Engineering Structures 29 (2007) 3345

www.elsevier.com/locate/engstruct

Momentcurvature relationship of flexure-dominated walls with partially

confined end-zones
Honggun Park a, , Su-min Kang b , Lan Chung c , Do-bum Lee b
a Department of Architecture, Seoul National University, San 56-1, Shinlim-dong, Kwanak-gu, Seoul, 151-742, Republic of Korea
b Daelim Industrial Co., 146-12, Susong-Dong, Jongno-Gu, Seoul, 110-732, Republic of Korea
c Department of Architectural Engineering, Dankook University, San 147, Hannam-ro, Yongsan-gu, Seoul, 140-714, Republic of Korea

Received 6 September 2005; received in revised form 8 March 2006; accepted 29 March 2006
Available online 27 June 2006

Abstract
Nonlinear numerical analysis was performed to evaluate the momentcurvature relationship and deformability of flexure-dominated walls
with partially confined end-zones. Through the numerical analysis, the effects of design parameters on the behavioral characteristics and the
failure mechanism of the walls were studied. The results showed that the lateral confinement in the end-zones of the cross-section increased the
deformability of the wall by reducing the depth of the compressive zone as well as by increasing the deformability of the confined concrete. The
deformability of the wall significantly increased as the compressive strength of the confined end-zones approaches the applied axial force. Based
on the results, design equations for predicting the momentcurvature relationships and the deformability of the walls were developed.
c 2006 Elsevier Ltd. All rights reserved.

Keywords: Confinement; Ductility; Reinforced concrete; Seismic design; Structural wall

1. Introduction
Recently, inelastic design methods such as the Capacity
Spectrum Method [1] and the Displacement-Based Design [2]
were developed to ensure the seismic safety and economical design of structural systems. To use such inelastic analysis/design
methods, the overall behavior of the structural members including their strength and deformability must be estimated accurately. Fig. 1 shows a generalized loaddeformation curve specified in FEMA 273 [3] (1997).
Generally, a flexure-dominated wall subject to a lateral load
is damaged at the wall base. Therefore, the overall behavior
of the wall is determined by the behavior of the plastic hinge
region at the wall base. The deformability of the wall which
has no end-zone confinement in its cross-section is determined
by the ultimate compressive strain of concrete u in the plastic
hinge region (see Fig. 2). However, current design codes and
researchers have proposed different u values. Wallace [2] and
Corresponding author. Tel.: +82 2 880 7055; fax: +82 2 871 5518.

E-mail address: parkhg@snu.ac.kr (H. Park).

c 2006 Elsevier Ltd. All rights reserved.
0141-0296/$ - see front matter
doi:10.1016/j.engstruct.2006.03.035

Priestley [4] proposed u = 0.004 for use in the ductility

design of the wall. The UBC [5] and New Zealand design
code [6] specify u = 0.003, and FEMA [3] uses u = 0.005.
Wu et al. (2004) [7] defined u as the compressive strain of
concrete at zero stress cu . Therefore, the deformability of a
wall can be evaluated differently according to the design codes
and recommendations used for the design of the wall.
The deformability of a wall with laterally confined endzones is more difficult to evaluate. Previous UBC [8] and
ACI provisions [9] prior to ACI 318-99 [10] required that
a wall categorized as a special reinforced concrete wall has
end-zone elements resisting all gravity and earthquake loads.
However, recently, based on good structural performance of
walls shown in the 1985 Chilean Earthquake, many researchers
(Wood (1991) [11], Wallace and Moehle (1992) [12], Wallace
(1996) [13], Wallace and Orakcal (2002) [14]) reported that
the previous UBC [8] and ACI [9] provisions are excessively
conservative for the earthquake design of slender walls. Based
on the results, current design codes including ACI 31802 [15], UBC-97 [5], the New Zealand design code [6], and
the Canadian design code [16] permitted a partially confined
compressive end-zone.

34

H. Park et al. / Engineering Structures 29 (2007) 3345

Notations
Area of confined end zone
Area of gross section
Area of tensile re-bars
Areas of lateral re-bars placed parallel to the
directions of short edge and long edge of a wall
cross-section, respectively
A0s
Area of compressive re-bars
A0s1
Area of compressive re-bars in yield state
A0s2
Area of compressive re-bars in elastic state
c
Total depth of compressive zone
C
Compressive force of concrete
Ccc
Compressive force of confined concrete
Cs
Compressive force of vertical re-bars
Cs1
Compressive force of vertical re-bars in yield
state
Cs2
Compressive force of vertical re-bars in elastic
state
Csb
Compressive force of concentrated end-zone rebars
Csw
Compressive force of web re-bars
Cuc
Compressive force of unconfined concrete
c B , cC , c D , cG , c F Depth of compressive zone at the
reference steps
Ec
Elastic modulus of concrete
Es
Elastic modulus of steel
f cc
Strength of confined concrete
f c0
Concrete strength
fy
Yield strength of re-bar
h
Wall height
l
Wall depth
lc
Depth of confined zone
l B , lC , l D , l G , l F Depth of unconfined zone at the reference steps
P
Axial compressive load
Pws
Yield strength of web re-bars
s
Vertical spacing of lateral re-bars
t
Wall thickness
tc
Thickness of cover concrete
Ts
Tensile force of vertical re-bars
Ts1
Tensile force of vertical re-bars in yield state
Ts2
Tensile force of vertical re-bars in elastic state
Tsb
Tensile force of concentrated end-zone re-bars
Tsw
Tensile force of web re-bars
,
Coefficients corresponding to rectangular stress
block of unconfined concrete
l
Depth of region where tensile end-zone re-bars
are located
0l
Depth of region where compressive end-zone rebars are located
1
Total curvature increment
11
Curvature increment due to enhanced deformability of confined concrete
12
Curvature increment due to enhanced strength of
confined concrete
Acc
Ag
As
Asl ,Ast

