ACI STRUCTURAL JOURNAL
Title no. 96-S73
Cyclic Load Behavior of Low-Slenderness Reinforced
Concrete Walls: Design Basis and Test Results
by Thomas N. Salonikios, Andreas J. Kappos, Ioannis A. Tegos, and Georgios G. Penelis
The present study addresses the problem of cyclic shear in squat
reinforced concrete walls and attempts to assess the validity of
current design provisions, both in Europe (Eurocode 8) and in the
U.S. (ACI 318). The paper describes a comprehensive experimental
program involving 11 wall specimens, six with shear span ratios of
1.5 and five with 1.0, detailed to the provisions of EC8; problems in
applying these provisions are pointed out and comparisons with the
corresponding ACI 318 provisions are also made. The wall speci-
mens are reinforced against shear, either conventionally (orthogonal
grids of web reinforcement), or with cross-inclined bars; the effects
of web and edge reinforcement ratio, of axial load level, and of the
quality of construction joints are also investigated. The reported test
results clearly show that properly designed and reinforced walls can
reach their flexural capacities, even when their aspect ratio is as low
as 1.0, that sliding shear in this category of walls is not a major
problem, and that cross-inclined (bidiagonal) web reinforcement
can effectively and economically control sliding and the subsequent
pinching of the hysteresis loops, particularly when these bars inter-
sect close to the critical section.
Keywords: cyclic loads; earthquake-resistant structures; reinforced con-
cretes; shear properties; span; walls.
The investigation reported herein addresses the behavior
of reinforced concrete (RC) walls with low slenderness
under reversed cyclic loading. RC walls with low slenderness
are common in low-rise construction, characterized by
normalized moment M to shear V ratios as = M/(V lw) 1.5,
where lw is the wall length, as is commonly referred to as
shear-span ratio or shear ratio. Whereas the behavior of prop-
erly designed walls with as 2 is dominated by flexure and
that of walls with as < 1 is dominated by shear, moment-to-
shear ratios around 1.5 typically result in the least predict-
able behavior, as either flexure or shear, or in fact, a mixed
mode of failure may result under seismic loading.1
Probably as a result of the foregoing uncertainties, code
procedures2,3 for shear design of low slenderness walls
present some notable differences. The ACI Code2 recognizes
the increased shear strength of walls with low hw/ lw (hw is the
height of the wall) by specifying a concrete contribution of
0.25 f c MPa (3 f c psi) for hw/ lw 1.5, reducing to
0.17 f c MPa (2 f c psi) for hw/ lw 2.0. In contrast to this,
the new European prestandard (Eurocode 8)3 does not
provide for increased concrete contribution in low slender-
ness walls. However, Eurocode 8 differentiates the design
equation for shear carried by web reinforcement on the basis
of the shear ratio as; the corresponding equation that applies
for as 1.3 is
V wd = [ h ( a s 0.3 ) + v ( 1.3 a s ) ] f yd b w z (1)
ACI Structural Journal/July-August 1999
TECHNICAL PAPER
where h and v are the ratios of horizontal and vertical web
reinforcement, respectively, fyd is the design strength of this
reinforcement, bw is the thickness of the web, and z 0.8lw is
the effective internal lever arm (distance between the resultants
of tensile and compressive stresses at the section considered).
Eq. (1) implies that for as = 1.3, only horizontal reinforcement
is contributing to shear strength, whereas for as 0.3, only the
vertical reinforcement is resisting shear; both types of web
reinforcement are considered effective for 0.3 < as <1.3. The
ACI Code simply requires that for aspect ratios hw/ lw < 2.0,
the vertical reinforcement ratio should not be less than the
horizontal one. It is worth pointing out that it is more rational
to base the distinction between slender and squat walls on the
shear ratio, rather than on the aspect ratio;4 the two ratios are
the same only in the case of cantilever walls loaded at the tip,
whereas in most cases of practical interest as is less than the
aspect ratio.
Another important difference between the American and
European practice concerns the possibility of sliding shear
failure of squat walls, which is explicitly recognized in the
Eurocode 8 (and a corresponding design procedure is
suggested, as discussed in the next section), but is not explic-
itly accounted for in the ACI Code, which controls it indirectly
by specifying an upper bound of 0.65 f c MPa (8 f c psi) at
the nominal shear stress.
Previous experimental work regarding the strength and
deformability of low slenderness walls subjected to reversed
cyclic loading has been reviewed in Reference 5 and 6. Tests
involving realistic specimens (scale of at least 1/3, two
curtains of web reinforcement) with shear ratios between 1.0
and 1.5, subjected to displacement-controlled reversed
cyclic loading are relatively few; of particular relevance to
the present study are those reported in Reference 7 to 10 that
involve walls with rectangular sections, and without trans-
verse flanges.
A review of the literature showed that previous studies do
not provide conclusive information with respect to the cyclic
shear behavior of squat (low slenderness) walls, particularly
in the case where a sliding shear mode dominates. More
specifically, only in the work by Paulay et al.7 has the effect
of bidiagonal reinforcement on delaying sliding shear failure
been investigated; this type of reinforcement is required by
the Eurocode 8 for walls with hw /lw 2. Moreover, most of
the aforementioned studies focused on a single aspect ratio,
hence the effect of this crucial parameter on the failure mode
ACI Structural Journal, V. 96, No. 4, July-August 1999.
Received January 14, 1998, and reviewed under Institute publication policies. Copy-
right 1999, American Concrete Institute. All rights reserved, including the making of copies
unless permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published in the May-June 2000 ACI Structural Journal if
the discussion is received by January 1, 2000.
649

