EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2002; 31:353–378 (DOI: 10.1002/eqe.112)
Eect of masonry inÿlls on seismic performance of a 3-storey
RC frame with non-seismic detailing
Han-Seon Lee
∗. †
and Sung-Woo Woo
Department of Architectural Engineering. Korea University. Seoul 136-701. Korea
SUMMARY
The objective of this study is to investigate the eect of masonry inÿlls on the seismic performance of
low-rise reinforced concrete (RC) frames with non-seismic detailing. For this purpose, a 2-bay 3-storey
masonry-inÿlled RC frame was selected and a 1 : 5 scale model was constructed according to the
Korean practice of non-seismic detailing and the similitude law. Then, a series of earthquake simulation
tests and a pushover test were performed on this model.
When the results of these tests are compared with those in the case of the bare frame, it can
be recognized that the masonry inÿlls contribute to the large increase in the stiness and strength of
the global structure whereas they also accompany the increase of earthquake inertia forces. The failure
mode of the masonry-inÿlled frame was that of shear failure due to the bed-joint sliding of the masonry
inÿlls while that of the bare frame appeared to be the soft-storey plastic mechanism at the ÿrst storey.
However, it is judged that the masonry inÿlls can be beneÿcial to the seismic performance of the
structure since the amount of the increase in strength appears to be greater than that in the induced
earthquake inertia forces while the deformation capacity of the global structure remains almost the same
regardless of the presence of the masonry inÿlls. Copyright ? 2001 John Wiley & Sons, Ltd.
KEY WORDS: masonry inÿlls; reinforced concrete frame; earthquake simulation test; pushover test;
strength; stiness; failure mode
1. INTRODUCTION
Most building structures, which are normally medium- to low-rise reinforced concrete (RC)
frames in Korea, have not been engineered to resist major or moderate earthquake. Masonry
inÿlls have usually been used in these structures for architectural or aesthetic reasons, and
they have been normally considered as non-structural elements, and their presence has been

Correspondence to: Han-Seon Lee, Department of Architectural Engineering, Korea University, Seoul 136-701,
Korea.

E-mail: hslee@korea.ac.kr
Contractgrant sponsor: Ministry of Construction and Transportation, the Republic of Korea, private companies
including SSangYong Engineering and Construction Corp., DongBu Construction Corp., Hyundai Construction Corp.,
and Dong Yang Structural Safety Consultants Corp.
Received 18 January 2001
Revised 2 April 2001
Copyright
?
2001 John Wiley & Sons, Ltd. Accepted 31 May 2001
354 H.-S. LEE AND S.-W. WOO
ignored by engineers. However, even though they are considered non-structural elements, they
tend to interact with the frame when the structures are subjected to lateral loads. In most of
the current seismic codes, the inuence of non-structural masonry inÿlls is ignored.
The performance of the structure, however, can be greatly improved by the increase of
strength arising from the masonry inÿlls. On the contrary, this increase in strength also ac-
companies the increase of the initial stiness of the structure and thus may result in an adverse
increase of the inertia force. The damage of the structure may be reduced by dissipating a
considerable portion of the input energy in the masonry inÿlls or in the interface between
the inÿlls and the frame. But, if the conÿguration of the inÿlls is irregular, they can induce
signiÿcant local damages to the structural elements [1].
The behaviour of the masonry-inÿlled frame under lateral loads has been investigated by
several researchers. Bertero et al. [2] performed the experimental investigation of a series of
quasi-static cyclic and monotonic load tests on 13-scale models of the lower 3-12 storeys
of an 11-storey 3-bay RC frame inÿlled in the outer two bays. Dierent panel material and
reinforcement combinations were tested. In this study, the eective interstorey lateral stiness
of inÿlled frames was 5.3–11.7 times the lateral stiness of the bare frame depending on
the type of inÿll. The maximum lateral resistance of inÿlled frames was 4.8–5.8 times that
obtained for the bare frame. Mehrabi et al. [3] tested 12 12-scale, single-storey, single-bay,
RC frames inÿlled with concrete block masonry that were designed in accordance with code
provisions [4]. Their objectives were to evaluate the inuence of the relative strength and
stiness of inÿll panels with respect to those of the bounding frame, the lateral load history,
the panel aspect ratio, the magnitude and distribution of vertical loads, and the adjacent inÿlled
bays on the performance of these frames. The experimental results indicate that inÿll panels
can signiÿcantly improve the performance of RC frames. However, specimens with strong
frames and strong panels exhibited a better performance than those with weak frames and
weak panels in terms of the load resistance and energy-dissipation capability. The lateral
loads developed by the inÿlled frame specimens were always higher than that of the bare
frame.
Negro et al. [5] conducted a series of pseudo-dynamic tests on a full-scale 4-storey RC
building designed according to Eurocodes 2 and 8 [6. 7]. The tests were conducted on the
bare frame, as well as on the frame with two dierent conÿgurations of non-structural ma-
sonry inÿlls. The experimental results indicate that the presence of light non-structural ma-
sonry inÿlls can change the response of the structure to a large extent and the presence of
a regular pattern of inÿlls to a large extent prevents energy dissipation from taking place in
the frame and that irregularities in the panels result in unacceptably larger damage to the
frame.
Fardis and Calvi [8] demonstrated through the analysis of the damage resulting from recent
earthquakes that in-plan irregular distribution of inÿlls does not modify the behaviour of the
building to a very great extent. Fardis et al. [9] presented that the inÿlls in the transverse
direction have an important beneÿcial eect on seismic response and performance under both
unidirectional and bidirectional motions in the response of a 2-storey RC frame structure with
two adjacent sides inÿlled through shaking table tests and non-linear dynamic analyses.
The studies of several researchers [2. 3. 5] are generally based on the results of quasi-static
or pseudo-dynamic tests. However, the eect of energy dissipation by the dynamic interaction
between inÿll panels and the frame cannot be simulated with sucient accuracy as far as
these quasi-static experimental techniques are used. Also, the information on the behaviours
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 355
of the masonry-inÿlled RC frames with non-seismic detailing when subjected to earthquake
ground motions is rather scarce worldwide. Therefore, this study concentrates on the ac-
tual responses of masonry-inÿlled RC frame with non-seismic detailing under the simulated
earthquake ground motions. After earthquake simulation tests, the monotonically-increasing
lateral-load test or the pushover test was performed to ÿnd out the ultimate capacities of the
model. By comparing the results of these tests with those in the case of the bare frame [10],
the signiÿcance or the eect of masonry inÿlls are evaluated.
2. EXPERIMENT
2.1. Test model and experimental set-up
First, a 3-storey RC frame, which has been in use as a police oce building in Korea
was selected as a prototype and the seismic responses of a typical portion of this building
structure with the masonry inÿlls were studied. Then, considering the capacity of the shaking
table used in earthquake simulation tests, a 1 : 5 scale model was constructed. The plan
and elevation of the 1 : 5 scale model are shown in Figure 1(a) and (b). The compressive
strength of concrete, !

