You are on page 1of 10

Comparative Study of Seismic Behavior of Multistory

Reinforced Concrete Framed Structures


Yong Lu1
Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Abstract: A comparative study of the nonlinear behavior of reinforced concrete 共RC兲 multistory structures is carried out on the basis of
measured response of four six-story, three-bay framed structures, namely a regular bare frame, a discontinuous-column frame, a partially
masonry-infilled frame, and a wall-frame system. The structures were designed for similar seismic requirements in accordance with
Eurocode 8, and their 1:5.5 scaled models were subjected to similar earthquake simulation tests. Experimental observations and numerical
analyses show that the distribution of the story shear overstrength is a rather stable indicator of the general inelastic behavior of frames,
and hence, can be employed as a characteristic parameter to quantify the frame irregularity for design purposes. Abrupt discontinuity of
the geometry or arrangement of structurally effective elements, where unavoidable, may be compensated by strength enhancement
targeting a smoothed overstrength profile to allow for distributed inelastic deformation, and this principle applies as well to nonuniformly
masonry infilled frames. For the wall-frame system, adequate countermeasures against rocking of the RC wall is shown to be a key to
maintaining the effectiveness of the system at advanced inelastic response.
DOI: 10.1061/共ASCE兲0733-9445共2002兲128:2共169兲
CE Database keywords: Comparative studies; Buildings, multistory; Concrete, reinforced; Building frames; Seismic response.

Introduction senting vertical irregularity; a partially masonry-infilled frame


with an open first story 共frame PIF兲; and a structural wall-frame
The seismic behavior of multistory structures has been the subject system 共SWF兲. Fig. 1 shows their geometry. All these structures
of extensive study over the last several decades. Generally speak- were designed to satisfy the seismic requirements for ductility
ing, typical multistory framed systems can be divided into three class ‘‘Medium’’ in accordance with EC8, with design peak
categories concerning their seismic behavior, namely reinforced ground acceleration 共PGA兲 of 0.30 g. For the experiment, models
concrete 共RC兲 frames, masonry-infilled frames, and wall-frame of 1:5.5 reduced scale were constructed and tested under the same
systems. Apart from the development of the general ductile and scheme of simulated earthquakes on an earthquake simulator.
capacity design methodologies, of particular interest in recent This paper summarizes the experimental program and presents a
studies are the effects of the structural configurations and the comparative study of the different response characteristics and
arrangement of nonstructural members on the seismic perfor- failure mechanisms based on the experimental observations and
mance 共Lu et al. 1999; Moehle and Sozen 1980; Negro and Ver- associated numerical evaluation. Some recommendations for fur-
zeletti 1996; Schultz 1990; Wood 1992兲. From numerous investi- ther improvement of the design of multistory structures are also
gations of individual cases, however, it is felt that there is a lack given.
of information regarding the comparative behavior of different
systems. The individually investigated structures seldom allowed
for a more rigorous comparative evaluation. Design of Test Structures
An experimental investigation conducted at the National Tech-
nical University of Athens in support of the Eurocode 8 共EC8兲 EC8 adopts the concept of the behavior factor, q, to factor down
关1988 共draft兲; 1994兴 has provided comprehensive data allowing the elastic spectrum for the determination of the seismic design
for an interpretation of the seismic performance of various mul- lateral load. The structures are classified into three ductility
tistory systems using a comparative approach. Four six-story, classes, namely ductility ‘‘high,’’ ‘‘medium,’’ and ‘‘low.’’ The
three-bay framed structures were selected to represent a variety of value of the q factor is determined according to the ductility class
multistory systems, including a benchmark regular bare frame chosen, taking into account the structure type and the irregulari-
共frame BFR兲; a discontinuous-column frame 共frame DCF兲 repre- ties involved. Specific detailing requirements are stipulated to sat-
isfy the respective ductility demand.
1
Assistant Professor, School of Civil and Environmental En- Ductility class ‘‘medium’’ was chosen for all the structures
gineering, Nanyang Technological Univ., Singapore 639798. under investigation. The corresponding q factor was 3.5 for the
E-mail: cylu@ntu.edu.sg regular frame BFR. For frames DCF and PIF, adjustment to the
Note. Associate Editor: Brad Cross. Discussion open until July 1, above basic q value was made to account for the irregularities
2002. Separate discussions must be submitted for individual papers. To
associated with them. To do so, it was necessary to first quantify
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted the irregularity, and for this purpose, the concept of ‘‘regularity
for review and possible publication on February 1, 2001; approved on index’’, as will be discussed in a greater length later, was adopted.
August 29, 2001. This paper is part of the Journal of Structural Engi- In the case of masonry-infilled frame PIF, the contribution of infill
neering, Vol. 128, No. 2, February 1, 2002. ©ASCE, ISSN 0733-9445/ walls was not considered in the final strength design but was
2002/2-169–178/$8.00⫹$.50 per page. accounted for when evaluating the regularity index and determin-