Thomas N. Salonikios was a graduate research assistant and doctoral candidate in
the Department of Civil Engineering, Aristotle University of Thessaloniki, Greece. His
research interests include seismic behavior of reinforced concrete walls.
Andreas J. Kappos is a reader in the Department of Civil Engineering, Imperial
College, London, UK, and a part-time associate professor in the Department of Civil
Engineering, Aristotle University of Thessaloniki. His research interests include the
behavior of reinforced concrete members and structures subjected to seismic actions.
Ioannis A. Tegos is a professor in the Department of Civil Engineering, Aristotle
University of Thessaloniki. His research interests include the behavior of reinforced
concrete members in flexure and shear under seismic conditions, fiber reinforced
concrete in torsion, and concrete technology.
Georgios G. Penelis is a professor in the Department of Civil Engineering, Aristotle
University of Thessaloniki, where he is Head of the Laboratory of Concrete Struc-
tures. His research interests include several aspects of the behavior of reinforced
concrete and masonry structures, particularly those related to seismic loading.
could only be estimated by comparing results for similar
walls tested in different programs.
The main purpose of the present study was to assess the
validity of current design provisions for cyclic shear in squat
RC walls, focusing on those of EC8 and ACI 318. This paper
reports on a comprehensive experimental program involving
walls with aspect ratios of 1.0 and 1.5, detailed to the provi-
sions of EC8; problems in applying these provisions are
pointed out. The wall specimens are reinforced against shear,
either conventionally (orthogonal grids of web reinforce-
ment), or with additional cross-inclined bidiagonal bars. Test
parameters also include the web reinforcement ratio, the
flexural reinforcement ratio, the level of axial loading, and
the quality of detailing at the construction joint.
RESEARCH SIGNIFICANCE
Currently available experimental data concerning the
behavior of squat RC walls subjected to high cyclic shear are
rather inconclusive, especially with respect to the sliding
shear failure mode and the role of inclined bars in preventing
it. The experimental program presented herein aims at clarifying
these points and shedding new light on the understanding of
cyclic shear behavior of RC walls with an aspect ratio
between 1.0 and 1.5. Unlike most previous studies, the
present study focuses equally on both aspect ratios, thus
making meaningful comparisons possible.
The reported test results clearly show that properly
designed and reinforced walls can reach their flexural capacities
(even when their aspect ratio is as low as 1.0) that sliding
shear in this category of walls is not a major problem, and that
cross-inclined (bidiagonal) web reinforcement can effectively
control sliding and the subsequent pinching of the hysteresis
loops, particularly when these bars intersect close to the
critical section.
DESIGN OF WALL SPECIMENS
Code provisions for shear design of walls
The walls tested were designed according to Eurocode 8,3
which is probably the first document of a regulatory character
that makes a clear distinction between the different modes of
shear failure in RC walls subjected to seismic loading,
providing a different design equation for each of the
following modes:
diagonal tension;
diagonal compression (such as web crushing);
sliding shear.
Resistance against diagonal tension failure is calculated by
superimposing a concrete contribution to a web reinforcement
contribution, which, for squat walls, is given by Eq. (1). In
650
contrast with the ACI provision, the concrete contribution in
the Eurocode is independent of the aspect ratio. Assuming an
average value of 0.06 f c MPa (0.72 f c psi) for the design
shear stress Rd specified in the Eurocode,1 the equation for the
concrete contribution inside a plastic hinge can be written
V cd = 0.06 f c ( 1.2 + 40 1 )b w z (2)
where 1 is the ratio of tension reinforcement. Eq. (2) applies
only when analysis does not indicate the presence of tensile
axial loading in the wall (although not clearly stated in the
code, Vcd = 0 should also be taken for zero axial load). As
typical values of 1 range from 0.01 to 0.02, it is clear that Eq.
(2) is significantly more conservative than the ACI value calcu-
lated from
V n = A c ( c f c + h f y ) (3)
which results in a concrete contribution ranging from 0.17 to
0.25 f c MPa (2 to 3 f c psi), depending on the aspect ratio,
and also involves Ac =bw lw instead of bw z. Outside the poten-
tial plastic hinge, an additional concrete contribution propor-
tional to a stress equal to 15 percent of the stress due to
compressive axial loading can be added to Eq. (2); no such
increase is provided for in the ACI procedure.
Unlike ACI 318 and other codes, Eurocode 8 requires an
explicit calculation of the resistance of walls against sliding
shear failure; this resistance is assumed to be made up from
the dowel resistance of vertical bars, the shear resistance of
cross-inclined bars, and the frictional resistance. For squat
walls (hw/lw < 2.0), it is suggested in Eurocode 8 that at least
half the design shear at the base should be carried by bidiag-
onal bars (at higher levels, the contribution of these bars may
be reduced to one-fourth the total shear). Cross-inclined bars
must be properly anchored beyond the sliding plane and
should cross all sections of the wall within a distance of 0.5lw
or 0.5hw (whichever is smaller) above the critical section.
The experimental program was designed in such a way as
to insure that either a diagonal tension (combined with flex-
ural cracking and spalling) or a sliding shear mode were
expected. Two shear ratios were selected, namely, as = 1.0
and as = 1.5. These ratios are below and above the limit of
1.3 specified by Eurocode 8 for the applicability of the truss
model for shear design [Eq. (1)]. The program includes a
total of 11 specimens, five specimens with as = 1.0, and six
specimens with as = 1.5 since an additional parameter
(quality of construction joint) is considered in the latter
case. All specimens are of the cantilever type (Fig. 1), hence
the aspect ratios coincide with the shear ratio as = M/(Vlw).
Design of wall specimens
Since the number of parameters involved is larger than the
available number of equations, an appropriate design
strategy had to be devised to insure a realistic design,
respecting the EC8 minimum reinforcement requirements.
The dimensions of the specimens were selected taking into
account the capacity of the reaction frame (Fig. 2); this
resulted in a 100 x 1200 mm (3.9 x 47.2 in.2) wall section,
corresponding to a scale of about 1:2.5. To achieve the
required aspect ratios, specimen heights of 1200 and 1800
mm were selected, for as equal to 1.0 and 1.5, respectively.
The longitudinal reinforcement ratio in the boundary
ACI Structural Journal/July-August 1999
 ties<~e/Hl ~4 tie546/'10 44114
"A'
MS\o/2 II II H MS\o/3 II }f
I
'I
~
..a 1 .
1
2#W2L!O 2ft4.:Z{!O ~
.. 0
~
IJ il
6\-_
1- 9011!1
<!116
. 6'- .t...r pug
~*16
61111" fiolr1fi
Ill II n II
...fi::EITTT T 44112
If I I II ..W:IIIII TY Till
Ill I 1/1 II II rn II
6'*16 5+16 ---1.20 ~ieS4e;lO

Fig. 1Reinforcement layout in wall specimens with aspect ratios of 1.5 and 1.0 (1 mm = 0.039 in.).

elements was kept close to the minimum required (1 percent)
by the code in most specimens.
The EC8 provisions for shear (see previous section) have
been applied to the wall specimens of the present study.
Unlike other codes, EC8 adopts different equations for design
against each possible shear failure mode, i.e., diagonal
tension (truss mechanism), diagonal compression (crushing
of concrete struts), and sliding. The values of shear resistance
calculated by introducing measured material strengths in the
EC8 equations are given in Table 1. It is worth pointing out

