SHEAR LINKS IN ECCENTRICALLY BRACED FRAMES

By James O . Malley 1 and Egor P. Popov, 2 F. ASCE

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT: Eccentrically braced steel framing in seismic applications can pro-

vide high elastic stiffness and large inelastic energy dissipation capacity. The
performance of this framing system depends to a great extent on the behavior
of short active link sections of the beams. The results of an experimental in-
vestigation of the effects of inelastic loading history, connection details, and
web stiffener details on active link behavior are presented. The test results are
then evaluated using energy dissipation as the basic parameter. A design pro-
cedure for active links which yield primarily in shear is then outlined. This
procedure includes recommendations on the determination of structural con-
figuration, member sizes, link connection details, and web stiffener details.
Suggested connection and stiffener details are illustrated.

INTRODUCTION

Seismic design of structures must satisfy two basic criteria. A structure

must have sufficient stiffness to keep deflections below the limit of non-
structural damage during minor seismic events, and possess sufficient
ductility to prevent collapse in the event of a rare overload which might
occur during a major earthquake.
In the past, moment resisting and concentrically braced framing sys-
tems have been the dominant structural systems employed in the se-
ismic design of steel structures. But, neither of these systems efficiently
satisfies both of the seismic design requirements. Moment resisting
framing systems typically have sufficient energy dissipation capacity to
provide the required ductility, but may require large member sizes and
costly panel zone doubler plates to meet the drift requirement. In con-
trast, concentrically braced framing systems efficiently meet deflection
limitations through truss action, but do not provide a stable mechanism
for energy dissipation.
Recently, the concept of eccentrically braced steel framing has been
advanced as an alternative to the two conventional framing systems (15).
Examples of four types of eccentrically braced systems are shown in Fig.
1. In eccentrically braced frames, the axial forces in the diagonal braces
are transferred to columns or to other braces by bending and shear in
a portion of the beam called the active link. With proper choice of the
structural geometry, an eccentrically braced frame can have an elastic
stiffness near that of a similar frame with concentric braces (6). On the
other hand, during severe overloads the active links deform inelastically,
dissipating energy in a manner analogous to that of a moment resisting
frame. Past experimental research on one third scale frame subassem-
blages demonstrated the excellent hysteretic behavior that can be achieved
by providing eccentric bracing (10,15). With proper design, then, the
R e s e a r c h Asst., Univ. of California, Berkeley, Calif. 94720.
2
Prof. of Civ. Engrg., Davis Hall, Univ. of California, Berkeley, Calif. 94720.
Note.—Discussion o p e n until February 1, 1985. To extend the closing date one
m o n t h , a written request m u s t be filed w i t h t h e ASCE M a n a g e r of Technical a n d
Professional Publications. The manuscript for this p a p e r w a s submitted for re-
view a n d possible publication on June 7, 1983. This p a p e r is p a r t of t h e Journal
of Structural Engineering, Vol. 110, N o . 9, September, 1984. ©ASCE, ISSN 0733-
9445/84/0009-2275/$01.00. Paper N o . 19159.

2275

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

(b) (c) (d)

FIG. 1,—Alternative Arrangements of Eccentric Bracing (5)

eccentrically braced framing system can efficiently meet both the drift
and ductility requirements of seismic design.
The ability of active links to dissipate large amounts of energy during
extreme overloads is of critical importance to the overall ductility of any
eccentrically braced structural system. With this impetus, a recent ex-
perimental investigation isolated the link beam from the remainder of
the framing system (7). The results of this investigation led to the fol-
lowing conclusions: (1) Active links which yield primarily in shear (shear
links) are more effective energy dissipators than those which yield pri-
marily in bending (moment links); (2) web buckling causes both load
carrying capacity and energy dissipation capability to drop significantly;
(3) substantial strain hardening is possible in shear links; and (4) equal
spacing of shear link web stiffeners is optimum.
The results of previous investigations provided a good deal of infor-
mation on the cyclic behavior of active links. However, some aspects
critical to the economical design of an eccentrically braced framing sys-
tem have yet to be addressed, including the sensitivity of link behavior
to: the imposed loading history, the link-column connection detail, and
the web stiffener details. The purpose of the investigation reported herein
was to examine these features of active link behavior.
This paper includes the results of twelve tests performed on full size
active link sections. The conclusions drawn from these and previous tests
lead to a set of recommendations for active link, design. Suggested typ-
ical details are also included as an aid to the design process.