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002 / 169

J. Struct. Eng., 2002, 128(2): 169-178


The design peak ground acceleration was 0.3 g. Other design
assumptions included: Soil profile A, Concrete class C20 共charac-
teristic compressive strength 20 MPa兲, Flexural reinforcement
Grade S400, Transverse reinforcement, and RC wall reinforce-
ment Grade S220, Floor dead load 1.5 kN/m2, and live load 2.0
kN/m2.
The proportioning of the structural members was carried out
following capacity design procedures for a weak-beam, strong-
column design and to avoid premature shear failure. In particular,
the column design moment effects were obtained by observing
equilibrium around a joint, taking into account the actual bending
Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

capacity of the beams framing into the joint. A relaxation was


then considered on the above column moment effects through a
‘‘moment reversal factor’’ 共EC8 1994兲. The transverse confining
reinforcement within the column critical regions was detailed ac-
cording to the required mechanical volumetric ratio to satisfy a
curvature ductility of not less than 9 for the medium ductility
frames. Table 1 lists the cross section dimensions and longitudinal
reinforcement ratios of the structural members.

Test Model Structures and Experimental Program


Each test model comprised a single planar frame, constructed at
1:5.5 reduced scale. At floor levels, a 300-mm wide flange was
provided on both sides of the frame to represent the floor slab
effects and to accommodate the additional masses required by the
similitude laws 共Tassios 1992兲. The model reinforcement, shown
Fig. 1. Geometry of structures under investigations 共model scale in Fig. 2, was deduced from the prototype design on a one-to-one
1:5.5兲 basis.
Microconcrete having an average cylinder compressive
strength of 30 MPa and tensile-to-compressive strength ratio of
ing the design seismic load. A minimum regularity index of 0.55 1:10 was used for casting the model structures. The model flex-
was imposed. In fact, it was this minimum requirement that actu- ural reinforcement in beams and columns were of conventionally
ally governed the proportioning of the open first-story columns in deformed 6-mm bars, which had average yield strength of 430
frame PIF as will be discussed later as well. The final design q MPa after annealing treatment and a bond strength of 25–30% the
factors were 2.7 for DCF and 2.4 for PIF. The q factor adopted for concrete compressive strength when embedded in the model con-
the design of the wall-frame system SWF was 3.0. crete. The remaining model reinforcement was of 3-mm smooth

Table 1. Cross-section Dimensions 共full-scale兲 and Longitudinal Reinforcement Ratios 共␳, %兲


Column Beam
Story/ Section ␳a Section ␳b
Model level 共mm兲 共mm兲
BFR 5– 6 350⫻350 2.76 350⫻400 1.32
3– 4 450⫻450 1.68 400⫻450 1.02
2 500⫻500 1.37 400⫻500 0.91
1 600⫻600 1.87 400⫻500 0.91

DCF 5– 6 350⫻350 2.76 350⫻400 1.32


3– 4 450⫻450 1.68 400⫻450 1.02
2 500⫻500 1.37– 4.10 400⫻500 0.91–1.37
1 600⫻600 0.93–2.80 400⫻500 0.91–1.37

PIF 5– 6 350⫻350 2.76 350⫻400 1.32


3– 4 450⫻450 1.68 –3.36 400⫻450 1.02–1.53
2 500⫻500 1.37–3.41 400⫻500 0.91–1.37
1 700⫻700 1.67–3.35 400⫻500 0.91–1.82

SWF 1– 6 450⫻450 1.68 400⫻450 1.02


Wall vertical ␳a 0.95
Wall horizontal ␳a 0.63
a
Total area of steel/Gross section area.
b
Area of tension steel/Effective section area.

170 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002

J. Struct. Eng., 2002, 128(2): 169-178


Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Typical test setup and input base motion

of ‘‘g’’. The following consecutive simulations were performed:


