Engineering Structures 23 (2001) 1480–1490

www.elsevier.com/locate/engstruct

Steel-concrete composite coupling beams — behavior and design

Binginan Gong a, Bahram M. Shahrooz b,*

a
S&B Infrastructure, Ltd., 3535 Sage Road, Houston, TX 77056-7011, USA
b
University of Cincinnati, Department of Civil and Environmental Engineering, 741 Baldwin Hall, PO Box 210071, Cincinnati,
OH 45221-00171, USA

Received 12 September 2000; received in revised form 16 January 2001; accepted 10 April 2001

Abstract

Structural steel/composite beams provide a viable alternative for coupling individual reinforced concrete wall piers. Well-estab-
lished guidelines for shear links in eccentrically braced steel frames form the basis of current design guidelines. However, these
provisions ignore the effects of nominally reinforced concrete encasement which typically surrounds the coupling beam, and are
based on overly conservative assumed deformation demands. A coordinated analytical and experimental research program at the
University of Cincinnati has focused on cyclic response of steel/composite coupling beams, their connections to reinforced concrete
walls, and overall behavior of composite coupled wall systems. Using the results from this study, guidelines for proper design and
detailing of steel/composite coupling beams and beam-wall connections have been developed. This paper summarizes the research
program, and highlights the basic concepts, important findings, and recommendations.  2001 Elsevier Science Ltd. All rights
reserved.

Keywords: Composite construction; Coupling beams; Coupled walls; Cyclic testing; Mixed construction; Seismic design

1. Introduction relatively large diameter bars which have to be

adequately confined to avoid buckling at advanced limit
An efficient structural system can be achieved if the states. Anchorage of the reinforcing bars in wall piers
openings in structural walls are arranged in a regular can pose difficulties.
pattern. In this manner, a number of individual wall piers Structural steel or steel–concrete composite beams
can be coupled together to produce a system with large provide a viable alternative, particularly for cases with
lateral stiffness and strength. The structural performance floor height restrictions. In contrast to conventionally
at or near ultimate state can also be optimized by proper reinforced concrete members, steel/composite coupling
detailing of the coupling beams (i.e., the beams that con- beams can be designed as a flexural-yielding or shear-
nect the individual walls). Coupling beams should be yielding member. Therefore, a desirable mode of energy
proportioned to avoid over coupling (i.e., a system that dissipation is achieved depending on the particular case.
acts as a single pierced wall) and light coupling (i.e., The main design issues are (a) proportioning and
a system that performs as a number of isolated walls). detailing of steel/composite coupling beams, and (b)
Extensive past research [1–8] has led to well established beam–wall connections. Depending on whether the wall
seismic resistant design guidelines for reinforced con- boundary element consists of structural steel columns or
crete coupling beams. Current design provisions [9] typi- reinforced concrete elements, the coupling beam–wall
cally result into diagonally reinforced deep beams in connection is different. In the former case, the connec-
order to satisfy the stiffness, strength, and energy dissi- tion is similar to steel beam–column connections. The
pation demands. The diagonal reinforcement consists of connection in the latter case, which is the focus of this
paper, is achieved by embedding the coupling beam
inside the wall piers and interfacing it with the wall
* Corresponding author. Tel.: +1-513-556-3677; fax: +1-513-556- boundary element. The embedment length will clearly
2599. have a major influence on the performance of
E-mail address: [email protected] (B.M. Shahrooz). steel/composite coupling beams.

0141-0296/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 4 2 - 6
 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490 1481

A number of recent studies at the University of Cin-

cinnati have been focused on examining seismic
response of such composite systems. An overview of the
research program is provided herein. Current design
guidelines [10] are evaluated, and a number of changes
are recommended.

