You are on page 1of 13

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S14

Behavior of Gravity Load Designed Reinforced Concrete


Buildings Subjected to Earthquakes

by Adel G. El-Attar, Richard N. White, and Peter Gergely

Two small-scale reinforced concrete building models were tested on the designed and detailed to resist primarily gravity loads
Cornell University shake table. The models were a 1/6 scale two-story office (termed hereafter as GLD, an acronym for gravity load
building and a 1/8-scale three-story one-bay by three-bays office building.
Both structures were designed to resist purely gravity loads without regard
design). Additional experimental evidence is clearly needed
to lateral loads (wind or earthquake forces). The reinforcement details were to understand the behavior of this class of buildings and to
based on typical reinforced concrete frame structures constructed in the develop more reliable analytical tools to predict their seismic
central and eastern United States over the past 50 to 60 years, as charac- damage potential. Based on both the new experimental
terized by (a) low reinforcement ratio in the columns, (b) discontinuous pos- results given here and supporting analytical information, risk
itive moment reinforcement in the beams at the column locations, (c) little
or no confining reinforcement in the joint regions, and (d) lap splices located
assessment can be conducted for existing GLD reinforced
immediately above the floor level. Both models were tested using the time- concrete buildings located in various seismic zones.
compressed Taft 1952 S69E ground motion scaled to increasingly large Typical reinforcement details for GLD buildings have
peak ground accelerations. been identified through a careful review of earlier versions
Test results indicated that gravity load design (GLD) reinforced of the ACI 318 code and related ACI detailing manuals
concrete buildings without walls will experience very large deformations
published since 1940 and by consulting several design
associated with a considerable stiffness degradation during a moderate
earthquake. The high flexibility produced significant P-Δ effects in the offices with long design experience in the central and eastern
three-story building model. Although the nonseismic details associated United States. This type of detailing is characterized by (a)
with the gravity load design philosophy forms a source of damage, the low reinforcement ratio in the columns, (b) discontinuous
experiments indicate that these details will not necessarily lead to positive moment reinforcement in the beams at the column
collapse or to a complete failure mechanism.
locations, (c) little or no confining reinforcement in the joint
Comparison with analytical results indicated that inclusion of the slab
contribution to beam flexural strength is a vital step in the assessment of the
regions, and (d) lap splices located immediately above the
performance of GLD reinforced concrete structures since it has the potential of floor level.
altering the relatively ductile strong column-weak beam mechanism to a The study presented here is part of a comprehensive
more brittle soft-story mechanism. research effort conducted by the National Center for Earth-
quake Engineering Research (NCEER) on the seismic
Keywords: earthquake-resistant structures; frames; joints (junctions); office damage assessment and performance evaluation of GLD
buildings; reinforced concrete. reinforced concrete buildings.
The major objectives of the current investigation were:
RESEARCH SIGNIFICANCE 1. Develop improved small-scale modeling techniques for
The behavior of gravity load design (GLD) reinforced reinforced concrete structures subjected to dynamic loading.
concrete structures during earthquakes is still not well under- 2. Investigate the performance of GLD reinforced concrete
stood. This paper reports the results of one of the first exper- buildings by testing two complete small-scale structures on
imental tests on three-dimensional GLD reinforced concrete the shake table.
model buildings on the shaking table. Test results presented 3. Evaluate the reliability of one of the recently developed
here will help designers to identify the major weaknesses in analytical modeling techniques to predict the response of
this type of building and will provide background data for these buildings.
proper strengthening strategies.

INTRODUCTION ACI Structural Journal, V. 94, No. 2, March-April 1997.


Recent seismic studies and historical records indicate that Received Jan. 26, 1995, and reviewed under Institute publication policies. Copy-
right © 1997, American Concrete Institute. All rights reserved, including the making
damaging earthquakes can occur in the central and eastern of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion will be published in the January-February 1998 ACI Structural Journal if
United States, where many thousands of buildings have been received by Sept. 1, 1997.