c
, in the prototype structure is assumed to be 20.6MPa and the nominal
yield strength of reinforcement, 294.2 MPa. The typical sections of members and the details
regarding transverse steel, anchorage and splice are shown in Figure 1(c)–(h). It is essential to
maintain the similitude in the material properties between prototype and model reinforcement.
However, it was dicult to make the cross-sections of the model reinforcement conform
exactly to the similitude law. So, the yield forces rather than yield stresses were selected as
the target to be achieved in annealing the model reinforcement. An electric furnace with a
3-zone vacuum tube was designed and used. The deformation on the surface of the model
reinforcements was made using a deforming device. Reinforcing bars D22 and D10 in the
full-scale structure match D3 and D2 in the model, respectively. The target yield forces
derived from similitude requirements are shown in Table I. The achieved average yield force
is approximately 10 per cent less than the target yield force in the case of D3. The model
concrete was made using type I Portland cement. The average strength of the model concrete
at the time of testing was about 29.6MPa. The techniques to construct this model according to
the similitude requirements, and the process and detailed results of the earthquake simulation
tests of the models of the bare and masonry-inÿlled frames are presented in the report [10].
The test model has non-seismic details as follows: (1) a lap splice at the bottom of the
column, (2) large spacings of hoops, (3) no hoops in beam-column joints, (4) no use of 135

seismic hooks, and (5) the special style of anchorage in the joints. That is, the length of
tension and compression anchorage are usually 40d
b
and 25d
b
, respectively, from the critical
section, where d
b
means the nominal diameter of reinforcement. Moreover, the length of
the tail in the hook is included in this anchorage length and the tails of the anchorage of
the bottom bars in beams usually direct downward into the exterior columns as shown in
Figure 1(h).
The used test facilities are the shaking table and actuator system in the laboratory at Hyundai
Institute of Construction Technology. Two layouts of masonry inÿlls were used for earthquake
simulation tests: that is, fully inÿlled frame (FIF) and partially inÿlled frame (PIF). The ex-
perimental set-ups for the shaking table tests are shown in Figure 2(a)–(c) and Figure 3.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
356 H.-S. LEE AND S.-W. WOO
Figure 1. Plan, elevation and details (unit: mm).
Two displacement transducers and accelerometers were installed at each oor to measure the
eect of torsion due to accidental unsymmetry of two frames. To measure storey drifts, a
reference frame was used as shown in Figure 2. The white-noise (random-vibration) test,
which was performed before the earthquake simulation tests, indicated that the natural fre-
quency of the reference frame is approximately 40 Hz. Therefore, the reference frame was
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 357
Figure 2. Test model and experimental set-up (unit: mm).
considered to have the sucient rigidity to measure accurately the drifts of the model. A
load cell was installed in the mid-height of the column at the ÿrst storey to measure the
shear force of each column. To measure the local responses such as the end angular rota-
tions in the possible plastic hinge regions, 16 displacement transducers were used. And also, to
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
358 H.-S. LEE AND S.-W. WOO
Table I. Similitude requirement for reinforcements.
Member and use Prototype (stress) 15 Model (force)
Yield strength Tensile strength Yield strength Tensile strength
Beam Bar Stress Force Stress Force Bar Stress (1) ×125

Stress (2) ×125
and column (MPa) (kN) (MPa) (kN) (MPa) (kN) (MPa) (kN)
(1) (2)
Stirrup D10 358 25.5 530 37.8 D2 325 1.02 481 1.51
and hoop
Main D22 358 138.5 530 205.3 D3 784 5.54 1161 8.21
reinforcement

