Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Influence of connection and constructional details on masonry-infilled RC T

frames under cyclic loading
⁎
Quanmin Penga,b, , Xiaojie Zhoua,b, Chenghao Yanga
a
Tianjin Chengjian University, Tianjin 300384, China
b
Tianjin Key Laboratory of Civil Structure Protection and Reinforcement, Tianjin 300384, China

A R T I C L E I N F O A B S T R A C T

Keywords: Infill walls significantly affect the dynamic characteristics and the seismic performance of reinforced concrete
Masonry infill wall (RC) frames. To evaluate the influence of connection type and constructional details of infill on the cyclic
Reinforced concrete frame response of RC frames, a quasi-static test was performed on four infilled frame specimens and a bare frame
Cyclic behavior specimen. The infilled frames were constructed with rigid connection or flexible connection between frames and
Finite element modeling
infills, and the infill walls were constructed with or without core columns. The test results show that the presence
of the masonry infill wall increased the lateral strength, stiffness, ductility and energy dissipation capacity of the
RC frame. The contribution of connection type and core column were analyzed and compared. Additionally,
finite element (FE) modeling was conducted and validated with the experimental observations.

1. Introduction a strong beam) infilled with strong panels (solid concrete masonry)
exhibited better performance than weak frames (strong columns and a
Masonry-infilled reinforced concrete (RC) frames are commonly weak beam) infilled with weak panels (hollow concrete masonry) in
used in many parts of the world, even in areas of high seismic risk terms of load resistance and energy-dissipation capability. The data
including China and Mediterranean countries. Although infill walls, from Kakaletsis and Karayannis [11] indicated that RC frames with
serving as interior and exterior partitions, are usually considered non- strong infills (vitrified ceramic brick) showed higher initial stiffness and
structural in design, they can significantly affect the dynamic char- higher ductility than those with weak infills (clay brick), but infill
acteristics and the seismic performance of RC frames [1–5]. The pre- strength did not substantially influence strength or energy dissipation.
sence of infill walls may increase the strength, stiffness and ductility of Cavaleri and Trapani [10] reported that frames infilled with light-
RC frames. However, infill may also increase the base shear, the un- weight concrete masonry and calcarenite masonry show improved
desired soft-story mechanisms of structures, brittle damage imposed on dissipative properties compared to the frames infilled with clay ma-
the surrounding frame components, and serious damage to the infill sonry. Regarding the influence of masonry openings on the seismic
walls themselves. These advantages and disadvantages result from the performance of infilled RC frames, Kakaletsis and Karayannis [11,15]
complex interaction between the surrounding frames and the infill investigated single-story, single-bay scaled specimens under cyclic
walls when they are subjected to earthquake loads. horizontal loading. The results showed that for low lateral displace-
Extensive experimental and analytical studies on the seismic per- ment, the energy dissipation of specimens with openings was higher
formance of infilled RC frames have been conducted to understand the than that of the bare frame; for high lateral displacement, the energy
interaction between frame and infill and to assess the effects of infill dissipation of specimens with openings was reduced and that of the
under different conditions. Studies have mainly focused on strength, bare frame remained constant. The window openings with a width from
stiffness, ductility, energy dissipation and failure mode of the structural 25% to 50% of the infill length led to an average reduction of 18.7% in
system. The variables investigated have included a variety of infill lateral resistance, 26.3% in initial stiffness, and 4.3% in cumulative
materials from concrete [6–8] to different kinds of masonry [6,9,10], energy dissipation capacity. The door openings with a width from 25%
strength of the frame or of the infill [9,11], frame aspect ratio [8,9,12], to 50% of the infill length led to an average reduction of 28.7% in
geometry of infill panel opening [11,13–15], connection between the lateral resistance, 30.3% in initial stiffness, and 27% in cumulative
frame and the infill [8,12,16], and constructional details of infill such as energy dissipation capacity. Quasi-static experiments conducted by
the presence of a concrete lintel beam [12] and constructional columns Asteris and Kakaletsis et al. [17] revealed that the failure modes of
[16]. Mehrabi et al. [9] reported that strong frames (weak columns and masonry-infilled RC frames with openings were more complex than

⁎
Corresponding author at: Tianjin Chengjian University, Tianjin 300384, China.

https://doi.org/10.1016/j.soildyn.2018.02.009
Received 10 August 2017; Received in revised form 13 December 2017; Accepted 4 February 2018
0267-7261/ © 2018 Elsevier Ltd. All rights reserved.
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

