You are on page 1of 14

Engineering Structures 24 (2002) 8598

www.elsevier.com/locate/engstruct

An analytical model to predict the inelastic seismic behavior of


shear-wall, reinforced concrete structures
P.A. Hidalgo *, R.M. Jordan, M.P. Martinez
Department of Structural and Geotechnical Engineering, Catholic University of Chile, Av. Vicuna Mackenna 4860, Casilla 306, Santiago, Chile
Received 10 July 2000; received in revised form 5 June 2001; accepted 11 June 2001

Abstract
The development of an analytical model to predict the inelastic seismic response of reinforced concrete shear-wall buildings,
including both the flexural and shear failure modes is presented. The use of shear-wall buildings is quite common in a number of
seismic countries as a result of their successful seismic behavior during past severe earthquakes. The objective of this study has
been to develop a computer model capable of predicting the seismic behavior of shear-wall buildings. Such model would allow
better estimations to be obtained of both the ultimate lateral strength of these buildings as well as their inelastic deformation demand
under severe ground motions. Such information may be used in the implementation of performance-based design procedures, and
to improve present code design procedures. To fulfill this objective, a shear failure mode model based on experimental results has
been added to the computer program larz. This paper discusses the most relevant problems and solutions devised during the
development of this model. Validation of the model proposed to predict the inelastic seismic response of shear-wall structures was
carried out by comparing its results with the actual response of two real buildings during the March 3, 1985 Chilean earthquake. In
spite of the fact that the model is two-dimensional and, hence, it ignores the torsional response, the results obtained are satisfactory.
2001 Elsevier Science Ltd. All rights reserved.
Keywords: Shear-wall buildings; Reinforced concrete buildings; Inelastic behavior; Shear failure model

1. Introduction
Properly designed multistory R/C shear-wall buildings
should behave in a ductile flexural manner when subjected to severe earthquake ground motions. Consequently, design forces are usually much smaller than
those required to design a structural system without the
characteristics of ductility and toughness typical of
buildings with predominant flexural failure mode.
Nevertheless, there are cases where this ductile failure
mode may not be achieved due to the large flexural
strength as compared with the shear strength of the
walls. In such cases, an undesired shear failure mode is
likely to develop. This may be the case of structural systems that have a large wall area relative to the floor plan
area. This situation may also happen in shear walls

* Corresponding author. Tel.: +56-2-686-4207; fax: +56-2-6864243.


E-mail address: phidalgo@ing.puc.cl (P.A. Hidalgo).

coupled by stiff lintels that may induce bending moment


to shear force ratios in the wall too small relative to the
length of the wall. On the other hand, even in the case
that ultimate strength is controlled by a ductile flexural
behavior of the shear walls, structural damage in the
form of mild or extensive shear cracking may affect the
objectives of performance-based design. Several
examples of these situations have been found in the
response of R/C shear-wall buildings after severe earthquakes like the March 3, 1985 Chilean earthquake.
The use of shear-wall buildings is quite common in
some earthquake-prone countries such as Chile; their
seismic behavior has been successful during past severe
earthquakes, both, from a serviceability as well as a
safety standpoint [1]. Therefore, their use has been recommended in earthquake-resistant design [2] as long as
its true behavior is included in building modeling.
Consequently, the objective of this study has been to
develop a computer model capable of predicting the seismic behavior of such buildings. The model proposed
allows better estimations to be obtained of both the ulti-