B , C , D , G , F Curvature at the reference steps

c
Compressive strain of unconfined concrete
c0
Compressive strain of unconfined concrete at
peak stress
cc
Compressive strain of confined concrete at peak
stress
c max Maximum compressive strain of unconfined
concrete
cu
Ultimate compressive strain of concrete at zero
stress
u
Ultimate compressive strain of concrete
s
Volumetric ratio of lateral re-bars in confined
zone
v
Ratio of uniformly distributed re-bars
v0
Ratio of concentrated end-zone re-bars
uc
Average stress of unconfined concrete

Fig. 1. Generalized loaddeformation (back-bone) curve (FEMA-273).

Fig. 2. Variation of curvature ductility according to ultimate compressive

strain.

The overall behavior of the wall with a partially confined

end zone is more complicated than that of the wall with
a completely confined compressive zone. Confined and
unconfined compressive zones showing different behaviors
coexist in the cross-section of a wall. Furthermore, the behavior
of the confined concrete varies according to the amount and

H. Park et al. / Engineering Structures 29 (2007) 3345

the details of the lateral re-bars. ACI 318-02 [15] and the
New Zealand design code [6] require that at least half of
the depth of the compressive zone be laterally confined. The
Canadian design code [16] requires that a depth not less than
c(0.4 + 1.5c/l) be confined. c = total depth of the compressive
zone. l = the wall depth. Wallace [2] and Priestley [4] proposed
that the region in which the vertical compressive strain exceeds
0.004 be laterally confined. Though such design codes and
recommendations can be used to improve the deformability of a
wall, they cannot be used to directly evaluate the deformability
of the wall. Furthermore, the depth of the compressive zone and
the ultimate compressive strain, which are used to determine the
depth of the confined zone, are values based on the assumption
that the wall is not laterally confined.
2. Research scope and objective
The failure of a flexure-dominated wall (h/l 3, h = the
wall height) with partially confined end-zones is caused
by the crushing of the confined concrete, crushing of the
unconfined concrete, fracture and buckling of the flexural rebars, and fracture of the lateral re-bars. Current design codes
strictly prescribe the details of the lateral confinement for
the compressive end-zones. Generally, the design provisions
require that the details of the lateral confinement conform to
those used for the earthquake design of columns. In Fig. 3, the
stressstrain relationship of the concrete confined according to
the ACI seismic details [15] was predicted by the Manders
Model [17,18]. The results show that the current details of
lateral confinement significantly increase the strength and
ductility of the confined concrete, without early crushing of the
confined concrete and buckling of the vertical re-bars. For this
reason, in the present study, a wall with partially confined endzones was assumed to fail due to the crushing of unconfined
concrete. Other failure modes such as the crushing of confined
concrete and the fracture and buckling of vertical re-bars were
out of scope of this study. The fracture of lateral re-bars will be
discussed later.

35

(a) Wall with partially confined end zones.

(b) Detail of confined zone.

(c) Stressstrain relationships.

Fig. 3. Stressstrain relationships of confined and unconfined concrete.

(a) Wall A with concentrated end-zone re-bars.

3. Numerical analysis
For numerical analysis, UcFyber [19], a computer program
for nonlinear analysis of the cross-sections of reinforced
concrete members, was used. UcFyber can generate nonlinear
momentcurvature relationships of a cross-section with
confined zone and unconfined zone. In UcFyber [19], a crosssection is discretized into small elements. For a given curvature
and neutral axis, the strain of an element is calculated using
the assumption of linear strain-distribution. Then, the stress of
each element and the resultant force of the cross-section are
calculated from the stressstrain relationship of the confined or
unconfined concrete. Iterative calculations are performed until
the force-equilibrium in the cross-section is satisfied.
Fig. 4(a) and (b) show the prototype walls used for numerical
analysis: Wall A with concentrated end-zone re-bars and Wall
B with uniformly distributed re-bars. In Wall A, the minimum
web re-bars, which are required in the design codes, were

(b) Wall B with uniformly distributed re-bars.

Fig. 4. Prototype walls for numerical study.

neglected to clearly evaluate the effect of the concentrated endzone re-bars on the behavior of the wall. The effect of the
web re-bars was investigated through analyzing the behavior of
Wall B. For numerical analysis, the cross-section of the wall
was divided into the unconfined zone and confined zone, in
which different stressstrain relations of concrete were used.
The stressstrain curve proposed by Mander [17] was used
to describe the behavior of the confined concrete. Fig. 3
compares the stressstrain relations of the confined concrete
and unconfined concrete used for numerical analysis.