(FE) models, in which the structure is discretized into numerous ele-

ments. For example, beam or continuum elements are used for the
surrounding frame, continuum elements for the infill or masonry
blocks, and interface or contact elements for the interaction between
the frame and infill or between masonry blocks [2,13,18–24]. Chiou
et al. [18] modeled infilled RC frames by discretizing the brick units
and concrete members into blocks that were interconnected with con-
tact springs to simulate tensile and shear failure. Asteris developed a
new FE technology, estimating the infill/frame contact lengths and
stresses as an integral part of the solution, to investigate the stiffness
reduction of the infill wall with opening under monotonic loading [13],
and to simulate the response of a masonry infilled RC frame under a
lateral static load [23]. Asteris and Cotsovos [24] conducted a non-
linear FE analysis of the effect of infill walls on the response of RC
frames under static and seismic loading. The concrete and masonry
were modeled by 27-node Lagrangian brick elements, and the re-
inforcement bars were modeled by 3-node truss elements. Stavridis and
Shing [19] combined the smeared-crack continuum elements with in-
terface elements to capture the different failure modes of masonry-in-
filled RC frames. Koutromanos et al. [20] extended this work by means
of with a developed cohesive interface model and an improved
smeared-crack model to capture the cyclic behavior of such structures.
Mohyeddin et al. [21] constructed a 3D discrete-finite-element model
Fig. 1. Geometry and reinforcement details of frame specimens. using ANSYS, and RC and masonry materials were modeled using the
Solid65 element. The mortar joint thickness was halved into two parts
that attached to adjacent masonry units and could interact with each
those with solid infill panels. The presence of an opening upon the di-
other through interface (contact) elements.
agonal of an infill panel led to the abolishment of the failure modes of
Macro-models are typically simplified models that use a single (or
Diagonal Compression (DC) and Diagonal Cracking (DK). The experi-
multiple) diagonal strut (or struts) connected to the frame through
mental findings also showed that frames with infill walls connected to
beam-column joints to represent the entire infill panel. Many equivalent
both the column and beams of the frame were superior to less con-
diagonal strut models have been developed, differing in amount, posi-
nected frames [8,12]. Jiang et al. [16] compared the effect of connec-
tion, cross-section width and cyclic nonlinear behavior of the equiva-
tion between infill walls and frames on the seismic performance of
lent diagonal strut [1,2,10,25–28]. Aiming to simply reproduce the
aerated concrete blocks infilled RC frames. They found that the addition
effects of infills in the global and local response, Fiore et al. [2] pro-
of a masonry infill wall with rigid connection contributed more than
posed a two-non-parallel-strut model, and the positions of the two struts
one with flexible connection to the observed increase in the lateral
were determined according to the geometry of the infill and the story
strength, stiffness and energy-dissipation capacity of the bare RC frame,
level. Asteris et al. proposed an analytical equation for the reduction
and to the observed decrease in the displacement ductility.
factor of the equivalent strut width that considered the opening [1], and
Both micro-models and macro-models have been developed to si-
then proposed an equation for dimensionless width to consider both the
mulate the performance. Micro-models usually mean finite element
opening and the vertical load [28]. Attempts have also been made to

Fig. 2. Configuration of infilled frame specimens.

97
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Table 1
Characteristics of tested specimen.

Specimen Frame type Masonry infill type Connection between frame and infill

PF Bare frame None –

GWF1 Infilled frame fly ash hollow block masonry without RC core columns rigid
RWF1 Infilled frame fly ash hollow block masonry without RC core columns flexible
GWF2 Infilled frame fly ash hollow block masonry with RC core columns rigid
RWF2 Infilled frame fly ash hollow block masonry with RC core columns flexible

Table 2
Material properties of specimens for experiment.

Material Compressive strength (MPa) Material Yield strength (MPa) Ultimate strength (MPa)

Frame concrete 24.5 Φ6 rebar 381 474

Fine aggregate concrete of core columns in masonry 22.3 Φ8 rebar 366 430
Fly ash hollow block masonry 1.02 Φ14 rebar 484 607

Fig. 3. Specimens test setup.

Fig. 4. Lateral loading history curve.

extend the applications of simple macro-models. For example, Uva et al. study results derived from the simplified model with the exact FE
[5] performed pushover analyses on different strength infilled frames model, and then presented an approach to correct the local shear forces
versus a bare frame by adopting lumped-plasticity models in which the for the use of the simplified model. Moreover, Asteris et al. [4] used a
infills modeled with a single strut approach. Celarec et al. [3] in- simplified strut model to perform a parametric study on the funda-
troduced an iterative pushover-based procedure based on simplified mental period of infilled RC frames and then proposed an estimation
nonlinear models. The objective of this procedure is to simulate the equation.
potential shear failure of columns due to infill effect and to quantify the The simplified model can predict the global behavior of the struc-
effect through a probabilistic seismic performance assessment proce- ture with relatively low computation cost, but cannot accurately model
dure. To accurately predict the shear forces in the frame using a simple the infill-frame interaction and capture the local failure. The FE model
single-strut model, Cavaleri and Trapani [22] compared parametric can accurately model the damage behavior and reveal failure

Fig. 5. FE models for some specimens.

98
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Table 3 Table 4
Material properties of USteel 02 in FE models. Material properties of UConcrete 02 in FE models.

Rebar Elastic modulus Yield stress Post yield Frame f′c (MPa) εc fu (MPa) εcu d cu ft (MPa) Ets (MPa)
E0 (MPa) f y (MPa) stiffness α member

Longitudinal rebars in 2×105 484 0.015 Beam 29.2 0.0039 20.3 0.0152 0.18 3.08 2476
frame members Column 31.4 0.0048 22.9 0.0197 0.18 3.08 2476
Longitudinal rebars in core 2×105 381 0.015
columns Note: f′c is compressive strength; εc is strain at compressive strength; fu is ultimate
compressive strength; εcu is ultimate compressive strain, i.e., strain at ultimate com-
pressive strength of confined concrete; d cu is ratio of unloading stiffness at ultimate
compressive strain to initial elastic modulus; ft is tensile strength; and Ets is tension
Table 5 softening modulus.
Material properties of masonry in FE models.