Fig. 2. Dimensions and reinforcement arrangement of test model EL0.10, EL0.30, EL0.60, EL0.90, and EL1.20 共EL1.20 was not
structures performed for model BFR due to an accidental setup problem兲.
Fig. 3共b兲 shows a typical achieved base acceleration time history
and the corresponding 5%-damped response spectrum. A comple-
wire having yield strength 195 MPa. A single layer of 3-mm-wire mentary random vibration test was performed after each earth-
mesh at 50-mm spacing was used in the floor slabs. quake simulation, and subsequently, the transfer functions of the
The model bricks for frame PIF were fabricated to the scaled measured floor accelerations to the random input signal were
size with cement-lime mortar. The strength of the model bricks computed to determine the changes of the natural frequencies of
was made higher to allow for cracking to occur along brick joints the models during the course of the tests.
for which similitude to the normal mortar joints was maintained.
The compressive strength of the mortar for joining the model
bricks was 3.5 MPa. Observed Response and Comparison
The models were mounted on the earthquake simulator by
fixing the base girders to the simulator platform. To ensure that
Cracking Pattern and Failure Modes
the model frame responded to the unidirectional excitation within
its plane, a side-support system was provided. A sketch of the test For test EL0.30 which corresponded to the design intensity, only
setup is shown in Fig. 3共a兲. The additional masses, weighing, a few scattered minor cracks were detected on the models, except
respectively, 6,900 kg 共BFR兲, 6,850 kg 共DCF兲, 7,800 kg 共PIF兲, for DCF where more cracks appeared on beams of the middle and
and 7,400 kg 共SWF兲, were distributed and fixed on the floor slabs. right-hand side spans, due to the relatively small overstrength
In an attempt to minimize the additional restraint on the flexural margin in these particular beams 共Lu 1996兲. Rather distinctive
deformation of the beams and slabs, the masses were separated crack patterns appeared after test EL0.60, and further developed
into several blocks and a layer of ‘‘soft’’ lime mortar was placed through test EL0.90, leading to the eventual failure mechanisms
between the mass blacks and the floor slabs. which became apparent during test EL1.20. Fig. 4 illustrates the
The instrumentation was organized so that both overall and crack patterns recorded after test EL0.90. The main features of the
critical local responses of interests were measured. Totally, about different damage patterns are discussed in what follows.
30 channels of measurements were recorded for each model. Frame BFR exhibited a rather uniformly distributed cracking
The models were tested for several levels of simulated earth- pattern. The severe damage at the fifth story is attributable to the
quakes resembling the motion wave form of the El Centro 1940 abrupt reduction 共over 20%兲 of the column cross section size and
N-S component, with a time scale of 冑5.5. In the description the subsequently intensified higher mode ‘‘whipping’’ effects. A
which follows, each earthquake simulation is labeled by ‘‘EL’’ similar phenomenon was also seen in frame DCF and this will be
followed by a numeral indicating the peak acceleration in terms further discussed later. For test EL0.90, the general response of

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002 / 171

J. Struct. Eng., 2002, 128(2): 169-178


Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Measured base shear versus top displacement relationships

the wall bottom was almost cut free from the base, while connec-
tion failure developed at the beam-wall adjoining regions. Some
horizontal cracks also appeared on higher levels of the wall, a
phenomenon which could be attributed to the higher-mode par-
ticipation as well as the effects of the opposing displacement
shapes between the wall 共bending兲 and the frame 共shear兲.

Overstrength and Displacements


Fig. 5 shows the measured envelope base shear versus the top
displacement relationships for the test models. The design base
Fig. 4. Crack patterns recorded after test EL0.90 shears are also indicated in Fig. 5 for a comparison.
The variation of the design base shear in terms of the total
weight 共W兲, in the range of 0.15– 0.25W as shown in Fig. 6共a兲,
BFR remained stable, however, cracking widened substantially was the result of considering the varying natural period and the
共maximum crack width on the model exceeded 1.0 mm兲, indicat- pertinent irregularities of different frames. The consideration
ing extensive yielding, while spalling of concrete occurred in the proved to be adequate as one can observe the same variation trend
lower stories. Plastic hinges appeared to also occur in several from the measured base shear demands during the initial EL0.10
columns, and this may be attributed to the ‘‘relaxation’’ of column test, also shown in Fig. 6. Because of the increased demands
design moments from satisfying equilibrium around joints, as (V D,B ) in PIF and SWF, despite that these two models achieved a
mentioned in the description of the design, as well as some en- much higher base shear strengths (V R,B ) than BFR, the base shear
hancement of the beams and slabs due to the fixing of additional overstrength factor, i OS,B ⫽V R,B /V D,B , was found to range
masses. It is noted that the diagonal cracks that occurred at some closely, between 3 and 3.5, in these three models 关Fig. 6共b兲兴.
beam–column joints were generally light and did not become Relatively low base shear overstrength was achieved in DCF due
critical thanks to the heavy joint reinforcement as shown in Fig. 2. to the lower available base shear strength compared to BFR.
The cracking pattern of DCF demonstrated the combined ef- Similar base shear overstrength, on the order of 3, was ob-
fects of a soft first story and the soft middle and right-hand side served in many previous experimental investigations under rather
spans. During EL0.90, severe spalling of concrete and buckling of deterministic laboratory conditions 共Charney and Bertero 1982;
reinforcement were observed on the top and bottom of the first- Shahrooz and Moehle 1990兲. The known contributing factors in-
story columns, indicating the formation of a soft-story mecha- clude the material overstrength, the surplus amount of reinforce-
nism. A permanent mechanism set was observed during test ment, as well as structural redundancy. With such overstrength,
EL1.20. the earthquake intensity which would generate the design antici-
In frame PIF, the enlarged first-story columns appeared to be pated inelastic response is expected to increase. For this reason,
effective in preventing what would usually be anticipated a soft the test EL0.90 (PGA⫽3⫻0.3 g⫽0.9 g) is considered herein as
story mechanism. Instead, the cracking spread into the beam- the reference simulation for evaluating and comparing the inelas-
column members as well as the masonry walls throughout the tic response of the models.
frame. 共The shear-looking cracks above the midheight of the first- Based on Fig. 5, an estimation of the overall ductility demands
story exterior columns, appeared at the late testing stage, was at different test stages can be obtained 共Lu 1996兲. Fig. 6共c兲 com-
found to result from an imperfect cast joint. This problem was pares the ductility demands on the four models for test EL0.90. It
avoided during the casting of other models.兲 can be observed that for all models except DCF the displacement
The cracking pattern of model SWF represented an over- ductility demands were in the range of 3 to 4, which correlates
whelming wall rocking response. Major horizontal cracks oc- very well with the design anticipated medium ductility response.
curred across the wall bottom section during test EL0.60 and Frame DCF exhibited an approximately 35% higher ductility de-
further developed to cause a rupture of the vertical reinforcement mand. The normalized ductility demands, obtained by multiplying
during test EL0.90, while intense cracking appeared in the beam– the measured ones with an overstrength correction factor equal to
wall adjoining regions due to the large rotation demands associ- (i OS,B /3.0), show a fair independence from the frame configura-
ated with the intensified wall rocking. At the final stage 共EL1.20兲, tion and all fall in the range from 3 to 4.