EXPERIMENTAL PROGRAM

Experimental Model.—Since this investigation dealt specifically with

the behavior of active links, the experimental model isolated the link
from the remainder of the structure. The model employed, shown in
Fig. 2(b), was identical to that of the previous study (7). The large welded
end plates served to restrain warping at the ends of the link, while the
welded flanges fixed the link against end rotation. The loading model
consisted of constant shear force, reverse curvature bending with equal
end moments, and zero axial force. These conditions are intended to
portray the state of links in the collapse mechanism of the prototype
structure. The actual loading conditions in the prototype structure may
2276

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

(c)

FIG. 2.—Active Link Model Extracted from Two Possible Prototype Configura-
tions (6)

differ, therefore care should be exercised in extrapolating the results pre-

sented here to such cases.
The testing apparatus, shown in Fig. 3, was designed to load the spec-
imens with the conditions stated previously. The loading ram trans-
ferred the loads to the specimens through a rigid L-shaped member. The
end plates were secured to the testing apparatus and the concrete re-
action block to prevent end rotations. The frictional resistance of the Tef-
lon coated supports and the axial force imposed by the system stabiliz-
ing sidearm were determined to be negligible.
Instrumentation of each specimen included strain gages on the flanges
and web. Specimen displacements were measured by a series of linear
potentiometers and linear variable differential transformers. A calibrated
loading cell monitored the applied shear force. The data were recorded
at specified intervals throughout the tests and stored in digital form for
reduction.
Design of Test Specimens.—All specimens in this sequence of tests
were 18 in. (457 mm) deep, 36 in. (914 mm) long rolled wide flange
sections of ASTM A36 steel. The actual yield stress of the material was
near 48 ksi (331 MPa), while the ultimate strength was approximately
68 ksi (469 MPa). All connection details were designed to meet AISC
specifications (1). Table 1 groups the specimens according to the major
2277

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

-LOADING RAM

- 3 5 0 k LOAD CELL

SIDE ARM

Jin=cp=WL
125 k
LOAD CELL

SUPPORTS

CONCRETE
REACTION
BLOCK SPECIMEN

FIG. 3.—Experimental Setup (6)

feature studied, and lists some of their important properties. The stiff-
ener details and end connections employed in these tests are shown in
Figs. 4 and 5, respectively, and are identified in Table 1 by reference to
these figures.
In the previous tests (7), the loading history consisted of quasi-stati-
cally applied cycles of relative end displacements incrementally increas-
ing at 1/2 in. (13 mm) intervals. Two cycles at each displacement level
were performed until specimen failure. Since structural response is path
dependent in the inelastic range, the sensitivity of link response to four
significantly different loading histories was investigated, as shown in
Table 1. Further, since in the previous tests the stiffeners were fully welded
on both sides of the link web, more cost effective details with stiffeners
attached only on one side of the web were tested. The shear link energy
dissipation mechanism and the web buckling phenomenon coupled to
motivate the testing of a series of four specimens with conventional mo-
ment resisting connections (13). For simplicity of specimen fabrication,
the other specimens employed all around fillet welding.
Test Results.—Specimens 16, 18, 20, and 24 had loading programs
which differed from the incrementally increasing displacements of pre-
viously reported tests. Specimen 16 initially was subjected to two large
displacement pulses, which induced buckling before subsequent incre-
£dia

J. Struct. Eng., 1984, 110(9): 2275-2295

 TABLE 1.—Details of Test Specimens
Speci-
men Num- Stif- Con-
Test num- ber of fener nection
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

feature ber Section panels Loading history detail detail

(1) (2) (3) (4) (5) (6) (7)
Loading 16 W 18 x 60 1 Large initial displ. 5(a)
history W 18 x 60 1 Nine 1 in. (25 mm) 5(a)
18 cycles
W18 x 40 3 One cycle at each 4(«) 5(a)
20 level
W 18 x 40 3 Monotonic 4(a) 5(«)
24
Stiffener 17 W 18 x 40 3 Two cycles at each 4(a) 5(a)
details 21 W 18 x 40 3 displ. level, in- 4(6) 5(a)
26 W 18 x 40 4 creasing in 1/2 4(c) 5(b)
27 W 18 x 40 4 in. (13 mm) 4(c) 5(b)
increments
Connection 22 W 18 X 40 3 Two cycles at each 4(a) 5(c)
details 23 W 18 x 40 3 displ. level, in- 4(a) 5(b)
25 W 18 X 40 3 creasing in (13 4(a) 5(d)
28 W 18 x 40 4 mm) 1/2 in. 4(c) 5(c)
increments

mental cycling. Nine 1 in. (25 mm) cycles were imposed on Specimen
18 before applying the incrementally increasing loading pattern. In these
two unstiffened specimens, the links experienced early buckling which
caused reductions of both load carrying and energy dissipation capacity.
Both specimens withstood extensive cycling after web buckling, a de-
sirable characteristic of links with large unstiffened panels. Specimen 20
was subjected to only one cycle at each displacement level. This less
severe loading program, coupled with multiple panel buckling, allowed
this specimen to dissipate a tremendous amount of energy. The classical
monotonic loading history was imposed on Specimen 24. This specimen
was laterally displaced 7.2 in. (183 mm) before reaching the limit of the
testing apparatus. At maximum displacement, the load resisted by this
specimen had only dropped 6% from the peak load, demonstrating the
large ductility of shear links.