Material Compressive Tensile Elastic Poisson's

strength (MPa) strength modulus ratios In an attempt to reduce unfavorable interaction and to provide a
(MPa) (MPa) reliable assessment of the interaction under different conditions, two
types of wall-frame connections were designed and tested under re-
Masonry 1.02 0.021 443 0.315
Core column 22.3 2.3 14740 0.25 versed cyclic loading in this study. Since the use of reinforcing masonry
with core columns to form confined masonry is a conventional practice
in China [30,31], conditions with and without core columns in the infill
mechanism, but at the expense of high computation cost. A compre- wall, representing strong infill and weak infill respectively, were also
hensive review of the two groups of models can be found in the lit- considered. Furthermore, FE models using software ABAQUS were es-
erature [26,27,29]. tablished and the simulation results were compared with the

Fig. 6. Final damage conditions of specimens.

99
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Table 6
Drift ratio (δ) corresponding to characteristic points of damage evolution.

Specimen First cracking/ yielding of frame element First cracking of Infill wall

Beam concrete Column concrete Longitudinal rebar in beam Longitudinal rebar in column Wall-frame mortar joint Masonry mortar joint Block

PF 1/500 1/300 1/200 1/150 – – –

GWF1 1/1000 1/500 1/700 1/200 1/1000 1/700(stepped) 1/500
RWF1 1/700 1/500 1/300 1/200 – 1/200(horizontal) 1/200
GWF2 1/1000 1/500 1/700 1/200 1/1000 1/1000(vertical) 1/700
RWF2 1/700 1/700 1/200 1/200 – 1/700(stepped) 1/700

Fig. 7. Numerical axial stress of frame elements in GWF1 during loading.

experimental results to evaluate the validity of the models for further respectively.
numerical analysis. All specimens were primarily subjected to a vertical constant load of
460 kN on each frame column to simulate the gravity load from the
upper stories. A lateral load was then applied at one end of the top
2. Experimental Program
beam using a MTS hydraulic actuator in displacement control mode
with gradually increasing amplitude. The lateral load and the corre-
Four single-story, single-bay, fully infilled RC frames and a bare
sponding displacement of the top beam with respect to the base beam
frame as a reference were tested under reversed cyclic horizontal
were monitored. Failure was assumed to occur when strength of the
loading. All frames had the same design with a strong column-weak
specimen was reduced by more than 15% of the peak strength. Figs. 3
beam scheme, as shown in Fig. 1. Infill walls were built after the con-
and 4 show the test setup and the lateral loading history.
struction of the frames with one wythe of fly ash hollow block masonry.
During the infill wall building, the horizontal tie steel bars, with one
end anchored in the frame column, were embedded in bed joins of the 3. FE models of specimens
wall. As shown in Fig. 2, two types of connections were used: rigid
connection, with a layer of masonry mortar between the frame and 3.1. Model establishment
infill, and flexible connection, a 30 mm-wide-gap infilled with poly-
styrene plate between the frame and infill. Additionally, two types of To reduce calculation time, two-dimensional FE models of the spe-
infill wall were studied, with and without RC core columns at the two cimens for numerical analysis were established using ABAQUS. The RC
ends of the wall. The core column was constructed by insertion of frame was modeled as the linear beam element B21, and the infill wall
vertical steel bars into the holes of fly ash hollow blocks, in which fine and the core column was modeled as the plane stress element CPS4R.
aggregate concrete was poured. The typological characteristics and the The longitudinal reinforcements of the RC frame elements were defined
material properties of the specimens are summarized in Tables 1 and 2, by editing keyword “*rebar” as the beam element property, and the

100
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 7. (continued)

Fig. 8. Numerical equivalent plastic strain (PEEQ) of infill wall in GWF1 during loading.

101
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 9. Numerical equivalent plastic strain (PEEQ) of infill wall at the last loading.

potential confinement effect from stirrups was considered by the con- et al. [35] was adopted to define the stress-strain relation of the com-
crete constitutive relation instead of by constructing stirrups. The pressive skeleton curve.
longitudinal rebar of the core columns in the infill wall was modeled by The complex material behaviors of the masonry infill and the fine
the truss element T2D2 and embedded in the wall. The element aggregate concrete of core column were described by the Concrete
SpringA, a one-dimensional nonlinear spring with large compressive Damaged Plasticity model of ABAQUS, which utilizes damage variables
stiffness and zero tensile stiffness, was used to model the interface be- to consider the damage of the masonry. For the fly ash hollow block
tween the frame and the infill. The compressive stiffness of SpringA was masonry, the compressive stress-strain relation was determined based
defined as a larger value for the rigid connection than the value for the on the experimentally determined fitting result [36], as given in Eq. (1);
flexible connection. The bottoms of the frame column and the infill wall while the stress-strain relation in tension was adopted according to the
were fixed as the initial boundary condition. In the first analysis step, a reference [37], as given in Eq. (2).
concentrated force of 460 kN was exerted vertically on each column to 2 3 4 5
take the gravity load of the RC frame into account. In the second step, a σ ε ε ε ε ε
= 3.7 −6.7632 ⎛ ⎞ +7.6377 ⎛ ⎞ −4.9878 ⎛ ⎞ +1.6197 ⎛ ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟

lateral displacement was applied according to the loading history curve fcm εc ε
⎝ c⎠ ε
⎝ c⎠ ε
⎝ c⎠ ε
⎝ c⎠
of the experiment at one end of the beam. Fig. 5 illustrates the FE 6
ε
models for some typical specimens. −0.2023 ⎛ ⎞⎜ ⎟