36

H. Park et al. / Engineering Structures 29 (2007) 3345

Variables for numerical analysis were the arrangement of

the vertical re-bars, the depth of confined end-zones, and
magnitude of axial load. The cross-section was 6000 300 mm
(depth thickness). Concrete strength f c0 = 24 MPa. Yield
strength of the re-bars f y = 400 MPa. Volumetric ratio of
the lateral re-bars in the confined zone s [=Asl /((t 2tc )
s) + Ast /(lc s)] = 0.04 (see Fig. 3). Vertical spacing of the
lateral re-bars s = 50 mm. Ratio of the uniformly distributed
re-bars v = 0.004. Ratio of the concentrated end-zone re-bars
v0 = 0.012. The depth of the end zone where concentrated
vertical bars was placed was 0.1l. The applied axial force was
0.2 f c0 A g or 0.5 f c0 A g , where A g = area of the gross section.
Fig. 5(a), (b) and (c) show the variations of the
momentcurvature relationships according to the re-bar
arrangement, the depth of the confined end-zone, and the
magnitude of axial force. Generally, the stiffness of the walls
decreased as the yielding of the reinforcement propagated in
the tensile zone. The walls failed due to the crushing of the
unconfined compressive zone. The figures present the yield
point B (Step B) and the maximum deformation C (Step C),
which will be defined in Behavioral characteristics.
As expected, the end-zone confinement has a significant
effect on the deformability of the cross-section. As the
confinement depth lc increases, the deformability (curvature
capacity) increases significantly. As shown in Fig. 3, the lateral
confinement increases both the strength and the deformability
of the confined concrete. As a result, the lateral confinement
not only enhances the deformability of the confined zone but
also reduces the depth of the compressive zone itself. As shown
in Fig. 6, the total curvature increase 1 is composed of 11
due to the enhanced deformability of confined concrete and
12 due to the enhanced strength of confined concrete (in
Fig. 6, 2 = (1 + 11 ) + 12 ). However, current design
codes and recommendations [2,4,5,15] address only the former
(11 ). As a result, they underestimate the effect of the endzone confinement on the deformability of the wall. It should be
noted that to estimate the depth of the compressive zone c with
reasonable precision, the uniform stress-distribution f cc of the
confined zone must be considered even though conservatively
the compressive strength of unconfined concrete is used for the
confined concrete : f cc = f c0 (see Fig. 6).
As shown in Fig. 5(a) and (b), the behavioral characteristics
of the walls with end-zone confinement differ according to the
re-bar arrangement. In Wall A with concentrated end-zone rebars, the load-carrying capacity was maintained after yielding,
and a brittle failure occurred regardless of the depth of the
confined end-zone. On the other hand, in Wall B with uniformly
distributed web re-bars, the load-carrying capacity decreased
gradually after the maximum load-carrying capacity. The effect
of the end-zone confinement on the deformability of the walls
was more pronounced for Wall A with the concentrated endzone re-bars.
As shown in Fig. 5(c), the deformability of the walls subject
to high axial force (P = 0.5A g f c0 ) significantly decreased,
compared to that of the walls subject to low axial force (P =
0.2A g f c0 ). This result verifies the validity of the UBC-97 [5]
which restricts the axial force level of walls to P = 0.35A g f c0 .

Fig. 5. Variations of momentcurvature relationship of wall according to re-bar

arrangement, depth of confined end-zone and magnitude of axial force.

In the walls subject to high axial force, unlike the walls subject
to low axial force, the tensile reinforcement did not yield.
According to Wu et al. [7], in the compression members subject
to high axial force, the yield point is defined by the onset of
compressive strength of the unconfined concrete.
4. Behavioral characteristics
Based on the results of numerical analysis shown in
Fig. 5(a), the momentcurvature relation of Wall A with
concentrated end-zone re-bars can be idealized as the curve
shown in Fig. 7(a). The momentcurvature curve is defined
by the reference points: Yield moment (B), Maximum moment
(C), and Ultimate moment (D). Fig. 8(a)(d) show the stress
and strain profiles of Wall A.
Fig. 8(a) shows the stress and strain profiles at Yield
moment B. Fig. 8(b) shows the stress and strain profiles at Step

H. Park et al. / Engineering Structures 29 (2007) 3345

37

Fig. 6. Curvature enhancement due to lateral confinement.

(a) Simplified momentcurvature curve.

(b) Variation of depth of compressive zone (results of numerical

analysis).
Fig. 7. Simplified momentcurvature curve and variations of depth of
compressive zone for wall with concentrated end-zone re-bars.

F, the time when the load-carrying capacity is maintained after

yielding. At this step, the equilibrium of vertical force is defined
as
P = Cs + Ccc + Cuc Ts

(1)

where P = axial compressive load; Ts = tensile force of

the vertical re-bars; Cs = compressive force of the vertical
re-bars; Ccc = compressive force of the confined concrete;
and Cuc = compressive force of the unconfined concrete. At
Step F, the vertical re-bars located at the end-zones yield.

The confined concrete maintains uniform stress because of its

ductile behavior (see Fig. 3). Consequently, P, Ts , Cs and Ccc
are constant values. Therefore, to satisfy the force-equilibrium
in Eq. (1), the compressive force of the unconfined concrete
Cuc should be maintained as a constant value. The compressive
force of the unconfined concrete can be defined as Cuc =
uc tl F , where uc = average stress of the unconfined concrete,
t = wall thickness and l F = depth of the unconfined concrete
at Step F (see Fig. 8(b)). Fig. 9 shows the values of the average
stress uc varying according to the reference steps. In Fig. 9,
c max = the maximum compressive strain of the unconfined
concrete (see Fig. 8).
As the curvature increases, initially, the average stress uc
of the unconfined concrete increases (Fig. 9). Therefore, to
maintain Cuc (=P + Ts Cs Ccc ), the depth of the
compressive zone decreases. Fig. 7(b) shows the variation of
the depth of the compressive zone according to the curvature
of the cross-section. Here, Step F is defined as the time
when the average stress of the unconfined concrete reaches its
maximum (see Fig. 9). Therefore at Step F, the depth of the
compressive zone reaches the minimum value c F (Fig. 7(b)).
After Step F, the average compressive stress decreases due
to the softened compressive stress of the unconfined concrete
(Fig. 9). Therefore, the depth of the compressive zone increases
from c F to cC , to maintain Cuc (Fig. 7(b)). However, after Step
C, Cuc cannot be maintained by increasing the depth of the
compressive zone, because a part of the unconfined concrete
starts to lose its compressive strength completely (c max
cu ). For this reason, the load-carrying capacity of the wall
cannot be maintained. Step G in Figs. 7(a) and 8(c) cannot exist.
After Step C, as the curvature increases, Cuc decreases
because a part of the unconfined concrete loses its compressive
strength. Therefore, to satisfy the force-equilibrium in Eq. (1)
under the condition of increasing curvature, the force of the rebars in the tensile zone Ts must decrease. For this, the neutral
axis must shift to the tensile zone of the cross section. The
stress and strain profiles are shown in Fig. 8(d). As shown in
Fig. 7(a), the shifting of the neutral axis significantly decreases
the moment-carrying capacity because the compressive force

