EARTHQUAKE RESPONSE OF BUILDINGS WITH ROBUST WALLS
Mete A. Sozen
Department of Civil Engineering, University of Ilinois
Urbana,IL 61801, USA
SUMMARY
Assimple device is presented for organizing the critical structural parameters of
buildings with robust wells to relate drift response to intensity of ground motion in
an effort to answer design questions about the needed amount of detail and wall.
RESUMEN
Se presenta una herramienta sencilla para organizar los parametros estructurales
ctiticos de edificios con muros robustos y correlacionar las derivas de respuesta con
la intensidad de movimientos sismicos en un esfuerzo para resolver aspectos de dis-
eno tales como la cantidad necesaria de detalles de refuerzo y de muros.
Fifth Chilean Conference on Earthquake Engineering, Santiago, Chile, August, 1989Introduction
This note is concerned with the observed behavior of a class of reinforced concrete
buildings in Vina del Mar during the March 1985 earthquake. The buildings of inter-
est include 322 structures, documented by Riddell, Wood, and de la Llera (1), A.
dominant characteristic of these reinforced concrete buildings, which will be
referred to as the Vina del Mar inventory of buildings or simply as the inventory, was
the generous use of structural walls to resist gravity loads and to stiffen the buildings
for lateral loading. The typical value of the wall index, ratio of total cross-sectional
area of wall oriented in one direction to the floor area, was 3 %, As summarized by
Stark (2), calculated and measured
proximately N/20 (where N is the number of stories). Although there were cases of
ial periods for the inventory were ap-
severe damage (3,4) in this group of buildings, their overall behavioral record during
the earthquake was admirable (1).
The object of this paper is to assemble a simple, if approximate, analog for the
dynamic response of the inventory buildings using elementary information from
mechanics, ground-motion response and structural behavior, in order to identify the
critical parameters affecting their response to strong ground motion.
‘The response of an individual building to an earthquake is the result of a compli-
cated set of interactions among the nature of the energy source, the geophysics of
the region, soil conditions at the site, the foundation, and the incremental stiff-
ness/strength characteristics of the building which will vary with displacement and
may vary with height. In defining a single and simple analog for a large inventory of
buildings, the intent is not to suggest that their individual responses are essentially
identical or to calculate post-facto what they might have experienced during the
1985 earthquake but to show that certain dominant structural properties would have
led to the observed results for a given type of strong motion.Measured Dynamic Response of A Structural Wall Model
Results of a particular earthquake-simulation test (5) will be used to provide a
perspective of acceleration and displacement responses of buildings with slender
walls, Here "slender" is used in the sense that the lateral displacement is generated
primarily by flexural and not shear deformations. The inventory buildings would be
in this class.
‘The small-scale structure shown in Fig.1 was subjected to a strong motion at its base
(6)in one horizontal direction parallel to the planes of the walls. To excite the small-
scale structure, the actual base motion was "compressed! by a scale of 2.5, but the
times shown in the figure refer to"real" time. (A duration of 30 sec in real time cor-
tesponded to a test with a duration of 12 sec). The small-scale reinforced concrete
walls carried story weights of 1000 Ib (453 kg) placed at equal intervals of nine in.
coe (35 cm) to result‘in a calculated
“real" time period of 0.5 sec (actual
period of 0.2 sec).
In the particular test reported (Test
tree DI-1, Reference 4) the structure ex-
perienced extensive cracking and
= yielding, but the acceleration his-
tories at the three levels shown in
Fig. 2 accomodate interpretation as
combination of responses in several
modes of a structure responding
; - linearly. This interpretation is rein-
Fig. 1 Ten-Story Test Structure (Arstizabal, Ref) s.-04 bythe observation thatthe
e response appeared to be dominatedby one mode at Level 8. Level 8 corresponded closely to the node point of the
second node.
° 5 10 5 20 25 Jo
a ‘er 10
ns
J
os
ew! 6
as
°
S
ae
Law! 6
; os
a
Bose
as
os if |
o 5 10 15 20 25 30
“Real” Time, sec.