⎝ εc ⎠ (1)

3.2. Material constitutive model

⎧ σ = E0 ε , ε≤εt

PQ-FIBER user subroutine developed by Tsinghua University [32]

⎩
(
⎨ σ = ftm 1 −
ε − εt
εm − εt ) ε, ε>εt
(2)
was used to describe the material properties for both the RC frame
modeled by B21s and for the rebars in core columns modeled by T2D2s. where and ftm are the compressive strength and tensile strength of the
This subroutine, based on a fiber line element model, is a collection of masonry, respectively; εc and εt are the strains at compressive strength
user-defined uniaxial hysteretic constitutive models for concrete and and tensile strength of the masonry respectively; εm is the ultimate
steel bar in ABAQUS. The material properties of the rebars in the RC tensile strain of the masonry, and is assumed to be ten times value of εt ;
frame and in the core columns of masonry were defined by User Ma- and E0 is the elastic modulus of the masonry. For the concrete of core
terial USteel 02, a kinematic hardenging model with degraded linear columns, the stress-strain behaviors during compression and tension
reloading stiffness according to Clough [33]. The concrete properties of were specified in accordance with the Chinese national standard
the RC frame were defined by User Material UConcrete 02 considering GB50010-2010 [38], based on the tested value of the compressive
tensile strength and damage-induced deterioration, similar to Concrete strength of the fine aggregate concrete in the core column. The relations
02 in Opensees [34]. Since stirrups were not constructed in the FE are as follows:
model, the confined-concrete constitutive model proposed by Mander For compression behavior:

102
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 10. Experimental force-displacement hysteresis curves.

σ = (1−dc ) Ec ε (3a) elastic stress-strain behavior before reaching the compressive strength
and the tensile strength. The values of the material properties inputted
ρc n
⎧ 1− n − 1 + xn
, x≤1 for FE models are summarized in Tables 3–5.
dc = ρc
⎨1 − , x>1
⎩ α c (x − 1)2 − 1 + x (3b) 4. Experimental and FE Results

fc Ec εc ε
ρc = ,n= ,x= 4.1. Damage behavior
Ec εc Ec εc − fc εc (3c)
Fig. 6 presents the final damage conditions at the conclusion of the
where α c is the parameter controlling the descending branch of uniaxial
experiment. For all specimens, the concrete crushed and spalled with
compression; f c is the uniaxial compressive strength; εc is the strain at
exposure and buckling of the longitudinal reinforcement exposing and
the uniaxial compressive strength; and Ec is the elastic modulus of
buckling at the beam ends and the column bottoms. Thus, the failure
concrete.
mode of all the frame elements was flexural failure, except the frame
For tensile behavior:
columns of GWF2, due to the formation of plastic hinges. There were
σ = (1−dt ) Ec ε (4a) two differences in the failure mode for the GWF2 frame columns. One
difference was that the longitudinal rebar initially yielded not at the
5
⎧1 − ρt (1.2 − 0.2x ), x ≤ 1 bottom of the columns but higher up the columns. The other was that
dt = ρt
⎨ 1 − α (x − 1)1.7 + x , x > 1 there was significant development of diagonal cracks at the column
⎩ t (4b)
bottoms and several stirrups yielded under the drift ratio of h/50.
ft ε Therefore, the failure mode of the GWF2 frame columns was flexural-
ρt = ,x= shear failure due to the exhibited characteristics of both flexure and
Ec εt εt (4c)
shear. However, the damage condition of the infill wall was different
where αt is the parameter controlling the descending branch of uniaxial for different constructs. The infill wall of GWF2 was seriously damaged
tension; ft is the uniaxial tensile strength; and εt is the strain at the with half of the blocks crushing and falling off, whereas those of GWF1
uniaxial tensile strength. and RWF2 were damaged slightly less extensively than that of GWF2,
To simplify the calculation, the materials of the masonry and the with several blocks crushing and spalling. The infill wall of RWF1 was
core column were assumed to have no damage and to exhibit linear less damaged for only horizontal and stepped cracks, together with

103
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 11. Comparison of experimental and numerical force-displacement hysteresis curves.