38

H. Park et al. / Engineering Structures 29 (2007) 3345

(a) Step B (yielding of tensile re-bars).

(b) Step F (Step C).

(c) Step G (virtual step).

(d) Step D.

Fig. 8. Stress and strain profiles for wall with concentrated end-zone re-bars.

Fig. 9. Variation of average stress of unconfined concrete.

of the unconfined concrete develops a negative moment. In

Fig. 7(a), the reduced moment-carrying capacity was assumed
to be maintained as a constant value after Step D. However,
in reality, since the shifting of the neutral axis significantly
increases the compressive strains of the confined concrete and
the unconfined concrete, crushing of concrete and buckling of
vertical re-bars occur. As a result, the reduced flexural strength
cannot be maintained, and a brittle failure occurs.
Based on the results of numerical analysis shown in
Fig. 5(b), the momentcurvature relationship of Wall B with
uniformly distributed re-bars was idealized as the curve shown
in Fig. 10(a). The stress and strain profiles at the reference

steps are shown in Fig. 10(b)(d). Like the behavior of Wall

A with concentrated end-zone re-bars, the moment-carrying
capacity of Wall B decreases at Step C, the time when a
part of the unconfined concrete starts to lose its compressive
strength completely (c max cu ). However, the momentcarrying capacity of Wall B decreases gradually. In Wall B,
vertical re-bars are arranged uniformly along the cross-section.
Therefore, as the neutral axis moves towards the tension side
after Step C, Cs increases and at the same time Ts decreases
(see Fig. 10(d)). As a result, the force-equilibrium in Eq. (1)
can be satisfied by the gradual shifting of the neutral axis
after Step C. As mentioned, the results of numerical analysis
showed that the moment-carrying capacity of Wall B decreased
gradually. However, since crushing of the unconfined concrete
is accompanied by buckling of the vertical re-bars in the web,
the moment-carrying capacity of Wall B might decrease more
rapidly than expected in Fig. 10(a). Therefore, in the present
study, the maximum curvature of the wall with uniformly
distributed web re-bars was defined as the curvature at Step
C, like Wall A with concentrated end-zone re-bars. This result
corresponds also to other compression members dominated by
flexure (Wu et al. [7]).

39

H. Park et al. / Engineering Structures 29 (2007) 3345

(a) Simplified momentcurvature curve.

(c) Step C.

(b) Step B.

(d) Step D.

Fig. 10. Simplified momentcurvature curve and stress and strain profiles for wall with uniformly distributed re-bars.

5. Momentcurvature relationship
Based on the behavioral characteristics and failure
mechanism of walls found from the numerical analysis, design
equations for predicting the momentcurvature relationships
were developed. The momentcurvature curves were developed
for three wall types: a wall with concentrated end-zone rebars (Wall A), a wall with uniformly distributed web rebars (Wall B) and a wall with both the concentrated endzone re-bars and uniform web re-bars. Since the majority
of slender walls are subject to axial compressive force less
than 35% of their maximum compressive strength, the present
study focused on the investigation of the behavior of walls
subjected to low axial force (P 0.35A g f c0 , UBC-97 [5]). The
behavioral characteristics of the compression members which
are subjected to relatively high axial force can be found in the
work by Wu et al. [7].
5.1. Wall with concentrated end-zone re-bars
The momentcurvature curve of the wall with concentrated
end-zone re-bars is shown in Fig. 7(a). The moments and
curvatures at the reference steps B, C and D can be calculated
based on the stressstrain profiles shown in Fig. 8(a), (b) and
(d) respectively.

At the yielding point B in Fig. 8(a), all end-zone re-bars yield

in tension. Therefore, B can be calculated as
B =

fy
E s [(1 )l c B ]

(2)

where l = depth of the cross section; l = depth of the region

where the tensile end-zone re-bars are located; and c B = depth
of the compressive zone at Step B.
At Step B, since the strain of the concrete is small, the
compressive stress of the concrete can be calculated based on
the assumption that the concrete behaves elastically. Under the
assumption, the force-equilibrium can be defined as
(3)

P + Ts = Cs + C

where Ts = As f y , Cs = A0s E s (c B 0.5 0l) B and C =

0.5E c c2B t B . C = compressive force of concrete; As = area
of the tensile re-bars; A0s = area of the compressive re-bars;
E c = elastic modulus of concrete; and 0l = depth of the region
where the compressive end-zone re-bars are located. By using
Eq. (3) and the definitions of Ts , Cs and C, B can be derived
as
B =

P + As f y
A0s E s (c B 0.5 0l) + 0.5E c c2B t

(4)

40

H. Park et al. / Engineering Structures 29 (2007) 3345

From Eqs. (2) and (4), c B and B are calculated.