0141-0296/02/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 6 1 - X

86

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

mate lateral strength as well as the inelastic deformation


demand of shear-wall buildings under severe ground
motions. The use of this information helps in the
implementation of performance-based design procedures
and the improvement of code design procedures.
A model capable of predicting the seismic behavior
of this type of structures has to include the possibility
of developing flexural as well as shear failure modes in
the shear walls. Most research studies have addressed
the problem of including flexural failure modes in the
modeling of structural shear-wall elements. For instance,
Saiidi and Sozen [3] studied several hysteretic models
associated to the flexural failure mode; in one of them,
pinching effects, typical of shear behavior, were
included and labeled as the SINA model. Later, Linde
and Bachmann [4] developed an element to represent the
inelastic seismic behavior of shear walls controlled by
flexure, with a modest influence of shear cracking in the
hysteretic response.
The model developed in this study has been
implemented in the larz computer program [5]. The
shear-wall model has been modified to allow the wall to
develop a shear failure mode when its lateral shear
strength is smaller than the strength required to develop
a flexural failure mode. The shear model characteristics
are determined from experimental results obtained from
cyclic test of shear walls and beams, as described below.
larz is a standard inelastic analysis software for
reinforced concrete buildings, like drain or other available software, and was developed by Saiidi [5] twenty
years ago. Nevertheless, the available software do not
recognize the possibility of developing a shear mode of
failure in the structural elements, which precludes shear
cracking or shear failure in these elements.
The definition of the uncoupled hysteresis models for
shear and flexure described previously allows for the
update of the flexural and shear tangent stiffnesses of
any wall element in a step by step integration procedure.
A wall element as shown in Fig. 1, defined at each story,
is divided into several sub-elements in height, each of
them having the lateral and vertical degrees of freedom

Fig. 1.

Typical wall element.

Fig. 2. Hysteretic model for flexural mode of failure (beams, columns, shear walls).

at each node and the end rotations. By assuming a constant inter-story shear and a linear variation of bending
moments along the height of the wall, the shear and
moment at the center of each sub-element are computed.
Hence, the tangent flexural stiffness EI and shear stiffness AG are determined from the moment-curvature and
the shear-deformation relationship for the wall as defined
in Figs. 2 and 3. The tangent stiffness matrix for each
sub-element is then computed and assembled into the
global stiffness matrix of the wall. By static condensation of the interior degrees of freedom associated with
the wall sub-elements, the tangent stiffness matrix
related to the external degrees of freedom is computed.
Thus, only the horizontal and vertical displacements and
the rotation at the ends of the wall are used to assemble
this element into the structural model.
The paper discusses the development and implementation of this analytical model. The model proposed is
validated by evaluating the results predicted by the
model for two buildings that underwent shear cracking
during the 1985 Chilean earthquake. The true horizontal
acceleration components recorded at nearby sites and

Fig. 3.

Hysteretic model for shear mode of failure (shear walls).

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

similar soil conditions as for the buildings were used as


input for the computer model.

2. Model for flexural failure mode


The SINA hysteresis model implemented in the larz
computer program (Fig. 2) was adopted in this study to
model the non-linear flexural behavior and the momentcurvature hysteretic relations for wall elements. As
shown in Fig. 2, pinching effects and stiffness and
strength reductions due to repeated cycles at the same
deformation level were not implemented in the model
for flexural behavior.
The model operates on a primary Mf envelope curve
consisting of four linear segments for positive and negative bending as shown in Fig. 2. The primary curves need
not be symmetric about the origin, but a single straight
line must be specified for moments below the cracking
moment in both directions (line CC in Fig. 2). Points
Y and U (and Y and U) correspond to moments associated with first yielding and a concrete compressive
strain of cu=0.003, respectively. An horizontal line following point U is assumed, and collapse is defined by
a maximum curvature fmax, associated to a concrete
compressive strain of cmax=0.01.
For moments below the cracking level, loading and
unloading follow the primary curve. For moments above
the cracking moment, unloading follows a line connecting the unloading point with the cracking point in
the other direction (line PC in Fig. 2). If the yield
moment is exceeded and unloading takes place at point
P1, the slope of the unloading branch P1P2 is taken as

KunKyc

fy
fm

(1)

where fm is the maximum curvature attained in the loading direction and Kyc the slope of the line connecting
the yield point in the loading direction with the cracking
point in the opposite direction. The exponent a controls
the slope of the unloading branch after yielding, and was
taken equal to 0.5 as suggested by Saiidi and Sozen [3].
A detailed description of all the hysteresis rules can be
found elsewhere [3].
The bending moment and curvature values were
defined using the standard theory for reinforced concrete
elements; both the boundary reinforcement and the distributed vertical reinforcement are taken into account in
defining the primary Mf curve for wall elements.
Further, the axial load force values due to gravity,
assumed to remain constant throughout the seismic
response, are considered in the calculation of moment
and curvature associated with points C, Y and U of the
primary curve. This constitutes an approximation for
shear walls coupled by spandrel beams, since these