minor sliding, were detected along mortar joints. the concrete cracking and rebar yielding that occurred at the column
The drift ratios during the damage evolution were recorded and are bottoms appeared nearly at the same drift ratio for the infilled frames,
listed in Table 6. It can be seen that concrete cracking and rebar which was seemly unrelated with the connection. Similarly, the first
yielding, whether happening in the beam or in the column, occurred cracking of infill wall occurred earlier in rigid specimens GWF1 and
late in the bare frame specimen PF due to a lack of thrust from others. GWF2 than in flexible specimens RWF1 and RWF2.
At the beam ends, the initial cracking of concrete and initial yielding of Take GWF1 for instance, Figs. 7 and 8 present the developments of
rebar occurred earliest in the rigid specimens GWF1 and GWF2, which the axial stress along the frame elements and the equivalent plastic
was associated with the connection type of wall-frame. Nevertheless, strain in the infill wall, respectively，as obtained from the FE

104
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 12. Force-displacement envelope curves.

simulation. The numerical equivalent plastic strain values in all the specimens are plotted in Fig. 12. As shown in Fig. 12, the strength of
infill walls at the final backward loading of − 55.2 mm are shown in each specimen increased until reaching peak strength. In contrast,
Fig. 9, except the value of GWF2 is shown at the second forward loading stiffness degradation was evident for all specimens during the loading
of 55.2 mm due to the program termination. As shown in these figures, process as the curve progressively inclined towards the displacement
the FE modeling results basically illustrated the damage evolution. axis. The numerical envelope curves are asymmetric due to the asym-
metry of the hysteretic force-displacement responses. All specimens
4.2. Hysteretic force-displacement response showed almost the same calculated initial stiffness since the FE models
overestimated the stiffness for the low stiffness specimens, PF and
The experimental results for the infilled/bare frame comparisons of RWF1. After a small displacement loading, however, the numerical
hysteretic force-displacement response are shown in Fig. 10. The higher envelope curves in the forward direction become distinct, due to the
and steeper hysteresis loops of each infilled frame demonstrate re- influence of connection type and core columns. Fig. 13 gives the nu-
markable gains of strength and stiffness compared to the hysteretic merical force-displacement envelopes compared with the experimental
response of the bare frame PF, particularly before reaching peak data. For the flexible connection specimens RWF1 and RWF2, the nu-
strength. It can also be seen from Fig. 10 that the area bounded by the merical envelopes are close to the experimental results. For the bare
hysteresis loop of each specimen increases as the lateral load increases. frame PF, the numerical envelope overestimates strength and stiffness
In the large displacement stage, the hysteresis loops of the infilled until reaching a moderate displacement of 19.7 mm (i.e. drift ratio
frames area are richer than those of the bare frame PF, resulting in a δ ≤ 1/70 ). For the rigid connection specimens GWF1 and GWF2, the
mitigated pinching effect. These observations indicate that the infill numerical results match well with the experimental at displacement less
wall significantly contributes to the strength, stiffness, and energy than 9.2 mm (i.e. drift ratio δ ≤ 1/150 ).
dissipation of the structure. The highest and steepest hysteresis loop of Tables 7 and 8 list the strength and the stiffness at characteristic
GWF2, implying the highest strength and stiffness, suggests that the points of the experimental and numerical envelopes, as the mean of the
infill wall with core columns and the rigid connection to the sur- two reversal loading directions. The data show that the presence of
rounding frame has a significant influence on strength and stiffness. masonry infill wall increases the strength and the stiffness of the RC
However, the minimal pinching effect on hysteresis loops of RWF2 frame. The two rigid connection specimens, GWF1 and GWF2, ex-
seems to indicate that incorporation of an infill wall with core columns hibited higher increases of strength and stiffness compared with the
and flexible connection has a significant effect on energy dissipation corresponding specimens constructed with flexible connection speci-
capacity. mens, RWF1 and RWF2. Similarly, GWF2 and RWF2, the specimens
The experimental/numerical comparisons of the hysteretic force- containing the core columns, showed greater increases in strength and
displacement responses of specimens are shown in Fig. 11. The nu- stiffness than the corresponding specimens without core columns,
merical results seem to be acceptable except for the asymmetry between GWF1 and RWF1. These observations indicate that the contribution of
forward loading and backward loading, resulting in an underestimation rigid connection to the enhancement of strength and stiffness is greater
of strength and an overestimation of the pinching effect on hysteresis than that of the flexible connection for direct load transferring via the
curves in backward loading, especially for the rigid connection speci- rigid connection between the frame and the infill. Similarly, the con-
mens GWF1 and GWF2. The more and more obvious asymmetry with tribution of infill with core columns is greater than that of infill without
the increased cycles may be related to the concrete damaged plasticity core columns for the reinforcement of infill. Overall, GWF2 exhibited
model which overestimates the cumulative damage in backward the highest strength and stiffness of all specimens, whereas RWF1 only
loading following repeated forward loading. exhibited slightly higher strength and stiffness than the bare frame PF.
Moreover, the contribution of the rigid connection to strength and
4.3. Strength and stiffness stiffness is likely greater than that of the core column, since the strength
and the stiffness of GWF1 are greater than those of RWF2.
To clearly reveal the strength and stiffness of different specimens, For strength prediction, the FE models underestimated the strengths
the experimental and numerical force-displacement envelopes of all of all specimens except the yield strength of PF. The relative errors of

105
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 13. Comparison between experimental and numerical force-displacement envelope curves.

the calculated yield strength were less than 22%. With the accumula- overestimated the initial stiffness and the secant stiffness at the peak
tion of the errors in damage and other aspects, the relative errors of strength of the bare frame PF.
peak strength and failure strength were less than 29%. For stiffness
prediction, there were some significant deviations between the FE 4.4. Deformability and ductility
model predictions and the test results, particularly for the initial stiff-
ness. The FE models evidently underestimated the initial stiffness of the Based on the hysteretic force-displacement envelope, the deforma-
rigid connection specimens GWF1 and GWF2, and significantly tion capacity and ductile behavior of the specimens can be analyzed. A

106
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Table 7
Lateral strength at characteristic points.