Fig.2 Acceleration Response Histories
Figure 3 shows the measured displacement histories at levels 6, 8, and 10 of the test
structure, The waveforms of the displacement records strengthen the impression
that response of the structure, which had experienced yielding, could be simulated
analytically by a linear analog with a longer period than that of the initial uncracked
structure, The response observed would fit that of a damped structure with an effec-
tive period of approximately 1 sec. which, in this case, was twice the calculated initial
period.
For the structure considered, the relative simplicity of the displacement response
records with respect to the acceleration records may be rationalized given that the
wall responds as a linear system with mode shapes that are reasonably well
preserved during strong motion. From the view point of the designer, this providesan opportunity. Even though conventional design procedures focus on forces, the
more important issue of design for ordinary building structures is to control the dis-
placement. As long as displacement is controlled, accelerations or local forces are
unimportant given thet the probability of a shear or bar anchorage failure i$ reduced
by proper detailing of the structure based on static limits of internal forces, as re-
quired by most building codes(6).
10 15 Fd 5 x»
10 Lave! 10
10 WV Wer V v
10 h Leva 8
10) WVy
10) Level 6
10) Wwyyy vy
az
§ Bose
az
o $ 10 18 7 2 ”
“peer” Teme, sec.
Dnt Rati,
Fig3 Displacement Response Histories
‘The displacement histories (Fig. 3) suggest that, short of a complete nonlinear
response analysis, the maximum displacement may be estimated iteratively from a
response-spectrum analysis using a linear model and finding compatible values of
matching effective stiffness, damping, and strength . Although this approach is
plausible and appears realistic except for its being limited to the frequency domain,
it does not lend itself conveniently to use for making design decisions. Considering
the lack of precision in a mandated earthquake intensity, the labor involved in cal-
culating a precise displacement for an approximate intensity violates the principle of
Sproportionality. A simpler if heuristic option is the method proposed by Shimazaki
(7). From a study of experimental and computational evaluations of nonlinear
response of reinforced concrete structures, Shimazaki concluded that lateral-dis-
placement response for moderate-rise structures calculated using an effective stiff-
ness based on half of the initial stiffness (effective period = V2T, where Ti is the
initial calculated period) and the displacement response spectrum determined for a
damping factor of 0.02 would provide a safe bound to the nonlinear drift. Shimazaki
limited his procedure to buildings which satisfied the condition
SR+TR =< 10
where SR = Base Shear Strength/Linear Base Shear Response
TR = Effective Period of Building/Characteristie Period for Ground Motion
‘The base shear strength refers to the calculated limit to the yield strength of the
building for an assumed linear mode shape, provided that the base force so caleu-
lated does not exceed the shear strength of the structure. The base shear response is
that calculated for the lowest translational mode on the basis of a response spectrum
(damping factor = 0.02). The effective period for the structure is that for the lowest
translational mode based on one half the initial uncracked stiffness, Shimazaki
defined the characteristic period for ground motion to be the period at which the cal-
culated energy response ceased to increase with increase in period, In the domain of
practical applications, itis sufficient to recognize that the characteristic period is ap-
proximately 0.6 sec for stiff and 1.2 for soft soil. Clearly, the procedure is applicable
to buildings of which mode shapes do not change drastically because of local failure.
All calculations and experiments supporting the conclusions referred to regular
structures with negligible torsional response.Mean Drift Response of Buildings with Robust Walls
‘The following device is developed to assimilate part of the experience of the Vina
~ Wall Index
—Area of Wall Section
PP "Tributary Floor Areo
Fig.4 Idealized Floor Plan
del Mar inventory of buildings. Components of knowledge from elementary
mechanics, ground-motion response, and structural behavior are assembled to
develop an approximate analog to help identify the major parameters and sen-
sitivities. The device may be said to comprise four components.