ACI Structural Journal/July-August 1999 651

Table 1Calculated and measured shear strengths for all specimens
Shear Vy, fl at flexural yield,
Diagonal kN, from analysis Ultimate
Diagonal tension compression Sliding shear shear, kN,
Shear strength, strength, kN, strength, kN, strength, kN, Shear Vu,fl at peak flexural from
Specimen kN, to ACI 318-95 to EC8 to EC8 to EC8 strength, kN, from analysis experiment
fc , MPa Vn VRd3 VRd2 VRd,s Vy, fl /Vu, fl Vu,exp Vu,exp /Vu, fl
MSW1 26.1 553 324 571 222 141/202 197 0.98
MSW2 26.2 350 162 573 137 94/127 124 0.98
MSW3 24.1 344 233 536 165 130/171 176 1.03
MSW4 24.6 552 327 545 296 102/156 158 1.01
MSW5 22.0 545 327 498 268 120/169 187 1.10
MSW6 27.5 557 324 594 231 141/202 202 1.00
LSW1 22.2 541 324 498 195 205/294 262 0.89
LSW2 21.6 342 162 492 117 136/184 191 1.04
LSW3 23.9 343 233 533 163 190/249 268 1.07
LSW4 23.2 548 327 521 288 146/226 232 1.03
LSW5 24.9 553 327 551 283 174/245 247 1.01
Note: 1 MPa = 0.145 ksi; 1 kN = 0.225 kips.

boundary elements to the width bo of the confined core,

namely s bo /3, which results in a spacing of only 27 mm
(1.1 in.), hence creating serious problems of concreting.
Under laboratory conditions, it was indeed possible to
obtain this unrealistically low spacing, but it was deemed
appropriate to use a spacing of 42 mm (1.7 in.) in some of
the specimens (MSW1-3; refer to Table 2) to see the effect
of violating this particular rule. Note that the 42-mm
spacing in combination with the selected hoop diameter
(4.4 mm) does provide the required confinement ratio.

TEST PARAMETERS AND SETUP

Fig. 2Side view of reaction frame with specimen in position
(1 m = 3.28 ft).
here that, unlike ACI 318, which introduces capacity reduction
factors multiplying the nominal strength of members, EC8
calculates the design strength on the basis of characteristic
strengths of materials divided by partial safety factors; more
specifically, a safety factor of 1.5 is used for concrete (applied
to its characteristic strength fck , which is practically the same
as f c in the ACI Code), and a factor of 1.15 is applied to the
steel yield strength fy .
In addition to the design against shear, the remaining
provisions of EC8 were applied, including those for the
confinement of the wall boundary elements. The amount of
confinement in boundary elements depended on the ductility
class adopted for the structure including the wall; EC8
adopts three ductility classes (high, medium, and low) that
are roughly equivalent to the special, intermediate, and ordi-
nary frames specified in the American Uniform Building
Code;11 a comparison of several aspects of seismic design in
European and U.S. practice can be found in a recent paper by
Booth, Kappos and Park.12 The intermediate ductility class
(DC M) that is commonly adopted in the new national
codes of earthquake-prone countries in Europe and else-
where was used in the present study. These detailing provi-
sions are generally similar to those of the ACI Code2
(Section 21.6). A requirement that was found difficult to
implement was the one relating the spacing of hoops in the
Test parameters and specimen selection
The main parameters examined in the experimental
program are the following:
1. Aspect (or shear) ratio (as = 1.0 and 1.5);
2. Use of bidiagonal bars, considering both the quantity
and the distance ld of the intersection point of the bars from
the wall base;
3. Web (grid) reinforcement ratio (0.28 and 0.56 percent);
4. Longitudinal reinforcement ratio in the boundary
members (1.3 and 1.7 percent);
5. Axial loading (N = 0 and N = 0.07fc Ac , where Ac is the
gross area of the wall section); and
6. Presence of construction joints (with and without
roughening).
A complete list of the selected specimens is shown in
Table 2, where the relevant test parameters are indicated and
defined in a legend. Also given as an insert to the table are
characteristic cross sections at the base of specimens,
including those with inclined bars intersecting close to the
base or at a distance ld equal to lw/2.
In the series of specimens with as = 1.0, LSW2 and LSW3
have the same web reinforcement, consisting of two curtains
(orthogonal grids) of 4.2-mm-(0.17-in.)-diameter bars spaced
at 100 mm (4 in.); this gives a reinforcement ratio of 0.28
percent, which is slightly more than the minimum requirement
of 0.2 percent of the Eurocode (and very close to the ACI
minimum ratio of 0.25 percent). The 100 x 240-mm2
boundary elements are reinforced with six bars of 8 mm (0.31
in.) diameter, giving a reinforcement ratio of 1.3 percent
(minimum code requirement is 1 percent). Confinement in the
boundary elements is provided by 4.4-mm-(0.17-in.)-diameter
closed stirrups (hoops) spaced at 26 mm (1 in.). The only

