Engineering Structures 24 (2002) 11651175

www.elsevier.com/locate/engstruct

The displacement capacity of reinforced concrete coupled walls

Tom Paulay
Department of Civil Engineering, University of Canterbury, Private Bag 4800, Christchurch, New Zealand
Received 20 December 2001; received in revised form 8 January 2002; accepted 8 April 2002

Abstract
With the identification of criteria of performance-based seismic design, the need to focus on estimations of displacement capacities
of ductile system emerges. This involves redefinitions of some properties of reinforced concrete structures. A system comprising
components with very different characteristics, a coupled wall structure, is used to demonstrate how displacement and ductility
capacities, satisfying specific performance criteria, can be predicted simply, even before the required seismic strength of the system
is established. An attractive feature of this approach is that the strengths of components, which contribute to the required seismic
strength of the system, may be freely chosen. The astute designer may advantageously exploit this freedom. 2002 Elsevier Science
Ltd. All rights reserved.
Keywords: Displacements; Coupling beams; Ductility; Stiffness; Strength

1. Introduction
The prediction of displacement demands imposed on
structures by earthquake motions has been one of the
important issues, challenging the earthquake engineering
research community. Relatively few studies addressed
explicitly the displacement capacity of reinforced concrete ductile structures. A rational evaluation of displacement capacities, associated with both elastic and postelastic response, satisfying specific performance criteria,
should enable acceptable seismic displacement demands,
relevant to local seismic scenarios, to be more convincingly established.
To allow displacement capacities to be realistically
estimated, some traditional definitions of structural
properties, particularly those applicable to homogeneous
materials, need to be redefined. Relevant principles are
presented first. Subsequently applications are illustrated
using a coupled wall example structure. It is postulated
that the displacement capacity of such a system should
be controlled by that of its component with the smallest
displacement capacity. Therefore, instead of commonly
specified or judgement-based global displacement ductility factors, the deliberate evaluation of these for each

Corresponding author. Tel.: 64-3-364 2249; fax: 64-3-364 2758.

E-mail address: t.paulay@civil.canterbury.ac.nz (T. Paulay).

specific system is advocated. The approach relies thus

on the hierarchy of the displacement ductility capacities
of constituent components.
The procedure is claimed to be rational, realistic and
simple. It is design oriented. Redefined properties of
components, as constructed, may then be used, to analyze, if necessary, a structural system comprising
components with different characteristics but known
strengths.
In this presentation abstract definitions of quantities
are, in general, immediately followed by their numerical
evaluations relevant to a particular example structure.
2. The traditional treatment of coupled walls
Some 50 years ago the analysis of elastic coupled
walls structures was a challenging topic for researchers
in several countries. With the arrival of computer technology this pioneering work, based on innovative modeling [16], has become also accessible to the structural
design profession. Even though during significant seismic events, reinforced concrete structures are expected
to perform in the inelastic domain, the assignment to
components of lateral design strength is still widely
based on elastic structural response. However, in recognition of ductile behaviour, within specified limits, a
redistribution between components of internal design
actions, so derived, has been considered acceptable [7].

0141-0296/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 2 ) 0 0 0 5 0 - 0

1166

T. Paulay / Engineering Structures 24 (2002) 11651175

Nomenclature
Ae
Asd
Awe
Awm
Awi
db
Db
Dwi
Ec
fy
h
he
hm
Ie
Ig
l
lp
ki
M
Mni
Mo
Myi
M1,M2
s
T
Vb
Vnb
Vni
a
b
dp
du
dy
ey
h
qb
qby
qw
qwy
mb
mw
m
x
fby
fyi
fwyi
a
by
c
e
T
p
u
y
yi

effective area of cracked concrete section

area of diagonal reinforcement in one direction
aspect ratio of a wall in terms of he
aspect ratio of a wall in terms of hm
aspect ratio of a wall based on its full height
diameter of a bar
overall depth of a beam
length (overall depth) of a wall
modulus of elasticity of concrete
yield strength of reinforcing steel
total height of structure
height where maximum storey drift occurs
height above base of center of accelerated mass
second moment of effective area of cracked reinforced concrete section
second moment of an area of gross concrete section
internal lever arm of coupling system
length of equivalent plastic hinge
stiffness of component
overturning moment at a level
nominal flexural strength of a section
overturning moment at the base of the structure
flexural yield strength of a section
moments assigned to components (1) and (2)
clear span of coupling beam
lateral force-induced axial load on coupled walls
total base shear for the structure
nominal shear capacity of coupling beam
nominal strength of a wall component in terms of its base shear
inclination of diagonal reinforcement
a moment ratio
post-yield storey drift
maximum acceptable storey drift
storey drift at the nominal yield displacement of a wall
yield strain of reinforcing steel
coefficient relevant to nominal yield curvature
beam chord rotation
chord rotation of beam associated with nominal yield curvature
wall slope (storey drift)
wall slope associate with nominal yield curvature
displacement ductility imposed on a beam
displacement ductility relevant to a wall
system displacement ductility
coefficient defining the position of the neutral axis relative to the tension edge
nominal yield curvature in a beam
nominal yield curvature at the critical section
nominal yield curvature of a wall section
anchorage deformation
nominal yield displacement of coupling beam
diagonal shortening of coupling beam
lateral displacement of elastic elements
elongation of diagonal bars in tension
post-yield displacement
maximum limit displacement
nominal yield displacement of a ductile system
nominal yield displacement of a wall component