The first component comes from mechanics and is the definition of the fundamental
period in terms of “smeared! properties. Dynamic characteristics of a slender can-
tilever beam of uniform thickness and mass represent a satisfactory analog for es-
timating the initial stiffness of a regular reinforced concrete building if ts lateralresistance is dominated by a wall of uniform dimensions, The lowest translational
period of such a wall is (10)
2a ut
354 ET _
T = lowest translational period
ye = mass per unit height
H = total height
I = moment of inertia of cross section
Referring to Fig. 4, Eq. 1 may be rewritten as
H
T=6.18*—+N (2)
“5° [Foe (2)
w = unit weight of building (in terms of unit floor area)
p = wall index, ratio of wall area in a given direction to tributary floor area
g = acceleration of gravity
h = typical story height
N = number of stories
The second component of the device is a definition of the ground motion in terms of
a displacement response spectrum, Because it is generally used, response spectrum
for a stiff soil is chosen and because it provides algebraic convenience, the one by
Shibata (8) is used although any other description would do. The general shape of
the Shibata displacement spectrum is shown in Fig.'5. As perceived by Newmark (9),
it may be divided into period ranges of almost-constant acceleration, almost-con-
stant velocity, and almost-constant displacement. For the application on hand,
response in the third range of almost-constant displacement may be ignored, not
only because it is out of range but also because displacement response for structures
in that range may not be stated simply in terms of the response of one mode. It isEffective Peak Acc. = 1 g
100
Displacement, em.
w
8
0 1
Period, sec.
Fig 5 Idealized Displacement Response Spectrum
plausible, in the interest of simplicity, to ignore the differences between response in
the second and first ranges. Assuming that the response in the first range varies as
that in the second one is conservative and the absolute error is practically of little im-
portance. Thus, focusing on Shibata’s "mid-range" expression for stiff soil, the
spectral drift response at a damping factor of 0.002 is
Dex Fm ®
Ds = spectral displacement response
A max = effective peak acceleration of ground motion
‘The third component of the device is the observation by Shimazaki (7) that a satisfac-
tory bound to nonlinear lateral displacement of reinforced concrete structures may
be obtained by modal analysis using a linear model of the structure with effective
stiffness equal to one half the initial stiffness and a response spectrum for a damping
factor of 0.02. Using the mechanics model for a uniform cantilever beam, the ratio of
the top-story displacement to the spectral displacement is approximated as 1.5 andthe mean drift ratio, MDR, is defined as the ratio of the top story displacement to
the total height. Then,
gatdan @
a
MDR=
Substituting 0.5g for A max and T from Eq. 2,
oH, | we 6)
which provides a measure of the damage to be expected. The measure is not ab-
solute in that implicit in it are assumptions of a uniform building and of effective
peak acceleration and frequency content reflected by the displacement spectrum.
But the equation does serve to assimilate the experience and to identify the relative-
ly important parameters in design.
Equation 5 emphasizes the obvious: the slenderness ratio H/D is the dominant
parameter. Any change in the slenderness ratio is reflected directly in the mean drift
ratio, If the slenderness ratio remains constant, it is seen that changes in typical story
height, b,(for constant total height, H) and unit weight, w are of equal and relatively
indirect influence. This is also true of the wall index, p, although it must be kept in
mind that changing p while holding H/D constant reduces the change to simply a
change in wall width and number. In any case, as opposed to the saying "Less mass is
its own reward” in automotive engineering, less mass and more wall appear to be of
equal importance in earthquake engineering, at least in the range of medium-rise
buildings with robust walls.
10‘Slenderess
Ratio, H/D
2
(Hf) 4
58 | enton romios/ A 3
a {2
Drift Ratio, %
2g
Wall Area / Floor Areo, %
Fig.6 Calculated Mean Drift Ratio vs. Wall Index
Figure 6 shows the variation of the calculated mean drift ratio with the wall index, p,
for four different slenderness ratios ranging from two to five. At a wall index of 3 %,
the calculated mean drift ratio is less than 1 96 even for a slenderness ratio of five. It
follows that such buildings would be successful for a more demanding earthquake
than that indicated by the Vina del Mar record of March 1985 and that their success
during the 1985 event can be explained by the fact that the mean drift ratios were
limited to below those indicated in Fig. 6.
Itis also interesting to note that the wall index of 3 %, which appears to have been
selected by Vina del Mar builders without coercion by code and without explicit
rationalization, occurs at approximately where adding more wall would start to be-
come inefficient.