652 ACI Structural Journal/July-August 1999

Table 2Details of test specimens
difference between Specimen LSW2 and LSW3 is that the
latter is subjected to a (constant) axial load N = 0.07 Ac fc .
Specimen LSW1 is similar to LSW2, but the web reinforce-
ment is increased by adding an extra grid of 8 mm bars at
170-mm (6.7 in.) spaces, as shown in Fig. 1 and in the last
sketch below Table 2. This results in doubling the web
reinforcement ratio. Moreover, the longitudinal reinforcement
ratio in the boundary elements is increased to 1.7 percent by
adding two more 8-mm bars. Hence, LSW1 is representative
of a wall subjected to higher seismic forces than the previous
two specimens.
Specimen LSW4 and LSW5 both include bidiagonal
reinforcement. Specimen LSW5 is similar to Specimen
LSW2 in every respect, except that a total of three 8-mm bars
inclined by 45 deg is added in each direction, as shown in
Fig. 1. The selection of the bidiagonal reinforcement was
made in such a way as to achieve an effective reinforcement
ratio for bars crossing the potential (horizontal) sliding plane
almost identical to that of vertical web reinforcement in
Specimen LSW1 (0.58 percent compared to 0.56 percent in
LSW1). This selection was expected to yield a meaningful
comparison of the relative effectiveness of the two types of
shear reinforcement. It is also noted that since the bidiagonal
bars pass through the boundary elements (and are anchored
into the foundation beam, as shown in Fig. 1), the effective
ratio of longitudinal (flexural) reinforcement is the same in
both LSW5 and LSW1.
Specimen LSW4 has the same amount of bidiagonal
reinforcement as Specimen LSW5, but the intersection of
ACI Structural Journal/July-August 1999
the 45-deg inclined bars is at the base of the wall (rather than
lw/2 above it), hence no increase in flexural capacity is
expected in the critical region. Using the Eurocode proce-
dure,3 it is found that the bidiagonal reinforcement can carry
58 and 56 percent of the shear developed at flexural strength
for Specimen LSW4 and LSW5, respectively; the different
percentage results from the fact that bidiagonal bars in
LSW5 also increase the flexural strength slightly.
Specimen MSW1 to MSW5 of the as = 1.5 series are rein-
forced in a similar way as Specimen LSW1 to LSW5, as shown
in Table 2 and Fig. 1. Note that, in the case of Specimen MSW4
and MSW5, bidiagonal reinforcement can carry 86 and 82
percent of the shear developed at flexural strength; these
percentages are higher than in the corresponding specimens
with as = 1.0. An extra specimen, MSW6, is included in this
series, which is identical to MSW1 in every respect except for
the casting procedure. In Specimen MSW6, the foundation
beam was cast first, with starter bars in place, at the position of
the vertical bars in the wall; starter bars extended a total of 560
mm (70 bar diameters) above the base. The rest of the spec-
imen was cast after 4 days, without taking any specific
measures regarding curing at the construction joint and without
any increase of reinforcement at the joint area. Hence, this
specimen was representative of poor construction practices in
an actual RC building site (with the exception, of course, of
possible adverse environmental conditions).
Average values of measured yield strengths of the Grade 500
[specified strength fyk = 500 MPa (72 ksi)] were equal to 585
MPa (84.8 ksi) for the 8-mm bars, and 610 MPa (88.5 ksi) for
653

Fig. 3Instrumentation of typical specimens.
the 4.2 mm bars (used in the web grids and for hoops).
Concrete strengths were found to vary from 22 to 27 MPa (3.5
to 3.8 ksi) at the time of testing.
On the basis of the Eurocode equations for sliding shear, it
was anticipated that the strength of the squat wall LSW1,
LSW2, and LSW3 would be governed by this failure mode
rather than by flexure, as shown in Table 1, which summa-
rizes the strength calculations for all specimens. Flexural
strengths in Table 1 were calculated using actual material
properties and standard fiber modeling, with due account for
concrete confinement (model suggested by Kappos13) in the
boundary members. The shears Vy, fl and Vu, fl corresponding
to the yield and ultimate flexural strength were calculated
assuming that the loading was applied exactly in the middle
of the top beam.
It is noted that the specimens generally have a (theoretical)
resistance to diagonal tension higher than the shear corre-
sponding to the development of their flexural strength (Table
1). This is much more clearly the case if the ACI 318 equation
for shear strength is used, whereas if the Eurocode Eq. (1) and
(2) are used, they predict diagonal tension resistances that
are a little lower than the corresponding (ultimate) flexural
strength in the case of Specimen LSW2 and LSW3; never-
theless, diagonal tension strengths are always predicted to be
higher than the shears corresponding to the (flexural) yield
strength, regardless of the code equation used.
Experimental setup
The reaction frame used in the present study is shown in
Fig. 2. The four columns of the frame have a IPB 600 section
with additional stiffening plates at their bases. The columns
are fixed to the 1.0-m-thick prestressed floor of the Laboratory
of Concrete Structures, Aristotle University of Thessaloniki.
Resting on two of the columns is a horizontal beam to which
one of the hydraulic actuators is attached; this beam can be
displaced vertically at intervals of 150 mm, thus allowing the
application of loading at various heights. The axial loading is
applied through a vertical actuator attached to a beam system,
as shown in Fig. 2. The horizontal actuator has a capacity of
+1000 (compression) and 600 (tension) kN, and the vertical
actuator has a capacity of +250 or 250 kN.
654
Fig. 4Sequence of internally imposed (continuous line)
and externally recorded (dotted line) top displacements for
typical specimen (MSW3) (1 mm = 0.039 in.).
The anchorage block of the specimen (Fig. 1 and 2) is
fixed to the laboratory floor with prestressing bars having a
total prestressing force of 1200 kN that prevents uplifting of
the specimen during the application of the horizontal
loading. Moreover, a blocking mechanism involving two
tubes eccentrically fitted within each other is used for
preventing the horizontal sliding of the specimen. Measured
displacements due to uplifting and to sliding were found to
be negligible (less than 0.08 mm).
Instrumentation
The data acquisition system consists of six internal control
and recording channels (three for each actuator) and a total
of eight channels for monitoring data from external instruments
[linear variable displacement transducers (LVDTs, strain
gages)].
In addition to the load cells at the end of the hydraulic
actuators, a series of LVDTs were used for measuring crit-
ical response quantities. As shown in Fig. 3, one LVDT was
installed at the top of the specimen to monitor the top
displacement. A second one was installed at a distance of
200 mm from the wall base to measure the sliding of the
base. Two LVDTs were installed close to the boundary
elements to measure their axial deformations within a length
equal to lw (the horizontal length of the wall) that corre-
sponded to the expected plastic hinge length. Finally, two
further LVDTs inclined at 45 deg were installed in the lower
region (Fig. 3) of the wall to measure shear deformations.
The foregoing system of measurements made it possible to
estimate the flexural, shear, and sliding components of the
wall deformation, as discussed in the following sections.
These estimates were also based on a series of strain gage
measurements at various positions along the reinforcing bars.
Application of loading
The horizontal loading was applied at the top beam of the
specimens (Fig. 2) wherein two metal plates were attached,
connected through a prestressing force of 300 kN to avoid
possible elongation of the bars. The use of a stiff beam at the
top of wall specimens has been questioned by some
researchers8 who pointed out that it tends to overestimate the
shear capacity of the walls; it is a fact, however, that in most
practical situations, beams are framing into the walls at the
level of floor slabs, although these beams are usually less
stiff than those used in test specimens.
ACI Structural Journal/July-August 1999
 (a) (b)
(d)
Fig. 5Cracking patterns in walls with as = 1.0 at end of tests: (a) Specimen LSW1; (b) Specimen LSW2; (c) Specimen LSW3;
(d) Specimen LSW4; and (e) Specimen LSW5.
To investigate the effect of applying the horizontal loading
at the edge or in the middle of the specimen, and of
prestressing the plates at the top of the specimen, a finite
element study of the specimen under various loading condi-
tions at the top was carried out. Two-dimensional shell
elements were used for constructing the model; both static
and dynamic (modal) analyses (using a spectrum, which
gave the same base shear as in static analysis) were carried
out. The analysis of the two specimen types for various
loading conditions has shown that, although the stress distri-
bution is quite different at the top of the specimen
(depending on the position of the loading point), the stress
pattern at the base is very similar in all cases, for both
dynamic and static loading; as expected, dynamic stresses
are slightly lower than the equivalent static ones. It was thus
concluded that as long as the critical region (plastic hinge) is
the one close to the base, the load may be applied either at
the edge or at the middle of the specimen head; for reasons
of practicality, the former solution was preferred. More
details of the FE study may be found in Salonikios et al.14
As the effect of loading history was not selected as a test
variable, the typical procedure of applying three loading
cycles at each ductility level until failure was used in the
present study. Displacement control was used throughout the
test, with the exception of the first cycle in the elastic range.
The typical displacement history consisted in three initial
single cycles at 2, 4, and 6 mm, followed by three cycles strengths; hence, the previous value is substantially higher
ACI Structural Journal/July-August 1999
(c)
(e)
at the same amplitude with increments of 2 mm (total drift of
0.1 percent for as = 1.5) up to a displacement of 16 mm (if
the specimen had not failed earlier), and of 4 mm thereafter,
up to the failure point, defined as that corresponding to 75
percent of the maximum strength. A typical sequence of
displacement cycles is shown in Fig. 4, where both internally
imposed and externally recorded values are shown for
comparison.
The axial loading (whenever present) was kept constant
during the entire test by applying load control to the vertical
actuator (Fig. 2). Hinged connections at the tips of both the
vertical and the horizontal actuator prevent any substantial
restraint to the rotation of the top of the wall, thus insuring
cantilever behavior.
DISCUSSION OF RESULTS
Walls with aspect ratio of 1.0
The cracking pattern at the end of the test of each specimen
with as = 1.0 is shown in Fig. 5. Cracking in these specimens
initiated with hairline flexural cracks in the lower part of the
walls that appeared at a base shear of approximately 120 kN,
was almost horizontal within the length of the boundary
elements and slightly inclined (downwards) in the web. The
first shear cracks appeared later for a base shear of approxi-
mately 190 kN, corresponding to a nominal shear stress of
0.3 f c MPa (3.6 f c psi). Note that the shear stress is calcu-
lated using measured, rather than specified, concrete
655
Fig. 6Load versus top displacement curves for walls with as = 1.0: (a) Wall LSW1; (b) Wall LSW2; and (c) Wall LSW3; (d)
Wall LSW4; and (e) Wall LSW5. (1 mm = 0.039 in.; 1 kN = 224.8 lbf.)