(a) (b) (c)

FIG. 4.--Stiffener Details Employed in Tests

2279

J. Struct. Eng., 1984, 110(9): 2275-2295

 Pr/T
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

it
it

(a) (b) (O (d)

FIG. 5.—Connection Details Utilized In Tests

Specimen 17, the first stiffened specimen in this series of experiments,

had two fully welded web stiffeners located on only one side of the web.
The hysteretic curves of this test were virtually identical to those of a
previous specimen with two-sided stiffeners (7). Because of this result,
all stiffened specimens in this series of tests employed one-sided web
stiffeners. The stiffeners of Specimen 21 were only attached to one flange,
a detail that reduces fabrication costs. The prebuckling response of Spec-
imens 17 and 21 were quite similar, but after buckling had occurred the
energy dissipation capacity decreased more rapidly in the link with the
stiffeners welded to the web and one flange. Specimens 26 and 27 were
tested with stiffeners which were only welded to the beam web. Both
of these specimens dissipated a great deal of energy prior to buckling,
and experienced strain hardening levels well above those of specimens
with larger panels. The peak load resisted by Specimen 27 was over 80%
above the initial yield level. However, neither of these two specimens
demonstrated extensive postbuckling resistance. Specimen 26 (Fig. 6) failed
one cycle beyond the onset of stiffener buckling, and a defect in a flange
weld caused the failure of Specimen 27 before noticeable web displace-
ments ever occurred.
Specimens 22, 23, 25, and 28 investigated the response of shear links
with typical moment resisting connections. An all-welded connection,
with full penetration flange welds and a fillet web weld to a shear tab
was employed in Specimen 23. This Specimen performed in a manner
quite similar to that of Specimen 17. The excellent postbuckled strength
of this specimen demonstrated the integrity of this fully welded con-
nection. Specimen 25 was designed to investigate the effects of con-
necting a shear link to the web of a column. Full penetration flange welds
to the gusset plates which were fillet welded to the column section formed
the connection of this specimen. Specimen 25 underwent extensive cy-
cling after the initiation of buckling, but the dissipation capacity dete-
riorated more rapidly than similar specimens connected to column flanges.
In previous tests on beam connections to column webs (5,12), stress con-
centrations caused flange weld failures. No such failure occurred in

2280

J. Struct. Eng., 1984, 110(9): 2275-2295

 (a) r
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

DISPLACEMENT (IN)

(b)

FIG. 6.—Force-Displacement Hysteretic Loops and Photo of Specimen 26 at End

of Testing

Specimen 25, probably due to the reduced flange forces which shear
links must resist. Full penetration flange welds and a web bolted to a
shear tab formed the connection for Specimens 22 and 28. Specimen 22
exhibited prebuckling response similar to that of tests with all welded
specimens, but experienced more rapid loss in capacity after buckling
occurred. Specimen 28 (Fig. 7) never buckled, but failed earlier than any
previous test. Examination of the hysteretic curves and failed specimen
photos for Specimens 26 and 28, shown in Figs. 6 and 7 respectively,
demonstrates the superior dissipation capacity of active links with fully
welded end connections. In both specimens with bolted web connec-
tions, bolt slip transferred large shear forces to the flange welds, re-
sulting in sudden flange failures. Therefore, the all-welded connections
appear to be preferable for such applications.
A more complete review of the test results which includes the hys-
teretic curves for each specimen can be found in Ref. 9.
Test Conclusions.—The ductile behavior of eccentrically braced frames
during severe seismic excitations relies on the ability of the active links
to dissipate energy. For this reason, the experimentally determined en-
ergy dissipation capacities of links become the primary basis for quan-
titative comparisons. When combined with qualitative observations, en-
ergy dissipation comparisons can lead to design recommendations.
A normalization process is necessary to compare the energy dissipa-
tion of test links with varying size, material properties, etc. Fig. 8 shows
plots of the normalized energy dissipation, X, versus the normalized
maximum or cumulative ductility, \i or 2(i. These normalized values are
defined as:
2281

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

DISPLACEMENT (IN)

FIG. 7.—Force-Displacement Hysteretic Loops and Photo of Specimen 28 at End

of Testing

energy dissipated by test specimen

X= (1)
energy dissipated by equivalent elasto-plastic system

(2)

X(A = summation of JJL'S for all cycles. (3)

in which vmax = the maximum relative link end displacement, measured
from the most recent zero load, and vy = the relative end displacement
at initial yield.
Table 2 provides a further comparison of the eight specimens which
were subjected to the same loading history. This table lists the values
of EJQT , the total energy dissipated before failure; Ef, the energy dis-
sipated before buckling; ETOT — E | , the energy dissipated after buckling;
^max> the maximum normalized energy dissipation; (jimax, the maximum
displacement ductility; and %\x, the cumulative ductility at failure.
Evaluation of the information presented in Fig. 8 and Table 2 leads to
the following conclusions:

1. For the type of loading used in these experiments, properly de-

signed and detailed shear links can dissipate large amounts of energy
regardless of loading history. Monotonic displacements of up to 20% of
the link length can be resisted without significant loss of capacity.
2. One-sided web stiffeners are structurally equivalent to two-sided
stiffeners for beams of moderate depth, such as those used in these tests.