T. Paulay / Engineering Structures 24 (2002) 11651175

1167

In the design of coupled walls the fact that they are

simply cantilever structures, is often overlooked. As Fig.
1 shows, only the mode of the resistance to lateral forcegenerated overturning moments is different in coupled
walls. The well established equilibrium requirement is
Mo M1 M2 lT

(1)

where the components of flexural resistance are shown

in Fig. 1(c). The axial force at any level, T, results from
the summation of the shear forces transferred by coupling beams above that level. The distance between the
centroidal axes of the two walls is usually taken as the
lever arm, l, on which the axial forces, T, operate. These
3 moment contributions are traditionally assigned proportionally to component stiffness. The latter are based
on flexural rigidities, EcIe, of prismatic components,
where Ec is the modulus of elasticity of the concrete and
Ie is the second moment of effective area of the cracked
reinforced concrete section. This is usually expressed in
terms of a fraction of the second moment of the gross
concrete sectional area, Ig. Values of Ie/Ig, recommended
in some codes or used in publications [7,8] or design
practice, vary in a wide range of 0.2 to 1.0. While the
allocation of design strength to various components is
not sensitive to such assumptions, predicted displacements of elastic coupled walls may involve errors of the
order of several hundred percent. A particular disadvantage of the use in seismic design of crudely estimated
values of Ie, is the inability to predict realistic values of
yield deformations of both components and the system.
Fig. 2(a), showing 3 interconnected rectangular cantilever walls, is used to summarize the force-displacement
relationship based on traditional bi-linear modeling. The
relative lengths, Dwi, of the rectangular walls with identical thickness are 1.00, 1.59 and 2.00, respectively.
Consequently the relative flexural rigidities of the wall
sections, EcIe, being proportional to D3wi, are 1, 4 and 8.

Fig. 2. Bi-linear idealisation of the response of interacting cantilever walls.

Lateral design strength to components are routinely

assigned in the same proportions. These stiffness-proportional strengths, associated with a given displacement, e, are shown in Fig. 2(b). It is then commonly
assumed that, having developed these strengths, components will simultaneously enter the inelastic domain of
response. This fallacy [9], relevant to ductile behaviour
shown in Fig. 2(b), is discussed in the next section.
Assumptions with respect to the stiffness of coupling
beams were considered [1,8,10,11] to affect both the
intensity and the variation with building height of the
shear forces generated in coupling beams. With minor
modifications [7,10,11] stiffness-dependent strengths
have been routinely adopted in conventional seismic
design. The ratio
lT / Mo b

Fig. 1. Comparison of flexural resisting mechanisms in structural

walls.

(2)

quantifies the degree of coupling. Figs 1(b) and (c) illustrate examples of relatively high and low degrees of
coupling. This ratio has been the subject of differing
views in the relevant literature [8]. Some studies suggested [12] that there is an optimal value for , which
promises favourable dynamic seismic response. Others
held the view that large lateral force-induced axial
forces, T, would be difficult for the foundations to
absorb. However, it is not likely that separate foundations for each coupled wall, i.e., a foundation structure
different from that required for a cantilever wall, shown

1168

T. Paulay / Engineering Structures 24 (2002) 11651175

in Fig. 1(a), would be contemplated. For some 25 years

the use of squat coupling beams, if possible, was advocated in New Zealand design practice [7,11]. It was perceived that for ductile systems, a high degree of coupling
could be an efficient and in many cases the major source
of energy dissipation and hence hysteretic damping. For
example a relevant code [10] specifies system displacement ductility capacities, m , in the range of 6m5
for values of 2 / 3b1 / 3. A value of m 5 was considered [10] applicable to appropriately detailed ductile
cantilever walls.