than the ACI value for the concrete contribution in squat walls
(3 f c psi), and approximately four times the corresponding
value of the Eurocode (0.08 f c MPa from Eq. (2), for a spec-
ified concrete strength f c = 20 MPa). As shown in Fig. 5,
shear cracks extended up to the top of the specimens, and their
inclination was quite higher than that of cracks at the bottom
part, although the inclination reduced towards the vertical
boundaries. Apparently due to the presence of the properly
reinforced and heavily confined boundary elements, the width
of the shear cracks tended to decrease towards the edges of the
specimens; the shear crack width did not exceed 0.2 mm in all
specimens.
In the lower part of the wall, flexural cracks originating
from one edge were intersected by inclined shear or flexure-
shear cracks originating from the opposite edge, resulting in
a characteristic criss-cross pattern, as shown in Fig. 5. With
cycling to increased deformations, the rhomboidal pieces of
concrete between the intersecting cracks gradually deterio-
rated and spalling of cover concrete occurred; as expected,
the spalling zone extended further upwards in the case of
Specimen LSW3 that was subjected to axial compression
[Fig. 5(c)]. At about the same displacement amplitude, a
major horizontal crack running through the entire base of the
wall formed. This crack is clearly visible in the case of Spec-
imen LSW2, which was lightly reinforced (Fig. 1) and had no
axial loading, but also formed in the other specimens.
However, in the case of the specimen with bidiagonal rein-
forcement intersecting in the middle of the base section