2. Summary of current design guidelines

Steel coupling beams are designed according to the

provisions outlined in the 1997 NEHRP Recommended
Provisions for the Development of Seismic Regulations
for New Buildings [10]. These provisions are based on
seismic detailing of steel links in eccentrically braced
frames. The coupling beam rotation angle is arbitrarily
set equal to 0.09 rad. Note that for link beams this angle Fig. 1. Mattock and Gaafar’s model for computing embedment
is computed. The assumed coupling beam rotation is length.
rather conservative. For example, if the beam length is
less than 1.6 Mp/Vp (Mp=plastic moment capacity;
Vp=plastic shear capacity, i.e., 0.6 Fy (d⫺2tf)tw in which
Fy=yield strength, d=beam depth, tf=flange thickness,
tw=web thickness), web stiffeners at (30 tw⫺d/5) will loading, this model calculates the required embedment
have to be provided. Considering that most coupling length (Le) according to Eq. (1) in which twall=thickness
beams are typically short, current design leads to closely of wall pier, bf=beam flange width, and b1=ratio of the
spaced web stiffeners. Steel coupling beams are often average concrete compressive strength to the
encased inside nominally reinforced concrete elements, maximum stress.
e.g., in door lintels. However, due to lack of information
about the effects of concrete encasement, current design
guidelines are tacit about cases in which the coupling

冉 冊 冤 冥
beam is encased. Accordingly, most engineers ignore the
Vu⫽4.05冑f⬘c
0.66
twall 0.58−0.22b1
influence of encasement apparently on the basis that (a) b1bfLe (1)
the encasement is nominally reinforced and hence not bf a
0.88+
expected to significantly contribute towards strength and Le
stiffness, and (b) the design will be more conservative
by not including the contribution of the encasing element
around the steel coupling beam. As a result, details for In this equation, the inflection point is assumed to be
preventing flange and web buckling are identical to those at the midspan; hence, the value of “a” is taken as one
used for unencased coupling beams, and the embedment half of the coupling beam span. The model proposed by
length is calculated to develop the design capacity of the Marcakis and Mitchel generally results in a slightly
steel beam only. longer embedment length; however, the difference
No specific guidelines are provided for computing the between the two models is negligible [13,14]. Although
required embedment length, but references are made to Marcakis and Mitchell [15] and Mattock and Gaafar [16]
previous studies [11–14] for further information. These originally developed their respective equations for
studies examined the applicability of two models pro- design of steel brackets attached to reinforced concrete
posed by Marcakis and Mitchell [15] and Mattock and columns, previous studies at the University of Cincinnati
Gaafar [16]. In Fig. 1, Mattock and Gaafar’s model is and elsewhere [11–14] have shown that these models
illustrated. The applied shear (Vu) is resisted by mobiliz- result in acceptable performance for steel coupling
ing an internal moment arm between bearing forces Cf beams which are subjected to cyclic shear. Moreover,
and Cb. A parabolic distribution of bearing stresses is the calculated embedment length from either model is
assumed for Cb, and Cf is computed by using a uniform adequate to ensure a desirable mode of energy dissi-
stress equal to 0.85 f⬘c where f⬘c=concrete compressive pation for steel, unencased coupling beams by forming
strength in MPa. The bearing stresses are assumed to be the plastic hinges in the beam rather in the connection
distributed uniformly over the beam flange width. Fol- region. Note that the value of Vu in Eq. (1) should be
lowing these assumptions and calibration against experi- taken as the plastic shear capacity of the steel beam (i.e.,
mental data for steel corbels subjected to monotonic Vp=0.6 Fy (d⫺2tf)tw) to ensure adequate performance.
1482 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490