3. Principles of displacement estimates for ductile

wall systems
Bi-linear modeling of force displacement relationships
for reinforced concrete components or systems, is generally accepted as being adequate for purposes of seismic
design. Implications of a more realistic use of this simple
technique, studied recently [9,1315], are briefly summarized here. Fig. 2 is used to complement this review.
3.1. Nominal yield curvature
Using first principles, it has been shown [13] that the
nominal yield curvature at the critical section of a
reinforced concrete wall component i, associated with its
nominal flexural strength, Mni, can be very satisfactorily
approximated by
fyi hey / Dwi

(3)

where ey and Dwi are the yield strain of the reinforcing

steel used and the length (depth) of the wall, respectively. The coefficient h quantifies the combined effects
of the ratio of the nominal to yield flexural strength,
Mni/Myi, and the distance, xDwi, of the neutral axis of the
section from the extreme tension fiber, thus
h (Mni / Myi) / x

(4)

Typical values of these parameter are presented in Fig.

3. It has been found [15,16] that the ratio of flexural
reinforcement and the intensity of axial compression
loads, usually encountered to act on walls of multistorey
buildings, are responsible for only negligible variations
in eq. (4). When axial forces are significantly larger or
smaller than those anticipated to act on cantilever walls,
as in the case of coupled walls, acceptable estimates of
the corresponding changes of the relevant parameters,
listed in Fig. 3, can be readily made. Important features
of nominal yield curvature to be noted are, that it is
inversely proportional to wall length, Dwi, and that, contrary to traditional usage, for design purposes, it is independent of strength.

Fig. 3.

Parameters affecting the nominal yield curvature of sections.

3.2. Nominal yield displacement

With the assumption that neutral axes at all levels of
a wall are located approximately as at the critical base
section, for a given pattern of moments, the displacement
at any level can be readily obtained. The assumption
implies that the extent of cracking over the height of the
walls is similar and that shear and anchorage deformations are neglected. When warranted, these additional
sources of displacements may, however, be included.
Under repeated reversing lateral displacements, effects
of tension stiffening may also be considered negligible.
Of particular interest are displacements of walls at
specific levels, such as that of the center of horizontally
accelerated mass, hm, associated with the nominal yield
curvature at the base of a wall. This, when combined
with the nominal strength of a wall, expressed in terms
of the base shear or moment sustained, enables component stiffness to be defined. Because displacements are

T. Paulay / Engineering Structures 24 (2002) 11651175

proportional to curvatures, the relative value of the nominal yield displacement of a wall is
yifyih2meyAwmhm

(5)

where Awm hm / Dwi is the effective aspect ratio of the

wall. Eq. (5) emphasizes the 3 important parameters to
be considered when attempting to estimate wall displacements.
3.3. Assignment of lateral strength
Because, as eq. (5) demonstrates, the nominal yield
displacement is independent of strength, in contrast with
traditional usage, the latter can be assigned arbitrarily to
interacting components of a wall system. This freedom
in the choice of component contributions to total
required strength can be advantageously exploited by the
astute designer. Of course equilibrium requirements
must not be violated. With the knowledge of the nominal
strength of a component, Vni, its stiffness is uniquely
defined as
ki Vni / yi

(6)

1169

changes in the moment and shear patterns of the walls

may occur.
Fig. 2(d) illustrates similar relationships when component strengths were chosen arbitrarily. In this example
wall strengths were made proportional to D2wi rather than
D3wi, used in the previous examples. The appeal of this
choice is that it results in approximately identical
reinforcement ratios in all walls. A slight reduction of
system stiffness leads to a correspondingly small
increase of the nominal yield displacement of the system. The examples used demonstrate also the relationship between the displacement ductility capacities of the
components and that of the system [9,13,15]. In this
example it was assumed that adequately detailed walls
have a displacement ductility capacity of 4. Wall (3)
being critical (y3 0.5), the seismic displacement
demand on the ductile system must be limited to
max 4 0.5 2.0 displacement units. This corresponds to system displacement ductility capacities of
m 2.0 / 0.5573.6 or m 2.0 / 0.5783.5, respectively. In existing strength-based seismic design procedures these values will control the required design
strength of the system.

3.4. Stiffness and ductility relationships

4. A 12 storey service core
The application of the above principles, controlling
the compatibility of the yield displacements of different
wall components, is illustrated here with the aid of a
simple example. The walls shown in Fig. 2(a) will be
considered again. As eq. (3) stated, nominal curvatures
at the base of these walls are inversely proportional to
their length, Dwi. Hence the relative yield curvatures of
walls (1), (2) and (3) are 1.00, 1/1.59=0.63 and 1/2=0.50,
respectively. If, as one of the possibilities, the strength
allocation recorded in Fig. 2(b) is adopted, the bi-linear
force-displacement simulations, presented in Fig. 2(c),
are established. Therefore, the relative stiffness of all 3
components are determined. For example k2
(4 / 13) / 0.63 0.488.
As Fig. 2(c) shows, nominal strengths of components
are attained at different displacements. The superposition
of the idealized component responses describes the nonlinear system response. However, in seismic design this
can also be modeled using a bi-linear relationship. The
equivalent nominal yield displacement of the system is
then
y Vni / ki

(7)

In this specific example this corresponds to 0.557 displacement units.