ACI Structural Journal/March-April 1997 133


again built with no walls.3 A similar model was tested on the
Adel G. El-Attar is an assistant professor in the Structural Engineering Department,
Cairo University, Egypt. He received his BSCE and MSCE degrees from Cairo Uni- shake table of the State University of New York (SUNY) at
versity and his PhD from Cornell University. His research interests include concrete Buffalo at a 1/3 scale.5,6 This structure also was designed for
plasticity and experimental and analytical modeling of building and bridge structures, dead load plus 50 psf live load. It represented a more realistic
with particular emphasis on seismic performance.
building than the two-story model as it had multiple floors
Richard N. White, FACI, is the James A. Friend Family Professor of Engineering at and bays, thus containing both interior and exterior joints.
Cornell University. He is a former vice president of ACI and a member of ACI’s board Geometry and reinforcement were again determined using
of directors and several ACI committees and task forces.
standard similitude requirements of true replica models.1
Peter Gergely, FACI, a faculty member in structural engineering at Cornell University Fig. 5 shows the reinforcement details for the prototype and
for 33 years, passed away in August 1995 after a long and courageous battle against the model buildings.
brain cancer. He served on a number of ACI committees and participated in many
other ACI activities.
MODEL MATERIALS
Model concrete
The overall research plan is shown in Fig. 1. This paper
focuses on results from the experiments on the two model One of the most challenging tasks met in the small-scale
buildings (bare frames with slabs but without walls) and modeling of reinforced concrete structures is the selection of
comparison of these results to numerical analysis results.1-3 a microconcrete mix that can faithfully reproduce a three-
dimensional failure criterion of the prototype concrete.
DESCRIPTION OF MODEL STRUCTURES Given that such a criterion is typically not known for a given
Selection of the geometric scales was based on the prototype concrete, this similitude condition is usually
dimensions and load capacity of the shake table, taking into relaxed to modeling four properties of the prototype
account the availability of the desired sizes of model rein- concrete: compressive strength f c′ , ultimate strain εcu, secant
forcement. It was decided that a geometric scale factor S1 modulus Ec, and tensile strength f t′ . Previous research
(where S1 = prototype dimension divided by the model conducted at Cornell University has indicated that by care-
dimension) in the range of 5 to 8 would be appropriate. fully selecting the water-cement (w/c) ratio, the aggregate/
The first model was a 1/6 scale two-story, one-bay by one- cement (a/c) ratio, and the aggregate grading, these parame-
bay office building (Fig. 2). A similar building with seismic ters can be effectively controlled. Detailed information on
details was tested at the University of California-Berkeley at a the microconcretes used for these two models is provided by
7/ scale.4 The main objectives of this pilot test were to check Kim et al.1
10
out the shake table control system and data acquisition system
and also provide some understanding of the response of a very Model reinforcement
simple GLD building. The prototype structure was designed Small-scale reinforcement is modeled by reproducing
and detailed for dead load plus 50 psf live load. The resulting the prototype values of yield strength fy, extent of yield
prototype building reinforcement details are shown in Fig. 3. plateau, post-yield behavior (strain-hardening), and bond
Model dimensions and reinforcement were obtained using characteristics as described in Reference 1. The model
standard similitude requirements of true replica models.1 bars were heat-treated to obtain the desired yield strengths
The second model (Fig. 4) was a 1/8 scale three-story, one- (60 ksi for the two-story building and 40 ksi for the three-story
bay by three-bay (in the shaking direction) office building, building) and yield plateaus.

Fig. 1—Flow chart of research program.