Target yield force in annealing.
Figure 3. Experimental set-up for shaking table tests.
measure the strains at the centre of the masonry inÿlls, strain gauges were diagonally attached
in the plane of masonry inÿlls. The amount of artiÿcial mass added to the model according
to the similitude law [11] was calculated, and steel plates, each of which has the dimension
W ×D×L=9 cm×4 cm×35 cm (0.097 kN), were used as the artiÿcial mass. But this is
not shown in Figure 2 for clarity. After the series of earthquake simulation tests have been
conducted on the FIF model, there appeared to be only minor cracks on the masonry inÿlls
with the frame itself remaining intact. Therefore, a portion of masonry inÿlls were removed
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 359
Figure 4. Experimental set-up for pushover test.
as shown in Figure 2(c) and then this model, deÿned as PIF, was again subjected to the same
series of earthquake simulation tests as the FIF.
The experimental set-up for the pushover test is shown in Figure 2(d) and Figure 4. The
roof drift was obtained by averaging the two measured values in both Frame A and Frame
B. Two dierent data acquisition systems were used due to the limitations in the number of
available channels in the data acquisition systems, but the measured data were synchronized
with each other system by comparing the displacements of transducers D5 and D6 installed
at the same location in Figure 2(d) but belonging to the dierent data acquisition system.
The experimental results were interpreted, assuming that the behaviour of Frame A represents
that of the whole model structure in both earthquake simulation test and pushover test. The
displacements at the second oor (D2) and third oor (D3) were those measured in the middle
of both frames, and the lateral force distribution was maintained in the shape of an inverted
triangle by using the whie tree.
2.2. Strength test of masonry unit
The style of brick construction was English and the thickness was assumed 1.0B. The proto-
type brick unit has the dimension of 190 mm×90 mm×57 mm and consists of cement and
coarse sand. The strength of 1 : 5 scale model brick was made as close to that of proto-
type as possible by adjusting the mix ratios. The test results for the compressive strength of
bricks, brick prisms and mortar are shown in Table II. The average compressive strength of
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
360 H.-S. LEE AND S.-W. WOO
Table II. Results of compressive strength tests (unit: MPa).
Type Brick unit Brick prism Mortar
Prototype Model Prototype Model Prototype Model
Average 27.2 23.3 19.6 22.2 13.6 13.7
Figure 5. Comparison of crack pattern at failure for shear tests of the masonry inÿll (unit: mm)
(the number in ( ) means dimension for model).
unit model brick turns out to be 14 per cent lower than that of prototype, and the average
strength of the model brick prism about 12 per cent higher. From the results of tests for
mortar strength, the prototype and the model reveal the similar strength.
Shear tests as shown in Figure 5 were performed on both the prototype and the corre-
sponding models of the masonry inÿlls and the values in Table III were obtained according
to ASTM E519 [12]. Comparing the test results in Table III with the adjustment according to
the similitude law, the average strength of the model (14.4 ×5
2
=360 kN) was found to be
smaller than that of the prototype (483.5 kN). The Poisson’s ratio (v) was estimated at a 13
level of the strength. The average Poisson ratio of the model (0.38) and that of the prototype
(0.29) turn out to be similar. Comparing the modulus of rigidity estimated at the 13 level
of the strength, the values of the model are found to be much smaller (about 44 per cent)
than that of the prototype. Their failure modes were found to be similar, as shown in Figure
5. Maximum strains in the model appear to be larger (about 50–100 per cent) than those in
the prototype.
2.3. Experimental program
The adopted input ground accelerogram is the Taft N21E component and the peak ground
acceleration (PGA) was modiÿed to 0.12q. 0.2q. 0.3q, and 0.4q as shown in Table IV while
the time scale has been compressed according to the similitude law. However, the actual peak
table accelerations turn out generally a little larger than the intended peak input accelerations.
Before and after each earthquake simulation test, free vibration tests were carried out to check
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 361
Table III. Results of shear tests for the masonry inÿlls.
Prototype Model
Strength Max. strain v

t
test
G

Strength Max. strain v

t
test
G

(kN) (10
−6
cmcm) (MPa) (MPa) (kN) (10
−6
cmcm) (MPa) (MPa)
Hori- Verti- Hori- Verti-
zontal cal zontal cal
Average 483.5 30 107 0.29 2.72 5088 14.4 61 161 0.38 1.86 2863

v is the Poisson’s ratio estimated at a 13 level of the strength.