J. Struct. Eng., 1984, 110(9): 2275-2295

 X
(a) 27
1
' x

X'
/J~?- ^17
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

0.9
y \2,
08 /ft
- ''//
07
-7, 1 i I ,

(c)

\24

400 600 -if-

FIG. 8.—Plots of Normalized Energy Dissipation: X versus Ductility, |x, for Test
Specimens

TABLE 2.—Energy Dissipation of Test Specimens

Specimen £TOT I £|, ETOT EJ. <
number kip-inches kip-inches kip-inches "•max M'triax 2(x
(1) (2) (3) (4) (5) (6) (7)
17 6,870 2,910 3,960 0.993 46.1 525
21 • 8,590 4,010 4,580 0.991 58.5 751
22 7,040 4,150 2,890 0.977 51.0 576
23 7,070 4,100 2,465 0.991 46.1 525
25 7,890 2,980 4,920 0.934 58.5 693
26 10,800 7,300 3,500 1.115 66.0 700
27 7,520 7,520 0 1.140 46.1 525
28 2,550 2,550 0 0.910 31.9 237
Note: 1 in. lb = 0.113 Nm.

2283

J. Struct. Eng., 1984, 110(9): 2275-2295

 Stiffeners which are not welded to both of the link flanges do not impair
the prebuckled response of shear links, although the postbuckled dis-
sipation capacity of the links is reduced.
3. An all-welded link-column connection develops the full capacity of
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

active links. Bolted web, welded flange connections may be inadequate

for links with large ductility demands.
4. Based on one experiment, it appears that the connection of a shear
link to the column web does not significantly alter the performance of
the link.
5. Web stiffeners are essential for shear links to exhibit satisfactory
behavior. Inadequate web stiffening leads to early buckling and subse-
quent loss of capacity.

DESIGN PROCEDURE FOR SHEAR LINKS IN ECCENTRICALLY BRACED

FRAMES

The results of completed research on eccentrically braced steel frames

together with common design practice leads to a procedure for the de-
sign of shear links. The basic procedure includes the following steps: (1)
Determination of structural configuration; (2) determination of member
sizes; (3) design of link connections; and (4) design and detailing of web
stiffeners.
Determination of Structural Configuration.—First, the designer must
choose a suitable structural configuration. The flexibility of brace loca-
tions inherent to the eccentrically braced framing system provides the
engineer with a choice from a number of arrangements which can meet
both the architectural and structural requirements. It is advantageous to
locate the active links such that they are not required to axially transfer
large seismic forces to the braces. A judicious location of the links, or
the use of parallel gathering beams can minimize this problem. The ef-
fect of such forces on active link behavior requires further investigation.
After the brace locations have been specified, the appropriate eccen-
tricities must be selected. This is a critical step in the efficient design of
an eccentrically braced frame since the amount of eccentricity provided
determines both the elastic stiffness of the frame and the ductility de-
mand on the active links. It is at this point in the design process that
the engineer can satisfy both the drift and the ductility requirements of
seismic design.
Drift control requirements are satisfied by providing the structure with
sufficient elastic stiffness. As the length of the active links increase, the
elastic stiffness of the frame decreases, reaching the stiffness of a similar
moment resisting frame in the limiting case. An earlier investigation (6)
demonstrated that for a simple structural configuration, as in Fig. 9(a),
there is little increase in elastic stiffness when the ratio of e/L > 0.5. As
the e/L ratio approaches zero, the frame stiffness increases toward that
of a concentrically braded frame. Conventional elastic analysis proce-
dures which include the effects of shear deformation in the links can be
used to determine the maximum eccentricity which provides sufficient
stiffness to meet drift requirements.
Structural ductility requirements of seismic design dictates that the
structural configuration limit member ductility demands below the avail-
OOO/I

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)
FIG. 9.—Simple Eccentrically Braced Frame and its Collapse Mechanism (5)

able supply. Earlier research using rigid-plastic analysis methods dem-

onstrated that the active link ductility demands in eccentrically braced
framing systems are reduced as the eccentricity of the braces is in-
creased. The collapse mechanism for a simple eccentrically braced frame,
shown in Fig. 9(b), illustrates this point (6). The collapse mechanism of
this system leads to an approximate geometrical relationship between
the structure and active link deformations:
0L = ye (4)
The structural deformation measure, 8, can be interpreted as an ultimate
story drift index, The angular deformation requirement of the active link,
7, measures the average shear strain for shear links, while for bending
links 7 is the plastic hinge rotations.
Minimum 6 values of 0.015-0.020 have been recommended by ATC
03-06 (17). The results of previous experiments and those described in
this paper indicate that properly designed shear links can attain 7 values
of up to ±0.10 for cyclic loading and 0.20 for monotonic loading. Once
the values of L, 9, and 7 have been specified, the minimum eccentricity,
e, can be estimated using Eq. 4. Relationships similar to that of Eq. 4
can be derived for other eccentrically braced systems.
Determination of the maximum eccentricity to meet elastic stiffness
(drift) requirements, and the minimum eccentricity to limit the active
link inelastic demand form bounds for the length of the shear links. A
suitable eccentricity within these bounds must then be selected.
Determination of Member Sizes.—Because of the importance of pro-
viding the desired energy dissipation mechanism during extreme over-
loads, preliminary member sizing should be performed using plastic
analysis techniques. The system of lateral forces employed in this initial
sizing can be determined by applying a load factor to the equivalent
static lateral forces specified in a building code (17,18). Several proce-
dures have been suggested for determining the member forces and mo-
ments for the desired collapse mechanism (10,15).