134 ACI Structural Journal/March-April 1997


Tests on two prototype and four 1/6-scale cantilever beam produced by the seismic loading as explained in References 2
specimens reported in Reference 1 indicated that best corre- through 4.
lation was obtained with threaded steel rods as model rein-
forcement. Model and prototype hysteretic load-deflection TEST RESULTS
responses correlated well up to a ductility factor of 6. Two-story building test results
A summary of the simulated earthquake test results is
MODEL LOADING AND INSTRUMENTATION presented in Table 1. Measured values of floor displace-
Self-weight similitude in both models was achieved by ments and base shear for a moderate earthquake (Taft 0.26g-
adding lead blocks, with special attention given to the attach- 1) and a strong earthquake (Taft 0.75g-1) are plotted versus
ment technique to avoid any possible stiffening of the model time in Fig. 7 and 8 respectively. The displacement records
floor slabs by the attached blocks. Floor accelerations were indicate that both floors were moving in phase even after the
measured using piezoresistive accelerometers. Displacements model was severely damaged. It was apparent that the first
were measured at two points (25 in. apart) on each floor to mode dominated the model response during all tests.
monitor any rotational motion during the seismic tests.
The three-story model also had internal force transducers
(load cells) at the midheight of the first and second-story
columns (Fig. 5). The load cells measured axial force,
bending moment, and shearing force in two perpendicular
directions (with a maximum interference of 3 percent), thus
providing considerable information about the state of forces
in the structure during the seismic tests. The internal config-
uration of the used load cells along with a sample calibration
curve are shown in Fig. 6. Details of the similitude require-
ments, model design and fabrication, loading techniques,
and instrumentation are given in References 2 and 3.

TEST PROCEDURE
Both models were tested using the time-scaled Taft 1952
S69E earthquake with peak ground acceleration set at
increasingly higher values. Each seismic test was preceded
and followed by a static test and a free vibration test to deter-
mine the change in the structural properties (such as the
flexibility matrix, fundamental period, damping ratio, etc.) (a)

(b)

Fig. 3—Reinforcement details of two-story prototype


Fig. 2—Layout of two-story building prototype (model) (model) building: (a) general reinforcement layout; (b) typi-
dimensions: (a) elevation; and (b) plan. cal joint detail.

ACI Structural Journal/March-April 1997 135


(a)

(c)

(b)

Fig. 4—Layout of three-story building (prototype dimensions given): (a) side view; (b) elevation; (c) plan.

Fig. 5—Reinforcement details of three-story prototype (model) building: (a) elevation; (b) exterior joint (prototype); (c) interior
joint (prototype).

136 ACI Structural Journal/March-April 1997


Hairline cracks appeared at the first-story column construc- Table 1—Summary of two-story model simulated
tion joints during Run Taft 0.26g-1. In subsequent runs, cracks earthquake test results
remained localized at these zones and did not spread over the Maximum story First Maximum story
displacement, in. story acceleration, g Maximum
columns’ height. After Run Taft 0.45g, wide, deep cracks drift, base shear,
were observed in the beams around the locations of discontin- Test First Second percent First Second kips
uous positive beam reinforcement (at the joint panels), Taft 0.26g-1 0.229 0.370 1.27 0.542 0.706 0.941
indicating bond failure and pending pullout of these bars. Taft 0.26g-2 0.244 0.377 1.36 0.745 0.914 1.005
Change of model properties including natural frequencies, Taft 0.36g 0.348 0.545 1.93 0.720 1.096 1.119
damping ratios, and flexibility matrix coefficients is shown in Taft 0.45g 0.405 0.600 2.25 0.540 1.167 1.156
Table 2. It can be seen that cracking during Run Taft 0.26g-1 Taft 0.75g-1 0.741 1.228 4.12 0.767 1.343 1.307
reduced the fundamental frequency by 29 percent. The next Taft 0.75g-2 0.789 1.036 4.38 0.643 0.920 1.104
major change was during Run Taft 0.75g-1, where a reduction Taft 0.9g 0.911 1.146 5.06 0.367 0.525 0.659

Fig. 6—Load cell internal configuration and sample calibration curve: (a) elevation
(Face 1); (b) plan; (c) Face 2; (d) Face 3; (e) Face 4; (f) typical interaction curve between
normal force N and moment M and shears Q1 and Q2.