The linear elastic response of components is an idealisation, which again is considered to be acceptable in
the design for systems for ductile response. After the
attainment of the nominal yield displacement of the critical element, such as component (3) in Fig. 2, some

To illustrate the application to a coupled wall structure

of the principles outlined in the previous sections, a specific example was chosen. While different aspects of displacement estimates are considered, as stated earlier, the
evaluation of relevant quantities will not only be given
in abstract terms, but will also be simultaneously
expressed in terms of the selected structural dimensions.
This should assist in the appreciation, particularly by
design practitioners, of the simplicity of the approach
employed.
The principal dimensions of a 12 storey service core,
comprising 2 channel shaped reinforced concrete
coupled walls, and its relevant details are shown in Figs
4(a) and (b). All dimensions are expressed in terms of
the total height, h, of the building. Because reference
displacements are strength-independent, only the pattern
of the lateral design forces need to be known. In terms
of a unit base shear, chosen for convenience, these are
given in Fig. 5(b).
Therefore, the overturning moments and shear forces
at each level of the cantilever structure with fully
restrained base, are readily determined. They are
presented in Figs 4(c) and 5.
4.1. Wall properties
The aspect ratio of the individual walls with respect
to the full height, h, is Awi 1 / 0.1357.4. As Wall (1),
shown in Figs 4(a) and (b), is expected to be subjected to

1170

T. Paulay / Engineering Structures 24 (2002) 11651175

Fig. 5. Bending moments and shear forces applicable to the walls of

a coupled wall structure.

4.2. Assignment of component strength

Fig. 4. Principal dimensions and bending moments relevant to a

coupled wall structure.

significant axial tension, the location of its neutral axis,

measured from the tension edge, is estimated as suggested in Section 3.1, with the aid of the information
provided in row 9 of Fig. 3, i.e. with the parameter
x 0.83 0.70. Considerations of Wall (2), subjected
to gravity and lateral force-induced compression (row 8
in Fig. 3), lead to a similar value of x 0.83 0.94.
With the assumption that the yield strain relevant to this
example structure is ey 0.002, we find from eq. (4)
that the yield curvature factor is h1 h2 1.55.
The important property of the walls, the nominal yield
curvature at the base, is thus from eq. (3)
fwy1 fwy2 1.55ey / 0.135h 0.023 / h

(8)

That nominal yield curvatures for two walls, although

subjected very different axial loads, are, in this rather
exceptional case, about the same.

As stated in section 3.3, the assignment of seismic

strengths to components of the coupled wall structure
should be the designers experience-based choice. For
the purpose of estimating wall actions, satisfying equilibrium criteria, the axes of the walls are assumed to
coincide with the centroidal axes of the gross concrete
sections. As Fig. 4(b) shows, the distance between these
axes is l 0.233h. In this example it has been decided
that b 0.56 (eq. (2)), i.e. lT 0.56Mo at the base.
Hence the lateral force-induced axial force in the walls
is Tmax (0.56 0.711hVb) / 0.233h 1.709Vb. Contrary to traditional procedures [7,8,11], identical
strengths are assigned to coupling beams at all levels,
i.e. 1.709Vb / 12 0.142Vb. The moment increment
introduced by the coupling beams at each level is
M 0.233h 0.142Vb 0.033hVb (Figs 4
and 5(a)).
As the lateral force-induced axial load on the walls,
T, increases (stepwise) linearly to its maximum at the
base, the corresponding (stepped) wall moments
(M1 M2), are derived. These are shown by the
shaded area in Fig. 4(c). With Vb 1.00, the sum of
the base wall moments is thus
M1 M2 (0.711 0.233 1.709)h 0.313h

(9)

i.e. 44% of the total overturning moment (eq. (1)). The

presentation in Fig. 4(c) of these moments is informative
because it shows clearly the effects of the chosen beam
strengths on the wall moment patterns. It is evident that

T. Paulay / Engineering Structures 24 (2002) 11651175

at approximately he 0.57h above the base, wall

moments became negligibly small. This enables critical
wall deformations to be readily estimated. The same wall
moments are presented in the more conventional form
in Fig. 5(a).
The final stage of strength assignment, not affecting
deformation estimates for the walls, involves the distribution of the required wall flexural strength, M1
M2, between components (1) and (2). This too can be
done arbitrarily. As wall (2) will be subjected to significant axial compression, it will be able to develop significant flexural strength with only a modest quantity of
tension reinforcement in the vicinity of the door openings. For example the designer may choose a strength
ratio of M2 / M1 7 / 3. Therefore, the total shear force
to be assigned to the walls should be V1 0.3Vb and
V20.7Vb, respectively. This is shown in Fig. 5(b). To
inhibit the interference of possible shear mechanisms
with the intended ductile response of walls, the nominal
shear strength of the walls, as constructed, needs to be
well in excess of that satisfying static equilibrium [7].
Fig. 5(b) also shows the chosen distribution over the
height of lateral static forces. In this case 92% of the
unit base shear was distributed in the traditional pattern
of an inverted triangle, while 8% of the base shear was
added to the lateral force at level 13. Modal shapes will
affect lateral force patterns relevant to elastic systems,
the displacements of which, as in the cases studied here,
are controlled by full height walls. Once walls entered
the inelastic domain of response, higher modes of
vibrations will have negligible effect on overall system
displacements, such a shown in Fig. 8. Therefore, any
type of commonly used lateral design force pattern, leading to displacements consistent with elastic first mode
response, should be considered to be adequate for the
purpose of displacement estimates.

