Professional Documents
Culture Documents
2, 53-74 (1993)
SUMMARY
This paper provides an overview of the behavior and design of seismic-resistant eccentrically braced frames
(EBFs). Basic characteristics of EBFs are first discussed. The important effects of link length on the
performance of EBFs are reviewed. The capacity design concept for EGFs is then addressed. The paper
addresses several design issues that appear to have been inadequately considered either in current practice
or in the emerging seismic code provisions. Some important observations are provided from pseudodynamic
tests of large EBFs and experimental studies of EBF subassemblages with link-to-box column connections
conducted recently. Future research needs are discussed.
1. INTRODUCTION
Within the last decade, the steel eccentrically-braced frame (EBF) has evolved' and been fully
codified' as an effective framing system for building structures to resist seismic forces. The
distinguishing characteristic of an EBF is that it possesses the stiffness of conventional
concentrically braced frames (CBF) and the ductility capacity of properly detailed moment
resisting frames (MRF). The high elastic stiffness of EBFs is provided by the braces, and the
high ductility capacity is achieved by transmitting the brace force to another brace or to a
column through shear and bending in a beam segment designated as a 'link'. Figure 1 indicates
three common EBF arrangements. In the figure, the link lengths are identified by e. When
properly detailed, these links provide a reliable source of energy dissipation. Brace and beam
buckling outside of the link can be inhibited by designing these members to resist forces
associated with the strength of the link.
Past experimental research has assessed the performance of isolated links3" as well as EBFs
or subassemblages of small The effects of axial force and bending moment on the beam
segment outside of the link have also been studied recently.' Extensive research and numerous
field applications have provided a good database for code provisions. However, most of these
developments are primarily based on studies of the EBF subassemblages connecting the links
to wide flange Because of a wide acceptance of dual framing systems where EBFs
and moment resisting frames using box columns are combined, and a lack of data on the behavior
of shear link-to-box column connections, two research programs were conducted recently at the
National Taiwan University.' '* '' Using pseudodynamic testing techniques, the effects of the
link web stiffener design on the behavior of long links in EBFs constructed as shown in Fig.
l(b) were also studied."
The purpose of this paper is to provide an overview of the basic design concepts of EBFs and
* Associate Professor.
t Research Assistant.
1062-8OO2/93/01Oo53-22% 16.00
0 1993 by John Wiley & Sons, Ltd.
54 K. C. TSAI, Y. F. YANG AND J. L. LIN
to discuss certain aspects of real EBF behavior and the implications of this behavior for the
designer. The paper addresses specific issues that appear to have been inadequately considered
in the current practice as well as in the emerging structural steel seismic building code
provisions.* Some important observations from recent experimental tests of large EBF sub-
assemblages with link-to-box column connections are presented. The paper concludes with the
future research needs for seismic eccentrically braced frame.
deformations without the development of excessive high local axial strains. In long links, where
flexural yielding dominates, very high bending strains can be developed locally at the link ends
in order to provide the required inelastic link rotations. These highly concentrated bending
strains can lead to premature failure of a long link by fracture of the link flange.' In short links,
the occurrence of shear yielding prevents the development of excessively high flexural strains.
The use of short, shear yielding links is therefore recommended for desired EBF behavior.'.'
e < 1.6-MP
v,
\-c = M i +JfB
Because of the very important influence of strain hardening on link behavior, both shear and
flexural yielding can occur over a wide transition range for links of intermediate length. The
transition from intermediate to very long links is somewhat more difficult to define. At present,
for purposes of computing allowable rotational demand, the latest seismic steel building code
define this transition length at e = 2.6Mp/V,. However, experimental data showed that the
transition to predominantly flexural yielding occurs at a link length of about 3Mp/V,.
e
From (2), it is obvious that link rotational demand grows as link length decreases. Therefore,
links cannot be made too short because the inelastic rotational demand on the link may be
excessive. The determination of the bay width, L, and the link length, e, for (2) will be discussed
in the section on experimental results.
