
tural Steel Conf.
Steel Tips, Structural Steel Education Council, Moraga, Calif. Roeder, C. W., Lehman, D. E., and Yoo, J. H. 共2005兲. “Improved seismic
Bjorhovde, R., and Chakrabarti, S. K. 共1985兲. “Test of full size gusset design of steel frame connections.” Int. J. Steel Struct., 5共2兲, 141–
plate connections.” J. Struct. Eng., 111共3兲, 667–684. 153.
Brown, V. L. S. 共1988兲. “Stability of gusseted connections in steel struc- Thornton, W. A. 共1991兲. “On the analysis and design of bracing connec-
tures.” Ph.D. thesis, Dept. of Civil Engineering, Univ. of Delaware, tions.” Proc., National Steel Construction Conf., AISC, Section 26,
Del. 1–33.
Chambers, J. 共2004兲. Brace frame gusset plate research—Phase I litera- Uriz, P., and Mahin, S. 共2004兲. “Seismic performance of concentrically
ture review, Univ. of Utah, Salt Lake City. braced steel frame buildings.” Proc., 13th World Conf. on Earthquake
Chen, C., Lai, J., Lin, M., Weng, Y., and Tsai, K. C. 共2004兲. “Pseudo- Engineering, Paper 1639.
dynamic test of a full-scale cft/bfb frame: Part 2—Construction and Whitmore, R. E. 共1952兲. “Experimental investigation of stresses in gusset
testing.” Proc., 13th World Conf. on Earthquake Engineering, Paper plates.” MS thesis, Univ. of Tennessee, Knoxville, Tenn.
2175, Canadian Association for Earthquake Engineering, International Yoo, J. H. 共2006兲. “Analytical investigation on the seismic performance
Assocaition of Earthquake Engineering. of special concentrically braced frames.” Ph.D. dissertation, Univ. of
Cheng, J. J. R., Yam, M. C. H., and Hu, S. Z. 共1994兲. “Elastic buckling Washington, Seattle.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2008 / 901

J. Struct. Eng. 2008.134:890-901.

Vous aimerez peut-être aussi