Subsequently, Ts , Cs and C are calculated using c B and B .
Then, the yield moment at the geometric center can be defined
as
M B = C(0.5l c B /3) + 0.5Cs (1 0 )l + 0.5Ts (1 )l. (5)
At Step C, the maximum compressive strain of the
unconfined concrete reaches the ultimate concrete strain
c max = cu (Fig. 8(b)). The stress profile of the unconfined
concrete can be idealized as a rectangular stress block shown
in Fig. 11. The compressive force of the unconfined concrete
is calculated as Cuc = f c0 t (cC lc tc ), where , [4,
20] = coefficients corresponding to the rectangular stress block
of the unconfined concrete, and tc = thickness of the cover
concrete. At this step, the compressive force of the confined
concrete Ccc can be defined as a uniform value because the
confined concrete shows ductile behavior: Ccc = Acc f cc , where
Acc = area of the confined zone and f cc = compressive
strength of the confined concrete. Since the compressive and
tensile re-bars at the end-zones yield at the ultimate state,
Ts = As f y and Cs = A0s f y . Then, the depth of the compressive
zone can be calculated using the force-equilibrium in Eq. (1).
cC =

P + (As A0s ) f y f cc Acc + f c0 t (lc + tc )

.
f c0 t

(6)

From the strain profile in Fig. 8(b), the curvature at Step C can
be estimated as
cu
cu
=
.
(7)
C =
lC
cC (lc + tc )
Inserting Eq. (6),
C =

f c0 tcu
.
P + (As A0s ) f y f cc Acc

(8)

For a wall with no end-zone confinement (Acc = 0), the

ultimate curvature is defined as
C0 =

f c0 tcu
.
P + (As A0s ) f y

(9)

Using Eqs. (8) and (9), the ratio of curvatures between the walls
with and without end-zone confinement can be defined as
C /C0 =

P + (As A0s ) f y
P + (As A0s ) f y f cc Acc
1

=
1

f cc Acc
P+(As A0s ) f y

1
1

f cc Acc
P

resist all factored gravity load to enhance the deformability

of the walls. At the same time, however, Eq. (10b) indicates
that the prescriptive ACI provision might cause excessively
conservative wall design.
Using the stress profile in Fig. 8(b), the moment-carrying
capacity at Step C can be calculated as
MC = Cuc [0.5l lc tc 0.5(cC lc tc )]
+ 0.5Cs (1 0 )l + 0.5Ts (1 )l
+ 0.5Ccc (l lc 2tc ).

(11)

As shown in Fig. 7(a), after Step C, the moment-carrying

capacity of the wall suddenly decreases because the neutral
axis moves to the tensile zone of the cross section to satisfy
the force-equilibrium. The flexural moment at Step D can be
calculated based on the strain and stress profiles shown in
Fig. 8(d). Assuming that the curvature at D is the same as that
at C,
M D = 0.5Cs (1 0 )l + 0.5Ts (1 )l + 0.5Ccc (l lc 2tc )


fy
cu
Cuc (0.5 )l

(1 0.5) . (12)
E s C
C
5.2. Wall with uniformly distributed re-bars

(10a)

If the area of the tensile re-bars is equal to that of the

compressive re-bars, Eq. (10a) can be simplified as
C /C0 =

Fig. 11. Stress profile of concrete at Step C.

Fig. 10(a) shows the momentcurvature curve of the wall

with uniformly distributed re-bars. The moments and curvatures
at the reference steps can be estimated using the stress and strain
profiles shown in Fig. 10(b), (c) and (d).
At the yield point B in Fig. 10(b), the re-bars in the endzone (0.25l) are assumed to yield [3]. Using this assumption,
the curvature B is estimated as

(10b)

In Eq. (10b), the curvature C increases significantly as the

strength of confined concrete f cc Acc approaches the applied
axial force P. Eq. (10b) verifies the validity of the former
ACI provision [9] which requires that the end-zone elements

B =

fy
.
E s (0.75l c B )

(13)

The equilibrium equation of the vertical forces is the same as

Eq. (3). However, for the wall with uniformly distributed rebars, Ts is defined as the sum of the tensile forces of re-bars

H. Park et al. / Engineering Structures 29 (2007) 3345

41

in yield state and in elastic state: Ts = Ts1 + Ts2 . Therefore,

the forces in Eq. (3) are redefined as Ts1 = 0.25v lt f y ,
Ts2 = 0.5v (0.75l c B )t f y , Cs = 0.5E s v c2B t B , and C =
0.5E c c2B t B . Using the force-equilibrium in Eq. (3) and the
definitions of the forces, the curvature at Step B is derived as
B =

2[P + v t f y (0.625l 0.5c B )]

c2B t (v E s + E c )

(14)

From Eqs. (13) and (14), c B and B are calculated.

Subsequently, Ts (=Ts1 + Ts2 ), Cs and C are calculated using
c B and B . Then, the flexural moment at the Step B is defined
as

Ts2 c B
cB 
M B = 0.375Ts1l +
+ (Cs + C) 0.5l
.
(15)
3
3
At Step C, the maximum compressive strain of the
unconfined concrete reaches the ultimate concrete strain cu
(Fig. 10(c)). The compressive forces of the unconfined concrete
and the confined concreteare calculated as Cuc = f c0 t (cC
lc tc ) and Ccc = Acc f cc , respectively (Fig. 11). Assuming that
all vertical re-bars yield at the ultimate state, Ts = v (l cC )t f y
and Cs = v cC t f y . Then, the depth of the compressive zone can
be calculated by using the force-equilibrium in Eq. (1).
P f cc Acc + f c0 t (lc + tc ) + v lt f y
.
cC =
f c0 t + 2v t f y