M . LINK
buckling of the web is the controlling ultimate failure mode. Web shear buckling can be
controlled by stiffening the web of the link. The relationship between stiffener spacing and the
available link inelastic rotation capacity up to the onset of web buckling has been fully recognized
and ~odified..~For a moment link connected to a column as shown in cases (a) and (c) of
Figure 1, premature fracture of the flanges is the primary failure mode.7 Additional failure modes
are severe flange buckling and/or lateral torsional buckling of the link or beam outside of the
link. Flange and lateral torsional buckling can be delayed through proper stiffening and lateral
bracing of the link. However, for moment links connected to columns, because of the very highly
concentrated bending strains developed at the ends of the links, fracture of the flange welds may
occur erratically at relatively low inelastic rotation^.^
Numerous tests on shear yielding links of length e ZS 1.6Mp/V,have shown that with proper
stiffening, these links can sustain cyclic inelastic rotations of y = kO.10 rad. The latest seismic
steel building code limits link rotation for e < 1-6Mp/V,to y = f0.09 rad. Recent tests indicate
that failure of long link-to-column connections using welded flanges and web details cannot be
reliably predicted. Therefore, long links attached to columns should be avoided in EBFs. There
is no experimental evidence that long links cannot be used when located between two braces,
shown in Figure l(b). The latest seismic steel building code limits link rotation for e > 2.6Mp/V,
to y = f0-03rad. In order to estimate the link rotation capacity for links of length between
1.6Mp/V,and 2.6Mp/V,,linear interpolation can be used.
i7!
where OL is the angle between the beam and the brace. Neglecting the link axial force, it is assumed
in (7) and (8) that a small amount of shear force of about O.lV,,, exists in the beam segment
outside the link. Since the beam segment outside the link and brace should remain essentially
elastic as ultimate link strength develops, the bending moments in the beam end and brace end
SEISMIC ECCENTRICALLY BRACED FRAMES 59
(4
Figure 6. Brace-to-link connection details
can be adequately estimated by elastically distributing the ultimate link end moment. Both the
beam and brace must be treated as beam-columns in design.I3 Note that the combination of
high axial force together with high bending moment in the beam segment may cause yielding
of the beam. However, if yielding of the beam occurs and the brace is rigidly connected to the
beam, inelastic redistribution of the moment from the beam to the brace may be unavoidable
and acceptable. In such cases, adequate lateral bracing of the beam is essential in order to
maintain the stability of the beam. The ultimate link end moment must be adequately resisted
by the combined flexural strength of the beam and brace with consideration for the axial force
and bending moment interactions. The effects of link axial force on the distribution of the internal
forces between the beam and brace will be assessed in the section on experimental investigation
described later (Section 3).
Figure 7. Sectional modulus ratio versus yield ratio relationships for various strain hardening effects
sustain the beam bending moment as inelastic rotation develops in the beam:
where Z , and Z are the plastic section moduli of the flanges and the entire beam section,
respectively;F, and F, are the minimum yield strength and ultimate tensile strength of the beam,
respectively;fl depends on the magnitude of the beam inelastic rotation and represents the effect
of strain hardening.