172 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002

J. Struct. Eng., 2002, 128(2): 169-178


Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Comparison of top displacements and critical first-story drifts

Comparison of Hysteretic Behavior


The hysteretic behavior reflects the energy dissipation capacity as
well as the inelastic stability of a structure. Fig. 9共a兲 shows the
base moment versus top drift 共top displacement divided by frame
height兲 hysteretic relations of the test models.
BFR maintained a stable hysteretic response throughout the
tests while the maximum top displacement reached 87 mm 共top
drift 2.4%兲. For model DCF, however, considerable strength and
Fig. 6. Comparison of base shear strength and top displacement stiffness degradation occurred with the formation of a mechanism
ductility demands during test EL0.90, while the top drift reached 3%.
Model SWF exhibited a good hysteretic behavior in contrast to
that of frame PIF where pronounced pinching occurred through-
With regard to the control of absolute drift, Fig. 7 compares out the tests. The pinching in PIF was characteristic and it can be
the top displacements and the first-story drifts attained in the four attributed to the imperfect integrity of the infill walls with the
models during various stages of testing. As can be seen, models surrounding RC frame as well as the behavior of the masonry
PIF and SWF effectively reduced the displacements, by 50 and
40%, respectively, compared to BFR, during low to moderate
responses 共up to EL0.60兲. However, the percentage reduction of
displacements notably declined with an increase of the inelastic
response, indicating more rapid deterioration of the RC wall and
the infill walls at advanced inelastic stages. As expected, frame
DCF exhibited much higher first-story drift.
To further examine the distribution of inelastic deformation
along the structure height, Fig. 8 shows the profiles of the dis-
placement and interstory drifts measured during various tests. The
first-mode dominance is readily observed. Nevertheless, all
frames exhibited more or less a ‘‘kink’’ in the displaced shapes at
the fifth story, indicating an appreciable higher mode contribu-
tion. The severe kink at the fifth story of model BFR and DCF
highlighted the adverse effects of abrupt changes to the column
dimensions and thereby the story strength. In model PIF, the pres-
ence of infill walls appeared to significantly increase the story
stiffness as long as the infills remained intact, subsequently, the
drifts in the infilled stories were smaller than the open first story
in spite of its substantially enlarged column cross section. The
infill walls deteriorated rapidly during advanced inelastic re-
sponse such that the second-story drift turned to be more critical
and approached 2.3% during test EL0.90. Frame SWF exhibited a
smooth drift profile due to the relative rigidity of the rotating RC
Fig. 8. Comparison of displacement and interstory drift profiles
wall.

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002 / 173

J. Struct. Eng., 2002, 128(2): 169-178


Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Measured and computed story shear envelopes and corre-
sponding lateral force distributions