see Fig. 3. A low-strength concrete (f⬘c=12 MPa) was

used for the encasement in order to accentuate nominal
confinement around the steel coupling beam. In all the
specimens, auxiliary transfer bars had been attached
(through the use of mechanical half couplers) to the top
and bottom flanges at two locations (Fig. 3) in an effort
to aid in the transfer of bearing forces to the surrounding
concrete [13,14]. Face bearing plates in specimens 6 and
7 consisted of a pair of 4.75 mm thick stiffeners welded
on both sides of the web. The face bearing plates were
located inside the wall boundary element transverse
reinforcement. Another pair of stiffeners were also added
at 125 mm from the face bearing plates under the auxili-
ary bars. The resulting concrete struts between these
plates (shown schematically in Fig. 3) are expected to
enhance the performance by reducing the contribution
of bearing stresses against the top and bottom flanges.
The benefits of face bearing plates have been demon-
Fig. 2. Plan view of selected prototype structure.
strated in past studies on steel beam-reinforced concrete
column connections [19], and as part of testing of speci-
men No. 6. Additional details regarding the test speci-
3. Response of steel–concrete composite coupling
mens and other aspects of the experimental program are
beams provided elsewhere [18]. The focus of this paper is on
specimens No. 1, 4, 5, and 7.
A recent study [17,18] has examined the effects of
concrete encasement. The test specimens were extracted
from a 20-story prototype building shown in Fig. 2. For 3.1. (a) Strength characteristics
the chosen span length of the steel/composite coupling
beams, current design guidelines [10] require stiffener The load-deflection responses of the unencased speci-
plates to be placed at (30tw⫺d/5) on one side of the web. men No. 1 and encased specimen No. 4 are plotted in
As part of the experimental phase of this study, which Fig. 4. The hysteresis loops are stable, and exhibit a sig-
was conducted in two phases, seven specimens were nificant level of energy dissipation. Sudden drops during
tested. The main test variables, summarized in Table 1, the last cycles are primarily due to fracture of wall trans-
were (a) presence or lack of encasement, (b) the amount verse reinforcement passing through the steel beam web,
of web stiffener in the steel beam, (c) presence or lack and fracture of the weld around the mechanical half
of face bearing plates at the wall–beam interface, (d) the couplers which were used to attach the auxiliary bars to
level of shear force for which the beam–wall connection the beam flanges [18]. Both specimens could develop
is designed, and (e) floor slab around the coupling beam. and exceed the capacity computed based on the meas-
The encasement was nominally reinforced with four 4.9 ured material properties [18]. For specimen No. 1, the
mm longitudinal bars and 4.9 mm transverse reinforce- capacity is equal to the plastic shear capacity, i.e., 0.6
ment placed at one-half depth of the encasing element, Fy(d⫺2tf)tw. The shear capacity of specimen No. 4 was

Table 1
Test specimens and test variables

Specimen I.D. Encasement Spacing of web stiffeners Force for connection design Face bearing plate Floor slab

Phase I
1 No L Vsteel section No No
2 Yes L Vsteel section No No
3 Yes 2L Vsteel section No No
4 Yes N/Aa Vsteel section No No
Phase II
5 Yes N/Aa Vcomposite section No No
6 Yes N/Aa Vcomposite section Yes No
7 Yes N/Aa Vcomposite section Yes Yes

a
No web stiffeners were provided.
 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490 1483

Fig. 3. Specimen details.

computed by (a) superposition of the shear capacities of hinges in the beam. This behavior is attributed to the
the steel section and concrete beam, or (b) a fiber cross provided embedment length.
section analysis incorporating flexure as well as shear.
The average value from these two analyses is plotted in 3.3. (c) Revised embedment length
Fig. 4.
The nominally reinforced encasement around the steel As indicated in Table 1, the required embedment
coupling beam is apparently adequate to prevent web length for specimen No. 1 and No. 4 was, according
and flange buckling at advanced yielding. Note that to current design guidelines and practice, computed to
specimen No. 4 could develop 107 kN at a shear angle develop the shear capacity of the steel coupling beam
of 0.057 rad. At this shear deformation, current NEHRP only, i.e., 0.6 Fy(d⫺2tf)tw. The experimental data shown
provisions [10] require web stiffener plates at 1.5 times in Fig. 4 suggest that nominally reinforced encasement
the spacing used for specimen No. 2. Despite having no can increase the capacity by as much as 23%. Clearly,
stiffeners, specimen No. 4 could develop shears corre- if the provided embedment length is calculated to
sponding to 1.10 Vp. Hence, nominally reinforced develop a smaller shear, plastic hinges form in the con-
encasement around steel coupling beams is sufficient to nection region before fully mobilizing the full capacity
prevent web and flange buckling; and web stiffeners are of the composite beam. This performance is not desir-
not needed. able. A capacity design methodology, in which the
embedment length is computed to develop the capacity
3.2. (b) Energy dissipation characteristics of the composite beam and not the steel beam alone, is
proposed. This method requires that the contribution of
The dissipated energies are compared in Fig. 5. In an concrete encasement towards shear capacity is taken
effort to examine the performance of the specimens, the into account.
dissipated energy was separated into two parts: (a) the In lieu of detailed techniques such as fiber based mod-
energy dissipated by plastic hinges in the beam els [18], a simple method based on combining the shear
(“Beam”), and (b) the energy dissipated by inelastic capacity of the steel beam (Vsteel) and encasement (VRC)
deformations in the connection region (“Connection”) appears to be a reasonable alternative, i.e., Vn=Vsteel+VRC
[18]. For specimen No. 1, the participation of beam in which Vsteel=0.6 Fy(d⫺2tf)tw and VRC is computed
towards energy dissipation was more significant than based on standard methods for reinforced concrete mem-
that for specimen No. 4. Although the beam capacities Avfyd
bers (e.g., [20:318–99]), i.e., VRC=0.166√f⬘cbd+
could be developed, the performance of the encased s
specimen was not as desirable because inelastic action in where b=width of the encasement, d=effective depth of
the connection region contributed more than the plastic the encasement, Av=total area of transverse reinforce-
1484 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490