Figure 7 illustrates the relationships of the section modulus ratio, Z,/Z, and the strength ratio,
Fu/Fy,for various fl in (9). From (4) and (3, it can be shown that the link end moment, MA,at
the link-column connection is 1.2Mpfor a link of length e = 1*6Mp/I/,.A fl of 1.2 in (9) is
therefore appropriate for evaluating the required strength for the shear link-to-column connec-
tion. From (9) or Figure 7, it is clear that for certain steel grades and beam sections where web
flexural capacity is significant (i.e. small ZF/Z ratio) and yield ratio (Fy/Fu)is relatively large,
the bolted web and welded flange link-column connection details may not be able to sustain
the link flexural demand. In such cases, i.e. Z,F, < BZF,, the beam web should be welded to
the wide flange column and the design strength of the welds should be at least j?ZF, less Z,F,
from the flexural strength point of view. Additional web welds should be provided to develop
the link shear and axial
However, for any beam-to-box column connection where column interior diaphragm plates
are provided opposite the beam flanges, beam flanges carry most of the beam bending moment
regardless of whether the beam web is fully welded to the column or not. This is illustrated in
Figure 8 for three different cantilever beams connected to box columns using finite element
analyse~.'~ From the ratios of the bending moment carried by beam flanges and the entire
beam moment, MF/M as shown in the figure, it can be found that beam moments are transferred
primarily through the beam flanges into the column. Since the stiffness at the beam web-to-
column region is not pronounced, as demonstrated in these finite element analyses, it appears
that, even for box columns with relatively thick plates, beam flange cover plates or stiffeners
may be required for the link-to-box column connection in EBFs when the connection flexural
strength requirement of (9) is violated. The stiffening requirement for the link flange should be
SEISMIC ECCENTRICALLY BRACED FRAMES 61
H900~250~15~18
r ) r b h ~ ZF/Z = 0.59
ms.w+a H900~250~15X25ZF/Z = 0.67
H900X250x15x35 ZF/Z = 0.75
90 r I 10
16 20 24 20 32
Column : BOX 55Ox550xt t (mm)
based on (9) in which 2, includes the effects of cover plates or stiffeners. The link web should
be adequately welded in order to develop the link ultimate shear and axial force.
Using the strength criterion (9) and the corresponding fi of 1.2 for the design of test specimens,
ductile behavior in recent rests' ',15 on several beam-column subassemblages as well as EBF
subassemblages discussed later in this paper have been achieved. The strength requirements for
the beam-column or link-column connections described in this section, which appear not to
have been adequately addressed in the current model steel seismic building code? have been
incorporated into the draft seismic building code of the Republic of China.16
I-----------
__.____A
(b)
Figure 9. Schematic diagram of experimental set-up: (a) location of subassemblages in frame; (b) details of test set-up
SEISMIC ECCENTRICALLY BRACED FRAMES 63
letter R following the specimen identification number designates that repairs have been made
after link flange weld failure occurred in the corresponding specimen.
Cyclically increased loads were applied using one 250-ton capacity actuator placed horizontally
between the reaction wall and the specimen. The specimens were laterally supported at various
locations. The internal force distributions in the link, the beam segment and the brace of each
specimen were computed based on load cell readings and the deformed geometry of the
subassemblage. The tests confirmed that the force distributions correspond reasonably well with
the conditions in the respective EBFs. Some important observations from this series of tests are
given as follows.
1.5
nection Offset
1.0
0.5
2 PI
0.0
P
-0.5
-1.0
-1.5
-0.04 -0.02 0.00 0.02
Y (rad)
3.0 I I I
1.5 -
<3
0.0
-1.5 -
(b)
Figure 10. Link shear versus link deformation relationships: (a) Specimen HE1; (b) Specimen HE2
Braces were subjected to a large axial force induced primarily by the fully yielded and
strain-hardened links. It was verified for each specimen that the shear force in the beam segment
outside of the link contributed about 10% of the total brace axial force, as suggested in (8).
Therefore, this shear should be included in the computation in order to estimate accurately the
design axial force for braces. Beam segments outside the link were subjected to large axial forces
and large bending moments. Design bending moment, as described previously, can be predicted
by distributing the ultimate link end moment based on elastic rotational stiffness of the brace
SEISMIC ECCENTRICALLY BRACED FRAMES 65
Figure 11. Wing plates at link-to-box column connections: (a) Specimen HEI; (b) Specimen HE2
0.30
Spec SEJR
0.15 -
$ 0.00
-0.15 -
-- Consider Deform
- - - Consider Undeform
-0.30 I 1
and the beam segment outside of the link. However, instead of using (7), the ultimate axial force
occurring in the beam segment must be approximated by the following equation (Figure 14):
where PE,I$ are the link axial force and the shear force, respectively, obtained from the elastic
frame analysis of the test structure; u is the angle between the beam and the brace, V,,, is the
ultimate link shear capacity. The design axial force and the design bending moment described
above should be used in the strength and stability comp~tations'~ for braces.