2285

J. Struct. Eng., 1984, 110(9): 2275-2295

 Energy dissipation data from previous tests indicated that shear hinge
behavior is preferred over moment hinge action (7). A shear-moment
interaction diagram for a wide flange beam (2,7,10), shown in Fig. 10,
can be combined with consideration of static equilibrium to ensure shear
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

rather than moment link behavior. The M*, V*, and Mp terms for this
figure are given by the following equations (7):
M*= uy{d - tf)(bf -t„)t, (5)
V*=ry(d-tf)ta (6)
Mv = vyZ (7)
In these equations the section properties d, tf, tw,bf and Z are the beam
depth, flange thickness, web thickness, flange width, and plastic section
modulus, respectively, such as given in the AISC Manual (1). The terms
ffy and Ty are the yield stresses in pure tension and pure shear, respec-
tively. Using the von Mises yield criterion, Ty = a y / V 3 .
Satisfying statics of the link model results in the following relation:
2M
e= (8)
K
V '
The existence of a rapid change in slope in the shear-moment inter-
action diagram at (M*,V*) is characteristic of wide flange sections. The
balance point locates where the entire web yields in shear while the flanges
yield simultaneously in uniaxial tension or compression. Using this point
on the shear-moment interaction diagram, the balanced length for a sec-
tion, e*, can be defined as:

BALANCE POINT

FIQ. 10.—Typical Moment-Shear Interaction Diagram for Wide Fiamge Sections

2286

J. Struct. Eng., 1984, 110(9): 2275-2295

 M* _ t,
e*-~-2^~2*v3bfr (9)

Active links should therefore act as shear links if e < e*. The test results
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

to date indicate that shear link action continues to predominate for lengths
up to approximately e = 1.15e*. Combining this experimental result with
Eq. 9 results in the following simplified expression for the maximum
length of a shear link:

emax « 4 6 ; - ^ (10)

It should be noted that Eq. 10 was developed from experimental data

which simulated links with reverse curvature and equal end moments
such as shown in Fig. 2. For links illustrated in Figs. 1(a), (b), and (d),
in the elastic range of behavior larger link moments tend to develop at
the column ends than at the brace ends. The extent to which these mo-
ments equalize under severe cyclic loading is the subject of a current
investigation. If an elastic analysis indicates that M* will be reached at
the column face before the initiation of shear yielding, either the eccen-
tricity should be reduced, or an alternative section should be selected.
Since significant strain hardening of shear links has been observed in
experiments, a safety factor should be used in the design of braces to
preclude the occurrence of brace buckling. The previous recommenda-
tion that the braces should be designed for 1.5 times the link yield shear
appears to be reasonable (15).
Design of Link Connections.—The structural configuration of ec-
centrically braced frames may be such that the active links are located
in the portion of the beam adjacent to the supporting column. In this
instance the integrity of the beam-column connection becomes critical
for developing the dissipation capacity of the active links. Without suf-
ficiently ductile connection details the active links may not be able to
withstand the large displacement requirements which they could be sub-
jected to during extreme seismic events.
The test results described previously showed that all-welded link-col-
umn connections develop the full capacity of shear links. As noted ear-
lier, in bolted web, welded flange connections the large shear forces can
cause bolt slippage which can lead to premature brittle flange failures.
To avoid such failures, all-welded shear link connections which can
achieve large energy dissipation and ductility should be employed. A
detail such as that shown in Fig. 11 was found to sustain a large number
of load reversals in the inelastic range. An alternative detail which pro-
vides for a more direct transfer of shear forces is shown in Fig. 12. Fab-
ricators indicate that this is a costlier detail. If an analysis shows that
the ductility and energy dissipation demands on an active link are large,
the bolted web, welded flange, connection shown in Fig. 13 should not
be used. Similar details can be adapted for connections of active links
to column webs. But, previous research showed such connections to be
inferior to those to column flanges (12). Lehigh University investigators
recommend (5) extending the connecting plates beyond the column
flanges, as shown in Fig. 14, to decrease the possibility of weld failure.