87

beams may develop significant seismic shear forces that


induce variable axial loads on the walls, but the resulting
error in the axial loads becomes smaller as the flexural
strength of the coupling elements decreases.
The evaluation of the bending moment for the collapse
point defined by fmax , assumes the same compressive
stress block in the concrete that was used for point U.
This is obviously an approximation since it always yields
to fmax=3.33fu in Fig. 2. This assumption is verified by
computing the moment-curvature relationship for some
of the walls using a realistic stressstrain curve for the
concrete. In all cases a slightly larger value of fmax is
obtained by using the more exact method. Nevertheless, when the model was used in this study to predict
the inelastic seismic behavior of real buildings, the
maximum curvature never exceeded the value of fu.
3. Model for shear failure mode
Shear dominated behavior was also modeled using the
SINA hysteresis model as shown in Fig. 3. Pinching
effects and strength reduction due to repeated cycles at
the same deformation level were now implemented in
the hysteresis model. The model for the shear failure
mode assumes independence of the shear strength of
walls on both the bending moment and the axial force
present in the wall. This is also an approximation, but
to neglect interaction between shear and axial force is
consistent with the current ACI design provisions for
walls [6].
The model was initially developed for squat shear
walls with an aspect ratio M/(VLw) of 1.0 or smaller,
where M is the bending moment at the base of the wall,
V the shear force, and Lw the length of the wall. It was
then extended to the case of slender shear walls with
aspect ratio larger than 1.0, as explained later. In Fig. 3,
point C represents the point where a change in the slope
of the envelope of the loaddisplacement relationship is
experimentally observed; the new value of the stiffness
of the specimens is about 60% of the initial stiffness.
Point C was generally very close to the point at which
the first diagonal crack from corner to corner of the walls
was developed during the tests. Point Y corresponds to
the largest value of shear load attained during the test,
while point U may be associated with the ultimate condition under which the element may still be considered
as an effective part of the resisting mechanism of the
structure. The definitions of points C, Y and U in the
envelope curve (Fig. 3) are based on the experimental
results obtained from the cyclic test of 26 full scale,
shear wall specimens. All these specimens were
designed to exhibit a shear mode of failure and had
aspect ratios M/(VLw) between 0.35 and 1.0. The loading
sequence of each test consisted of sets of two cycles at
a given displacement amplitude, which increased gradu-

88

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

ally and followed a sequence determined by the aspect


ratio of the specimen. All tests were finished when lateral strength of the specimen had dropped to 75% of the
maximum value, approximately. More details of this test
program may be found elsewhere [7]. On the other hand,
the model characteristics for slender shear walls were
obtained from test results of reinforced concrete beams
[8,9]. The main difference between the shear behavior
of beams and that of squat walls is that strength of beams
is lost after point Y (Fig. 3) has been attained.
The actual slope of branch YU obtained from tests of
squat walls showed decreasing shear strength with
increasing displacement. This fact leads to a problem in
the software that cannot handle structures with a degrading stiffness and, hence, with negative semi-definite tangent stiffness matrices occurring at some points in the
response. For this reason, the branch YU of the model
was taken as nearly constant, but keeping the actual ultimate displacement dult obtained from experimental
results. Once this displacement is eventually attained in
a wall during the response history analysis, the element
is removed from the structure and the stiffness matrix is
re-evaluated. Fig. 3 also shows the eight hysteresis rules
for this model, that also follow the SINA model [3]. In
the shear failure model, a crack closing point (dcg, Vcg)
is defined to account for the pinching effect always
present in the hysteretic behavior after a shear crack has
developed. In order to consider the shear strength
reduction due to repeated cycles at the same deformation
level observed in the test specimens, a strength reduction
factor of 0.15 was adopted, as shown by point E in Fig.
3, i.e. after unloading takes place from point D in Fig.
3, the target point for subsequent load cycles is set vertically under D at a shear V=0.85VD. The model for slender walls is similar to that for squat walls, with the only
difference that points Y and U (Fig. 3) become the same
point in the tests of beams when a shear failure is attained.
Figs. 411 show comparisons between experimental

Fig. 5.