656 ACI Structural Journal/July-August 1999

[LSW4, Fig. 5(d)] the width of the horizontal crack was
smaller. Significant loss of strength, leading to failure, was
observed when concrete started to deteriorate in the most
heavily stressed parts of the boundary elements, hoops began
to lose support and were moved away from each other, as
buckling of the outer longitudinal bars occurred.
Shown in Fig. 6 are (base shear) - (top displacement) hyster-
esis loops for specimens with as =1.0. The well-known char-
acteristics of RC members subjected to cyclic loading, such as
unloading and reloading stiffness reduction as the cyclic
displacement amplitude increases, and pinching of hysteresis
loops, can be clearly seen in the figure. However, the degree
of pinching, which is a manifestation of crack closing and
consequent loss of stiffness during reloading in the opposite
direction, varies significantly. Specimen LSW1 and LSW2
exhibit the maximum degree of pinching, which is very (a)
similar in both specimens, although the amount of web rein-
forcement in LSW1 is double that in LSW2; this is a strong
indication that pinching was controlled by bond-slip and hori-
zontal sliding, rather than by inclined shear crack opening.
The degree of pinching is slightly lower in Specimen LSW3,
apparently due to the effect of the small axial compression,
which is known1 to favorably affect the bond-slip and sliding
mechanisms. A much lower degree of pinching is observed in
the loops of the two specimens with bidiagonal reinforcement
[Fig. 6(d) and 6(e)], whose hysteretic behavior is similar to
that of well-designed, flexure-dominated RC members. It
appears that the bidiagonal bars were effective in restricting
sliding at the critical crack close to the base, confirming
previous similar remarks by Paulay et al.7 These bars have
also contributed to better control of the inclined shear cracks
in the web of Specimen LSW4 and LSW5 (measured
maximum crack width at the end of the test of 0.14 mm,
compared to 0.2 mm in the other specimens). The main reason (b)
for this better control is that bidiagonal bars intersect the
inclined shear cracks almost at right angles (refer to the Fig. 7Base shear versus shear sliding for specimens: (a)
cracking pattern in Fig. 5), hence, they work essentially in LSW2; and (b) LSW4. (1 mm = 0.039 in.; 1 kN = 224.8 lbf.)
direct tension, whereas the bars in the orthogonal grid intersect
the shear cracks at 35 to 45 deg and tend to work primarily as percent higher than that by LSW4), they are subjected to
dowels. This remark has previously been confirmed analyti- inelastic strains that render them less effective in resisting the
cally by Sittipunt and Wood;15 indeed the hysteresis loops of sliding deformations. The participation to the inelastic flex-
Fig. 6(d) and 6(e) are quite similar to those calculated in the ural mechanisms is also the reason of the reduced effective-
aforementioned study. ness of the vertical bars of the web reinforcement in the
To further clarify the role of the bidiagonal reinforcement, conventionally reinforced specimens. Note that Specimen
recourse is made to the measured (base shear) - (base sliding) LSW1 and LSW5 have effectively the same amount of
relationships shown in Fig. 7 for two of the specimens. The vertical reinforcement, but as the loops in Fig. 6(a) and 6(e)
significant difference between the behavior of the specimen indicate, the control of sliding shear is much more effective
without bidiagonal bars (LSW2) and the corresponding spec- in the latter specimen.
imen with bidiagonal bars (LSW4) is clearly seen in the While the foregoing discussion clearly points to the fact
figure. For the former, the sliding at the base accounts for up that inelastic stiffness and energy dissipation characteristics
to 60 percent of the total displacement at the stage close to of walls with aspect ratio of 1.0 are improved if shear sliding,
failure (the contribution of sliding increases with the level of as well as diagonal crack propagation, are controlled through
inelasticity). For LSW4, which had bidiagonal bars inter- the use of bidiagonal reinforcement, the question still remains
secting at the level of the base section, maximum sliding was whether these members can indeed develop their flexural
approximately 2 mm, accounting for only 20 percent of the strength prior to failure, which is a prerequisite of adequate
top displacement. The corresponding value for Specimen seismic performance. Table 1 summarizes the calculated and
LSW5, where bidiagonal bars intersected above the critical measured maximum shear forces for all specimens, wherein
section (Fig. 1), was equal to 37 percent of the total, indi- the shear resistances have been calculated from the Eurocode 8
cating that this type of reinforcement arrangement is some- and ACI 318 equations, but using actually measured material
what less effective with respect to shear sliding control; a strengths; for the walls with no axial loading, Vcd = 0 has been
possible explanation for this could be that since the bidiag- taken in the Eurocode design [Eq. (2)]. It is seen that
onal bars reached the boundary elements in the most critical measured forces are greater than those corresponding to
region of the wall, obviously contributing to its flexural flexural yield and, with the exception of Wall LSW1, greater
strength (the maximum load sustained by LSW5 was 7 than or approximately equal to those corresponding to the