ACI Structural Journal/March-April 1997 137


Fig. 7—Two-story model response, Run Taft 0.26g-1: (a) first floor displace-
ment; (b) second floor displacement; (c) base shear.

Fig. 8—Two-story model response, Run Taft 0.75g-1: (a) first floor displace-
ment; (b) second floor displacement; (c) base shear.

138 ACI Structural Journal/March-April 1997


of 47 percent was observed. Table 2 also indicates that the stiffness degradation (18 percent reduction of fundamental
flexibility matrix coefficients increased gradually with the frequency corresponding to 50 percent reduction in stiff-
intensity of the ground motion. ness). The maximum base shear during this run (1.25 kips)
The model did not collapse under even the highest level
dynamic loading (0.75 g). However, significant stiffness Table 2—Variation of two-story model properties
deterioration was recorded (40 percent reduction in stiffness due to seismic tests
after Run Taft 0.36g). After Run Taft 0.75g-1, the model Natural Damping ratio, Flexibility matrix
stiffness was about 17 percent of the uncracked stiffness. frequency, Hz percent of ζcr coefficients, in./lb × 103
Few more cracks appeared after this run at the top and Test f1 f2 ζ1 ζ2 f11 f12 f22
bottom column sections, while old cracks at the discontin- Uncracked 6.25 17.19 — — 0.100 0.113 0.225
uous positive beam reinforcement became wider, indicating Taft 0.26g-1 4.44 13.18 2.64 2.44 0.254 0.292 0.532
a possible pullout of these bars (Fig. 9). Taft 0.26g-2 4.07 12.10 3.06 3.14 0.300 0.350 0.660
It is useful to recognize that even a complete pullout of the Taft 0.36g 3.66 11.70 4.19 3.85 0.345 0.406 0.733
discontinuous positive moment beam reinforcement in a Taft 0.45g 3.61 11.48 4.76 3.72 0.426 0.473 0.773
GLD building cannot be the primary cause of failure. When Taft 0.75g-1 2.93 10.31 4.49 3.46 0.544 0.682 1.192
a discontinuous bottom reinforcing bar starts to pull out of Taft 0.75g-2 2.93 10.30 4.50 3.43 0.640 0.790 1.338
the joint region, the accompanying beam flexural cracks at Taft 0.90g 2.45 9.48 4.53 3.43 0.775 0.955 1.578
the face of the column become visibly larger. The lateral
stiffness of the frame reduces, but a collapse mechanism
cannot develop, provided that the negative reinforcement at
the other end of the beam is below yield. A state of
“controlled” hinging action occurs, with the discontinuous
bars at the positive moment carrying a reduced tensile force.
When the frame moments reverse, the bottom cracks tend to
close (but not completely) as slip of the discontinuous bars is
recovered. Cyclic loading further degrades the peak tensile
capacity and axial stiffness of the discontinuous bars.

Three-story building test results


General behavior—Top story displacements and base
shears recorded during Run Taft 0.18g are given in Fig. 10.
Similar data for Run Taft 0.35g are plotted in Fig. 11 and 12
for the story displacements and story shears respectively. As
previously observed for the two-story model, all three stories
were moving in phase, indicating the domination of the first
mode. A brief summary of the seismic tests is given in Table 3,
where it can be seen that during Run Taft 0.18g the model Fig. 9—Cracking pattern of two-story model after Run Taft
showed a large degree of flexibility associated with a high 0.75g-1.

Fig. 10—Three-story model response, Run Taft 0.18g: (a) third floor displace-
ment; (b) base shear.

ACI Structural Journal/March-April 1997 139


Fig. 11—Three-story model story displacements, Run Taft 0.35g: (a) first story; (b)
second story; (c) third story.

Fig. 12—Three-story model story shears, Run Taft 0.35g: (a) base shear; (b) second
story; (c) third story.