G is the modulus of rigidity estimated at a 13 level of the strength.
Table IV. Test program.
Identiÿcation PGA Remark (return period)
TFT 012 0.12q Design earthquake in Korea (475 years)
Earthquake TFT 02 0.2q Max. earthquake in Korea (1000 years)
simulation test TFT 03 0.3q Max. considered earthquake in Korea (2000 years)
TFT 04 0.4q Severe earthquake in high-seismicity regions of the world
Pushover static test PUSH To ÿnd ultimate capacities of the structure
the changes of the natural period. The signiÿcance of each simulated earthquake regarding
the seismic activity and the Korean seismic code [13] is shown in the column of remarks in
Table IV. Pushover test was performed to observe the structural behaviour in the full range
covering both the elastic and inelastic behaviours and the ultimate capacity of the structure
after earthquake simulation tests.
3. TEST RESULTS AND INTERPRETATION
3.1. Earthquake simulation tests
3.1.1. Global responses. The natural periods of FIF and PIF models are compared with those
of the bare frame (BF) model in Figure 6. The period of FIF model (0.06s) was found to be
the shortest while the PIF model (0.17s) shows shorter period than the BF model (0.23s). The
natural period of the FIF model did not change signiÿcantly except the small increase after
TFT 03 test whereas that of the PIF model was found to increase gradually as the applied
peak ground acceleration (PGA) increased.
Table V summarizes the measured maximum response quantities. In Figure 7, maximum
interstorey drift indices (IDI) in the FIF and PIF models under the varying peak input
accelerations are shown and compared with those measured in the case of BF. It can be
seen from the table and ÿgure that the drifts of the PIF are greater than those of the FIF un-
der the same level of input ground motions. However, IDI of neither FIF nor PIF exceeds the
maximum value of 1.5 per cent allowed in the Korean seismic code [13] even under TFT 04.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
362 H.-S. LEE AND S.-W. WOO
Figure 6. Change of natural periods.
Table V. Summary of measured maximum response amplitude.
Test Table Roof IDI Roof Base Dynamic JW
acceleration drift (%) acceleration shear ampliÿcation
(q) (mm) (q) (kN) factor
TFT 012 0.183 0.72 0.04 0.33 32.0 1.8 0.26
TFT 02 0.316 1.50 0.11 0.69 54.7 2.18 0.44
FIF TFT 03 0.372 1.78 0.11 1.02 91.4 2.74 0.74
TFT 04 0.529 2.55 0.19 1.19 94.3 2.25 0.76
TFT 012 0.210 3.12 0.24 0.55 37.3 2.62 0.30
TFT 02 0.250 4.16 0.28 0.76 49.0 3.03 0.40
PIF TFT 03 0.344 5.76 0.30 0.90 68.8 2.62 0.56
TFT 04 0.426 7.32 0.51 1.04 72.8 2.44 0.59
TFT 012 0.138 4.5 0.26 0.28 17.6 2.03 0.14
TFT 02 0.21 14.06 0.78 0.53 30.8 2.52 0.25
BF TFT 03 0.31 17.87 1.08 0.61 35.1 1.97 0.28
TFT 04 0.4 29.88 1.68 0.69 37.1 1.73 0.30
Under TFT 012 representing the design earthquake in this code, the maximum IDI of all the
models reveal the value less than 0.3 per cent. Time histories of oor drifts under the TFT 04
test are shown in Figure 8. Generally, the ÿrst mode governs in both FIF and PIF. Proÿles of
drift envelopes are shown in Figure 9. The drifts in PIF appear to be approximately 3 times
those in FIF. In turn, the drifts in BF are about 3 times those in PIF.
When the maximum response accelerations at the roof under the shaking table motions
with the varying peak input accelerations are compared, it can be found that all three models
have revealed the trend of decrease in dynamic ampliÿcation factors with the increase of the
level of shaking table motions. It is interesting to note in Figure 10 that the accelerations
of the third and second oors in PIF show the eect of the higher modes almost at every
peak accelerations of the roof. The reason for this eect is considered to be the sliding at the
bed-joint cracks of the second-storey inÿll which occurred before the shaking table tests of
PIF due to the impact imposed by the mis-control of the shaking table.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 363
Figure 7. Change of maximum interstorey drift indices.
Figure 8. Time histories of oor drifts for TFT 04.
Figures 11–13 show the hysteretic relations between the base shear and the interstorey
drift at the ÿrst storey of FIF, PIF, and BF, respectively. It can be seen that FIF, PIF, and
BF all behave linear elastically under TFT 012 which is assumed to represent the design
earthquake in Korea. The stiness of FIF: PIF: BF turns out to be 147 : 33.3 : 7.94 kNmm.
The FIF model had more energy absorption through the friction within the inÿlls or between
the inÿlls and the bounding frame with the stiness remaining almost constant as the intensity
of earthquake ground motions became higher. In Figure 12, on the other hand, the PIF model
revealed the phenomenon of varying drift with almost constant base shear. The reason for this
phenomenon is again conceived due to the prior occurrence of the bed-joint sliding cracks
at the second-storey inÿll masonry. The amount of energy absorption in PIF is found to be
the smallest. Finally, in Figure 13, the BF model reveals clear yielding under TFT 04 and
therefore a large amount of input energy could be dissipated by this yielding. The maximum
base shear of FIF, PIF, and BF under TFT 012 were 32.0, 37.3, and 17.6 kN, respectively.
These are 2.5–5.3 times the design base shear, 7.03kN, according to the Korean seismic code,
which will be shown later.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
364 H.-S. LEE AND S.-W. WOO
Figure 9. Proÿles of drift envelopes in earthquake simulation tests.
Figure 10. Time histories of oor accelerations for TFT 04.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 365
Figure 11. Relation between base shear and interstorey drift at ÿrst storey (FIF).
Figure 12. Relation between base shear and interstorey drift at ÿrst storey (PIF).
Figure 13. Relation between base shear and interstorey drift at ÿrst storey (BF).
Figure 14 compares the time histories of the total base shear (the sum of the oor inertia
forces) and the sum of column shears (the sum of shears measured at load cells) in FIF
and PIF under TFT 04, respectively. It can be observed here that the time histories of base
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
366 H.-S. LEE AND S.-W. WOO
Figure 14. Time histories of base shear and column shear (TFT 04).
shear and column shear are nearly in phase and that the shear carried by the columns are
very small compared with the total base shear (7 per cent in FIF and 23 per cent in PIF).
This implies that the remaining portions, in other words, the contributions of the masonry
inÿlls to the strength of the global structure are signiÿcant. It is also interesting to note in
Figure 14 that there is some directional bias in the sum of column shears in the case of PIF
while there appears no such bias in FIF. Similarly, Figure 15 shows the hysteretic relation
between the interstorey drift at the ÿrst storey and the base shear obtained by summing the
shears measured from load cells at the ÿrst-storey columns. When Figure 15, Figure 11(b),
and Figure 12(b) are compared, it can be found that (1) relatively large energy is dissipated
in the inÿll panels or the interface between the inÿll panels and the bounding frame in FIF
and (2) the whole stiness of FIF at the ÿrst storey is about 17 times the stiness due to
the columns (117.6 : 6.9 kNmm), whereas that of PIF is about 6 times (31.4 : 5.2 kNmm).
These ÿndings mean that the masonry inÿlls serve as the ÿrst line of defence that prevents the
damage of the frame by dissipating a considerable portion of energy as well as by increasing
the whole stiness of the structure. The time histories of the total absorbed energy for the
case of TFT 04 in FIF, PIF, and BF are depicted in Figure 16. The amounts of the total
absorbed energy for the tests of TFT 04 are 2917, 1990, and 4237 kNmm for the FIF, PIF,
and BF, respectively. It is worthwhile to note that the energy absorption of FIF or PIF mainly
originated from the internal friction within the inÿll panels or between the inÿll panel and
the bounding frame, therefore, does not imply the damage in the main frame, whereas that
of BF was developed by the inelastic deformations, in other words, the damage in the frame.
Though the building structures built before 1988 or lower than 6 in the number of storeys
need not be designed against the earthquake in Korea, it is worthwhile to evaluate the test
results with regards to the current Korean seismic code [13]. The design seismic coecient,
C
c
, that is, the ratio of the design base shear to the eective weight of the structure according
to the Korean seismic code is as follows:
C
c
=J
D
W =
AISC
R
=0.057 (1)
where J
D
is the design base shear, W the eective weight of the structure, A=0.12 the
zone factor, I =1.0 the importance factor, C =11.2