weakness introduced by an abrupt reduction 共more than 20%兲 of


the column cross section size compared to the story below. While
the damage could be predicted by a nonlinear dynamic analysis, it
could not be anticipated with an equivalent static analysis adopt-
ing either an inverted triangle or uniform distribution of lateral
Fig. 9. Comparison of hysteretic behavior forces 共Lu 1996兲. This indicates that there existed a significant
dynamic amplification effect at the top of these frames, and an
unsafe design of the upper stories could result if such an effect is
wall itself against shear. Despite the poor hysteretic loops, how- not appropriately taken into account. To extend the evaluation of
ever, the frame members were better protected in PIF than in BFR this phenomenon, Fig. 10 plots the profiles of the envelope story
due to the reduced interstory drifts. shear forces measured during the elastic 共EL0.10兲 and an inelastic
A quantitative comparison of the general hysteretic energy dis- 共EL0.60兲 test, as well as the corresponding distributions of the
sipation capacities is given in Fig. 9共b兲 in terms of the equivalent lateral forces, for all four test models. Note that the magnitude of
hysteretic damping coefficient, which is regarded as a key param- the forces is normalized so that the base shear is equal to the
eter in a performance-based design procedure 共Priestley 2000兲 design base shear in each model. The design lateral force distri-
and is evaluated herein by the following expression, butions and those predicted by linear and nonlinear dynamic
frame analyses are also shown in Fig. 10.
E hys
␰ eq⫽ (1) As can be observed, for frames BFR and DCF at the elastic
4␲E el stage, both the equivalent static and dynamic lateral force distri-
where E hys and E el⫽hysteretic and elastic strain energy of the butions reasonably resemble the measured distributions. For the
system, respectively, at peak displacement, i.e., the secant stiff- inelastic response, however, the measured lateral force distribu-
ness of the equivalent system with the above hysteretic damping tion exhibits a substantial amplification at the top level and the
corresponds to the peak displacement on the base moment versus magnitude appears to be predictable only with a nonlinear dy-
top drift curves. Generally speaking, a 10% hysteretic damping namic analysis. An equivalent static analysis could resemble the
coefficient was achieved in all models at advanced inelastic re- measured response if a concentric lateral force, in this case equal
sponse, except for PIF for which the coefficient was approxi- to 20% of the total base shear, is added to the top level while the
mately 6%. rest is distributed over height. It is interesting to note that for
models PIF and SWF where no significantly weaker stories were
present, no apparent top amplification is observed and the equiva-
Discussion on Performance and Evaluation lent lateral force distribution appears to result in an adequate
of Design story-shear envelope.
The general interpretation of this observation may be as fol-
lows: 共1兲 For frames with significant vertical irregularity, for ex-
Vertical Irregularity and Distribution of Lateral Inertial
ample with a regularity index 共described in the next section兲 less
Forces
than 0.70 as in the case of DCF, a reliable prediction of the in-
Frames DCF and BFR both exhibited concentrated damage at the elastic response would necessitate a nonlinear dynamic analysis.
fifth-story columns. The direct cause of this appeared to be the As an approximation, an equivalent static analysis may be accept-

174 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002

J. Struct. Eng., 2002, 128(2): 169-178


Fig. 11. Story-shear overstrength profile of frame discontinuous col-
Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

umn frame 关dashed lines: modified design corresponding to Fig.


12共b兲兴

able provided the top amplification effect is appropriately taken


into account. 共2兲 The fact that the distribution of the elastic story
shear demands shows much less effect of the discontinuity of the
initial story stiffness indicates that essentially it was the story
strength 共and thereby the postyield secant stiffness兲 distribution
that governed the development of the general inelastic behavior in
the frames. Further discussion follows. Fig. 12. Comparison of responses between original DCF and its
modified design
Evaluation of Discontinuous-Column Irregularity and
Soft Story Mechanism
The removal of a first-story column created a significant draw- The subsequent increase of the first-story drift was so substantial
back in the topology of frame DCF. A general countermeasure in that it would render any enhancement of the local ductility mean-
dealing with all vertical irregularities, as explicitly prescribed in ingless.
EC8 共1988兲 is a reduction of the design q-factor 共thus an increase The improvement of the inelastic behavior of the DCF system
in the design base shear兲. The reduction coefficient is taken as a appears to require an effective strengthening of the first-story col-
function of the regularity index, ␣ r , which is defined as the ratio umns. To explore this, a nonlinear dynamic response analysis was
of the minimum story-shear overstrength factor to the average, carried out using the program DRAIN-2D with a modified Takeda
共 i OS 兲 min restoring force model 共Kanaan and Powell 1973; Lu 1996兲. The
␣ r⫽ (2) response during EL0.60 was selected to represent a significantly
共 i OS 兲 ave
nonlinear and yet stable situation which is deemed adequate for
and the story shear overstrength factor, i OS , is calculated as an inelastic evaluation. In order to apply the Takeda model for
VR columns, each column member was approximated to undergo a
i OS ⫽ (3) constant axial force during the response such that an envelope
VS
moment–rotation relationship could be prescribed. Other assump-
where V S ⫽design story shear force and V R ⫽available story shear tions included a section stiffness to be 40% of the uncracked
strength taking into account all vertical members in the story. gross-section stiffness for all beams and columns, and a viscous
For bare RC frames, V R is computed by equilibrium assuming damping coefficient of 3%. Fig. 12共a兲 shows a good agreement
all columns yield concurrently at both ends. Fig. 11 depicts the between the computed and measured interstory drifts for the test
story shear overstrength (i OS ) profile for frame DCF 共solid lines兲. frame DCF. The above calibrated modeling scheme was then ap-
From this profile, the ␣ r index is calculated to be 0.67 with the plied to compute the response of a modified design, in which the
first story being the weakest. The frame is thus considered to be first-story column section was increased from 110⫻110 to 120
irregular but within an acceptable range 共␣ r is not allowed to be ⫻120 mm 共600⫻600 to 650⫻650 mm at full scale兲 together with
smaller than 0.55 while ␣ r ⭓0.85 is classified as regular兲. an increase of the longitudinal reinforcement as shown in Table 2.
Comparing the general response between DCF and BFR This modification leads to an increase in the story-shear strength
shown in Figs. 8 and 9, it may be fair to say that the two frames by approximately 30%, resulting in a smoothed overstrength pro-
exhibited practically the same performance up to a moderate in- file as shown in Fig. 11 共dashed lines兲. The weak fifth-story col-
elastic response 共EL0.60兲. However, frame DCF abruptly became umns are also modified to have the same dimension and rein-
unstable following the formation of the first-story mechanism. forcement as the fourth story.