140 ACI Structural Journal/March-April 1997


represented 8.8 percent of the total vertical load on the struc- source of damage to this particular GLD structure. The
ture, corresponding to 85 percent of the maximum calculated cracking pattern recorded after Run Taft 0.35g is shown in
model base shear capacity. Assuming viscous damping, the Fig. 13.
first mode damping ratio, expressed in Table 3 as a 2. P-Δ effect: The large flexibility of the tested skeleton
percentage of the critical damping, increased significantly structure with no frame in-fill walls resulted in a pronounced
during Run Taft 0.18g as new cracks developed. P-Δ effect. At the instant of maximum story drift, during Run
After the Taft 0.35g run, the model fundamental frequency Taft 0.18g, the sum of the column shears recorded by the
decreased to 1.65 Hz (25 percent reduction), indicating a 78 internal force transducers installed at their midheights was
percent stiffness reduction ([(2.2/1.65)2 – 1] × 100). No 27 percent higher than the base shear computed from the
significant change was detected in the damping ratio, story accelerations. This discrepancy was due to the fact that
suggesting that few new cracks developed during this run. the load cells could capture the vertical load component
The maximum base shear of 1.38 kips recorded during this run developed in the columns as produced by the side-sway of
represented 97 percent of the calculated base shear capacity. the structure, while the accelerometers on the story floors
Story shears and mode shapes recorded at the instant of could only record their horizontal acceleration (Fig. 14).
maximum base shear are shown in Table 4. It can be seen This phenomenon was observed in all subsequent runs.
that in spite of considerable reduction in the model stiffness, 3. Columns: During all seismic tests, it was noticed that
the mode shapes remained essentially unchanged for all the average measured shear force in interior columns was
three complete runs. Furthermore, except for the first low- much higher than that of exterior columns (Table 5). This
amplitude seismic test, the shear force distribution over the can be attributed to (a) the large axial force in interior
three stories remained the same for all subsequent runs. columns that delayed their cracking and consequently
The model collapsed during Run Taft 0.80g (after 7 sec) in increased their relative stiffness, and (b) the higher degree of
a soft-story mechanism in the first-story columns. The peak
base shear was only 3.3 percent higher than that measured in
the previous run at 0.35g peak ground acceleration. Failure Table 3—Summary of three-story model seismic
was initiated at one of the interior columns, followed by test results
failure of the remaining first-story columns. This failure Maximum Fundamental First mode
mode is consistent with the fact that interior columns are Top story base shear, frequency, damping ratio,
Test drift, percent kips Hz percent of ζcr
subjected to a higher axial force and consequently are less
Taft 0.05g 0.19 0.338 2.20 1.30
ductile than exterior columns. In the current test, the
Taft 0.18g 2.02 1.252 1.80 2.74
average static reaction of interior columns was equal to 2.61
Taft 0.35g 2.84 1.384 1.65 2.76
kips (0.3Ag f c′ ) while for exterior columns the average reac-
Taft 0.80g — 1.430 — —
tion was equal to 1.05 kips (0.12Ag f c′ ) (Reference 3).
Local behavior—
1. Cracking: Cracks were detected at the top and bottom Table 4—Mode shapes and shear distribution at
sections of the first and second-story columns, within a maximum base shear
distance equal to the column dimension of 1.5 in. (plastic
hinging region), after Run Taft 0.18g. These cracks were
localized and did not spread over the columns’ height even
after the Taft 0.35g run. No damage was observed in
beams, joint regions, or splice termination areas, indicating
that the nonseismic reinforcement details were not a critical

Fig. 13—Cracking pattern of three-story model after Run


Taft 0.35g.