66 K. C . TSAI, Y. F. YANG AND J. L. LIN
100
fl 50
I
t
4
v
4 0
$
s -50
-100
-100 -50 0 50 100
Link Moment MB(ton-m)
Figure 13. Brace end moment and beam end moment versus link end moment relationships (Specimen SE3)
400.0
,
4
v
s 200.0
0)
0
k 0.0
0
k
4
3
(d
-r(
w
4 -200.0
In the test of each specimen, upon extending the actuator the beam bending moment interacted
with the axial force producing a larger compressive stress at the beam top flange than that at
the bottom flange. It can be found from (10) that the beam segment outside the link in Specimen
HE2 violates the code prescribed stability and strength requirements when the strain hardened
yield capacity of the link is reached. During the test the beam segment buckled laterally, and an
additional lateral support was subsequently installed near the buckled beam top flange after the
beam had been straightened under the reversed loading. As shown in Figures 10(b) and 15,
Specimen HE2 sustained several large inelastic cycles without further failure. In Figure 16,
SEISMIC ECCENTRICALLY BRACED FRAMES 67
1.2
0.6
ah 0.0
\
!
I4
a
-0.6
-1.2
-1.5 -1.0 -0.5 0.0 0.5 1.0 1
Klsun/MP
Figure 15. Beam axial force versus bending moment relationships (Specimen HE2)
0.12
i
Spec HE2
0
b
h 0.06 o o
*,
3
-20 0.00
J
3
(d
&
5 -0.06
rd
G
I
-0.12 I I I
I I I I
the Specimen HE2 test also showed that lateral forces as large as 10% of the flange axial strength
developed in the lateral supports at the link end after the occurrence of beam flange buckling.
However, in tests of the other specimens in this study, the largest force developed in the lateral
supports at the link end was generally less than 2% of the flange axial strength, as shown in
Figure 17.
The lateral drifts of the subassemblage and the link rotations are compared. As illustrated in
Figure 18 for Specimen HE2, it is found that the experimental results correlated satisfactorily
with (2) when the center-to-center line dimension of members are used.
68 K. C. TSAI, Y. F. YANG AND J. L. LIN
0.010
Spec SE3 I
I
h
0.005
E4k
t
4
v
2d 0.000
s
&
d
5 -0.005
3 i
iI
rd
;
I ooooo Top Flange I
AAAAA Bottom Flange 1
-0.010
-0.12
I I
-0.06
I
I
0.00
I I
0.06
I i
0.12
Y (rad)
Figure 17. Lateral force versus link rotation relationships (Specimen SE3)
14
Figure 18. Link rotation versus EBF lateral drift relationships (Specimen HE2)
3.3. Conclusions
Based on this limited experimental investigation, the following tentative conclusions can be
reached.
0 The current code provisions for EBFs can be applied to EBFs using box columns if the
link-to-box column connections are properly detailed using the strength criterion proposed
in (9). When there is a horizontal offset between the center lines of the link and the box
column, stiffeners for the conventional link-to-column connection may be required at the
SEISMIC ECCENTRICALLY BRACED FRAMES 69
link-to-box joint in order to reduce the stress concentrations and sustain large inelastic link
deformations.
0 It may not be possible to eliminate completely the link axial force in an ABF; small axial
forces tend to develop in links. It is confirmed that if the link axial force is smaller than 15%
of the beam axial capacity, its effects on the buckling behavior of the link can be neglected.
However, the link axial force should be considered in the design of the beam segment outside
the link in order to avoid buckling.
0 When the members in an EBF are properly proportioned, the link will yield and the behavior
of the brace and the beam outside the link will remain essentially elastic. Force distributions
between the brace and the beam segment outside the link can be accurately predicted using
elastic analysis.