2287

J. Struct. Eng., 1984, 110(9): 2275-2295

 BRACE ON THIS LINE
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

„ /BOTH SIDES
P^TNQFGUSSET

FIG. 11 .—All-Welded Link-Column Flange Connection with Fillet Welded Web

Showing Stlffener Spacing

Web strain hardening must be considered in the design of the shear

connections.
Since active links are susceptible to lateral torsional buckling, the link
ends at the eccentric braces must be laterally supported. A brace-beam
gusset connection which is T-shaped in plan, shown in Figs. 11, 12, 13,
and 14, reduces the load to the lateral bracing beam (10).
Design and Detailing of Web Stiffeners.—Shear link web buckling
leads to a loss in both load carrying capacity and energy dissipation ca-
pability. Web buckling can be delayed and controlled by the addition of
web stiffeners. Stiffener design can be separated into three stages: spac-
ing, sizing, and detailing.
Stiffener Spacing.—An earlier paper presented a set of relevant em-
pirical equations for determining the appropriate stiffener spacing based

BRACE ON THISLINE

/ BOTH SIDES
V \OFGUSSET

FIG. 12.—All-Welded Link-Column Flange Connection with Full Penetration Web

Weld Showing Suggested Link-Brace Connection

2288

J. Struct. Eng., 1984, 110(9): 2275-2295

 BRACE ON THIS LINE
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

_y BOTH SIDES
y \OFGUSSET

J
EQ. EQ. EQ. EQ.

-v
FIG. 13.—Bolted Web, Welded Flange Link-Column Flange Connection Showing
Stiffener Spacing

on the amount of energy dissipation required of a shear link (7). The

two basic equations are:
a t?
— = 90 - 9 In — (11)
t F
a E*
— = 94 - 14 In — (12)
t E
The topological parameters in these equations are a, the minimum panel
dimension, and tw, the web thickness. The energy dissipation param-
eters are £ | , the total energy dissipated prior to buckling; Ee, the elastic
energy stored by a link at yield; and E *, the energy absorbed during the
largest prebuckling cycle in an experiment.
If the link energy dissipation requirements are unknown, two ap-
proximate formulas can be used (7). Assuming an elasto-perfectly plastic

BRACEON THIS LINE

BOTH SIDES
y \OFGUSSET

FIG. 14.—Suggested Link-Column Web Connection Showing Stiffener Spacing

2289

J. Struct. Eng., 1984, 110(9): 2275-2295

 material undergoing predominantly shear deformations, these relation-
ships are:

:
<13>
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

F = 2l>.-1)

— = 2|i - 1 (14)

in which (x, = the ductility demand in excursion i, and (JL is as defined

in Eq. 2.
The a/t w ratio chosen should be equal to or less than the smaller value
found from Eqs. 11 and 12. Stiffener spacing smaller than 20tw is not
recommended, since a closer spacing of stiffeners seems to indicate an
excessive energy dissipation requirement (7).
For shear links which are adjacent to columns, the stiffeners should
be located to decrease the likelihood of buckling in the panel zone con-
tiguous to the link-column connection. Therefore, for the detail with the
web fillet welded to a gusset plate (Fig. 11), the stiffeners can be spaced
from the erection bolt line. For bolted or full penetration welded web
connections it appears preferable to space the stiffeners from the face of
the column (Figs. 12 and 13).
Sizing of Web Stiffeners.—Web stiffeners in shear links and plate gir-
ders are subject to similar design requirements. First, the stiffeners must
have enough axial strength to develop tension field action. Second, the
bending rigidity of the stiffeners must be sufficient to prevent buckling
of the entire link web as a single panel.
The AISC Specifications provide design equations for plate girder web
stiffeners based on the work of Basler (3). These equations do not apply
to shear links since they are based on elastic buckling solutions. Because
of the inelastic nature of shear link web buckling, an exact solution of
the stiffener design problem would be extremely complex and imprac-
tical for applications. Therefore, an approximate procedure has been de-
vised for sizing shear link web stiffeners.
In order to design the web stiffeners for axial forces, the tension field
theory used by Basler is modified using a formulation similar to that
employed by Adams, Krentz, and Kulak (2). As shown in Fig. 15, the
tension field action is analogous to that of a Pratt truss with the web
acting as the diagonal member. Using the notation of this figure, the
shear, V,, due to this tension field action is:
Vt = <yttw(h cos 6 - a sin 6) sin 9 (15)
Since the tension field will orient itself in the most efficient manner,
one determines the maximum Vt, which is attained when
1/2

2h
sine = (16)

From vertical force equilibrium of the free body shown in Fig. 15(b), the
2280

J. Struct. Eng., 1984, 110(9): 2275-2295

Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

V2

(b)
FIG. 15.—-Free Body Diagram For Determining Stiffener Forces

stiffener axial force, Fs, can be expressed as:

Fs = atta sin 2 6 (17)
Because of the extensive strain hardening shown by well stiffened ex-
perimental links, it can be reasonably assumed that cr, is approximately
equal to the ultimate stress, cru, of the steel. Using this conservative
approach, the stiffener axial force can be found from the following
expression:
a
h
1 (18)

- ^
Shear link tests indicate that stiffener yielding does not impair their
behavior, and that local stiffener buckling generally does not occur.
Therefore, allowing yielding of the stiffeners and assuming web partic-
ipation equal to half the flange width, the required area, Ast, for a two-
sided stiffener becomes:
F t
A$t = bf—. (19)