ACI Structural Journal/March-April 1997 141


restraint provided by the interior beams (from four sides of reinforcement in either direction. The response of Section
the joint) as compared to that of exterior beams. B5 was quite different: it cracked and yielded in both direc-
Column moment is plotted against axial force in the inter- tions. Several reasons may have contributed to the difference
action curves shown in Fig. 15 for the first-story interior and in response between Sections B2 and B5, among them (a)
exterior columns with all measurements taken during Run Taft accidental overstrength of Section B2 associated with the
0.35g. It can be seen that first-story interior columns [Fig. 15(a) difficulty of controlling the slab thickness, and (b) depen-
and (b)] yielded in the two directions, while exterior columns dence of the section moments on the change in the column
[Fig. 15(c) and (d)] yielded only in the high axial force reactions, which may contain some errors.
direction, suggesting that their stiffness (and hence their ability Sections B3 and B4 were also cracked from self-weight
to attract more shear) is directly related to their axial force level. effects. Both sections did not yield in the negative direction
A similar conclusion was reached by Abrams.7 during Run Taft 0.35g. It can be seen that only Section B4
4. Beams: Bending moments at the first-story longitudinal cracked and yielded in the positive direction once during
beam critical sections were calculated using both the column this run.
static reactions measured before seismic tests and the changes The high force level in the positive moment beam rein-
in those reactions measured by the internal load cells during forcement, as they reached yield in many occasions
each table excitation. (Sections B1, B4, B5, and B6), in addition to the minor
Time histories of these moments from Run Taft 0.35g are cracking observed at the beam-column interface at their
shown in Fig. 16(a) to (f) for Sections B1 to B6 respectively. level provide more evidence that these bars did not pull out
Values of cracking and yielding moments calculated using during this run.
an effective tee-section width of span/4 = 6.75 in.8 for both 5. Longitudinal frame bending moment: Fig. 17(a) and
positive and negative bending are also shown on each plot. (b) show the bending moments of the longitudinal frame
Inspection of this set of curves along with visual inspection during Run Taft 0.35g at the time of maximum positive and
of the beam critical sections indicates that no pullout of the maximum negative base shear respectively. It can be seen
discontinuous positive beam reinforcement occurred during from both figures that the increase in the exterior column
this run. It can be seen from Fig. 16(a) to (f) that Sections B1 axial force resulting from the overturning moment effect
and B6 were subjected to a similar bending history and produced increases in bending moments. This phenomenon
reached their positive yield point once. Both sections did not was observed to a lesser degree for the interior columns.
yield in the negative direction during the entire run. Beam moments at the interior joints were always negative
Sections B2 and B5 were already cracked from self- (except for Section B2); consequently, no pullout of
weight effects. During Run Taft 0.35g, the bending discontinuous positive beam reinforcement was detected.
moment at Section B2 was not enough to yield the beam Although the situation was different for exterior joints,
where positive moments were developed, it is believed that
positive beam reinforcement yielded at these regions
Table 5—First-story column shears at time of
without pulling out.
maximum base shears (three-story building test3)
Exterior columns Interior column
average ANALYTICAL RESULTS USING PROGRAM IDARC
shears, kips shears, kips
interior
Shear average The two tested models were analyzed using Program
Test direction C1 C4 Average C2 C3 Average exterior IDARC (Inelastic Damage Analysis of Reinforced Concrete
Taft Max. +ve 0.099 0.187 0.143 0.249 0.259 0.254 1.776 Structures), developed at SUNY/Buffalo. The objective of
0.18g Max. –ve 0.176 0.085 0.130 0.265 0.225 0.245 1.885 the analysis was to investigate the suitability of IDARC in
Taft Max. +ve 0.104 0.198 0.151 0.281 0.294 0.288 1.901 predicting the response of this type of building. The program
0.35g Max. –ve 0.170 0.091 0.131 0.279 0.241 0.260 1.984 is based on macroconcrete elements in which a three-parameter

Fig. 14—Simplified computation of P-Δ effect in first-story columns of three-story model


during Run Taft 0.18g.