1171

placement limits for the ductile system subsequently to

be established.
4.4. Beam deformations
4.4.1. Conventionally reinforced coupling beams
A convenient form of expressing beam deformations
is by defining the chord rotation at the development of
nominal yield curvatures, shown in Fig. 6(a) as qby
by / s, where by is the relative vertical displacement
of the ends of the beam with clear span s. The nominal
yield curvature for such a beam, with depth Db
0.018h, is estimated as
fby hey / Db 1.7 0.002 / (0.018h)

(12)

0.189 / h
The corresponding transverse beam displacement is
by fbys2 / 6 (0.189 / h)(0.045h)2 / 6

(13a)

0.064 103h
However, due to steel strain penetrations at the beam bar
anchorages, particularly after a few elastic displacement
reversals, additional beam displacements must be
expected. It is assumed that this anchorage deformation,
a, is in the order of yield strain over 8 times the diameter, db, of bars in tension [7].
In the example structure db0.55 103h, and
hence a
8 0.002 0.55 103h 9 106h. The corresponding beam deflection is
by (s / Db) a
(0.045 / 0.018)9 106h

(13b)

4.3. Wall deformations

The typical moment pattern, applicable to the walls
and shown in Figs 4(c) and 5(a), suggests that for the
purpose of displacement estimates, linear variation over
the height he 0.57h may be considered. This is
shown by the dashed line in Fig. 5(a). Hence the nominal
yield deflection of the walls (eq. (5)) at that height may
be estimated by
y1 y2 fwyi h2e / 3

(10)

(0.023 / h)(0.57h) / 3 0.0025h

2

The slope of the walls, i.e. the drift in the 8th storey, is
qwy fwyihe / 2 (0.023 / h)(0.57h) / 2

(11)

0.0066 rad
These are two important quantities which enable dis-

Fig. 6. Sources of nominal yield displacements in reinforced concrete

coupling beams.

1172

T. Paulay / Engineering Structures 24 (2002) 11651175

0.023 103h

The nominal yield chord rotation of the beam, as

shown in Fig. 6(b), is thus

Therefore, the nominal yield chord rotation of a conventionally reinforced coupling beam will be of the order of

qby by / s

qby (by by) / s (0.064

which is found for the example structure to be

(14)

qby

0.023)103h / (0.045h) 0.0019 rad

1.3 0.112 103h / (0.045h 2 sin 18)

4.4.2. Diagonally reinforced coupling beams

Early studies [17,18] indicated that to avoid sliding
shear failures in the inelastic regions of conventionally
reinforced coupling beams, which are subjected to high
shear demands, diagonal reinforcement could be used.
Typical details are shown in Fig. 6(b). Such beams have
been first used in New Zealand and subsequently in
many other countries. Their design and behaviour has
been extensively reported [7,11].
Occasionally claims are made [8] that, because of the
reduced inclination of the diagonal reinforcement in
slender beams, shear resistance becomes inefficient.
Such claims fail to recognize the simple equilibrium-dictated fact that, irrespective of the inclination, a, (Fig.
6(b)) diagonal steel forces can resist simultaneously the
total moment and shear generated by earthquakeimposed chord rotations. Test beams with bar inclinations as small as a 6, exhibited [19], as expected,
Ramberg-Osgood type of hysteretic response with moderate stiffness degradation and displacement ductility
capacities in the order of 14.
The properties of such beams [7,15,17] are:The nominal strength, in terms of the shear force sustained by
diagonal bars with area Asd, as shown in Fig. 6(b), is
Vbn 2Asdfysina

(15)

where fy is the yield strength of the steel used.