0 As the code requires, the width-thickness ratios of the link section complied with the stringent
compact section requirement. The tests showed that well detailed shear links sustained large
inelastic deformations with controlled web buckling, but no sign of local buckling developed
in the web of the beam segment outside the link despite its violation of the compact section
requirement as large axial force developed in the web. Although not addressed by the current
building codes, it appears that less stringent width-thickness ratios could be tolerated for the
beam segment outside the link.
0 In typical EBSs, the axial force and the bending moment in the beam segment outside the
link can induce larger compressive stress in the beam top flange than that in the bottom flange.
In this situation, lateral torsional buckling of the beam segment is more likely to occur in the
top flange than in the bottom flange. Therefore, the presence of a composite floor slab can
be expected to contribute substantally to maintaining the stability of the beam. Similar
observations were made in tests by Engelhardt and Popov.
0 These experiments have also shown that a large force can develop in the lateral bracing at the
link end. The current code provisions for the design of lateral support are adequate only if
the stability of the beam segment is maintained.
0 For the purpose of estimating the link rotational demand, experimental results suggest that
link length e is best defined using member center-to-center lines dimensions.
design) of another 50-storey office building in Taiwan. In both specimens, the beam is a
420 x 140 x 8 x 10mm wide flange section of A36 material and the braces are 200 x 100 x
10 x 16 mm wide flange sections of A572 Grade 50. All columns are 340 x 340 x 16 x 16 mm
box shapes of A572 Grade 50. The link length in both specimens is equal to 1240mm or
2.58MP/V,, computed based on the actual strength of the tensile coupons. Therefore, the link
length in both Specimens GE2 and GE3 are in the transition range near the limiting length for
moment links as discussed previously. Specimen GE3 is identical with Specimen GE2, except
in the arrangement of the link web stiffeners. In Specimen GE3, the locations of the link web
stiffener adhere to the latest seismic provisions' noted earlier. In order to study the effects of
the link web stiffeners on the overall performance of the EBF, the code prescribed link web
stiffener requirements for the intermediate long link are intentionally violated for Specimen GE2.
Three equally spaced link web stiffeners are provided between the link ends in Specimen GE2.
In order to study the elastic and inelastic responses of the model EBF structures, various
ground acceleration records, including an artificial earthquake simulated from the design
response spectrum for the prototype building, of different intensities were used for each specimen.
Here, only some important observations from the tests are presented.
5.3. Conclusions
Based on this limited experimental investigation, the following tentative conclusions can be
reached.
0 When link axial force is not significant, the effects of the different stiffener designs noted in
Specimen GE2 and GE3 on the overall behavior of EBFs tested appears insignificant.
0 When link axial force is substantial, link web and flange local buckling, which occurred at a
distance of about one flange width measured from the link ends in Specimen GE2, can be
successfully delayed by providing link web stiffeners at a distance of 1-5 times the flange width
measured from the link ends.
0 As shown in Figure 20, the overall dynamic responses of the EBF, but not the buckling of
the beam, can be accurately predicted by a general purpose nonlinear frame response analysis
computer program.
0-
0 --
-8-
7:
Scale Factor = 0.010903 r,
V
-15 I I I I -16 I I I
0 10 20 30 0 10 20 30 3
Time (sec) Time (see)
h
100
4
v
50
4
b
0
F
.+
5 -50
w
p:
-100
Artificial Earthquake
Scale Factor = 0.010903
-100 I I
1 I I I I -150 ! I I
-15 -10 -5 0 5 10 l !5 -16 -0 0 8 16
Displacement (mm) Displacement (mm)
Figure 20. Time-history responses and force-deformation relationships for Specimens GE2 and GE3
SEISMIC ECCENTRICALLY BRACED FRAMES 73
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support of the National Science Council of the Republic
of China (Grant No. NSC80-0414-P002-15B), the China Steel Structures Company Ltd., and
the Chun Yuan Steel Industry Co. Ltd. Valuable suggestions provided by Professor Egor P.