161.5 the dynamic factor, S =1.2 the
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 367
Figure 15. Hysteretic relations of column shear versus interstorey drift at the ÿrst storey (TFT 04).
Figure 16. Time histories of absorbed energy in earthquake simulation test (TFT 04).
soil factor.
1 =0.06h
34
n
(h
n
: height of structure. m) =0.36 (natural period. s) (2)
SC (61.75): 1.67. R=3.5 the response modiÿcation factor: ordinary moment frame.
When the measured natural periods of the models are converted to the values corresponding
to the prototypes, the values become 0.134, 0.380 and 0.514 s for the FIF, PIF, and BF,
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
368 H.-S. LEE AND S.-W. WOO
Figure 17. Angular rotations under TFT 04.
respectively. Here, the calculated natural period, 0.36 s, in the above appears to be similar to
the natural period, 0.38 s, measured in the case of the PIF. The estimated eective weights
of the models, FIF, PIF, and BF are 123.4, 117.3 and 100.6 kN, respectively. However, it is
assumed here that the eective weight of all the models remains constant and has the value of
123.4 kN as for the case of FIF. The ratios of the base shear to this assumed eective weight
for the varying level of the input table motions appear in Table V. These values are much
larger than the design seismic coecient of 0.057. The base shear of the FIF is approximately
over twice that of the BF under the same level of earthquake ground motions. Even the PIF
has the base shear 96 per cent larger than that of the BF under the TFT 04. Though the
damages in the main frame and masonry inÿlls looked minor, the rate of increase in base
shear seems to decrease due to the energy dissipation within the inÿll panels and through the
frictions between the inÿll panels and the bounding frame with the increase of the level of
shaking table motions.
3.1.2. Local responses. Several time histories of angular rotations in the ends of members
and their maximum values are given in Figure 17 and Table VI, respectively. The angular
rotations were measured over the full depth in the ends of beams and over the cross-sectional
dimension parallel to the shaking direction in the ends of columns. The maximum angular
rotation occurred at the location R4 in FIF and BF while that occurred at location R9 in PIF,
as shown in Table VI. In any case of FIF and PIF, the maximum value of angular rotation
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 369
Table VI. Maximum angular rotation (unit: 10
−5
rad).
Test\Location

R1 R2 R3 R4 R5 R6 R7 R8 R9
(Interior (Interior (Beam) (Interior (Beam) (Exterior (Exterior (Beam) (Exterior
column) column) column) column) column) column)
FIF 29 4 46 15 24 26
TFT 012 PIF 30 83 36 121
BF 170 230 100 270 100 120 110 200 60
FIF 12 90 28 39
TFT 02 PIF 112 117 30 74 158
BF 600 680 220 1,000 330 400 340 700 210
FIF 54 65 64 15 22 21 27
TFT 03 PIF 156 52 140 226
BF 660 960 550 1,300 500 570 440 900 310
FIF 74 79 196 32 38 38 33
TFT 04 PIF 201 185 61 134 316
BF 1,340 1,540 880 2,410 760 750 500 1,410 420