(16)

From the strain profile in Fig. 10(c), the curvature at Step C is

estimated by Eqs. (7) and (16):
C =

tcu ( f c0 + 2v f y )
P f cc Acc + Pws

(17)

where the yield strength of the web re-bars Pws = v t f y [l

2(lc + tc )]. For a wall with no end-zone confinement (Acc = 0),
the curvature can be simplified as
C0 =

tcu ( f c0 + 2v f y )
.
P + Pws

(18)

Using Eqs. (17) and (18), the ratio of curvatures between the
walls with and without end-zone confinement is defined as
C /C0 =

P + Pws
1
=
.
f cc Acc
P f cc Acc + Pws
1 P+P
ws

(19)

As mentioned in Eq. (10b), for the wall with concentrated endzone re-bars, the curvature C significantly increases as the
strength of confined concrete f cc Acc approaches the applied
axial force P. On the other hand, for the wall with uniformly
distributed re-bars, the curvature C increases as the strength of
the confined concrete f cc Acc approaches the sum of the applied
external force P and the yield strength of the web re-bars Pws .
The flexural moment at Step C is calculated as
MC = Cuc [0.5l lc tc 0.5(cC lc tc )] + Cs (l cC )
+ 0.5Ccc (l lc 2tc ).

(20)

At Step D in the descending branch of the momentcurvature

curve (Fig. 10(a)), if an arbitrary curvature D is selected, the

Fig. 12. Variations of curvature capacity according to the ratio of compressive

strength of end-zone element to applied axial force.

corresponding compressive force of the unconfined concrete

Cuc can be defined as Cuc = f c0 tl D = f c0 tcu / D . Then,
the depth of the compressive zone c D and the flexural moment
M D can be calculated as
cD =

P f cc Acc f c0 tcu / D + v lt f y
2v t f y

M D = Cs (l c D ) + 0.5Ccc (l lc 2tc )
+ Cuc [0.5l lc tc 0.5(c D lc tc )]

(21)

(22)

where Cuc = f c0 t (c D lc tc ), Cs = v c D t f y , and

Ccc = f cc Acc . The flexural moments in the descending branch
can be calculated by interpolating the flexural moments at Steps
C and D.
The momentcurvature relationship of the wall with both
concentrated end-zone re-bars and uniformly distributed web
re-bars is presented in Appendix A.
Eqs. (8), (17) and (A.6) define the deformability of walls
with various re-bar arrangements: Curvature at Step C. In
the equations, the deformability of the walls increases as
the compressive strength of concrete, the wall thickness, the
ultimate strain of the unconfined concrete, and the area of the
confined zone increase, and as the applied compressive force
decreases.
Fig. 12 shows the variations of the deformability (curvature
capacity) C according to the ratio of the compressive strength
of the confined end-zone to the applied axial force Acc f cc /P.
The relationship was derived in Eqs. (10b) and (19). As shown
in Fig. 12, the deformability of the wall significantly increases
as Acc f cc /P [or Acc f cc /(P + Pws )] increases. For example,
if two confined end-zone elements are designed to resist all
gravity load (Acc f cc /P = 0.5), the deformability C becomes
twice as much as that of the wall without end-zone confinement
C0 . If f cc cannot be accurately estimated, conservatively,
f cc = f c0 can be used for design purpose. However, it
should be noted that to estimate the depth of compressive zone
and the deformability of wall with reasonable precision, the
uniform stress-distribution of the confined concrete must be
considered even though f cc = f c0 is used. If the confinement
details specified in the ACI seismic provision [15] are used,
approximately, f cc is greater than 1.4 f c0 .

42

H. Park et al. / Engineering Structures 29 (2007) 3345

(a) Wall A.

(b) Wall B.
Fig. 13. Wall models for verification of proposed method.

Since the deformability of the unconfined wall C0 can be

easily calculated, Eqs. (10b) and (19) are useful in estimating
the deformability of walls with confined end zones. In Fig. 12,
C must be limited by the deformability of the confined zone
which varies according to the details of the lateral re-bars.
The confined zone fails due to the fracture of lateral bars if
the lateral confinement conforms to the special seismic details
specified by current design codes. UBC-97 recommends that
the maximum strain of confined concrete be limited to 0.015 to
prevent the fracture of lateral re-bar. In the thin walls subject to
severe cyclic loading, the confined zone might fail due to local
buckling of the vertical re-bars and buckling of the end-zone
elements (Priestley [4]).
The momentcurvature relationships of the walls subject
to high axial force (approximately P 0.35A g f c0 ) can be
defined in the same manner. However, due to the effect of high
axial force, the tensile re-bars do not yield. Therefore, yield
point B is defined as the time when the maximum compressive
stress of the unconfined zone reaches the compressive strength
of concrete (Wu et al. [7]). In this case, nonlinear stressdistribution in the unconfined zone should be considered to
define the momentcurvature relationship at yield point B. The
momentcurvature relationships of columns subject to high
compressive force can be found in the work by Wu et al. [7].
6. Verification
To verify the momentcurvature curves proposed in the
present study, nonlinear numerical analysis using UcFyber [19]
was performed on the walls shown in Fig. 13. The size of
the cross-section was 6000 300 mm (depth thickness).
f c0 = 24 MPa, f y = 400 MPa, and P = 0.2A g f c0 . In Wall B, the
vertical reinforcement ratio v was 0.004. In Wall A, the ratio
of the concentrated vertical re-bars v0 was 0.012, and the depth
of the end-zone where the concentrated vertical re-bars were
placed was 0.1l. Fig. 14 compares the results of the nonlinear
numerical analysis and the proposed momentcurvature curves.
As shown in the figures, the proposed curves agree with the
results of numerical analysis.
Current design codes including the ACI 318 seismic
provision permit partial confinement of the compressive zone
as long as the details of lateral confinement conform to the
seismic details of columns. However, there have been very