Fig. 4. Hysteresis responses of encased and unencased coupling

beams.

ment, and s=spacing of the transverse steel. In this sim- Fig. 5. Distribution of dissipated energy.
plified approach, the constitutive models for concrete
and steel are idealized, and factors such as strain harden- as well as the ultimate strength, dimensions of the steel
ing of the longitudinal and transverse steel are ignored; sections, and the overall encasement dimensions were
the shear capacity of the reinforced concrete encasement changed systematically. The ratio between the two com-
is based on the ACI method in which the concrete contri- puted capacities ranges between 1.31 to 1.85 with an
butions towards shear capacity from aggregate interlock, average of 1.61 and a standard deviation of 0.075 [18].
dowel action of the longitudinal bars, and the uncracked Within the limitations of this parametric study, a “cor-
concrete are lumped together [21]; and the contribution rection factor” of 1.6 is recommended and hence the
of the concrete beyond cracking is not included [22]. shear capacity of composite coupling beams is taken as
Therefore, this simple model needs to be calibrated so Vn=1.6 (Vsteel+VRC). Additional studies covering a wider
that the computed capacity would be comparable to that range of parameters is necessary to further refine the rec-
obtained from a more detailed analysis such as fiber ommended correction factor.
modeling. A “correction factor” was determined by com-
paring the capacity computed from detailed fiber section 3.4. (d) Evaluation of revised embedment length
analysis based on the modified compression field theory
[22] and the capacity from the superposition method, i.e., In an effort to evaluate the performance of coupling
Vn=Vsteel+VRC. The parametric study included 24 cases beams for which the embedment length is calculated
in which the concrete compressive and the steel yield based on the proposed capacity design method, strength
 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490 1485

and energy dissipation characteristics of specimens No.

4 and 5 are compared. Specimen No. 5 was generally
similar to those tested in the first phase, except for a
longer embedment length that the revised procedure
would require, and the testing method which included
wall overturning moment in contrast to the first four
specimens for which the wall overturning was not simu-
lated.
Normalized load-shear angle envelope curves of
specimens No. 4 and 5 are compared in Fig. 6. The verti-
cal axis is normalized with respect to the shear at the
onset of web yielding (Vy). Shear angle at this stage (gy),
was used to normalize the measured shear angles. Prior
to yielding, the two specimens exhibit almost identical
load-deformation relationships. The response of speci-
men No. 5 shows a remarkable improvement in terms
of achieving higher ductilities and reduced strength
deterioration beyond the maximum load. For example, Fig. 7. Distribution of dissipated energy for specimen No. 5.
at shear angle of 0.0885 radians, the load had dropped
only to 96% of the peak value. Specimen No. 5 also
delayed the connection failure until a higher capacity
developed a larger capacity in the positive direction
could be developed, and hence the enhanced behavior.
when the wall overturning moment produced compress-
A similar observation is made by evaluating the
ive stresses around the connection. Under negative bend-
energy dissipation characteristics of specimen No. 5
ing, the two specimens developed rather similar loads,
shown in Fig. 7 in which the total dissipated energy
although specimen No. 5 reached a slightly lower load
(“Total”) is divided as discussed previously. The input
than specimen No. 4. This difference is attributed to the
energy was predominately dissipated by inelastic defor-
presence or lack of wall overturning moment. The wall
mations in the coupling beam. Beyond shear angle of
overturning moment in specimen No. 5 resulted in ten-
0.05 rad., the participation of the connection was gradu-
sile stresses that reduced the level of bearing stress trans-
ally increased although the plastic hinges in the beam
fer between the beam flanges and the surrounding con-
outside the connection continued to dissipate a reason-
crete in the connection region. Despite these stresses, the
able portion of the total energy. This trend is different
load carrying capacity did not drop significantly below
from that observed for specimen No. 4 (see Fig. 5). The
that for specimen No. 4 in which the connection region
longer embedment length in specimen No. 5 evidently
was kept under a constant gravity compressive stress.
enhanced the energy dissipation characteristics by reduc-
The longer embedment length in specimen No. 5, which
ing the contribution of the connection region.
is the result of using the revised design methodology,
Note that the contribution of floor slab is not included
in the proposed design method because (a) as seen in
Fig. 8 the contribution of slab is relatively negligible
because the additional tensile forces from the slab bars