See Figure 2(a) to identify the locations of R1–R9.
appeared less than 0.004 rad. It is noted in Figure 17 that the angular rotations in locations
R5 and R9 of PIF are about 10 times larger than those of FIF and that those of BF are in
turn 2–3 times larger than those of PIF. However, the amount of the maximum rotation in the
location of R8 is about 10 times larger in BF (0.0141 rad) than in PIF (0.0013 rad) though
the corresponding span is free from the masonry inÿll in both cases. It is also observed in
case of PIF that the history of angular rotation at location R5 has some bias in the direction,
whereas that at location R9 does not.
The time histories of shear forces in the ÿrst-storey columns, as shown in Figure 18, indicate
that the central column, whose section is the smallest, has relatively large shear force. Masonry
inÿlls reduce the column shears greatly in FIF and PIF in comparison with the case of BF.
However, the bias in shear due to the eect of the axial compressive forces can be clearly
noticed in columns (2) and (3) in case of PIF while that can be observed in columns (1) and
(3) in case of BF.
From the measured strains of the masonry inÿlls at the ÿrst storey for FIF in Figure 19,
compressive strains appear to be generally twice larger than tensile strains in both FIF and PIF,
while compressive strains in PIF are also approximately twice larger than those in FIF. The
maximum strain of the masonry inÿlls in the PIF model turns out to be 230 ×10
−6
cmcm,
which exceeds the failure strain, 161×10
−6
cmcm, obtained through the shear test. The reason
for this larger strain seems to be the eect of the conÿnement of the bounding frame to the
inÿll, which led to the prevention of the shear failure at the early stage.
3.1.3. Crack pattern. The model did not show any serious damage even in the case of TFT 04
simulating a severe earthquake in the high-seismicity region of the world. In FIF, minor cracks
occurred in the masonry inÿlls for TFT 04 as shown in Figure 20(a). As mentioned previously,
the bed-joint cracks occurred at the second-storey inÿll due to the impact imposed accidentally
by the mis-control of the shaking table before the earthquake simulation tests on PIF were
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
370 H.-S. LEE AND S.-W. WOO
Figure 18. Column shears under TFT 04.
Figure 19. Strains of the masonry inÿlls at the ÿrst storey under TFT 04.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 371
Figure 20. Crack pattern.
performed. However, several additional cracks occurred mainly at the interface of the masonry
inÿlls with the frame in PIF for TFT 03 and TFT 04, as shown in Figure 20(b).
3.2. Pushover test
3.2.1. Global responses. In Figure 21 the maximum response points of FIF, PIF, and BF ob-
tained through the earthquake simulation tests are superposed over the results of the pushover
tests and analysis for PIF and BF shown in solid and dotted lines. Comparing the curve of
the pushover test with that of the analysis for BF [14], it can be seen that the initial stiness
appears to be smaller, while the ultimate strength larger. The discrepancy in the initial stiness
means the damage caused by the earthquake simulation tests previously performed; the reason
for the discrepancy in strength in the latter half is attributed to the strain-aging phenomenon
[15. 16]. Accordingly, it is presumed that, if the pushover test had been performed without any
prior earthquake simulation tests, the result identical with the curve of analysis would have
been obtained. This presumption is supported by the fact that the maximum response points
obtained from the results of the earthquake simulation tests comply with the curve obtained
from the analysis. Therefore, it is now assumed that the curve of analysis are more reliable
than that of the test excluding the maximum drift of the roof and the failure mechanism.
Then, these analytical results of BF are compared with those of PIF.
First of all, it can be observed that the maximum response points of PIF obtained from the
earthquake simulation tests also comply well with the curve of the corresponding pushover
test. For TFT 0.4, PIF came just before the elastic proportional limit point while BF ex-
perienced considerable non-linear plastic behaviours. In order to calculate the displacement
ductility ratio for BF and PIF, an elasto-perfect-plastic (EPP) bilinear model was drawn in
Figure 21, based on the equal-energy concept. In the displacement capacity, PIF : BF shows
43.1 : 47.2mm. Therefore, the displacement ductility ratio of PIF : BF appears to be 4.2 : 2.4
since the yielding displacement is deÿned to be 10.1 : 20.0 mm using this EPP model. With
this bilinear model, the eective yielding strength of PIF : BF is 98.0 : 40.0kN. The summary
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
372 H.-S. LEE AND S.-W. WOO
Figure 21. Base shear versus roof drift in tests.
Table VII. Comparison of global responses in pushover tests using EPP model.
Bare frame (BF) Partially inÿlled frame (PIF) (2)(1)
(1) (2)
Ultimate strength 41.7 kN (analysis) 106.4 kN 2.55
Yielding strength 40.0 kN (analysis) 98.0 kN 2.45
Yield drift at roof 20.0 mm (0.9%)

10.2 mm (0.5%)

0.51
Initial stiness 2.0 kNmm 9.6 kNmm 4.80
Drift capacity 47.2 mm (2.13%)

43.1 mm (1.94%)