For one-sided stiffeners, by modifying the previous equation to account

approximately for eccentricity (3), the required area, Ast, is:

A'st = 2.4 (20)

2291

J. Struct. Eng., 1984, 110(9): 2275-2295

 In these equations, F s = the stiffener force given by Eq. 18, ay = the
yield stress of the stiffener material, bf = the flange width, and tw = the
web thickness. Even though over twice as much stiffener material is re-
quired, one-sided web stiffening is likely to be more economical than
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

the two-sided detail because of reduced welding costs. Web stiffeners

are usually detailed so that they do not protrude outside the longitudinal
edge of the flanges.
In addition to the axial force requirements, the web stiffeners must
have sufficient rigidity to prevent buckling of the whole link web as a
single panel. In the late phases of the experiment, Specimen 26, with
stiffeners welded to the web only, exhibited a tendency for this type of
a response, as may be discerned from the photograph in Fig. 6. This
problem was studied by a number of investigators who have presented
solutions for the required rigidity for elastic buckling (14,16). Their so-
lutions relate the critical buckling stress to the stiffener rigidity param-
eter, 7, as a function of the panel aspect ratio. This nondimensional pa-
rameter is defined as:
EI
7=— (21)
aD
In this equation, E = the elastic modulus, I = the moment of inertia of
the stiffener, a = the smallest panel dimension, and D = the plate stiff-
ness factor, D = Ef^/12(l - v2), in which v = Poisson's ratio.
Using the results of Stein and Fralich (16), Bleich (4) determined the
minimum stiffener rigidity required to obtain the largest critical shear
stress as a function of the panel aspect ratio, p, defined as the ratio of
the larger to smaller panel dimension. Using this parameter, the follow-
ing relationship was postulated (4):
7req = 28P2 - 20 (22)
This relationship is based on the assumption that the longitudinal edges
of the web are simply supported and therefore free to rotate. Rockey
and Cook (14) presented a solution assuming that the longitudinal edges
are fixed against rotation. Based on their solution, an equation similar
to that given in Eq. 2 results in the following relationship:
7req = 16p2 - 8 (23)
The previous two equations apply for 1 <. p < 5.
With some analytical and experimental justification, for general ap-
plications Massonnet and Maquoi (11) conservatively recommend that
the rigidity required at the critical shear stress be multiplied by a factor
of 4-5 to keep the stiffeners straight in the postbuckled range. Since
shear links are relatively short and are laterally braced at both ends, for
the usual rolled sections the flanges would appear to provide significant
restraint along the longitudinal edges of the web. It therefore seems rea-
sonable to make use of Eq. 23 rather than Eq. 22 in the design of shear
link web stiffeners. Moreover, since in seismic design some damage to
the links can be tolerated at extreme overloads, it seems that a multiplier
of 4 can be adopted, requiring the following stiffener rigidity for shear
links:

2292

J. Struct. Eng., 1984, 110(9): 2275-2295

 7 Kq = 64 [^ - 32 . . , (24)
Combining Eqs. 21 and 24 requires the stiffener moment of inertia, 7 req ,
to be:
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

a
3 -
'req - twlieq .^ . _ 2" (25)

Following the usual practice, for both one and two-sided stiffeners,
the moment of inertia can be taken about an axis in the plane of the
web (1). Again assuming that the stiffeners just reach the longitudinal
edge of the flanges, the required thickness of two-sided stiffeners, tst,
becomes:

t 12
^ JfJff^^)- (26)
For one-sided stiffeners, the required thickness, t'st, is:

ti = M
^2|^<r^j (27)
It is not recommended to use a stiffener thickness less than the thick-
ness of the web.
Stiffener Detailing.—In past design applications of eccentrically braced
frames, link web stiffeners have been fitted to allow fillet welding to
both beam flanges as well as to the web. The test results presented in
this paper indicate little energy dissipation capacity reduction of links
without fitted, fully welded stiffeners. Therefore, for shear links with
small ductility demand, it appears possible to relax the requirement of
fitted stiffeners, and to allow them to terminate a distance k below the
top beam flange, as shown in Figs. 11, 12, 13, and 14. The detail shown
is based on the assumption that the concrete floor can provide adequate
buckling restraint for the top flange. This may not be justified in some
applications due to the anticipated cracking of the floor slab. Under such
circumstances, the use of more conservative fitted stiffeners is advisable.
In all cases, stiffener fillet welds should be continuous on both sides of
the stiffeners.
The web stiffening requirements are usually more stringent if Eq. 24
indicates the need for two-sided stiffeners. However, fitted stiffeners may
again be avoided. For such cases it would appear reasonable to termi-
nate the stiffeners a distance k below the top flange on one side of the
web, and the same ditance above the bottom flange on the other side
of the web.