Fig. 8. Moment-curvature response of composite coupling beam in

Fig. 6. Normalized load-deflection curves. prototype structure.
1486 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490

The floor slab clearly increases the initial stiffness of

specimen No. 7. However, at shear angle of about 0.005
rad., the significance of slab is effectively lost. Beyond
this small level of deformation the stiffness of specimen
No. 7 drops to a level comparable to the initial stiffness
of the specimens without slab. When specimens were
loaded up to a shear angle of about 0.06 rad., all the
three specimens had reached about the same stiffness.
Therefore, although floor slabs increase the initial stiff-
ness of coupling beams, the contribution of the slab
deteriorates under small deformations and may be
ignored in seismic design and analysis.

Fig. 9. Distribution of dissipated energy for specimen No. 7. 4. Evaluation of impact of encasement on overall
structural response
are relatively small in comparison to the tensile force in
The effects of the additional stiffness due to nominally
the flanges of a typical steel beam, and (b) the floor slab
reinforced encasement around steel coupling beams,
wraps the connection region and reduces its participation
which is ignored in current design guidelines, were
as illustrated in Fig. 9 [18]. The distribution of dissipated
evaluated analytically by examining the overall response
energy clearly indicates that the slab and beam dissipated
of the prototype structure (Fig. 2). The larger stiffness
the majority of the input energy, and the connection
obviously results in smaller vibration periods. Such a
essentially did not participate. Therefore, the contri-
shift could influence the dynamic behavior if the fre-
bution of floor slab is neglected, and the embedment
quencies coincide with the frequency band of the design
length needs to be calculated to develop only 1.6
ground motion with high input energy content.
(Vsteel+VRC).
The demands in the walls and coupling beams are
affected by the changes in the coupling beam stiffness.
3.5. (e) Stiffness of composite coupling beams
The concrete encasement increases the coupling beam
stiffness which in turn increases the level of coupling
The variation of peak-to-peak stiffness against shear
action between the individual wall piers. For instance,
angle is plotted in Fig. 10. The initial stiffness of speci-
the wall axial load in the first floor of the prototype struc-
men No. 4 is 25% larger than that for specimen No. 1.
ture increases by 40% when the influence of encasement
The initial stiffness for specimen No. 5 is less than the
is taken into account [17]. Such a large increase could
stiffness of specimen No. 4 because this specimen was
pose stability problems if the wall boundary elements
accidentally loaded due to difficulties in the computer-
are designed for forces calculated based on ignoring the
based control of one of the actuators used for loading.
encasement. In addition, the foundation system needs to
Specimen No. 5 had apparently experienced some minor
be designed for the increased demands in the walls. The
cracks before the testing program was started.
increase in wall shear force, which is about 10%, is not
perhaps as critical. The increases in beam design forces
are offset by the corresponding additional capacity due
to encasement. Therefore, design of walls and foun-
dation systems needs to incorporate the effects of
encasement around steel coupling beams. The numerical
values stated above are particular to the prototype struc-
ture, and are based on elastic analysis. For other struc-
tures with different geometries and stiffness character-
istics, the increase in the stiffness and hence the changes
in the design forces may be more or less. Moreover,
cracking of the encasement around the steel coupling
beam will reduce the stiffness of the composite coupling
beam, and hence the increased coupling action will be
less than that predicted from a simple elastic analysis.
Nevertheless, in view of the potential higher design
forces, the engineer should consider the increased stiff-
Fig. 10. Stiffness degradation of various composite coupling beam ness of composite coupling beams as one of the variables
specimens. in the design model.
 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490 1487