0.91
Displacement ductility ratio 2.36 4.23 1.79

The number in ( ) means drift ratio (% of building height).
of the global responses in pushover tests on the BF and PIF is given in Table VII. From
this table, it can be found that the actual strength of PIF is 13.9 times the design base shear,
7.03 kN, with that of BF being 5.7 times.
Figure 22 compares the applied storey shear and the sum of the column shears measured in
the load cells as the functions of the interstorey drift at the ÿrst storey. For PIF in Figure 22(b),
the sum of the shear forces measured in the load cells is found to be approximately 20 per
cent of the applied lateral load (17 per cent at the ÿrst-storey drift of 2 mm and 22 per cent
at the ÿrst-storey drift of 8.4 mm). This means that the masonry inÿlls carry approximately
80 per cent of the total base shear and hence signiÿcantly contribute to the increase in the
total strength of the structure. The initial stinesses at the interstorey drift of 2 mm appear to
be 5.9 kNmm for the column shear and 28.7 kNmm only for the inÿll wall. Therefore, the
total stiness of 34.6 kNmm in PIF looks much higher than that of the BF (4.80 kNmm). It
can also be seen that the contribution of column (1) to the shear resistance in PIF is almost
negligible when compared with that in BF. This implies that the main action in column (1)
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 373
Figure 22. Column and total shears at ÿrst storey in pushover test.
for the resistance to the lateral load is the axial behaviour rather than the shear in the case of
PIF, which can be represented as tension chord in the truss model for inÿlled frames suggested
by Priestley et al. [17]. It can be noticed that the columns in the ÿrst storey developed only
about half of their capacities in deformation and strength in PIF when compared with the
case of BF. The masonry inÿlls show linear elastic behaviour up to 2.1 mm of the drift of
the ÿrst storey, but hereafter the non-linear behaviour of the masonry inÿlls governs the total
response of the ÿrst storey though there appeared no signiÿcant cracks as will be explained
later. Even after the resistance of the masonry inÿlls is degraded by their cracking at Point A
in Figure 22(b), the total resistance of the ÿrst storey is still maintained due to the increase
of resistance in the columns of the frame.
The comparison of the relations between the interstorey drift and the storey shear are made
in Figure 23. The maximum interstorey drift in BF occurred at the ÿrst storey, the value of
which is 20.2mm (2.6 per cent). The continuation of drift in the ÿrst storey after the strength
degradation indicates that failure mechanism was formed at the ÿrst storey. In case of PIF, the
largest interstorey drift occurred at the second storey, the value of which is 30.4 mm (4.2 per
cent). The initial stinesses at the ÿrst and second storeys in the case of BF appear to be 4.80
and 3.53 kNmm while those in case of PIF are 34.6 and 14.8 kNmm. The storey ductility
at the second storey of PIF appears to be about 5.0 based on the equal-energy concept in
deÿning yielding displacement while that of BF at the ÿrst storey is about 3.0.
The shear capacity at the second storey of PIF, J
2nd
, is 87.91kN, as shown in Figure 23(b).
The failure mechanism at the second storey of PIF as shown in Figure 24(b) consists of: (1)
plastic hinges at the top and bottom ends of the left column (C
3
), (2) plastic hinges at the
middle point and the bottom end of the middle column (C
2
), (3) the shear failure at the top
of the right column (C
1
), and (4) the bed-joint sliding at the middle of the masonry-inÿlled
wall. Also, it can be noticed in this ÿgure that most of the gravity load around the masonry
inÿlls was being transferred through the masonry inÿlls rather than through the columns since
the deformation of the columns became so large at the collapse state. See the photographs
in Figure 28(b). Therefore, with the estimated gravity load transferred through the masonry
inÿlls N
w
, being 57.28kN, the modulus of shear friction, j, can be predicted as 0.38 by using
the relations of J
w
=J
2nd
−(J
C
1
+J
C
2
+J
C
3
) =87.91−65.94kN=21.97kN and j =J
w
N
w
. The
failure mechanism in BF as shown in Figure 24(a) is the soft-storey mechanism. The failure
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
374 H.-S. LEE AND S.-W. WOO
Figure 23. Interstorey drift versus storey shear in pushover test.
Figure 24. Failure mechanisms in pushover test.
occurred in the mode of the exural compressive crushing at the upper end of the interior
column at the ÿrst storey.
3.2.2. Local responses. It is worthwhile to note in Figure 25 that, up to the proportional
limit, BF and PIF models have indicated the dierent trends in locations of measured angular
rotations at the interior joint of the second oor; the angular rotations in the ends of columns
(locations (1) and (2)) appear to be about twice those in the ends of beams (locations (3)
and (4)) in the case of BF. But the angular rotations in the locations (1), (3), and (4) are
approximately one-third the rotation in the location (2) in the case of PIF. This implies that
the sliding was being activated from the beginning at the bed-joint cracks of the second-storey
inÿll panel. However, in the range of initial linear proportional part, the amount of rotations
in the PIF model are generally less than one half of the values in the BF model. In the case
of PIF, after the proportional limit point, the load of which is 71.3 kN and corresponds to
the tension failure of the ÿrst-storey inÿll panel as shown in Figure 27(a), the non-linearity
between the rotation and applied load commenced. Above 100 kN of the applied lateral load,
the failure mechanism as shown in Figure 24(b) enforced the rotation at the location (2) to
reach the value of 0.081 rad, which is much larger than the rotational capacity of 0.037 rad
in the case of BF. The reason for this relatively large rotational deformability in the case of
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 375
Figure 25. Histories of angular rotations in pushover test.
Figure 26. Maximum angular rotations in the ends of members at collapse state in pushover test (rad).
PIF seems to be the increase in the ductility of the columns due to the large alleviation of
the axial compression force by the contribution of the masonry inÿll. Figure 26 shows the
maximum angular rotations in the ends of members at collapse state.
The histories of strains measured at the strain gauges attached diagonally to the surface of
the masonry inÿlls in the ÿrst storey are depicted in Figure 27. Inÿll A had lost its resistance
capacity abruptly when certain horizontal cracks occurred in the joint of the masonry under
the load of 71.3 kN. This load level is identical with the yield point of the ÿrst-storey drift
as shown in Figure 22(b). Inÿll B appears to have resisted the lateral load to the last in
compression and tension. The maximum strain appears to be 210×10
−6
cmcm in compression
and 110 ×10
−6
cmcm in tension.
The value of shear strength, t
test
, obtained from the shear test as shown in Table III is
1.86 MPa. When this value is applied to the horizontal section area of the masonry inÿll at
the ÿrst storey (1170 mm × 38 mm × 2 walls =88 920 mm
2
), the calculated ultimate shear
strength of these inÿll walls is approximately 165kN. But from Figure 22(b), it can be noticed
that the resistance of PIF at the ÿrst storey consists of the sum of column shears (23.4kN) and