CONCLUSIONS

Previous investigations have demonstrated that properly designed ec-

centrically braced frames can have both large elastic stiffness and large
inelastic energy dissipation capacity. This experimental program was
concerned with the sensitivity of shear link behavior to design consid-
erations such as the imposed loading history, the type of link-column
connection, and the detailing and spacing of web stiffeners. The ex-
2293

J. Struct. Eng., 1984, 110(9): 2275-2295

 perimental results presented here were combined with previous studies
to develop a shear link design procedure which includes suggested de-
tails as well as a m e t h o d for stiffener design. It is to be n o t e d that the
effect of axial forces a n d / o r unequal link e n d m o m e n t s on the behavior
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

of active links has not been explored, and is the subject of current research.
Based on the completed investigation, the design of stiffened shear
links can n o w be made on a more rational basis for cases of small axial
loads and essentially equal link end m o m e n t s .

ACKNOWLEDGMENTS

The support of the National Science Foundation, Grant N o . CEE 81-

07217 is gratefully acknowledged. However, t h e opinions expressed in
this paper are those of the authors and do n o t necessarily reflect the
views of NSF. Keith D. Hjelmstad, a doctoral student in Civil Engi-
neering, offered m a n y useful suggestions in b o t h the performance of the
experimental work and the preparation of this paper. Wes Neighbour
assisted with the experiments. Their contributions to this research are
greatly appreciated.

APPENDIX.—REFERENCES

1. AISC, "Specification for the Design, Fabrication and Erection of Structural

Steel Buildings, with Commentary," Manual of Steel Construction, American
Institute of Steel Construction, Chicago, 111., 1980.
2. Adams, P. F., Krentz, H. A., and Kulak, G. L., "Limit State Design in Struc-
tural Steel," Canadian Institute of Steel Construction, 1979.
3. Basler, K., "Strength of Plate Girders in Shear," Journal of the Structural Di-
vision, ASCE, Oct., 1961, pp. 151-180.
4. Bleich, F., "Buckling Strength of Metal Structures," McGraw-Hill, New York,
N.Y., 1952.
5. Driscoll, G. C , and Beedle, L. S., "Suggestions for Avoiding Beam-to-Col-
umn Web Connection Failure," AISC Engineering Journal, Vol. 19, No. 1, 1982.
6. Hjelmstad, K. D., and Popov, E. P., "Characteristics of Eccentrically Braced
Frames," Journal of Structural Engineering, ASCE, Vol. 110, No. 2, Feb., 1984,
pp. 340-353.
7. Hjelmstad, K. D., and Popov, E. P., "Cyclic Behavior and Design of Link
Beams," Journal of Structural Engineering, ASCE, Vol. 109, No. 10, Oct., 1983,
pp. 2387-2403.
8. Kasai, K., "A Plastic Design Method for Eccentrically Braced Frames," De-
partment of Civil Engineering, University of California, Berkeley, Calif., CE
299 Report, 1983.
9. Malley, J. O., and Popov, E. P., "Design Considerations for Shear Links in
Eccentrically Braced Frames," EERC Report 83-24, University of California,
Berkeley, Calif., Nov., 1983.
10. Manheim, D. N., "On the Design of Eccentrically Braced Frames," D. Eng.
Thesis, Department of Civil Engineering, University of California, Berkeley,
Calif., Feb., 1982.
11. Massonnet, Ch., and Maquoi, R., "Recent Progress in the Field of Structural
Stability of Steel Structures," I.A.B.S.E. Surveys, S.-5/78, May, 1978.
12. Popov, E. P., and Pinkney, R. B., "Cyclic Yield Reversal in Steel Building
Connections," Journal of the Structural Division, ASCE, Vol. 95, No. ST3, Pa-
per 6441, Mar., 1969, pp. 327-352.
13. Popov, E. P., and Stephen, R. M., "Cyclic Loading of Full-Size Steel Con-
nections," Steel Research for Construction, Bulletin No. 21, American Iron and
Steel Institute, New York, N.Y., Feb., 1972.
2294

J. Struct. Eng., 1984, 110(9): 2275-2295

 14. Rockey, K. C , and Cook, I. T., "Shear Buckling of Clamped and Simply
Supported Infinitely Long Plates Reinforced by Transverse Stiffeners," Aer-
onautical Quarterly, Vol. XIII, Feb., 1962.
15. Roeder, C. W., and Popov, E. P., "Inelastic Behavior of Eccentrically Braced
Downloaded from ascelibrary.org by Indian Institute of Technology Roorkee on 01/12/19. Copyright ASCE. For personal use only; all rights reserved.

Steel Frames Under Cyclic Loadings," EERC Report 77-18, university of Cal-
ifornia, Berkeley, Calif., Aug., 1977.
16. Stein, M., and Fralich, R. W., "Critical Shear Stress of an Infinitely Long
Simply Supported Plate with Transverse Stiffeners," N.A.C.A. Technical Note
1851, 1949.
17. "Tentative Provisions for the Development of Seismic Regulations for Build-
ings," Report No. ATC 03-06, Applied Technology Council, Palo Alto, Calif.,
1978.
18. Uniform Building Code, 1982 Ed., International Conference of Building Offi-
cials, Whitier, Calif.

2295

J. Struct. Eng., 1984, 110(9): 2275-2295