The elongation of the diagonal bars in tension is
T (s / cosa 16db)ey

(16a)

where, as in section 4.4.1, allowance was also made for

anchorage deformations. For the beams of the structure
shown in Fig. 4(a), a 18 and db0.55 103h.
Therefore,
T (0.045 / cos18
16 0.55 103)0.002h

(16b)

0.112 103h
The shortening of the diagonal compression chord,
C, depends on the ratio of diagonal reinforcement used.
An approximation, acceptable for seismic design purposes and in agreement with observed magnitudes [18],
results in C 0.3T.
The relative vertical displacement at the ends of the
this beam is
by 1.3T / (2sin a)

(17b)

(17a)

(17c)

0.00524 rad
If there is any effective horizontal reinforcement present,
for example in a flange formed by a floor slab (Fig. 6(a)),
the flexural resistance of the coupling beam will correspondingly increase at one end only. The contribution
of such reinforcement, subjected to tension only, can be
readily determined [7]. Strength enhancement will however, diminish during hysteretic response of the beam.
Such horizontal reinforcement will increase beam
strength only when the imposed ductility demand is
larger than any previously imposed one. The participation in strength development of such bars is similar to
those placed in tension flanges of beams in frames.
In some experimental studies [20] it has been found
that, when the elongation of coupling beam test specimens is prevented during cyclic and reversing loading
by artificial restrainers, other forms of diagonal
reinforcement are likely to result in better ductile
response. In real structures full restraint of beam elongations does not exist. Moreover, in axially restrained
beams, the contribution of shear forces by means of a
diagonal concrete compression field is significant. Under
reversing inelastic displacements the deterioration of the
compressed concrete eventually leads to drastic loss of
beam resistance. As the model in Fig. 6(b) suggests, in
the elastic range of response, diagonal forces associated
with shear transfer can be sustained predominantly, and
in many cases entirely, by the reinforcement without any
reliance on concrete compression strength.
4.5. Relationships between beam and wall
deformations
In this section the estimated displacements of the
walls and the critical pair of coupling beams, associated
with 3 distinct limit states, are compared. These states
refer to (i) the elastic limit of wall response, (ii) acceptable maximum storey drift and (iii) the displacement
ductility capacity of the walls.
4.5.1. At the attainment of the nominal yield
displacement of walls
Eq. (11) estimated the maximum wall rotation, i.e. storey drift, associated with the nominal yield curvature at
the wall base. The relationship between traditionally
evaluated [7] rotations of two identical rectangular walls,
based on EcIe, and the coupling beam chord rotations

T. Paulay / Engineering Structures 24 (2002) 11651175

1173

shortens by 2. However, this contradicts the fact that

after cracking walls expand vertically. Therefore, a more
realistic estimate of the differential axial deformations
of walls can be made if rotations are related to positions
of the neutral axes of the cracked elastic walls. In the
determination of these, the simultaneous actions of
moment and axial force need be taken into account. With
this simulation, shown in Fig. 7(b) the ratio of the beam
chord rotation, qb, and the wall rotation, qw, at a given
level is
qb / qw (Dw c1 c2) / s w

(18)

where the relevant dimensions are defined in Fig. 7(b).

This magnification factor, w, affects dramatically displacement demands on coupling beams. For the channel
shaped walls of the example structure it was estimated
that c1 c2 0.023. In this case the rotation magnification relevant to the beams is simply
w Dw / s 0.135 / 0.045 3
Fig. 7.

Relationships between wall and coupling beam rotations.

are shown in Fig. 7(a). To estimate the relative vertical

displacements of the walls, based on the traditional
definition of axial stiffness [7,11], EcAe, allowance is
also made for axial deformations of the walls, shown as
1 and 2 in Fig. 7(a). The simulation implies that, under
the lateral force-induced axial compression load wall (2)

(19)

Hence with the known nominal yield rotation of the

walls, given by eq. (11), the beam chord rotation at level
8 is estimated as
qb wqwy 3 0.0066 0.02 rad

(20)

This is significantly larger than the nominal yield chord

rotation of conventionally reinforced coupling beams,
given by eq. (14). The displacement ductility imposed
on the critical coupling beam at this elastic limit stage
of the walls is, therefore, of the order of
mb qb / qby 0.02 / 0.00191 10.5

(21)

a magnitude which would be difficult to sustain without

significant loss of beam strength.
However, if diagonally reinforced beams are used,
from eq. (17c) it is found that
mb 0.02 / 0.00524 3.8

(22)

It may be readily shown that at the development of

nominal yield curvatures at the wall bases, all diagonally
reinforced coupling beams would have yielded. Therefore, the development at this stage of all strength components, M1, M2 and lT, shown in Fig. 1(c), can be
expected.
4.5.2. At the attainment of the limiting storey drift
It is assumed that the adopted performance criterion
restricted the maximum storey drift to du 1.5%. Eq.
(11) established that at the nominal yield of the walls
the critical drifts was qwy dy 0.66%. The
additional drift, requires plastic hinge rotations at the
wall base. The recommended [7] effective length of a
plastic hinge of the wall is
lp 0.2Dw 0.044he (0.2 0.135
Fig. 8.

Wall deformations associated with three limit states.