ACI Structural Journal/July-August 1999 657

 (a)
(d)
Fig. 8Cracking patterns in walls with as = 1.5 at end of tests: (a) Specimen MSW1; (b) Specimen MSW2; (c) Specimen
MSW3; (d) Specimen MSW4; (e) Specimen MSW5; and (f) Specimen MSW6.
development of the full flexural capacities of the specimens.
It is worth pointing out that for both LSW2 and LSW3, the
theoretical resistances against sliding shear failure were
substantially lower than the flexural strengths. This is a clear
indication that the Eurocode equations are underestimating
the capacity of walls with respect to sliding shear failure.
Note that the maximum nominal shear stresses (V/blw) were
0.46 f c (fc as measured in cylinder tests) in Specimen
LSW1 and LSW3, and 0.34 f c in LSW2.
The ACI-based values shown in Table 1 always predict that
the shear strength of the walls is safely higher than the corre-
sponding flexural strength. This is generally confirmed by the
test results with the exception of Specimen LSW1, where it is
clear that the full flexural strength was not developed; for this
specimen, the Eurocode appears to better predict the (final)
failure mode. These important aspects are discussed in more
detail in a forthcoming companion paper.16
658
(b) (c)
(e) (e)
Walls with aspect ratio of 1.5
The cracking pattern at the end of the test of each specimen
with as = 1.5 is shown in Fig. 8. Cracking in these specimens
initiated with hairline flexural cracks in the lower part of the
walls, extending up to a height of approximately lw. These
cracks were initially horizontal and confined within the
length of the boundary elements, but as the loading
increased, they became slightly inclined downwards and
extended into the web. As shown in Fig. 8, the cracking
pattern in all specimens of the MSW series was of a
primarily flexural type. Some shear cracking appeared
(subsequent to flexural cracking) in the upper third of the
wall height in most specimens, but shear crack propagation
was minimal in most specimens. The only exception was
Specimen MSW3 [Fig. 8(c)], where, due to the combination
of higher shear force (due to the compressive axial loading)
and lower shear reinforcement, the shear crack width was
larger than in the other specimens.
ACI Structural Journal/July-August 1999
 In the lower part of the wall, up to a height of approximately
lw /2, flexural cracks originating from one edge were inter-
sected by flexure-shear cracks originating from the opposite
edge, resulting in a characteristic criss-cross pattern, as shown
in Fig. 8. This pattern is similar to that observed in the LSW
series (Fig. 5); however, the concrete pieces formed by cracks
have approximately 90-deg angles in the 1.5 slenderness case
(MSW, Fig. 8), whereas in the 1.0 slenderness walls (LSW,
Fig. 5) the upper and lower angles are more than 90 deg, and
the left and right angles less than 90 deg.
Subsequent to flexural crack propagation, spalling of cover
concrete initiated at the most-heavily compressed fibers of the
boundary elements, followed by spalling at the two sides of the
boundary elements (refer to typical patterns in Fig. 8).
Following the spalling of the concrete cover, large perma-
nent deformations of the bars in the boundary elements were
observed, and eventually buckling of the mostly compressed
bars took place (Fig. 8). This can be considered as the failure
stage, although no abrupt drop in strength was observed.
Subsequent to this stage, significant sliding took place at the
base of some specimens. The measured sliding at the base of
the walls with conventional shear reinforcement and no axial
loading was up to 40 percent the top displacement, whereas
the corresponding values for the walls with bidiagonal rein-
forcement were only 10 to 14 percent; the wall with axial
loading and conventional shear reinforcement (MSW3) had
a sliding of only 18 percent the top displacement.
Shown in Fig. 9 are (base shear)-(top displacement)
hysteresis loops for the six specimens with as = 1.5. The
loops for the three specimens with conventional shear rein-
forcement [Fig. 9(a), 9(b), and 9(c)] are generally similar,
exhibiting the well-known characteristics of gradual stiff-
ness and strength degradation, and a significant degree of
pinching; the latter is less pronounced, as expected, in the
specimen (MSW3) with axial compression. As shown in
Fig. 9, Specimen MSW2 is characterized by a lower initial
stiffness, which is mainly due to the fact that this was the pilot
specimen for the program and, as reported in detail previ-
ously,14 it has failed in an out-of-plane buckling mode during
initial testing without a lateral supporting frame. The spec-
imen was retested, as described in the previous section, and it
was found that the critical (flexural) crack was different from
the existing ones due to the out-of-plane failure, and the spec-
imen fully developed its anticipated flexural strength (Table
1). Hence, it was decided to include the results for this spec-
imen herein, as it was deemed that they were indeed relevant.
As can be shown in Fig. 9(b), the hysteresis loop for this spec-
imen is similar to that of the other specimens with comparable
characteristics, except that it is consistently more flexible.
The strength of MSW3 increased (as expected) due to the
presence of the compressive axial load, but the ductility was
slightly inferior to that of MSW1 and MSW2. Significant
strength degradation occurred at a displacement of 25 mm
(1.4 percent the specimen height) following extensive
concrete crushing and reinforcement buckling at the edges;
further cycling led to eventual fracture of some buckled bars.
Hence, the behavior of this axially loaded wall was domi-
nated by flexure and not by the diagonal shear cracks that
developed in the web. Using the EC8 recommended spacing
(27 mm) of hoops in the boundary members (instead of the
42 mm provided) should have improved the ductility of the
wall. As previously mentioned, sliding at the fixed base was
found to be smaller than in the case of Specimen MSW1 and
MSW2.
ACI Structural Journal/July-August 1999
Shown in Fig. 9(d) and 9(e) are the hysteresis loops
recorded for Specimen MSW4 and MSW5, which were
similar to MSW3, but, in addition, contained the bidiagonal
reinforcement indicated in Fig. 1 and Table 2. The strength
of Specimen MSW5 was roughly the same as that of MSW3,
which was subjected to axial loading. This was expected
since the inclined bars are positioned close to the member
edges in the region near the base and contribute to the flex-
ural strength of the wall (as also recognized by EC8). In
addition to the strength increase, the bidiagonal reinforce-
ment contributed to reducing pinching of the hysteresis loops
(essentially through the direct-tension mechanism described
in the previous section), thus resulting in increased energy
dissipation. Moreover, the ductility of the specimen was
superior to that of MSW2; the significant strength degrada-
tion occurring at a displacement of approximately 28 mm
(1.6 percent the specimen height). Diagonal tension cracks
appeared in both directions; however, the failure mode was
a flexural one, similar to that of Specimen MSW3. It has to
be pointed out that the increased strength due to diagonal
bars intersecting away from the wall base may also have
unfavorable effects, since it increases the shear stress in the
member; this was not the case with Specimen MSW5, where
the maximum shear stress was 0.33 f c , well below the
limits defining shear-dominated behavior.1 Similar charac-
teristics can be detected in the case of Specimen MSW4,
wherein the bidiagonal bars intersect at the base section. In
this case, the increase in strength is smaller (Table 1), as the
bars are much closer to the compression zone than in Spec-
imen MSW5. It is interesting to note, though, that in MSW4,
all bidiagonal bars contribute to the flexural strength, while
in MSW5, only one half of them contribute to flexural
strength since the other half lies within the compression
zone. It is clear from Fig. 9 that the presence of the bidiag-
onal bars leads to significant reduction in the horizontal
sliding (from over 35 to less than 15 percent), which results
in reduced pinching and increased energy dissipation capacity.
As can be seen in Table 1, all specimens with a shear ratio
of 1.5 have developed shears very close to those corre-
sponding to their ultimate flexural strengths; this is a further
indication of the flexure-dominated response of these walls.
The maximum discrepancy between calculated and measured
force was for Specimen MSW5, where the measured shear
exceeded by 10 percent the theoretical flexural capacity. The
main reason for this should be that the latter was estimated
on the basis of the vertical component of the 45-deg inclined
bars, whereas during cyclic testing the inclination of these
bars within the critical zone was larger.
From the economic point of view, it has to be pointed out
that in the specimens with bidiagonal bars, the total rein-
forcement was approximately 15 percent lower than in Spec-
imen MSW1 and LSW1; nevertheless, as discussed
previously, the overall behavior was superior. It appears,
therefore, that bidiagonal reinforcement might be an attractive
alternative to current practice for web reinforcement in walls,
especially due to the fact that the placement of a few inclined
bars in each direction (not surrounded by any special hoop
reinforcement) does not present any practical difficulties.
Finally, the loop for Specimen MSW6, which had a rela-
tively poorly detailed construction joint, should be compared
to that of MSW1, which had the same reinforcement but no
construction joint. It is seen that both specimens reach
approximately the same strength (corresponding to a shear of
approximately 200 kN), but strength degradation is more
659
 (a) (d)

(b) (e)

(c) (f)

Fig. 9Load versus top displacement curves for walls with as = 1.5 (see Fig. 8): (a) Wall MSW1; (b) Wall MSW2; (c) Wall
MSW3; (d) Wall MSW4; (e) Wall MSW5; and (f) Wall MSW6. (1 mm = 0.039 in.; 1 kN = 224.8 lbf.)

pronounced in the specimen with the construction joint
(MSW6). The latter can maintain at least 75 percent of its
strength for at least three cycles up to a displacement of 20
mm (1.1 percent its height), whereas significant strength
degradation in MSW1 starts at displacements larger than 25
mm. Specimen MSW6 is also characterized by an unsym-
metric behavior (maximum strength in one direction is 17
percent higher than in the other one), which should be attrib-
uted to the different behavior of the spliced bars in tension
and in compression. Nevertheless, as shown in Fig. 8(f), the
failure mode of MSW6 is at least initially flexural, and the
high sliding deformations (up to 40 percent the total displace-
ment) are recorded along the critical crack (formed exactly at
the level of splicing) only at a later stage.