Popov of the University of California at Berkeley and Professor Michael D. Engelhardt of the
University of Texas at Austin during the study were very much appreciated. The opinions
expressed in this paper are those of the authors and do not necessarily reflect the views of the
sponsors.
REFERENCES
1. M. D. Engelhardt and E. P. Popov, On design of eccentrically braced frames, Earthquake Spectra, 5
(3), Earthquake Engineering Research Institute, 1989.
2. AISC, Seismic Provisions for Structural Steel Buildings, American Institute of Steel Constructions
Inc., Chicago, Illinois, 1992.
3. K. D. Hjelmstad and E. P. Popov, Seismic behavior of active beam links in eccentrically braced frames,
Report No. UCB/EERC-83/24, Earthquake Engineering Research Center, University of California,
Berkeley, California, 1983.
4. K. Kasai and E. P. Popov, A study of seismically resistant eccentrically braced frames, Report No.
UCB/EERC-86/01, Earthquake Engineering Research Center, University of California, Berkeley,
California, 1986.
5. J. 0. Malley and E. P. Popov, A study of seismically resistant eccentically braced frames, Report No.
UCB/EERCC-83/24, Earthquake Engineering Research Center, University of California, Berkeley,
California, 1983.
6. D. N. Manheim and E. P. Popov, On the design of eccentrically braced frames, D.Eng. Thesis,
Department of Civil Engineering, University of California, Berkeley, California, 1982.
7. M.D. Engelhardt and E. P. Popov, Behavior of long links in eccentrically braced frames, Report
No. UCB/EERC-89/01, Earthquake Engineering Research Center, University of California, Berkeley,
California, 1989.
8. J. M. Ricles and E. P. Popov, Experiments on eccentrically braced frames with composite floors,
Report No. UCB/EERC-87/06, Earthquake Engineering Research Center, University of California,
Berkeley, California, 1987.
9. C. W. Roeder and E. P. Popov, Inelastic behavior of eccentrically braced steel frames under cyclic
loadings Report No. UCB/EERC-77/18, Earthquake Engineering Research Center, University of
California, Berkeley, California, 1977.
10. A. S. Whittaker, C. M. Uang and V. V. Bertero, Earthquake simulation tests and associated studies
of a O.3-scale model of a six-story eccentrically braced steel structure, Report No. UCB/EERC-87/02,
Earthquake Engineering Research Center, University of California, Berkeley, California, 1987.
11. K. C. Tsai and Y. F. Yang, Experimentalstudies of large EBF subassemblageswith link-to-box column
connections, Report No.CEER/NTU R80-6, Center for Earthquake Engineering Research, National
Taiwan University, 1991 (in Chinese).
12. K.C. Tsai and J. L.Lin, Pseudodynamic tests of eccentrically braced frames, Report No.CEER/NTU
R80-7, Center for Earthquake Engineering Research, National Taiwan University, 1991 (in Chinese).
74 K. C. TSAI, Y. F. YANG AND J. L. LIN
13. AISC, Manual of Steel Construction, Load and Resistance Factor Design, American Institute of Steel
Constructions Inc., Chicago, Illinois, 1986.
14. K. C. Tsai and E. P. Popov, Beam-column joints in seismic moment resisting frames, Report No.
UCB/EERC-88/19, Earthquake Engineering Research Center, University of California, Berkeley,
California, 1988.
15. K. C. Tsai, K. C. Lin and M.C. Liu, Seismic behavior of steel beam-to-box column connections,
Proceedings of the Tenth World Conference on Earthquake Engineering, 1992.
16. K.C. Tsai, I. C. Tsai and C. P. Chiou, Tentative seismic design provisions and commentary for building
structures, A Report to the Building Research Institute, Ministry of Interior of the Republic of China,
Report No. CSSE81-01, Chinese Society of Structural Engineering, September 1992 (in Chinese).