Fig. 14. Predictions by proposed method for wall models A and B.

few experiments on walls with partial end-zone confinement

because the ACI provisions prior to the ACI 318-99 required
end-zone elements which can resist all gravity loads, and
because very large specimens and loading equipment are
required to test the walls with partially confined compressive
end-zone.
In the present study, a T-shaped wall tested by Wallace [21]
was used to verify the proposed momentcurvature curves
(Figs. 15 and 16). Specimen TW2 had a partially confined
end-zone, and was expected to fail due to the crushing of the
unconfined concrete. To predict the behavior of the T-shaped
wall using the proposed method, the flange was regarded as
an end-zone element with a large thickness. Fig. 15 shows
the dimensions and details of the specimen, loading condition,
and the strain profile expected at failure. Conservatively, the
strength of confined concrete f cc was assumed to be f c0 . As
shown in Fig. 15, the depth of the confined zone was less
than the predicted overall depth of the compressive zone.
Therefore, the failure of the wall was expected to occur in the
unconfined concrete. Fig. 16 compares the momentrotation
relation (at the plastic hinge region) obtained by the experiment
and the proposed method. The height of the plastic hinge
region was assumed to be 0.5l. As shown in Fig. 16, when the
ultimate strain of the unconfined concrete cu was assumed as
0.004, the base rotation predicted by the proposed method was
0.0163 radian, which was greater than 0.014 rad obtained from
the experiment. When the ultimate strain of the unconfined
concrete was assumed as 0.003, the base rotation predicted
by the proposed method was 0.0134 rad. The base rotation

H. Park et al. / Engineering Structures 29 (2007) 3345

43

Fig. 15. T-Shaped Wall TW2 (Wallace, 1995).

Fig. 16. Prediction for TW2.

predicted by FEMA 273 [3] was 0.0175 which overestimated

the test result.
In the present study, the maximum curvature of a wall was
defined as the curvature at Step C, the time when the maximum
strain of the unconfined zone reached the ultimate strain of
concrete at zero stress: c max = cu (see Figs. 7(a) and 8(b)).
However, the test result showed that the maximum strain of
the unconfined zone ranged only from 0.003 to 0.004, which
was about the strain at Step F corresponding to the maximum
average stress of the unconfined concrete (Fig. 9). As shown in
Fig. 7(b), after Step F, the depth of compressive zone rapidly
increases. As a result, the unconfined concrete might fail earlier
than expected. Therefore, for design purposes, the maximum

strain of the unconfined zone at failure must be defined as the

strain corresponding to the maximum average stress: the strain
at Step F. For normal strength concrete, the strain corresponding
to the maximum average stress is conservatively 0.003.
Fig. 17 shows the dimensions and details of specimen RW1
tested by Wallace [21], the loading condition, and the strain
profile expected at failure. As shown in Fig. 18, the actual
deformability (rotational capacity) of the specimen reached
only 56% of that predicted by the proposed method. In
specimen RW1, the vertical spacing of the lateral re-bars was
76 mm, which was too large to assure the ductility of the
confined zone. According to ACI 318-02, the vertical spacing
of the lateral confinement must be less than t/4 (102/4 =
25.5 mm). Consequently, the failure of RW1 occurred at the
confined concrete, due to the insufficient lateral confinement.
The maximum strain of the confined concrete at failure was
0.008, which was less than the generally accepted maximum
strain of the confined concrete 0.0150.05 (Mander [17,18] and
UBC-97 [5]).
7. Conclusions
In the present study, numerical analysis was performed
to investigate the behavioral characteristics and deformability
of flexure-dominated walls with partially confined end-zones.
Based on the results of numerical analysis, the flexural failure
mechanisms of the walls were studied, and design equations for
predicting their momentcurvature curves were developed. The
proposed momentcurvature curves were compared with the
results of inelastic numerical analysis and existing experiments.

44

H. Park et al. / Engineering Structures 29 (2007) 3345

Fig. 17. Rectangular Wall RW1 (Wallace, 1995).

Acknowledgements
This research was financially supported by The Ministry
of Construction & Transportation of Korea (03R&D C0401, 04R&D C02-02) and the National Research Laboratory
Program of the Ministry of Science and Technology
(M10400000294-05J0000-29410). The financial support is
gratefully acknowledged.
Appendix A. Momentcurvature relationship of walls with
both concentrated end-zone re-bars and uniformly distributed re-bars
Fig. 18. Prediction for RW1.

Major findings obtained in the present study can be summarized

as follows.
(1) Walls with partially confined end-zones fail in a brittle
manner, due to the compressive crushing of the unconfined
zone. The compressive crushing of the unconfined zone
causes the neutral axis to shift to the tension side, which
significantly reduces the moment-carrying capacity of the
wall.
(2) The lateral confinement not only enhances the deformability of the confined concrete but also reduces the depth of
the compressive zone itself. To estimate the deformability
of the wall with reasonable precision, the effect of the latter
must be addressed.
(3) The deformability of the wall increases as the compressive
strength of concrete, the wall thickness, the ultimate strain
of unconfined concrete, and the area of the confined zone
increase, and as the applied compressive force decreases.
Particularly, the deformability of the wall significantly
increases as the compressive strength of the confined endzones approaches the applied compressive force.