Table 2. Design Modifications on Frame Discontinuous Column Frame 共Modified Values in Parentheses兲
Column reinforcement 共No. of bars per face兲
Column section
Story 共full scale, mm兲 Exterior 共left-hand side兲 Interior Exterior 共right-hand side兲
1, lower 600⫻600 (650⫻650) 4 共4兲 3 共3兲 3 共3兲
1, upper 600⫻600 (650⫻650) 2 共3兲 2 共3兲 2 共3兲
5 350⫻350 (450⫻450) 2 共2兲 2 共2兲 2 共2兲

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002 / 175

J. Struct. Eng., 2002, 128(2): 169-178


Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Comparison of strength and stiffness characteristics of noninfilled 共BFR兲 and infilled 共PIF兲 second stories

Considerable improvement of the nonlinear behavior is 1 1


achieved with the modified design, as clearly indicated in the V R ⫽⌺V RC⫹ 兺 V RW⫽ ␥ C 兺 A C ␶ RC⫹ ␥ W 兺 A W f v k (4)
computed interstory drift profile and the plastic hinge pattern
shown in Fig. 12共b兲. The concentration of inelastic deformation at where V RC and V RW⫽contribution of shear resistance from the
the first and fifth stories is effectively eliminated. This result il- columns and the infill walls, respectively; A C denotes the hori-
lustrates that, whereas the disadvantage in topology is sometimes zontal cross sectional area of columns and A W that of infill walls;
unavoidable, a generally satisfactory inelastic behavior may be ␶ RC⫽shear strength of concrete; f v k ⫽characteristic tensile
achieved if the geometric weakness is compensated such that a strength of the masonry wall; while ␥ C and ␥ W ⫽partial factors.
smooth distribution of the actual story-shear overstrength pre- The V R for the open first story is calculated as described in the
vails. previous section.
The design of frame PIF revealed that, following the afore-
mentioned procedure, it was essentially the requirement for ␣ r
Effects of Infill Walls and Design of Nonuniformly ⭓0.55, rather than the design seismic load, that governed the
Infilled Frames proportioning of the frame members in the open first story. The
Significant effects of the infill walls on the frame behavior have fact is that Eq. 共4兲 treats the infilled story like a composite shear
been demonstrated by the general displacement response of frame panel and hence tends to overestimate considerably the story-
PIF 共Figs. 8 and 9兲. To give a more quantitative estimation of the shear strength 共a rough comparison of the computed story-shear
extent to which the infill walls contribute in the story strength and strength using actual material properties and the measured value
stiffness, Fig. 13 compares the story shear versus interstory drift for the second story of PIF indicates an overestimation of ap-
relationships between the infilled second story in PIF and the bare proximately 100%兲. This leads to a substantial strengthening of
second story in BFR. Note that the total flexural reinforcement in the first-story columns than what is implied by satisfying the
the second-story columns of PIF was 1.7 times as much as that in minimum ␣ r requirement, and subsequently resulted in a much
model BFR 共cross section is the same兲, and this is estimated to more desirable overstrength profile in PIF that enabled a distrib-
contribute about 60% increase of the story strength and, to a uted inelastic deformation. As can be observed from Fig. 14, the
lesser extent, the stiffness. real story-shear overstrength profile, obtained by considering the
As can be seen from Fig. 13, the second story of PIF exhibited measured story shear strengths, does exhibit a satisfactory
an actual increase of more than 120% in both strength and stiff- smoothness comparing to the design-anticipated scenario. Note
ness as compared to BFR, about 60% of which can therefore be that the design overstrength profile shown here is evaluated using
attributed to the infill walls. As a result, the interstory drift was the design material properties to reflect what could be perceived
reduced by approximately 50%. The infilled story maintained a at the design stage.
stable hysteretic behavior at maximum interstory drift of 2.3%. This evaluation demonstrates again the significance of the
Marked strength degradation occurred when the maximum drift story-shear overstrength distribution. Bearing this in mind, further
approached 3%, at which level severe separation of mortar joints refinement of the design procedure for PIF frames can be straight-
was observed. Hence, the interstory drift of 3% may be regarded
as an overall stability limit for the infilled stories.
With such significant effects of the infill walls, of particular
interest is the fact that the open first story of PIF was effectively
prevented to develop into a soft story. The design preventive mea-
sures seemed to be effective, and a scrutiny of the proportioning
procedure is helpful for an adequate treatment of this widely used
frame configuration.
The essential point in the design of frame PIF was the mini-
mum requirement for the vertical regularity index, ␣ r ⫽0.55, and
for the infilled stories, the story-shear overstrength factor 关Eq. 共3兲兴
was calculated taking into account the contribution of the infill
Fig. 14. Story-shear overstrength profile of frame PIF
walls, such that