the contribution by masonry inÿlls (83 kN). It should be noted here that the failure occurred
at the second storey, instead of at the ÿrst storey and that the shear failure occurred partially
throughout the bricks and partially through the bed-joint sliding in the shear test whereas the
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
376 H.-S. LEE AND S.-W. WOO
Figure 27. Strains of masonry inÿlls in pushover test.
bed-joint sliding only occurred in the pushover test. The shear test for the bed-joint sliding
was not performed in this study. Therefore, the actual shear strength of the inÿll walls at the
ÿrst storey can be expected to be larger than 83 kN but probably less than 165 kN. However,
since the workmanship of the mason and the quality of bed-joint mortar are generally not
uniform, it seems to be very dicult in actuality to predict the shear strength in the failure
of bed-joint sliding for the masonry inÿlls with high reliability.
3.2.3. Crack patterns and failure modes. Development of cracks in both BF and PIF modes
are shown in Figure 28. The numbers in the ÿgure indicate the value of the roof drift at the
time when the crack was noticed for the ÿrst time. It can be found in Figure 28(b) that several
cracks occurred newly in the masonry inÿlls during the pushover test. Generally, the direction
of cracks is identical with the expected diagonal strut. In the case of PIF, the storey failure
mechanism described in Figure 24(b) has formed around the roof drift of 27 mm; while in
the case of BF, the soft-storey mechanism has been reached at the roof drift of approximately
24 mm. In the PIF, the masonry inÿlls have been crushed just beside the portion of the
column where the shear failure occurred. See the photographs in Figure 28(b).
4. CONCLUSIONS
There appeared neither signiÿcant damage on the masonry inÿlls, nor any damage on the frame
itself even under the severe earthquake ground motions. The contribution of masonry inÿlls to
the global capacity of the structure turns out to be 80 per cent in strength and 85 per cent in
stiness from the results of pushover test. However, the failure mode of the masonry-inÿlled
frame was that of shear failure due to the bed-joint sliding of the masonry inÿlls while that
of the bare frame appeared to be the soft-storey plastic mechanism at the ÿrst storey and the
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
EFFECT OF MASONRY INFILLS ON SEISMIC PERFORMANCE 377
Figure 28. Development of cracks in pushover test (the number in the ÿgure means the value of roof
drift (mm) at the time of the occurrence of crack).
deformation capacity of the global structure remains almost same regardless of the presence
of the masonry inÿlls. Therefore, it is essential to consider the eect of masonry inÿlls for
the practical evaluation of the seismic safety of moment-resisting RC frame buildings.
Masonry inÿlls behave beneÿcially on buildings as far as this experimental study alone
is concerned. The reason for their beneÿcial behaviour is that the amount of increase in
earthquake inertia force appears to be relatively small, when compared with the increase in
the strength by masonry inÿlls. Above all, masonry inÿlls appear to have a great eect on
the reduction of the global lateral displacement.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378
378 H.-S. LEE AND S.-W. WOO
The quality of masonry inÿlls, however, depends on the workmanship of masons and also
the credibility of the structural system depends in turn on the quality of masonry. In case
that there are openings in masonry inÿlls, or that panels are partially inÿlled with masonry,
a more complicated mode of failure can occur with the interaction to the bounding frame, as
already seen in many instances of earthquake damages. Further studies on the eects of the
construction quality of masonry and partial inÿlling on the structural behaviour are considered
to be necessary in the future.
ACKNOWLEDGEMENTS
The research stated herein was supported by the Ministry of Construction and Transportation, the Repub-
lic of Korea, and several private companies including SSangYong Engineering and Construction Corp.,
DongBu Construction Corp., Hyundai Construction Corp., and DongYang Structural Safety Consultants
Corp. The contributions of graduate students at Korea University, Dong-Woo Ko, Yun-Sup Heo, and
Kyi-Yong Kang, were crucial to the success of this research. The technique which is essential to the
manufacture of the reduced-scale model was developed by the support of advanced STructure RESearch
Station (STRESS) at Hanyang University. The authors express their deepest gratitude to all of these
supports. The authors are grateful for the thoughtful and constructive comments of the anonymous
reviewers.
REFERENCES
1. Fardis MN (ed.). Experimental and numerical investigation on the seismic response of RC inÿlled frames and
recommendations for code provisions. ECOESTPREC8 Report No. 6, Laboratorio Nacional de Engenharia
Civil, Lisbon, 1996.
2. Bertero VV, Brokken S. Inÿlls in seismic resistant building. Journal of Structural Engineering ASCE 1983;
109(6):1337–1361.
3. Mehrabi AB, Shing PB, Shuller MP, Noland JL. Experimental evaluation of masonry-inÿlled RC frames. Journal
of Structural Engineering ASCE 1996; 122(3):228–237.
4. UBC. Uniform building code 1991. International Conference of Building Ocials, 1991.
5. Negro P, Verzeletti G. Eect of inÿlls on the global behaviour of RC frames: Energy considerations from
pseudodynamic tests. Earthquake Engineering and Structural Dynamics 1996; 25:753–773.
6. Eurocode No. 2. Common uniÿed rules for concrete structures. Report EUR 8848 EN, Commission of the
European Communities, 1994.
7. Eurocode No. 8. Design provisions for earthquake resistance of structures, Parts 1–3: General rules–speciÿc
rules for various materials and elements. ENV 1988-1-3, CEN, Brussels, 1994.
8. Fardis MN, Calvi MG. Eects of inÿlls on the global response of reinforced concrete frames. Proceedings of
10th European conference on earthquake engineering, Rotterdam, Balkema, 1995.
9. Fardis MN, Bousias SN, Franchioni G, Panagiotakos TB. Seismic response and design of RC structures with
plan-eccentric masonry inÿlls. Earthquake Engineering and Structural Dynamics 1999; 28:173–191.
10. Kim SD, Lee HS, Kim YM et al. A study on the seismic evaluation and retroÿt of low-rise reinforced concrete
building in Korea. R&D 96-0057, Ministry of Construction and Transportation, Republic of Korea, 1997, 2000
(in Korean).
11. Harris HG, Sabnis GM. Structural modeling and experimental techniques. CRC Press: Boca Raton, FL, 1999.
12. ASTM standards on masonry (1st edn) ASTM: Philadelphia, PA, 1990.
13. Regulations concerning the structure standards of building, etc. Ministry of Construction and Transportation,
Republic of Korea, 1996.
14. Lee HS, Woo SW, Heo YS. Correlation of experimental and analytical seismic responses of a 1 : 5 scale 3-storey
reinforced concrete frame. Proceeding of US–Korea Workshop on New Frontier in InfrastructuralSeismic
Engineering, Techno-Press: Taejon, Korea, 1999.
15. Booth E (ed.). Concrete structures in earthquake regions: Design and analysis. Wiley: New York, 1994.
16. Erasmus LA. Cold straightening of partially embedded reinforcing bars—A dierent view. Concrete
International: Design and Construction 1981; 3(6):47–52.
17. Priestley MJN, Paulay T. Seismic design of reinforced concrete and masonry building. Wiley: New York, 1992.
Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:353–378