142 ACI Structural Journal/March-April 1997


hysteresis model is used to describe the stiffness degradation 1.0 were used in the current study. Also, a damping ratio ζ of
(α parameter), strength deterioration (β parameter), and 2 percent of the critical damping ζcr was adopted. The effec-
pinching and/or slip behavior (γ parameter).9 Based on proto- tive tee-section width was taken equal to span/4 for beams in
type connection tests,9 values of α = 2.0, β = 0.05, and γ = both positive and negative moments. A full description of the

(a) (c)

(b) (d)

Fig. 15—Moment-normal force response, first-story columns, Run Taft 0.35g: (a) interior column bottom section; (b) inte-
rior column top section; (c) exterior column bottom section; (d) exterior column top section.

Fig. 16—Bending moments in first-story beams, Run Taft 0.35g: (a) Section B1; (b) Section B2; (c) Section B3; (d) Section B4;
(e) Section B5; (f) Section B6.

ACI Structural Journal/March-April 1997 143


program, input data, and analysis results is given in Refer- base shear correlated reasonably well with the measured
ences 2, 3, and 9. Due to space limitations, only sample shear and that both records were in phase during the entire
analytical results of the 3-story building are presented here. run, indicating that the change in the mode period due to
The time history of the calculated base shear for the 3-story stiffness degradation was reasonably reflected in the analysis.
model is plotted against the measured base shear during Run The calculated top story displacement during Run Taft
Taft 0.35g in Fig. 18(a). It can be seen that the calculated 0.35g also correlated well with the measured values up to the
maximum positive value [Fig. 18(b)], after which the calcu-
lated displacements started to deviate slightly from the
measured ones.

SUMMARY AND CONCLUSIONS


The experiments presented in this paper are among the
very first to address the performance of gravity load design
(GLD) reinforced concrete buildings tested under realistic
seismic loading conditions. The results revealed many
aspects of the behavior of this important class of building
during earthquake loading.
Experimental evidence from the two tested model buildings
indicate that GLD reinforced concrete structures will experi-
ence significant reduction in lateral stiffness after the first few
cycles of a moderate earthquake. This reduction is mainly due
to (a) the lack of ductility caused by poor confinement, espe-
cially in the joint regions, and (b) incipient pullout of the
discontinuous positive beam reinforcement. The large model
flexibility associated with testing the skeleton structure only,
with no wall in-fill, resulted in a pronounced P-Δ effect (up to
27 percent of the measured base shear).
The data from both models indicate the domination of the
first mode of vibration in all seismic tests. The mode shape
was almost constant for all tests despite the significant
reduction in the models’ stiffness (increase in their funda-
mental period).
It was also found that each column’s share of the total
story shear is directly related to its axial force level. Interior
Fig. 17—Bending moments at maximum base shear, Run columns in the three-story building resisted about twice the
Taft 0.35g (all moments in in.-kips): (a) bending moments at shear force acting on the exterior columns.
maximum positive base shear (time = 1.86 sec); (b) bending Most of the deformation, damage, and energy dissipation
moments at maximum negative base shear (time = 3.36 sec). of the two tested models occurred in their first-story

Fig. 18—Computed versus measured response, Run Taft 0.35g: (a) base shear; (b)
third-story displacement.