660 ACI Structural Journal/July-August 1999

 A detailed discussion of the failure modes and the corre-
sponding analytical predictions of the strengths at each mode
will be presented, along with quantitative evaluations of the
various components of the total displacement, the strength
and stiffness degradation characteristics during cycling, and
the energy dissipation capacities (hysteretic damping), in a
forthcoming companion paper.14
CONCLUSIONS
Currently available experimental data concerning the
behavior of squat RC walls subjected to high cyclic shear are
rather inconclusive, especially with respect to the sliding
shear failure mode and the potential role of inclined bars with
respect to this mode, as well as to control of diagonal
cracking. The experimental program presented herein
attempted to clarify these points and shed new light on the
understanding of cyclic shear behavior of low slenderness
walls, as well as contribute to the evaluation of the code
provisions for shear design of walls.
All specimens tested failed in a predominantly flexural
mode, characterized by concrete crushing and reinforce-
ment buckling at the confined edges. Moderate diagonal
cracking of the web and sliding at the fixed base were also
observed, but did not significantly affect the failure mode.
Importantly, this was the case not only in walls with a shear
span ratio of 1.5, but also those with a shear span ratio of
1.0. With only one exception (a wall with low slenderness
and high web reinforcement ratio), the tested walls were
able to develop their expected flexural capacity, and they
all showed a satisfactory energy dissipation mechanism.
Pinching of the hysteresis loops caused by horizontal
sliding and bond-slip of vertical bars (rather than by shear
crack opening) was significant in the conventionally rein-
forced specimens, and considerably less in the specimens
with bidiagonal reinforcement. It was found that bidiagonal
bars intersecting at the critical (base) section (a detail first
tested in the framework of this study) were more efficient in
controlling the sliding mode than similar bars intersecting at
approximately lw/2 from the base, particularly in walls with
a shear span ratio of 1.0. Furthermore, bidiagonal reinforce-
ment was found to offer an attractive alternative to current
practice from the economic point of view since, for a lower
quantity of total web reinforcement an improved seismic
performance was achieved.
From the test data discussed herein, it appears that walls
with as = 1.5, although classified as squat in codes like the
EC8, may be flexure dominated (at least for the low rein-
forcement ratios typical in low-rise walls), and the critical
design parameter is not shear reinforcement, but mainly the
ACI Structural Journal/July-August 1999
confinement of the boundary regions. This also appears to be
the case in walls with as = 1.0, at least when the nominal
shear stress does not exceed approximately 0.45 f c MPa
(5.4 f c psi).
ACKNOWLEDGMENTS
The research presented herein was partly funded by the Research
Committee of the University of Thessaloniki, and by the Commission of the
European Communities through Contract CHRXCT930201 (ENESE project
of the HCM program). The authors gratefully acknowledge both supporters.
REFERENCES
1. Penelis, G. G., and Kappos, A. J., Earthquake-Resistant Concrete
Structures, E & FN Spon, London, UK, 1997.
2. ACI Committee 318, Building Code Requirements for Reinforced
Concrete (ACI 318-92) and Commentary (318R-92),American Concrete
Institute, Farmington Hills, Mich., 1992, 347 pp.
3. CEN Technical Committee 250/SC8, Eurocode 8: Earthquake Resis-
tant Design of StructuresPart 1: General Rules (ENV 1998 1-1, 1-2, and
1-3), CEN, Brussels, 1995.
4. Aktan, A. E., and Bertero, V. V., RC Structural Walls: Seismic
Design for Shear, Journal of Structural Engineering, ASCE, V. 111, No.
8, Aug. 1985, pp. 1775-1791.
5. Wood, S. L., Shear Strength of Low-Rise Reinforced Concrete
Walls, ACI Structural Journal, V. 87, No. 1, Jan.-Feb. 1990, pp. 99-107.
6. Farrar, C. R.; Reed, J. W.; and Salmon, M.W., Failure Modes of Low-
Rise Shear Walls, Journal of Energy Engineering, ASCE, V. 119, No. 2,
Aug. 1993, pp. 119-137.
7. Paulay, T.; Priestley, M. J. N.; and Synge, A. J., Ductility in Earth-
quake-Resisting Squat Shearwalls, ACI JOURNAL, Proceedings V. 79, No.
4, July-Aug. 1982, pp. 257-269.
8. Lopes, M. S., and Elnashai, A. S., Behavior of Reinforced Concrete
Walls Subjected to High Cyclic Shear, Proceedings of 9th European
Conference on Earthquake Engineering, Moscow, 1990, V. 6, pp. 80-86.
9. Saatcioglu, M., Hysteretic Shear Response of Low-Rise Walls,
Concrete Shear in Earthquakes, Elsevier, 1991, pp. 105-114.
10. Hidalgo, P. A., and Jordan, R. M., Strength and Energy Dissipation
Characteristics of Reinforced Concrete Walls under Shear Failure, 11th
World Conference on Earthquake Engineering, (Acapulco, Mexico, June
1996), Elsevier, 1996.
11. International Conference of Building Officials, Uniform Building
Code1997 Edition, V. 2: Structural Engineering Design Provisions,
Whittier, Calif., 1997.
12. Booth, E.; Kappos, A. J.; and Park, R., A Critical Review of Interna-
tional Practice on Seismic Design of Reinforced Concrete Buildings,
Structural Engineer, V. 76, No. 11, London, June 1998.
13. Kappos, A. J., Analytical Prediction of Collapse Earthquake for R/C
Buildings: Suggested Methodology, Earthquake Engineering and Struc-
tural Dynamics, V. 20, No. 2, Feb. 1991, pp. 167-176.
14. Salonikios, T.; Tegos, I; Kappos, A; and Penelis, G., Cyclic Shear
Behaviour of Low Slenderness R/C Walls, 5th SECED Conference on
European Seismic Design Practice, Chester, UK, 1995, pp. 293-299.
15. Sittipunt, C., and Wood, S. L., Influence of Web Reinforcement on
Cyclic Response of Structural Walls, ACI Structural Journal, V. 92, No. 6,
Nov.-Dec. 1995, pp. 745-756.
16. Salonikios, T. N.; Kappos, A. J.; Tegos, I. A.; and Penelis, G. G.,
Cyclic Load Behavior of Low-Slenderness Reinforced Concrete Walls:
Failure Modes, Strength Analysis, and Design Implications, ACI Structural
Journal, V. 97, No. 1, Jan.-Feb. 2000, pp. 132-141.
661