0.044 0.57)h 0.052h

(23)

1174

T. Paulay / Engineering Structures 24 (2002) 11651175

i.e. 40% of the length of the walls. The necessary lateral

post-yield displacement at the level of maximum drift
needs to be
p (he 0.5lp)dp
(0.57 0.5 0.052)0.0084h 0.0046h

(24)

The total displacement at level he, using eq. (10), is thus

u y1 p (2.5 4.6)103h

(25)

0.0071h
implying a displacement ductility demand on the walls
of
mw u / y1 7.1 / 2.5 2.8

(26)

The corresponding chord rotation of the diagonally

reinforced coupling beams at level 8 will be of the order
of qbm wdu 3 0.015 0.045 rad. From eq.
(17c) the displacement ductility demand on this pair of
beams is thus
mb qbm / qby 0.045 / 0.005248.6

(27)

which, with eq. (17a), translates into a steel tensile strain

ductility of the order of m 8.6 / 1.3 6.6. The
maximum steel tensile strain is thus esmax 1.3%.
4.5.3. At the attainment of the displacement ductility
capacity of the walls
Assuming that the displacement ductility capacity of
adequately detailed walls [7] is 5, the additional inelastic
displacement of (5-2.8)y1 2.2 0.0025h
0.0055h of the walls would be acceptable. However,
the associated lateral displacement near level 8 of
max 5 0.0025h 0.0125h (Fig. 8) would
increase the maximum storey drift to dmax 2.5%. The
displacement ductility on the critical coupling beam and
the maximum steel strain would increase to mb
14.3 and esmax 2.2%, respectively.
4.6. System response
The deformed shapes of the walls and lateral displacements just below level 8 and associated with the previously defined 3 limit states, are shown in Fig. 8.
For the purposes of seismic design, the bi-linear
modeling of the ductile response of this example structure, as shown in Fig. 9, was claimed to be entirely
adequate. Lateral displacements of the walls shown
relate to the level of accelerated mass at hm 0.71h
above the base. These displacements can be readily
extrapolated from those previously evaluated at a lower
level, i.e. at he,. Strength increase with post-yield deformations, having negligible influence on the response of
the system, have not been considered. Displacement ductilities, associated with the 3 selected limit states, are
also recorded in Fig. 9. This simple modeling, based on

Fig. 9. Bi-linear modeling of the ductile behaviour of a coupled

wall system.

realistic displacement estimates, for a single mass system, may well replace popular pushover analyses techniques.
As the data in Fig. 3 suggest, displacements during
the first elastic response of the structure can be predicted
by bilinear modeling only if strength demands on the
walls do not exceed approximately 80% of their nominal
strength. Under the same circumstances the onset of
yielding in some coupling beams can be expected at less
than 50% of the nominal strength of the structure.
The choice of the contribution to the total flexural
strength of the system by the coupling beams, that is,
the lT component seen in Fig. 1(c), determines the height
at which the maximum storey drift can be expected. If
the lT / Mo ratio would have been chosen 0.75, instead of
0.56, the maximum drift should have been expected in
the 5 storey, i.e., at he 0.38h. The corresponding
moments to be resisted by the walls are shown in Fig. 4
by the dashed stepped lines. This choice, requiring 34%
increase of beam strengths, would have led to 33%
reduction of the critical nominal yield drift. The inelastic
contribution of the walls at the 1.5% drift limit,
expressed in terms of their displacement ductility, mw,
would have increased from 2.8 to 4.3. This alternative
illustrates how more efficient utilization of energy dissipation and hysteretic damping could be achieved by
deliberate increase of the contribution of the coupling
system to the resistance of overturning moments.
It is re-emphasized that, as eq. (5) has shown, displacement limitations are strongly influenced by the
yield strength of the reinforcement used. For example if
steel with 25% larger strength, i.e. ey 0.0025, was
to be used, nominal yield displacements would correspondingly increase. At a drift limit of 1.5% the ductility
demand on the walls would reduce from 2.8 to 2.2. In
current force-based seismic design methods [21], the
corresponding design base shear for the system would
be increased, negating partly the economic advantages
which the use of higher strength steel would offer.