Fig. 6.

Shear model to estimate drift at maximum strength.

Shear model to estimate drift at ultimate strength.

Fig. 7. Comparison of VC (ACI) with experimental cracking strength


of squat walls.

Fig. 4. Shear model to estimate drift at first cracking.

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 8. Comparison of shear model to estimate cracking strength with


experimental values.

Fig. 9. Comparison of Vu (Arakawa) with experimental maximum


strength of squat walls.

Fig. 10. Comparison of Vu (New Zealand) with experimental


maximum strength for slender walls.

89

Fig. 11. Comparison of shear model to estimate maximum strength


with experimental values.

results and model envelope curve definitions for both


squat and slender shear-walls under a shear failure mode.
Experimental results for walls with aspect ratios less
than or equal to one were obtained from the tests performed by Hidalgo et al. [7], while those for elements with
aspect ratios larger than one were taken from the test of
beams [8,9]. Fig. 4 shows the definition of the drift ratio
at cracking point C (Fig. 3), DRcr as a function of the
aspect ratio of the wall. Similarly, Figs. 5 and 6 show
the estimations of the drift ratios at point Y, DRu, and
at point U, DRult, respectively. Experimental results for
beams may be obtained only for DRcr and DRu since
there is no residual shear strength after point Y (Fig. 3);
therefore, as shown in Fig. 6, DRult was assumed to be
0.014 for walls with aspect ratio larger than one.
Same information as before but for strength is shown
in Figs. 7 and 8 for shear strength at cracking point C,
Vcr. Different models were adopted for squat and slender
shear walls. Fig. 7 shows the comparison between the
experimental value Vcr for squat walls and the value of
Vc proposed in the ACI Code [6] to estimate the contribution of concrete to the shear strength of such walls.
As shown in Fig. 7, the Vc value prescribed in ACI for
shear walls underestimates the Vcr value obtained in the
tests conducted earlier [7]. In order to improve the correlation between Vc (ACI) and Vcr for walls with aspect
ratio less than or equal to one, the value of Vc has to be
amplified by a factor of 1.12. For more slender walls no
correction of the ACI Vc value is necessary, as shown
in Fig. 8.
Likewise as for point C, Figs. 911 show the estimation of maximum shear strength Vu at point Y (Fig.
3). The best correlation with experimental results for
squat walls (Fig. 9) was obtained for the contribution of
concrete Vc as proposed by Arakawa for beams, with the
contribution of shear reinforcement as proposed in the
ACI Code [7]. As before with the cracking value, the
correlation of the Vc value proposed by Arakawa and the

90

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 12.

Structural floor plans of buildings studied. (Dimensions in centimeters). (a) Villa Real building. (b) Sermena building.

experimental values obtained from the test of squat walls


[7] may be improved by multiplying Vc by a factor of
1.41. The estimation of Vu for slender walls was taken
as the ACI formula to predict the shear strength of
beams, but using as contribution of concrete the
expression proposed by Paulay and Priestley [10] (Fig.
10). This model has been labeled as N.Z. model in Fig.
11 that shows the complete correlation between model
values and experimental results. Finally, the value of Vult
shown in Fig. 3 was arbitrarily taken as 1.01 times Vu
to avoid instability in the numerical model analysis due
to negative stiffness.
4. Prediction of inelastic seismic behavior of real
buildings
The models of flexural and shear failure modes were
implemented into the larz computer program, such that

either a flexural or a shear failure mode could be


developed in each of the walls of the structure under a
severe ground motion. The failure mode predicted by the
program is based on the fact that the flexural response
of a particular wall element follows the hysteresis curve
for flexure (Mf curve, see Fig. 2) while the shear
response follows the hysteresis curve for shear (Vd
curve, see Fig. 3). Therefore, at each integration step,
the curvature f and shear displacement d demands associated with the calculated values of M and V are obtained
directly from the hysteresis curves, making possible the
development of either a flexural or a shear failure mode.
Since the shear model parameters depend on the aspect
ratio M/(VLw), the value of this ratio for each wall was
taken as the inter-story height h divided by the length
of the wall Lw. This assumption is based on time-history
results of M/(VLw) which was found to be closer to h/Lw,
than to the total wall height divided by Lw. The analytical