5. Evaluation of shear angle assumed to be distributed uniformly or triangularly, and

three ground motion records (1940 El Centro NS, 1989
Well established guidelines for shear links in eccentri- Loma Preita, and 1994 Northridge N60E) were used for
cally-braced frames [23] form the basis of current guide- the dynamic analyses. Three different analyses were
lines [10] for design and detailing of steel/composite conduced in which (a) the coupling beams were unen-
coupling beams. The expected coupling beam rotation cased steel members, (b) the influence of encasement
angle plays an important role in the required beam around the steel coupling beams was taken into account,
details such as the provision of stiffeners. Current design and (c) flexibility at the foundation level was approxi-
guidelines [10] arbitrarily set the coupling beam shear mately taken into account by placing vertical and
angle equal to 0.09 rad., and the beam is detailed accord- rotational elastic springs under the column and wall
ingly. Such a large angle results in closely spaced web elements as shown in Fig. 11(c). The spring coefficients
stiffeners. Note that the shear angle is computed for steel were computed based on basic principles by assuming a
link beams in eccentrically braced frames. modulus of sub-grade of 54,260 kN/m3.
In order to examine whether this level of shear angle The maximum computed coupling beam shear angles
(0.09 rad.) is reasonable and to understand the maximum for various analyses are summarized in Table 2. The
expected range of shear angle, the response of the proto- level of shear angle for encased and unencased coupling
type structure was evaluated. A pseudo three-dimen- beams is considerably less than 0.09 rad. Only when the
sional model of the prototype structure was constructed foundation flexibility was taken into account and lateral
(Fig. 11(a)). The torsional and vertical springs used in loads were assumed to be distributed triangularly did the
the model are intended to simulate the outrigger action coupling beam shear angle approach the value of 0.09
of the transverse members. The walls were modeled by rad. stipulated in NHERP provisions [10]. However, at
an element (Fig. 11(b)) that incorporates axial, flexural, this shear angle the roof lateral drift exceeds 10% of the
and shear deformations in the elastic and inelastic range building height, which is well above acceptable levels,
[24]. Both dynamic and static push-over analyses were and the base shear approaches 40% of the building
conduced [18]. For static analyses, the lateral loads were weight. For other cases with reasonable drifts, the com-

Fig. 11. Analytical modeling of prototype structure.

1488 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490

Table 2
Maximum coupling beam shear angle (rad.)a

Analysis Unencased fixed base Encased fixed base Encased “flexible base”

Static push over: Rectangular lateral loads 0.0177 0.0198 0.0269

(1.2%) (1.3%) (2.0%)
Static push over: Triangular lateral loads 0.0328 0.0378 0.0884
(2.7%) (3.2%) (10.2%)
Dynamic:1940 El Centro NS 0.00391 0.00355 0.00388
Dynamic:1989 Loma Prieta 0.00709 0.0075 0.00673
Dynamic:1994 Northridge N60E 0.00208 0.0188 0.00209

a
The values in the parentheses are calculated roof drifts at maximum reported shear angle.

puted shear angles are significantly less than 0.09 rad.