The device represented by Eq, 5 also offers a basis for rationalization about the re-
quirement of boundary elements. Boundary elements in walls are intended to enable
walls to sustain large drifts without structural failure. Unconfined concrete would be
relied on to develop at its base a compressive strain of 0.003 even under eyclic load-
iling, If the depth to neutral axis is assumed not to exceed 0.3D, where D is the depth
of the wall, a curvature limit of 0.01/D would be considered tolerable. Whether this
curvature is related to drift by assuming a distribution of curvature comparable to
that of a linear beam or whether it is assumed to exist over a finite height of the wall
as a function of its depth, a corresponding drift ratio of
can be defended as tolerable for a wall without boundary elements. For the
earthquake conditions assumed in Eq 5 and illustrated in Fig. 6, this would suggest,
that the use of a wall index of over 1.25 % would justify not calling for boundary ele-
ments.
Conclusions
Subjected to an earthquake motion with an effective peak acceleration of 0.5g and a
broad-band spectrum (comparable to that for El Centro 1940), reinforced concrete
buildings with robust walls (wall index = 3 %) are likely to develop tolerable drifts
and, therefore, undergo limited damage.
Itis inferred that the overall positive record of the buildings of the Vina del Mar in-
ventory was due to their having wall indices of typically 3 %.
‘Use of robust walls which controls drift also rationalizes not calling for boundary ele-
ments. Calculated drift is considered to be a better criterion for deciding whether
boundary elements are needed than a calculated stress for design forces.
12Acknowledgments
‘The work that led to this paper was supported by the National Science Foundation
Grant ECE 86-03789. The project was part of a network comprising Municipality of
Vina del Mar, Technical University of F. Santa Maria (Valparaiso), Catholic Univer-
sity of Chile (Santiago), University of California (Berkeley), University of Michigan
(Ann Arbor), and the University of Ilinois,(Urbana). The writer has benefited from
discussions and works of P. Bonelli, J.C. de la Llera, J. P, Moeble, R. Riddell, R.
Stark, J. W. Wallace, J. K. Wight, and S. L. Wood.
References
1, R, Riddell, S. L. Wood, and J. de la Llera, "The 1985 Chile Earthquake; Struc-
tural Characteristics and damage Statistics for the Building Inventory in Vina del
Mar," Civil Engineering Studies, Structural Research Series No. 534, University of Il-
linois, Urbana, April 1987, 265 p.
2. Roberto Stark, "Evaluation of Strength, Stiffness, and ductility Requirements of
reinforced Concrete Structures Using Data from Chile(1985) and Michoacan (1985)
Earthquakes," Ph.D Dissertation, Department of Civil Engineering, University of Il-
linois, Urbana, 1988.
3. L. Wyllie et al., "The Chile earthquake of 3 March 1985," Earthquake Spectra,
V.2, No.2, April 1986, pp. 249-510.
4,8. L. Wood, J. K. Wight, and J. P, Moeble, "The 1985 Chile Earthquake; Observa-
tions on Earthquake-Resistant Construction in Vina del Mar," Civil engineering
studies, Structural Research Series No, $32, University of Ulinois, Urbana, February
1987, 176 p.5.1. D. Aristizabal and M. A. Sozen,"Behavior of Ten-Story Reinforced Concrete
‘Walls Subjected to Earthquake Motions," Civil Engineering Studies, Structural Re-
search Series No. 431, University of Illinois, Urbana, October 1976.
6. American Concrete Institute Committee 318, " Building Code Requirements for
Reinforced Concrete, ACI318-83" Box 19150, Redford station, Detroit, Michigan,
1983.
7.K. Shimazaki and M. A. Sozen, "Seismic Drift of Reinforced Concrete Structures,”
Research Reports, Hazama-Gumi Ltd., Tokyo, 1984, pp.145-166.
8. A. Shibata and M. A. Sozen,"Substitute Structure method for Seismic design in
Reinforced Concrete," Journal of the structural Division, ASCE, Vol. 102, No. ST1,
January 1976, pp. 1-18.
9. N. M, Newmark, "Current Trends in Seismic Analysis and Design of High-Rise
Structures," Earthquake Engineering, R. Wiegel, Prentice-Hall, Englewood Cliffs,
1970, pp.403-424,
10.C. M. Harris,"Shock and Vibration Handbook," McGraw-Hill, NY, 1988, pp.1-15.