Specimen Yield strengtha Py (kN) Peak strength Pmax (kN) Failure strengthb Pu (kN)

Tested Py, t Tested strength ratioc Calculated Py, c Py, c Tested Pm, t Tested strength ratio Calculated Pm, c Pm, c Tested Pu, t Calculated Pu, c Pu, c
Py, t Pm, t Pu, t

PF 246.5 1 259.3 1.05 334.7 1 294.5 0.88 284.47 250.33 0.88

GWF1 334.3 1.35 264.2 0.79 421.5 1.26 304.5 0.72 358.35 258.83 0.72
RWF1 271.5 1.10 267.2 0.98 340.5 1.02 306.0 0.90 289.43 260.10 0.90
GWF2 384.9 1.56 323.8 0.84 495.5 1.48 371.5 0.75 421.21 315.78 0.75
RWF2 281.8 1.14 274.5 0.97 389.7 1.16 332.0 0.85 331.22 282.20 0.85

Note: .
a
Yield strength is determined by energy equivalent method on force-displacement envelope curve.
b
failure strength is the load of 85% of the peak strength in the descending stage.
c
tested strength ratio is the ratio of the tested strength of the infilled frame to that of the bare frame.

Table 8
Lateral stiffness at characteristic points.

Speci-men Initial stiffness Secant stiffness at yield strength Secant stiffness at peak strength
(kN/mm) (kN/mm) (kN/mm)

Tested Tested stiffness ratioa Calculated K 0, c Tested Tested stiffness ratio Calculated Ky, c Tested Tested stiffness ratiob Calculated Km, c
K 0, t K 0, c K 0, t Ky, t Ky, c Ky, t K m, t K m, c Km, t

PF 32.93 1 57.50 1.75 12.33 1 25.17 2.04 7.54 1 5.82 0.77

GWF1 124.06 3.77 63.08 0.51 31.24 2.53 35.70 1.14 9.61 1.27 9.27 0.96
RWF1 55.7 1.69 58.48 1.05 18.47 1.50 21.55 1.17 7.95 1.05 6.05 0.76
GWF2 140.61 4.27 64.53 0.46 44.75 3.63 33.04 0.74 17.96 2.38 9.92 0.55
RWF2 71.52 2.17 60.43 0.84 26.09 2.12 27.45 1.05 8.72 1.16 8.02 0.92

Note: .
a
Tested stiffness ratio is the ratio of the tested stiffness of the infilled frame to that of the bare frame.

Table 9
Lateral displacement at characteristic points.

Specimen Yield displacementa Peak displacementb Ultimate displacementc Ductility factor

∆y (mm) ∆m (mm) ∆u (mm) ∆u / ∆y

Tested Calculated Tested Calculated Tested Calculated Tested Calculated

PF 20.0 10.3 44.37 50.6 56.96 58.01 2.82 5.63

GWF1 10.7 7.4 43.88 32.85 50.41 38.26 4.71 5.17
RWF1 14.7 12.4 42.82 50.6 61.24 54.86 4.17 4.42
GWF2 8.6 9.8 27.60 37.45 47.57 70.53 5.53 7.20
RWF2 10.8 10.0 44.71 41.40 62.22 71.76 5.76 7.18

Note: .
a
Yield displacement is the displacement at yield strength.
b
peak displacement is the displacement at peak strength.
c
ultimate displacement is the displacement at failure strength.

yield-based ductility factor is defined as the ultimate displacement re-

lative to the yield displacement. Table 9 lists the average displacement
from the forward and backward directions and the ductility factor for
each specimen. The calculated ultimate displacement does not include
the backward displacement value because the FE model does not si-
mulate the descending trend in this direction. For this and other reasons
such as stiffness overestimation, there are significant errors in the cal-
culated ultimate displacement and ductility.
As can be seen from the experimental data in Table 9, the specimens
with the rigid connection showed smaller displacements than the cor-
responding specimens with flexible connection, except for the similar
peak displacements of specimens GWF1 and RWF1. The displacement
of GWF2 at every stage was the smallest of all the specimens due to the
greatest stiffness resulting from the infill wall with rigid connection and
core columns. In spite of this high stiffness, GWF2 still exhibited a re-
lative high ductility. Conversely, Specimen PF, with the lowest initial
stiffness, exhibited the largest yield displacement, 2.3 times that of
Fig. 14. Experimental energy dissipation-displacement curves.
GWF2. The deformability of PF gradually decreased, resulting in a

107
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Fig. 15. Comparison of experimental and numerical energy dissipation-displacement curves.

moderate ultimate displacement that was 1.2 times that of GWF2 and increasing strength, leads to a reduction or even reversion in the dif-
the lowest ductility. The results show that the strong stiffening effect ference of deformability between the infilled frame and the bare frame.
caused by the mildly damaged infill wall at yield load allows a sig- Finally, the ductility of the infilled frame was higher than that of the
nificantly smaller yield displacement of the infilled frame compared to bare frame, and the infilled frame with core columns showed notably
the bare frame. However, the stiffness degradation caused by increas- higher ductility than the infilled frame lacking the core columns. Thus
ingly damaged infill wall at the large displacement stage, together with Specimen RWF2 showed the highest ductility among all specimens.