It is deemed that similar observations are made for other
well-proportioned buildings employing coupled walls.
Despite an effort to perform a reasonably complete
analysis of the prototype structure, major simplifications
had to be made, e.g., (a) the three-dimensional behavior
of the structure is modeled rather crudely, (b) simulation
of the soil-structure action is very approximate, and (c)
the parameters in the wall and beam hysteretic models
were established based on subassembly tests with bound-
ary conditions different from those in an actual building.
The main reason behind the reported analytical study
was to explore the rationality of the assumed coupling
beam shear angle of 0.09 rad., in current design codes.
Additional detailed studies that overcome the stated limi-
tations of the current study are necessary.
Despite the simplicity of the analytical studies used
here, the current assumed shear angle of 0.09 rad., Fig. 12. Analytical model for computing shear angle of coupling
appears to be questionable and can lead to unnecessary beams.
conservative detailing of steel coupling beams. Note that
the reported test results (Fig. 4) show that stiffeners can
be eliminated for steel–concrete composite coupling L+0.6 Le in which Le is the embedment length of the
beams; therefore, the focus of this discussion is on steel coupling beam inside each wall pier. Note that with the
coupling beams. More rational techniques for computing exception of the assumed collapse mechanism and the
the value of shear angle are needed. relationship between the shear angle and drift angle, the
The coupling beam shear angle is computed with ref- proposed method is similar to that used for steel shear
erence to the collapse mechanism shown in Fig. 12 links in eccentrically braced frames.
which corresponds to the expected behavior of coupled
wall systems, i.e., plastic hinges at the base of walls and
at the ends of coupling beams. The value of plastic 6. Summary and conclusions
interstory drift angle (qp) is taken as Cdqe (Cd=deflection
amplification factor defined by NEHRP [10]), where the Seismic behavior of steel and steel–concrete com-
elastic interstory drift angle, qe, is computed under code posite coupling beams was evaluated through a coordi-
level lateral loads (e.g., [9,10]). Knowing the value of nated experimental and analytical research study. One of
Lwall the main objectives of the reported study was to scrutin-
qp, shear angle, gp, is calculated as gP=qP in which
L ize current design guidelines, and to recommend modi-
Lwall is the distance between center lines of the wall piers fications if necessary. Based on the reported study, the
and L is the clear span of the coupling beam. Previous following conclusions are drawn. These conclusions are
experimental data suggest that the “effective fixed point” clearly based on a relatively limited number of tests and
of steel or steel–concrete composite coupling beams is analytical studies. Additional test data from more com-
about 1/3 of the embedment length from the face of the plete subassemblies and more detailed analytical studies
wall [13,14,18]. Therefore, for consistency with experi- are recommended to supplement the results reported her-
mental observations it is recommended to take Lwall as ein.
 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490 1489