176 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002

J. Struct. Eng., 2002, 128(2): 169-178


forward provided a sound estimation of the story-shear strength inelastic deformation and ductility demand were attained at
involving masonry walls is available. The revised version of EC8 increased earthquake intensity approximately proportional to
共1994兲 adopts increased design action effects at a partially or the base shear overstrength, i.e., PGA⫽0.9 g. The hysteretic
noninfilled story according to the relative reduction of the story response remained essentially stable at this stage, while the
resistance due to the removal of the masonry walls. This method overall ductility demand ranged between 3 to 4. Thus, the
also aims at a smoothed overstrength distribution, and hence, the design provisions adopted for medium ductility class framed
effectiveness relies on the calculation of the masonry wall resis- structures proved to be generally adequate.
tance as well. 2. For the story-shear controlled frames including the infilled
frames, the distribution of the story-shear overstrength factor
is observed to be a quite stable indicator of the distribution
Critical Top and Interstory Drifts
of inelastic deformation irrespective of the frame configura-
Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

The horizontal displacement is commonly used as a criterion to tion, and hence, can be employed to quantify the frame regu-
quantify the structural performance and identify the ultimate state larity. In this respect, a rational distribution of the seismic
for collapse. The design critical drifts given by different codes lateral load is of particular importance for the design analy-
typically fall within 1.5 and 2.5% 共EC8 1994; Uniform Building sis. In lieu of a nonlinear dynamic analysis for frames with
Code 1997兲. Some previous observations based on simple mem- significant vertical irregularity, which may be defined as hav-
ber tests recommended a drift limit as high as 6% for actual ing a regularity index less than 0.70, an equivalent static
failure 共Roufaiel and Meyer 1983兲. The current test results sug- analysis may be acceptable provided a concentric lateral
gest that it is reasonable to expect a stable frame behavior if the force, tentatively taken to be 20% of the total base shear, is
displacement response falls within both 共1兲 top drift limit of added on the top level. For frames having a relatively uni-
2.5%, and 共2兲 interstory drift limit of 3%. Pertinent evidences are form overstrength along their height, a conventional equiva-
described in what follows. lent static load distribution 共inverted triangle type兲 is deemed
1. At a drift slightly lower than the limits just proposed, the to be adequate.
regular frame BFR maintained a stable hysteretic response 3. Merely increasing the design base shear force does not actu-
共Fig. 7, EL0.90兲. However, spalling of cover concrete oc- ally remove the adverse irregularity effects due to drawbacks
curred at several beam and column ends, while diagonal in topology such as in frame DCF, consequently a soft-story
cracking appeared on some beam–column joints. In conjunc- mechanism could develop leading to abrupt loss of structural
tion with the observation of a fully developed cracking pat- stability. Nonlinear analysis indicates that such vertically ir-
tern showing spread plastic hinge locations 共Fig. 4兲, it may regular frames could also behave in a regular manner if the
be deduced that the frame was approaching the mechanism geometrically weak-story columns are proportioned such
formation leading to instability. that a story-shear overstrength factor similar to the adjacent
2. In the case of the irregular DCF frame, the response was stories is achieved.
observed to remain perfectly stable during test EL0.60 for 4. Significant effects of the masonry-infill walls were observed.
which the maximum drifts were about 20% lower than the In the infilled stories, an increase of the story-shear strength
proposed limits. The frame apparently became unstable at and stiffness by 60% was attributable to the masonry walls
the maximum response during the subsequent test, with a up to an interstory drift of 3%. The prevention of a soft-story
sharp increase of the first-story drift as well as the top dis- mechanism to occur in the open first story of the PIF frames
placement. Although it is physically reasonable to lower requires a smoothed overstrength distribution taking into ac-
somehow the 2.5% top drift limit for such irregular frames count the infill walls in the adjacent stories, and for this
with localized damage, it does not appear to be practically purpose, more accurate estimation of the resistance of the
necessary since in these situations, the 3% interstory drift infilled stories under cyclic loading is necessary.
limit is expected to govern. 5. Comparing to the regular frame, the wall-frame system ex-
3. The maximum top and interstory drifts of the wall-frame hibited a good drift control 共reduction by 40%兲 up to a mod-
SWF were 1.9 and 2.3%, respectively, during test EL0.90 erately inelastic stage with a displacement ductility demand
and the hysteretic response was satisfactory 共Figs. 7 and 9兲, on the order of 2. Severe concentration of inelastic deforma-
however concentrated damage occurred at the wall base. tion occurred at the wall base during advanced inelastic re-
Marked strength degradation occurred during test EL1.20 sponse, subsequently, the rocking of the wall intensified, re-
关Fig. 5共c兲兴 while critical local damage occurred near the wall sulting in the system performance to deteriorate more rapidly
base and at the beam-wall adjoining regions. Since the maxi- than the regular frame. Further improvement of the inelastic
mum top and interstory drifts increased to 3 and 4%, respec- behavior of the wall-frame systems requires effective coun-
tively at this stage, by interpolation the proposed drift limits termeasures against the wall rocking.
can be obtained. 6. Based on the hysteretic response and damage observations, it
4. The proposed drift limits appear to apply also to the case of is recommended that as a general guideline a top drift of
the infilled frame PIF as discussed in the previous section. 2.5% and interstory drift of 3% may be considered critical
concerning the inelastic stability of multistory framed struc-
tures.
Conclusions and Recommendations
Acknowledgments
1. A general base-shear overstrength factor on the order of 3 The experimental work was conducted at the National Technical
was observed for the various multistory frames under the University of Athens, under the guidance of Professor T. P. Tas-
rather deterministic construction and testing conditions. As a sios. Dr. G.-F. Zhang participated in the shaking table tests. As-
result, the test structures remained virtually elastic with sistance received from Professor P. Carysis, Professor E. Vintzi-
minor cracking under the simulated earthquake of design leou, Dr. H. Mouzakis, and Dr. C. Zeris during the experiment is
intensity with PGA equal to 0.3 g. The design anticipated greatly acknowledged.