In a wall cross-section, the equilibrium of vertical forces is

defined as
P + Tsb + Tsw = Csb + Csw + C

(A.1)

where Tsb = tensile force of the concentrated end-zone re-bars;

Tsw = tensile force of the web re-bars; Csb = compressive force
of the end-zone re-bars; and Csw = compressive force of the
web re-bars.
At yield point B, the curvature and the flexural moment, B
and M B , are defined as
B =

fy
E s [(1 )l c B ]

(A.2)

B = {P + As f y + 0.5 f y v t[(1 )l c B ]}/[A0s E s (c B

0.5 0l) + 0.5E c c2B t + 0.5E s (c B 0l)2 v t] (A.3)
M B = C(0.5l c B /3)
+ 0.5Csb (1 0 )l + Csw [(0.5 0.67 0 )l c B /3]
+ 0.5Tsb (1 )l + Tsw (l/6 + c B /3 0.67 l) (A.4)
where Tsb = As f y , Tsw = 0.5v [(1 )l c B ]t f y , Csb =
A0s E s (c B 0.5 0l) B , Csw = 0.5E s v (c B 0l)2 t B and
C = 0.5E c c2B t B .

H. Park et al. / Engineering Structures 29 (2007) 3345

At step C, the depth of the compressive zone, curvature and

flexural moment can be calculated as
cC = [P + (As A0s ) f y f cc Acc + f c0 t (lc + tc )
+ (1 + 0 )v lt f y ]/( f c0 t + 2v t f y )

C =

(A.5)

tu ( f c0 +2v f y )
P+(As A0s ) f y f cc Acc +[(1 + 0 )l2(lc +tc )]v t f y

(A.6)
MC = Cuc [0.5l lc tc 0.5(cc lc tc )}
+ 0.5Ccc (l lc 2tc )
+ 0.5Tsb (1 )l + 0.5Tsw (cC l)
+ 0.5Csb (1 0 )l + 0.5Csw [(1 0 )l cC ]

(A.7)

where Tsb = As f y , Tsw = v [(1 )l cC ]t f y , Csb =

A0s f y , Csw = f y v t (cC 0l), Cuc = f c0 t (cC lc tc )
and Ccc = Acc f cc .
At Step D in the descending branch, if a curvature D is
selected, the depth of the compressive zone c D and the flexural
moment M D can be defined as

cD =

P+(As A0s ) f y f cc Acc f c0 t cu +(1 + 0 )v lt f y

D
2v t f y

(A.8)
M D = 0.5Ccc (l lc 2tc )
+ Cuc [0.5l lc tc 0.5(c D lc tc )]
+ 0.5Tsb (1 )l + 0.5Tsw (c D l)
+ 0.5Csb (1 0 )l + 0.5Csw [(1 0 )l c D ].

(A.9)

References
[1] Building Seismic Safety Council. NEHRP recommended provisions for
seismic regulations for new building and other structures. 1997.
[2] Wallace JW. Seismic design of RC structural walls. Part I: new code
format. J Struct Engrg, ASCE 1995;121(1).

45

[3] Building Seismic Safety Council. NEHRP guidelines for the seismic
rehabilitation of buildings (FEMA 273) 1997.
[4] Paulay T, Priestley MJN. Seismic design of reinforced concrete and
masonry buildings. Wiley Interscience; 1992.
[5] International Conference of Building Officials. Uniform building code
(UBC). 1997.
[6] Standards Association of New Zealand. Draft New Zealand code of
practice for the design of concrete structures, DZ 3101. Wellington; 1978.
[7] Wu YF, Oehlers DJ, Griffith RC. Rational definition of the flexural
deformation capacity rc column sections. Eng Struct 2004;26:64150.
[8] International Conference of Building Officials. Uniform building code
(UBC). 1991.
[9] American Concrete Institute. Building code requirements for structural
concrete (ACI 318-95). Special Provision for Seismic Design 21.6. 1995.
[10] American Concrete Institute. Building code requirements for structural
concrete (ACI 318-99). Special Provision for Seismic Design 21.6, 1999.
[11] Wood SL. Performance of reinforced concrete buildings during the 1985
chile earthquake: Implications for the design of structural walls. Earthq
Spectra 1991;7(4).
[12] Wallace JW, Moehle JP. Ductility and detailing requirements of bearing
wall buildings. J Struct Engrg, ASCE 1992;118(6):162544.
[13] Wallace JW. Evaluation of UBC-94 provisions for seismic design of RC
structural walls. Earthq Spectra 1996;12(12):32748.
[14] Wallace JW, Orakcal K. ACI 318-99 provisions for seismic design of
structural walls. ACI Struct J, ACI 2002;99(4):499508.
[15] American Concrete Institute. Building Code Requirements for Structural
Concrete (ACI 318-02). Special Provision for Seismic Design 21.7, 2002.
[16] Canadian Portland Cement Association. Concrete design hand book.
1985.
[17] Mander JB, Priestley MJN, Park R. Theoretical stressstrain model for
confined concrete. J Struct Engrg, ASCE 1988;114(8).
[18] Mander JB, Priestley MJN, Park R. Observed stressstrain behavior of
confined concrete. J Struct Engrg, ASCE 1988;114(8).
[19] Cross Section Analysis Software. http://www.imbsen.com/xtract.htm.
[20] Chang GA, Mander JB. Seismic energy base fatigue damage analysis of
bridge columns, part I : Evaluation of seismic capacity. Technical report
NCEER-94-006. Buffalo; 1994.
[21] Thomsen JH, Wallace JW. Displacement based design of RC structural
wall: An experimental investigation of walls with rectangular and t-shaped
cross sections. Technical report NSF-BCS-9112962. Clarkson University;
1995.