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 13. Acceleration record and base shear response, Villa Real
building. (a) Vin a del Mar acceleration record, S20W component. (b)
Base shear, X-direction analysis. (c) Base shear, Y-direction analysis.

model developed so far is two-dimensional, but a threedimensional version is currently being investigated.
The model was validated by predicting the inelastic
seismic behavior of two real buildings that developed
inelastic behavior and shear cracking during the March
3, 1985 Chilean earthquake. The 10-story plus a basement Villa Real building, designed in 1981, and the 7story plus a basement Sermena building, designed in
1971, are presented in Fig. 12(a) and (b), respectively.
Both are shear wall reinforced concrete structural systems located in the coastal cities of Vin a del Mar and
Valparaiso, respectively, about 30 km from the epicenter
of the 1985 earthquake. The distribution in plan of shear
walls for both buildings is quite symmetric with respect
to horizontal axes X and Y. The typical ratios of shear
wall area to floor plan area are 0.027 in the X-direction
and 0.033 in the Y-direction, respectively, for the Villa

91

Fig. 14. Acceleration record and base shear response, Sermena building. (a) El Almendral acceleration record, N50E component. (b) Base
shear, X-direction analysis. (c) Base shear, Y-direction analysis.

Real building, and 0.008 and 0.015, in the X- and Ydirections, respectively, for the Sermena building. The
latter, despite being three stories smaller than the Villa
Real building, has lower values of this ratio, which is
consistent with the more extensive shear cracking
developed by this building during the 1985 earthquake.
The earthquake response histories along both principal
horizontal directions of the Villa Real building were
obtained using the S20W component of the acceleration
record obtained in downtown Vin a del Mar. The site of
the record was at about 1000 m from the site of the
building. In the case of the Sermena building, the N50E
component of the El Almendral acceleration record was
used for the analysis of the building. This was recorded
at about 2600 m from the building. In both cases, the

92

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 15. Hysteretic response examples, Villa Real building. (a) Hysteretic shear behavior wall M5, axis 2, story 2. (b) Hysteretic flexural
behavior at base of wall M5, axis 2, story 2.

Fig. 16. Hysteretic response examples, Sermena building. (a) Hysteretic shear behavior wall M3, axis A, story 1. (b) Hysteretic flexural
behavior at base of wall M3, axis A, story 1.

soil at the recording site had the same characteristics as


those of the corresponding building site.
Structural and architectural drawings of both buildings, strength characteristics of both concrete and reinforcing steel obtained after the earthquake, and a detailed
documentation of damage and cracking patterns are
available to conduct this study [11,12]. In order to ascertain the validity of structural model assumptions associated with the elastic behavior of the buildings, the computed first mode translational periods of vibration are
correlated with the values measured after the earthquake.
For the Villa Real building the nominal periods are
0.74 s (0.71) and 0.53 s (0.51), in the X- and Y-directions, respectively; and for the Sermena building 0.48 s
(0.45) and 0.42 s (0.41), in these directions, respectively.
The numbers in parenthesis correspond to the periods
measured experimentally.
Let us consider first the prediction of base shear
obtained with the computer program along each of the
principal directions of resistance as shown in Figs. 13
and 14. It is interesting to note that the original design
base shear for the Villa Real building was 2390 kN
(11.5% of total weight), using the current Chilean Code
when the design was carried out. Maximum values predicted for the X- and Y-directions of the Villa Real build-