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002 / 177

J. Struct. Eng., 2002, 128(2): 169-178


References Negro, P., and Verzeletti, G. 共1996兲. ‘‘Effect of infills on the global be-
havior of R/C frames: energy consideration from pseudodynamic
Charney, F. A., and Bertero, V. V. 共1982兲. ‘‘An evaluation of the design tests.’’ Earthquake Eng. Struct. Dyn., 25共8兲, 753–773.
and analytical seismic response of a seven-story reinforcement con- Priestley, M. J. N. 共2000兲. ‘‘Performance based seismic design.’’ Proc.,
crete frame-wall structure.’’ Rep. No. UCB/EERC-82/08, Univ. of 12th World Conf. on Earthquake Engineering, Auckland, New
California, Berkeley, Calif. Zealand, Paper ID: 2831.
Eurocode 8 共EC8兲. 关1988 共draft兲兴; 共1994兲. ‘‘Design provisions for earth- Roufaiel, M. S. L., and Meyer, C. 共1983兲. ‘‘Analysis of damaged concrete
quake resistance of structures.’’ CEN 共European Commission for frame buildings.’’ Rep. NSF-CEE81-21359-1, Dept. of Civil Engi-
Standardization兲/TC250/SC8, Brussels, Belgium. neering and Engineering Mechanics, Columbia Univ., New York.
Kanaan, A. E., and Powell, G. H. 共1973兲. ‘‘DRAIN-2D: A general- Schultz, A. E. 共1990兲. ‘‘Experiments on seismic performance of RC
purpose computer program for dynamic analysis of inelastic plane frames with hinging columns.’’ J. Struct. Eng., 116共1兲, 125–145.
structures.’’ Rep. No. UCB/EERC-73/6, Earthquake Engineering Re- Shahrooz, B. M., and Moehle, J. P. 共1990兲. ‘‘Evaluation of seismic per-
Downloaded from ascelibrary.org by MANIPAL INST OF TECHNOLOGY on 08/24/19. Copyright ASCE. For personal use only; all rights reserved.

search Center, Univ. of California, Berkeley, Calif. formance of reinforced concrete frames.’’ J. Struct. Eng., 116共5兲,
Lu, Y. 共1996兲. ‘‘Study of seismic behavior of reinforced concrete frames 1402–1422.
having vertical irregularities.’’ Doctoral dissertation, National Techni-
Tassios, T. P. 共1992兲. ‘‘Modeling of structures subjected to seismic load-
cal Univ. of Athens, Athens.
ing.’’ in Small Scale Modelling of Concrete Structures, F. A. Noor and
Lu, Y., Tassios, T. P., Zhang, G.-F., and Vintzileou, E. 共1999兲. ‘‘Seismic
L. F. Boswell, eds., Elsevier, New York.
response of reinforced concrete frames with strength and stiffness
International Conference of Building Officials 共ICBO兲. 共1997兲. Structural
irregularities.’’ ACI Struct. J., 96共2兲, 221–239.
Moehle, J. P., and Sozen, M. A. 共1980兲. ‘‘Experiments to study earth- engineering design provisions.’’ Uniform building code, Vol. 2, Whit-
quake response of R/C structures with stiffness interruption.’’ Civil tier, Calif.
Eng. Studies, Structural Res. Series No. 4.82, Univ. of Illinois, Ur- Wood, S. L. 共1992兲. ‘‘Seismic response of R/C frames with irregular
bana, Ill. profiles.’’ J. Struct. Eng., 118共2兲, 545–566.

178 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2002

J. Struct. Eng., 2002, 128(2): 169-178

You might also like