108
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

Unfortunately, the test results do not explicitly reveal the influence of stage, but showed less energy dissipation in the small displacement
connection type to displacement ductility. stage. Due to the obviously large influence of core columns on en-
ergy dissipation, the superiority of flexible connection was not ap-
4.5. Energy dissipation parent until extreme large displacement occurred.
(4) The FE model, combining the available fiber model for a RC frame
The energy dissipated in each cycle was summed for each dis- and the concrete damage plasticity model in ABAQUS for a masonry
placement level for both the test and simulation. The resulting energy wall, can well simulate the pinching effect on hysteresis curves. The
dissipation versus displacement from the experiment is displayed in numerical hysteresis curves are acceptable in shape but are asym-
Fig. 14. The comparison of the tested and the numerical results is de- metrical resulting in large errors of strength, stiffness etc. in back-
picted in Fig. 15. The energy dissipation versus displacement curves of ward loading. The FE model overestimated the strength and stiff-
specimens PF and RWF1, evaluated by the FE models, show basic cor- ness for the low stiffness specimens such as PF and RWF1 during
respondence with the tested results. However, the calculated and tested early stage, but underestimated them for the high stiffness speci-
results for GWF1, GWF2 and RWF2 are significantly different either in mens such as GWF1, GWF2 and RWF2 in large displacement stage.
curve shape or in magnitude. The underestimation of the energy dis- It is difficult to accurately simulate the complex cyclic behavior of
sipation by the FE model results from strength underestimation and infilled frames as it involves several material models and numerous
pinching effect overestimation of the numerical hysteresis curves material parameters.
during backward loading. (5) Considering the uncertainties related to the number of specimens,
As shown in Fig. 14, the bare frame PF exhibited the worst dis- scale effects, variability of masonry's mechanical properties and
sipative property due to the absence of an infill wall, while infilled other factors, further experimental and analytical studies will be
frames GWF2 and RWF2 exhibited the best dissipative properties due to necessary to conduct.
the presence of infill and core columns. During initial loading, the en-
ergy dissipations of RWF1 and RWF2 were slightly lower than those of Acknowledgements
GWF1 and GWF2, as the flexible connection weakens the friction en-
ergy dissipation across the surrounding frame and retards the dissipa- The authors gratefully acknowledge the financial support by the key
tion of friction energy across the cracks of the infill. At a displacement technology R & D program of Tianjin under Grant No. 12ZCZDSF00500,
of about 13 mm, Specimen RWF1 began to dissipate more energy than the National Natural Science Fund of China under Grant No. 51678389,
the corresponding GWF1, owning to crack development in the infill. A and Tianjin Research Program of Application Foundation & Advanced
little more energy was dissipated in RWF2 compared to GWF1 but it Technology under Grant No. 15JCYBJC48900.
was less than that in GWF2. Finally, RWF2 showed a rapid increase in
energy dissipation and when the displacement exceeded 46 mm, it References
dissipated more energy than GWF2.
[1] Asteris PG, Giannopoulos IP, Chrysostomou CZ. Modeling of infilled frames with
5. Conclusions openings. Open Constr Build, Technol J 2012;6:81–91.
[2] Fiore A, Netti A, Monaco P. The influence of masonry infill on the seismic behaviour
of rc frame buildings. Eng Struct 2012;44:133–45.
The cyclic behavior of fully infilled RC frames with different con- [3] Celarec D, Dolšek M. Practice-oriented probabilistic seismic performance assess-
nection types and different constructional details of infills, compared ment of infilled frames with consideration of shear failure of columns. Earthq Eng
Struct Dyn 2013;42:1339–60.
with a bare frame, were investigated experimentally and analytically [4] Asteris PG, Repapis CC, Repapi EV, Cavaleri L. Fundamental period of infilled re-
using FE modeling. Based on the results, the following conclusions can inforced concrete frame structures. Struct Infrastruct Eng 2016;13:929–41.
be drawn: [5] Uva G, Porco F, Fiore A. Appraisal of masonry infill walls effect in the seismic re-
sponse of RC framed buildings: a case study. Eng Struct 2012;34:514–26.
[6] Al-Chaar G, Issa M, Sweeney S. Behavior of masonry-infilled nonductile reinforced
(1) The presence of the masonry infill wall significantly increased the concrete frames. J Struct Eng 2002;128:1055–63.
lateral strength and the stiffness of the RC frame in the small dis- [7] Turk AM, Ersoy U, Ozcebe G. Effect of introducing RC infill on seismic performance
of damaged RC frames. Struct Eng Mech 2006;23:469–86.
placement stage, but this increase gradually become smaller with
[8] Anil Ö, Altin S. An experimental study on reinforced concrete partially infilled
the increased lateral displacement due to damage development of frames. Eng Struct 2007;29:449–60.
the infill. The strength and the stiffness of the infilled frames with [9] Mehrabi AB, Shing PB, Schuller MP, Noland JL. Experimental evaluation of ma-
rigid connection were higher than those of the corresponding sonry-infilled RC frames. J Struct Eng 1996;122:228–37.
[10] Cavaleri L, Di Trapani F. Cyclic response of masonry infilled RC frames: experi-
frames with flexible connection. Construction of core columns can mental results and simplified modeling. Soil Dyn Earthq Eng 2014;65:224–42.
further increase the effect of infill wall. The connection type makes [11] Kakaletsis DJ, Karayannis CG. Influence of masonry strength and openings on in-
a greater contribution to strength and stiffness than does the pre- filled R/C frames under cycling loading. J Earthq Eng 2008;12:197–221.
[12] Stylianidis KC. Experimental investigation of masonry infilled RC frames. Open
sence of the core column. Constr Build Technol J 2012;6:194–212.
(2) Although the stiffening effect of the infill wall caused a considerable [13] Asteris PG. Lateral stiffness of brick masonry infilled plane frames. J Struct Eng
decrease in the yield displacement of the structure, there was a 2003;129:1071–9.
[14] Blackard B, Willam K, Mettupalayam S. Experimental observations of masonry in-
small drop or even an increase in the ultimate displacement for the filled reinforced concrete frames with openings. ACI Spec Publ 2009;265:199–221.
infilled frame with rigid connection or with flexible connection. As [15] Kakaletsis DJ, Karayannis CG. Experimental investigation of infilled reinforced
a result, the displacement ductility of the infilled frames with or concrete frames with openings. ACI Struct J 2009;106:132–41.
[16] Jiang HJ, Liu XJ, Mao JJ. Full-scale experimental study on masonry infilled RC
without core columns was more than 5 or 4 respectively for both moment-resisting frames under cyclic loads. Eng Struct 2015;91:70–84.
connection types, showing higher ductility compared to the bare [17] Asteris PG, Kakaletsis DJ, Chrysostomou CZ, Smyrou E. Failure modes of in-filled
frame. An infill wall with core columns showed greater effect on frames. Electron J Struct Eng 2011;11:11–20.
[18] Chiou Y-J, Tzeng J-C, Liou Y-W. Experimental and analytical study of masonry
ductility compared with an infill wall lacking core columns.
infilled frames. J Struct Eng 1999;125:1109–17.
(3) The energy dissipations of infilled frames showed substantial im- [19] Stavridis A, Shing PB. Finite-element modeling of nonlinear behavior of masonry-
provement over those of the bare frame, and the presence of core infilled RC frames. J Struct Eng 2010;136:285–96.
columns further improved energy dissipation capacity. The con- [20] Koutromanos I, Stavridis A, Shing PB, Willam K. Numerical modeling of masonry-
infilled RC frames subjected to seismic loads. Comput Struct 2011;89:1026–37.
nection between the frame and infill, however, did not show a [21] Mohyeddin A, Goldsworthy HM, Gad EF. FE modelling of RC frames with masonry
consistent influence on energy dissipation during the entire loading infill panels under in-plane and out-of-plane loading. Eng Struct 2013;51:73–87.
process. The infilled frame with flexible connection dissipated more [22] Cavaleri L, Di Trapani F. Prediction of the additional shear action on frame mem-
bers due to infills. Bull Earthq Eng 2015;13:1425–54.
energy than one with rigid connection in the large displacement