ing, were 5520 kN (26.6% of weight) and 7020 kN


(33.9% of weight), respectively, that is, 2.3 and 2.9 times
larger than the design base shear. Likewise for the Sermena building, the design base shear was 2830 kN (12%
of weight), while predicted maximum values were
8370 kN (35.5% of weight) and 7310 kN (31% of
weight), in the X- and Y-directions, respectively, i.e. 3.0
and 2.6 times the design base shear. Maximum values
of inter-story drift ratios predicted for these buildings
were 0.0073 and 0.0051, in the X- and Y-directions,
respectively, for the Villa Real building. Corresponding
values for the Sermena building were 0.0048 and
0.0044, respectively.
Examples of hysteretic behavior, both in flexure and
shear, developed at some critical shear-wall segments are
shown in Fig. 15 for the Villa Real building and in Fig.
16 for the Sermena building. Fig. 15(a) and (b) shows
the hysteresis loops developed in wall M5 of the Villa
Real building, shown in Fig. 12(a), both for the shear
behavior and the flexural behavior, respectively. It can
be observed from Fig. 15(a) that shear behavior almost
reached the cracking point, with very limited energy dissipated through this type of behavior. On the other hand,
the formation of a plastic hinge is apparent from Fig.
15(b), with larger energy dissipation through the flexural

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 17.

93

Experimental results for Specimen 23 (M/VLw=0.69, rh=0.25%, rv=0). (a) Actual cracking patterns. (b) Hysteretic shear behavior.

behavior than through the shear behavior. The same


information is shown in Fig. 16 for wall M3 of the Sermena building (Fig. 12(b)); in this case, more energy is
dissipated, through both the shear and the flexural
behavior. It is worthwhile to note that minimum values
of the aspect ratio M/(VLw) of walls that experienced
shear cracking in the Villa Real building, at any instant
of the time-history response, reached 0.72 and 0.60, for
the X- and Y-directions, respectively.
The capability of the model to predict the inelastic

seismic response was evaluated by comparing the cracking that would be obtained from the predicted internal
forces in each of the structural elements with the actual
cracking patterns due to flexure and shear. The cracks
due to flexure are sometimes difficult to observe after
an earthquake, but those due to shear may be easily recognizable. In order to correlate the predicted internal shear
forces with cracking patterns, the experimental information obtained from a previous research program was
used, [7]. Figs. 17 and 18 show two examples of the

94

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

Fig. 18.

Experimental results for Specimen 27 (M/VLw=0.50, rh=0.25%, rv=0). (a) Actual cracking patterns. (b) Hysteretic shear behavior.

relation between the hysteretic shear behavior of shear


walls and the resulting cracking patterns. A more
detailed analysis of the results of this comparison for
each building follows.
4.1. Villa Real building
The inelastic response of the structure was determined
by using the acceleration record indicated above, first as
base acceleration history along the X-direction, and then
along the Y-direction. To evaluate the inelastic shear

behavior of walls, the attention was mainly focused on


the walls that developed shear cracking during the 1985
earthquake. In order to validate the results from the
analysis with the actual performance of the structure, the
experimental results of cracking patterns, like those
presented in Fig. 17, are used as follows. Consider for
instance wall M14 (Fig. 19) which shows a maximum
shear of 58% of the cracking shear (Fig. 19(b)). Therefore, this wall has attained a point that is below point 1
in Fig. 17 and, hence, should present less cracking than
pattern 1 in that figure. This coincides with the actual

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

95

Fig. 19. Axis D, Villa Real Building. (a) Observed cracking pattern.
(b) Quality of shear behavior prediction.