1. Nominal encasement around steel coupling beams the expected coupling beam shear angle demands, and
provides an effective means for preventing web buck- hence the amount of stiffeners, if necessary, can be
ling. Hence, web stiffeners can be eliminated. Current established more rationally.
design codes need to be re-evaluated and relaxed for
the cases where the steel coupling beam is encased.
Available provisions appear to be overly conserva-
tive.
Acknowledgements
2. Although current design procedures result in a con-
servative design and detailing of encased steel coup-
ling beams, the increased strength and stiffness due The research presented herein is based on an investi-
to the surrounding concrete encasement could have gation sponsored by the National Science Foundation
detrimental effects on the overall performance if they under grant no. BCS-9319838, with Dr. Shih Chi Liu as
are not taken into account as part of the design. the program director. Any opinions, findings, and con-
3. Nominally reinforced encasement around steel coup- clusions or recommendations expressed in this paper are
ling beams is expected to increase the stiffness. The of those of the writers and do not necessarily reflect the
additional stiffness increases the level of coupling views of the sponsors.
between walls, which in turn affects the distribution
of design forces. Most importantly, the wall axial load
at the base could substantially increase. The increased References
stiffness of encased coupling beams has to be incor-
porated in design of coupled walls as well as foun- [1] Aktan AE, Bertero VV. The seismic resistant design of R/C
dation systems. coupled structural walls. Report No. UCB/EERC-81/07, Univer-
4. Unless design calculations consider the contribution sity of California, Berkeley: Earthquake Engineering Research
Center, 1981.
of encasement towards shear capacity of composite [2] Aristizabal-Ochoa JD. Dynamic response of coupled wall sys-
coupling beams, a significant portion of the input tems. ASCE J Struct Div 1982;108(8):1846–57.
energy will be dissipated by inelastic deformations in [3] Aristizabal-Ochoa JD. Seismic behavior of slender coupled wall
the connection region, which is not desirable. There- systems. ASCE J Struct Div 1987;113(10):2221–34.
fore, the embedment length has to be computed to [4] Paulay T. The design of ductile reinforced concrete structural
walls for earthquake resistance. Earthquake Spectra
develop the expected shear capacity of the composite 1986;2(4):783–823.
section. In lieu of refined fiber analyses, the shear [5] Paulay T. Coupling beams of reinforced concrete shear walls.
capacity may be taken as 1.6 times the sum of the ASCE J Struct Div 1971;97(3):843–62.
shear capacity of the steel beam and encasement. The [6] Paulay T, Santhakumar AR. Ductile behavior of coupled shear
energy dissipation characteristics, ductility, and load- walls. ASCE J Struct Div 1976;102(1):93–108.
[7] Shiu NK, Takayangi T, Corley WG. Seismic behavior of coupled
carrying capacity of composite coupling beams are wall systems. ASCE J Struct Div 1984;110(5):1051–66.
substantially improved by using the proposed meth- [8] Shiu NK, Barney GB, Fiorato AE, Corley WG. Earthquake resist-
odology which leads into a longer embedment length. ant walls coupled wall test. Report to NSF submitted by Portland
5. The contribution of floor slab towards stiffness and Cement Association, Research and Development, Skokie, Illi-
strength of composite coupling beams may be nois, 1981.
[9] International Council of Building Officials (ICBO). Uniform
ignored. The additional stiffness due to floor slab is building code, vol. 2. Whitier, CA: Structural Engineering Design
lost after rather small deformations. Contribution of Provisions, 1994.
floor slab is less than that expected for reinforced con- [10] Building Seismic Safety Council (BSSC). NEHRP Rec-
crete beams because the area of slab reinforcing bars ommended provisions for seismic regulations for new buildings
within the effective slab width is small in comparison and other structures (FEMA 302) and Commentary (FEMA 303),
1997 edition. Washington, DC, 1998.
to the flange area of the coupling beam. The [11] Harries KA. Ductility and deformability of coupling beams in
additional strength may be ignored when the reinforced concrete coupled walls. Proceedings of the Eighth
embedment length is computed as the slab prevents Canadian Conference on Earthquake Engineering, Vancouver,
the formation of plastic hinges in the connection June 1999, 1998:475–481.
region. [12] Harries KA. Seismic design and retrofit of coupled walls using
structural steel. PhD thesis, McGill University, 1995.
6. Relatively detailed inelastic static and dynamic analy- [13] Shahrooz BM, Remmetter MA, Qin F. Seismic response of com-
ses suggest that the maximum expected coupling posite coupled walls. Composite Construction in Steel and Con-
beam shear angle in reasonably proportioned coupled crete II, ASCE, 1992:429–441.
walls is probably less than the value assumed by cur- [14] Shahrooz BM, Remmetter MA, Qin F. Seismic design and per-
rent design provisions. Until the availability of further formance of composite coupled walls. ASCE J Struct Div
1993;119(11):3291–309.
data, a simple procedure, similar to a well established [15] Marcakis K, Mitchell D. Precast concrete connections with
method for link beams in eccentrically braced frames, embedded steel members. Prestressed Concrete Inst J
is proposed to compute a more reliable estimate of 1980;25(4):88–116.
1490 B. Gong, B.M. Shahrooz / Engineering Structures 23 (2001) 1480–1490

[16] Mattock AH, Gaafar GH. Strength of embedded steel sections as [20] American Concrete Institute (ACI) Committee 318. Building
brackets. ACI J 1982;79(2):83–93. code requirements for reinforced concrete and commentary (ACI
[17] Gong B, Shahrooz BM, Gillum AJ. Cyclic response of composite 318-99/ACI 318R-99), Farmington Hills, MI, 1999.
coupling beams. ACI Special Publication 174 — Hybrid and [21] MacGregor JG. Reinforced concrete mechanics and design.
Composite Structures, Farmington Hills, MI, 1998:89–112. Englewood, NJ: Prentice Hall, 1992.
[18] Gong B, Shahrooz BM. Seismic behavior and design of com- [22] Collins MP, Mitchell D. Prestressed concrete structures. Engle-
posite coupled wall systems. Report No. UC-CII 98/01, Cincin- wood, NJ: Prentice Hall, 1991.
nati Infrastructure Institute, Cincinnati, OH, 1998. [23] American Institute of Steel Construction (AISC). Seismic pro-
[19] ASCE Task Committee on Design Criteria for Composite Struc- visions for structural steel buildings, Chicago, IL, 1997.
tures in Steel and Concrete. Guidelines for design of joints [24] Kunnath SK, Reinhorn AM. IDARC-2D Version 3.1: Inelastic
between steel beams and reinforced concrete columns. J Struct damage analysis of reinforced concrete building structures. State
Div, ASCE 1994;120(8):2330–57. University of New York at Buffalo, 1994.