T. Paulay / Engineering Structures 24 (2002) 11651175

1175

5. Concluding remarks

References

To satisfy the intents of performance-based seismic

structural design, the importance of more realistic predictions of target displacement capacities should be more
widely recognized. For reinforced concrete structures,
addressed here, such displacement limits can be readily
and realistically predicted in a rather simple way without
the knowledge of the eventual seismic strength required.
Therefore, displacement estimates made during the preliminary stage of the design, can immediately expose
undesirable features of the contemplated structural system.
The use of a number of simple principles, often overlooked or ignored in seismic design, was demonstrated.
These include: (a) The stiffness of a reinforced concrete
component depends on the strength eventually assigned
to it. Therefore, element or system stiffness cannot be a
priory assumed. (b) The nominal yield curvature of a
reinforced concrete section, and all displacements of a
component associated with it, are insensitive to the
flexural strength of the section. (c) Because deformation
limits, applicable to components of ductile system, are
independent of the strength, the latter can be arbitrarily
assigned to them. This enables the astute designer to distribute the required seismic strength among components
so that more economical and practical solutions are
obtained.
The estimation of displacement capacities of components of a system, such as a coupled wall structure,
enables the critical component to be identified. Hence,
instead of assuming global ductility factors for structural
systems, their displacement and hence ductility capacity
should be made dependent on that of the critical component. Such relationships can be established before
strengths are assigned to components.
The approach, illustrated with the aid of an example
coupled wall structure, can be readily incorporated into
existing strength-based seismic design methods. Its
major appeal relates, however, to displacement-based
design strategies.
Coupled wall structures offer distinct advantages such
as: (i) very good displacement control, (ii) a strong coupling system allows the use of slender walls without jeopardizing drift limits, (iii) displacement limits during ductile response are not affected by higher mode dynamic
effects, (iv) with appropriate detailing of the reinforcement, they can be expected to deliver larger hysteretic
damping than any other conventionally constructed
reinforced concrete system.

[1] Chitty L. On the cantilever composed of a number of parallel

beams interconnected by cross bars. The London, Edinburgh and
Dublin Philosophical Magazine and Journal of Science
1947;38:68599.
[2] Beck H. Ein neues Berechnungsverfahren fu r gegliederter
Scheiben dargestellt am Beispiel der Vierendeltra gers. Der Bauingenieur 1956;31(12):43643.
[3] Albiges M, Goulet J. Contreventment des batiments. Annales de
lInstitute Technique du Batimants et des Travaux Public
1960;13(149):473500.
[4] Rosman R. Beitrag zur statischen Berechnung waagrecht belasteter Querwa nde bei Hochbauten. Der Bauingenieur
1960;35(4):133136,1962;37(1) :2426,(8);303308.
[5] Beck H. Contribution to the analysis of coupled shear walls. Proceedings ACI Journal 1962;59(8):105570.
[6] Coull A, Choudhury JR. Analysis of coupled shear walls. Proceedings ACI Journal 1967;64(9):58793.
[7] Paulay T, Priestley MJN. Seismic Design of Reinforced Concrete
and Masonry Buildings, p. 767, Wiley.
[8] Harries AK. Ductility and deformability of coupling beams in
reinforced concrete coupled walls. Earthquake Spectra
2001;17(3):45778.
[9] Paulay T. A simple displacement compatibility-based design
strategy for reinforced concrete buildings. Proceedings of the
12th World Conference on Earthquake Engineering, Auckland,
New Zealand, 2000. Paper No.0062
[10] NZS 3101:1995, Standards New Zealand. Concrete Structures
Standard Part 1 The design of concrete structures, p. 256, Part
2 Commentary on the design of concrete structures, p. 264
[11] Park R, Paulay T. Reinforced Concrete Structures, p. 786.
Wiley, 1975.
[12] Saatcioglu M, Derecho AT, Corley WG. Parametric study of
earthquake-resistant coupled walls. ASCE Journal of the Structural Division 1987;113(1):14157.
[13] Paulay T. A re-definition of of stiffness of reinforced concrete
elements and its implications in seismic design. Structural Engineering International 2001;11(1):3641.
[14] Paulay T. Some seismic design principles relevant to torsional
phenomena in ductile buildings. Journal of Earthquake Engineering 2001;5(3):273308.
[15] Paulay T. Seismic response of structural walls: recent developments. Canadian Journal of Civil Engineering 2002;28:in.
[16] Priestley MJN, Kowalsky MJ. Aspects of drift and ductility
capacity of rectangular walls. Bulletin of the New Zealand
Society for Earthquake Engineering 1998;31(2):7385.
[17] Paulay T. Diagonally reinforced coupling beams of shear walls.
ACI Special Publication SP-42 1972;1:57998.
[18] Paulay T, Santhakumar AR. Ductile behaviour of coupled shear
walls. Journal of the Structural Division, ASCE 1976;102:ST1.
[19] Paulay T, Spurr DD. Simulated seismic loading on reinforced
concrete frame-shear wall structures. Sixth World Conference on
Earthquake Engineering, New Delhi 1977;3:2216.
[20] Galano L, Vignoli A. Seismic behaviour of short coupling beams
with different reinforcement layouts. ACI Structural Journal
2000;97(6):87685.
[21] Priestley MJN. Myths and fallacies in earthquake engineering
Conflicts between design and reality. American Concrete Institute, (SP-157) Recent developments in lateral force transfer in
buildings, 231-257

Acknowledgements
The contribution of Rolando Castillo to some of the
data presented, using moment-curvature analyses, is
gratefully acknowledged.