Fig. 20. Axis E, Villa Real Building. (a) Observed cracking pattern.
(b) Quality of shear behavior prediction.

behavior exhibited by wall M14 (Fig. 19(a)). On the


other hand, wall M34 in Fig. 20 attained a maximum
shear of 1.04 times the cracking value, i.e. it is slightly
above cracking, say point 3, in the force-deformation
loop of Fig. 17. The cracking associated with this level
of shear (cracking pattern 3 in Fig. 17) is similar to that
observed in the structure after the earthquake (Fig. 20(a))
and, thus, there is a good correlation between prediction
of the model and the actual behavior. Similar analysis
is performed for each wall with actual cracking and the
results presented in Figs. 1922. The ratios of maximum
shear obtained from the inelastic, dynamic analysis and
Vcr and Vu defined for each wall are included in Figs.
19(b)22(b). Such a procedure would enable us to validate the analytical model developed in this study.
Figs. 19 and 20 illustrate the results of this damage
identification procedure for walls in two of the X-direction resisting planes. Given the uncertainties involved in
the analytical prediction of the inelastic shear behavior,
as will be discussed below, the quality of the prediction
is only rated in general terms as adequate, too high,
or too low. A quality of prediction rated as adequate,
means that shear cracking observed in the walls after
the 1985 earthquake is similar to that observed in the
experimental program for the same level of shear force.
On the other hand, too high or too low means that predicted shear force is above or below the values consistent

with the actual cracking pattern observed in the wall,


respectively. Figs. 21 and 22 do the same for the walls
in two of the Y-direction resisting planes. The rating of
the quality of the prediction has to consider several
aspects that do not allow the prediction of the cracking
pattern to be more exact. First, the actual base acceleration history experienced by this particular building is
not exactly known; the actual seismic excitation was bidirectional and not unidirectional as it has been assumed,
and the vertical seismic action has not been included in
the analysis. Secondly, the experimental information that
correlates the hysteretic shear force with the resulting
cracking patterns for shear walls generally shows a significant degree of scattering. Thirdly, the strong assumptions related to the definition of the structural model.
And finally, the fact that this analytical model is still
two-dimensional, and torsional effects have not been
considered to predict the earthquake response of the
structure.
The analytical model also showed the development of
bending moments larger than the yield moment in coupling beams and in some of the walls, as shown in the
example of Fig. 15(b). The prediction of bending
moments and the development of plastic hinges in the
coupling beams have a reasonably good correlation with
the actual amount of flexural cracking experienced by

96

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

this building. Nevertheless, the actual flexural cracking


is sometimes difficult to observe after an earthquake
event, particularly if yielding has been moderate.
Besides, the emphasis of the prediction capability of this
model has been placed in the inelastic shear behavior
rather than in the flexural behavior.
Generally speaking and based on the results presented
in Figs. 19(b)22(b), it may be stated that this analytical
model has been able to reproduce reasonably well the
inelastic shear behavior experienced by the Villa Real
building during the 1985 earthquake.
4.2. Sermena building
The procedure followed for this building was the same
as for the Villa Real building. Figs. 23 and 24 illustrate
the results of this procedure for walls in two of the Xdirection resisting planes, while Figs. 25 and 26 illustrate
the results of this procedure for walls in four of the Ydirection resisting planes. In general terms, limitations
associated with the analytical prediction of the earthquake inelastic behavior of the structure and quality of

Fig. 21. Axis 2, Villa Real Building. (a) Observed cracking pattern.
(b) Quality of shear behavior prediction.

Fig. 22. Axis 5, Villa Real Building. (a) Observed cracking pattern.
(b) Quality of shear behavior prediction.

Fig. 23. Axis A, Sermena Building. (a) Observed cracking pattern.


(b) Quality of shear behavior prediction.

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

97

Fig. 24. Axis B, Sermena Building. (a) Observed cracking pattern.


(b) Quality of shear behavior prediction.
Fig. 26. Axis 5b, Sermena Building. (a) Observed cracking pattern.
(b) Quality of shear behavior prediction.

this difference between the actual behavior of the building and the prediction of the model may be one of the
following: (1) the analytical model did not include the
structure for the elevator shaft adjacent to axis A; (2)
there is evidence of existing cracking in the first story
shear walls along axis A, prior to the 1985 earthquake
[12]; (3) some unknown facts related to the construction
of the building; and (4) inelastic torsion of the building
plan as reported elsewhere [13].