144 ACI Structural Journal/March-April 1997


columns. No significant damage was observed in the beams, ACKNOWLEDGMENTS
at the splice locations, or in the joint panels. At failure, This research was supported by the National Center for Earthquake Engi-
neering Research, with funding from the National Science Foundation, State
plastic hinges developed in the columns (strong beam-weak of New York, and industrial sponsors, under NCEER Grant No. ECE 86-
column behavior), producing a soft-story failure mode. 07591. The authors gratefully acknowledge the assistance provided by Tim
Based on the current experimental and analytical results, Bond, manager of the George Winter Laboratory, technicians Paul Jones,
the following conclusions may be drawn: Glenn Darling, and John Powers, and by undergraduate students Adam
Hoffman, Rebecca Frein, and William Druc.
1. Although the reinforcement details typically used in
GLD reinforced concrete structures may form a potential
REFERENCES
source of damage, they are probably not sufficient in them- 1. Kim, W.; El-Attar, A.; and White, R. N., “Small-Scale Modeling Tech-
selves to develop a complete failure mechanism. Available niques For Reinforced Concrete Structures Subjected to Seismic Loads,” Tech-
experimental evidence indicates that the lack of sufficient nical Report NCEER-88-0041, National Center for Earthquake Engineering
Research (NCEER), Nov. 22, 1988.
strength of columns as compared to beams in typical GLD 2. El-Attar, A.; White, R. N.; Gergely, P.; and Conley, C., “Shake Table
buildings usually leads to a premature soft-story mechanism Tests of a 1/6-Scale Two-Story Lightly Reinforced Concrete Building,”
before these details are subjected to significant demand. Technical Report NCEER-91-0017, NCEER, 1991.
3. El-Attar, A.; White, R. N.; Gergely, P.; and Bond, T. K., “Shake Table
2. GLD reinforced concrete buildings with no in-fill walls
Tests of a 1/8-Scale Three-Story Lightly Reinforced Concrete Building,”
may experience large lateral deformations during a moderate Technical Report NCEER-91-0018, NCEER, 1991.
earthquake, with deformations accentuated by substantial 4. Blondet, J. M.; Clough, R. W.; and Mahin, S. A., “Evaluation of a
P-Δ effects. The inclusion of the in-fill walls in GLD reinforced Shaking Table Test Program on Response Behavior of a Two-Story Rein-
forced Concrete Frame,” Report No. UCB/EERC-80/42, Earthquake Engi-
concrete buildings is indeed an important area for additional
neering Research Center, University of California/ Berkeley.
experimental and analytical investigation. 5. Bracci, J. M.; Reinhorn, A. M.; and Mander, J. B., “Seismic Resis-
3. Lateral stiffness of columns is a function of their axial tance of Reinforced Concrete Frame Structures Designed Only for Gravity
force level. This observation imposes a limitation on the use Loads: Part I—Design and Properties of a One-Third Scale Model Struc-
ture,” Technical Report NCEER-92-0027, SUNY/ Buffalo, 1992.
of many available elastic dynamic analysis programs, where
6. Bracci, J. M.; Reinhorn, A. M.; and Mander, J. B., “Seismic
such effect is not reflected. Resistance of Reinforced Concrete Frame Structures Designed Only
4. Accounting for the slab contribution to the beam nega- for Gravity Loads: Part III— Experimental Performance and Analyti-
tive moment flexural strength is a vital step in the assessment cal Study of a Structural Model,” Technical Report NCEER92-0028,
of the performance of GLD reinforced concrete frames since SUNY/Buffalo, 1992.
7. Abrams, D., “Influence of Axial Force Variations on Flexural Behav-
it has the potential of altering the relatively ductile strong ior of Reinforced Concrete Columns,” ACI Structural Journal, V. 84, No. 3,
column-weak beam mechanism to a soft-story mechanism. May-June 1987, pp. 246-254.
This will be true even when a reduction factor is applied to 8. ACI Committee 318, “Building Code Requirements for Reinforced
positive beam flexural strength to account for the effect of Concrete and Commentary (ACI 318-89/ACI 318R-89),” American Con-
crete Institute, Detroit, 1989, 353 pp.
possible pullout of the discontinuous positive moment rein- 9. Park, Y. J.; Reinhorn, A. M.; and Kunnath, S. K., “IDARC: Inelastic
forcement in the beams. Damage Analysis of Reinforced Concrete Frame-Shear Wall Structures,”
6. Small-scale reinforced concrete models can be used as a Technical Report NCEER-87-0008, NCEER, July 20, 1987.
10. Pessiki, S. P.; Conley, C. H.; Gergely, P.; and White, R. N., “Seismic
powerful tool in studying the seismic performance of complete Behavior of Lightly Reinforced Concrete Column and Beam-Column Joint
buildings or multimember structural subassemblies. Details,” Technical Report NCEER 90-0041, NCEER, Aug. 27, 1990.

ACI Structural Journal/March-April 1997 145

You might also like