5. Conclusions

Fig. 25. Axes 2, 3 and 5, Sermena building. (a) Observed cracking


pattern Axis 2. (b) Observed cracking pattern Axis 3. (c) Observed
cracking pattern Axis 5. (d) Quality of shear behavior prediction, Axes
2, 3 and 5.

the results are about the same as those found for the
Villa Real building. The distinctive feature of the results
obtained for the Sermena building relates to the results
obtained for the seismic analysis in the X-direction. The
analytical model assumes that the structure is longitudinally symmetric, and therefore, predicts the same inelastic behavior for resisting planes along both axes A and
D (Fig. 12(b)). Nevertheless, the resisting plane along
axis D showed much less damage than that along axis
A and was practically free of shear cracks in the first
story walls after the earthquake. The reasons to explain

This study has proposed and implemented a model


to include the shear failure mode for walls in available
computer programs. The model developed in this study
is a macro-model validated with the experimental results
of cyclic tests of shear walls. Though this model may
still be refined, it may be used to predict the inelastic
seismic behavior of reinforced concrete, building structures as long as they have nominally symmetric structural plans, thus providing a useful tool to estimate seismic demands on this type of buildings.

Acknowledgements
This research was funded by the Chilean Superior
Council for Technological Development under Grant
No. 1980953. The sponsorship of this institution is gratefully acknowledged. The authors also thank Professor
Juan C. De la Llera for his valuable comments and con-

98

P.A. Hidalgo et al. / Engineering Structures 24 (2002) 8598

tributions to the writing of the manuscript, and Professor


P. Bonelli from the Universidad Tecnica F. Santa Maria,
who provided the structural details and damage documentation of the buildings used in this study.

References
[1] Wood S. Performance of reinforced concrete buildings during the
1985 Chile earthquake: implications for the design of structural
walls. Earthquake Spec 1991;7(4):60738.
[2] Sozen M. Earthquake response of buildings with robust walls. In:
Proceedings of the Fifth Chilean Conference on Seismology and
Earthquake Engineering, Santiago (Chile), 1989.
[3] Saiidi M, Sozen MA. Simple and complex models for nonlinear
seismic response of reinforced concrete structures. Structural
Research Series No. 465. University of Illinois, Urbana (Illinois),
USA, 1979.
[4] Linde P, Bachmann H. Numerical modeling and design of earthquake-resistant walls. Earthquake Eng Struct Dynamics
1994;23:133150.
[5] Saiidi M. Users manual for the LARZ family. Structural
Research Series No. 466. University of Illinois, Urbana (Illinois),
USA, 1979.

[6] American Concrete Institute. Building code requirements for


reinforced concrete, ACI 318-99. Detroit (Michigan), USA, 1999.
[7] Hidalgo P, Jordan R, Ledezma C. Experimental study of
reinforced concrete walls under shear failure. In: Proceedings of
the Sixth US National Conference on Earthquake Engineering,
Paper No. 297, Seattle (Washington), 1998.
[8] Bresler B, Scordelis AC. Shear strength of reinforced concrete
beams Series III. Report No. 65-10, Structures and Materials
Research. University of California, Berkeley (California), USA,
1966.
[9] ASCE-ACI Task Committee 426 on Shear Diagonal Tension. The
shear strength of reinforced concrete members. J Struct Division
ASCE 1973;99:ST6.
[10] Paulay T, Priestley M. Seismic design of reinforced concrete and
masonry buildings. New York: Wiley, 1992.
[11] Gajardo J. Estudio de un edificio levemente fisurado por el terremoto del 3 de Marzo de 1985. Civil Engineering Thesis. Universidad Tecnica F. Santa Maria, Valparaiso (Chile), 1987.
[12] Yan es E. Estudio del edificio del Consultorio Externo del
Servicio Nacional de Salud de Valparaiso, dan ado en el sismo de
Marzo de 1985. Civil Engineering Thesis. Universidad Tecnica F.
Santa Maria, Valparaiso (Chile), 1987.
[13] De La Llera J, Chopra A. Evaluation of seismic code provisions
using strong-motion building records from the 1994 Northridge
earthquake. Report No. UCB/EERC-97/16, University of California, Berkeley, 1998.

You might also like