109
Q. Peng et al. Soil Dynamics and Earthquake Engineering 108 (2018) 96–110

[23] Asteris PG. Finite element micro-modeling of infilled frames. Electron J Struct Eng Architecture and Building Press; 2010.
2008;8:1–11. [32] Lu XZ, Ye LP, Miao ZW. Seismic elasto-plastic analysis of buildings: theory, model
[24] Asteris PG, Cotsovos DM. Numerical investigation of the effect of infill walls on the and practice in ABAQUS, MSC. MAC and SAP2000. Beijing, China: China
structural response of rc frames. Open Constr Build Technol J 2012;6:164–81. Architecture and Building Press; 2009.
[25] Madan A, Reinhorn AM, Mander JB, Valles RE. Modeling of masonry infill panels [33] Clough RW, Johnston SB. Effect of stiffness degradation on earthquake ductility
for structural analysis. J Struct Eng 1997;123:1295–302. requirements. In: Proceedings of the 2nd Japan Earthquak Engineering Symposium,
[26] Crisafulli FJ, Carr AJ, Park R. Analytical modelling of infilled frame structures-a Tokyo. 1966. p. 227–232.
general review. Bull N Z Soc Earthq Eng 2000;33:30–47. [34] Mazzoni S, McKenna F, Scott MH, Fenves GL. OpenSees Command Language
[27] Asteris P, Antoniou S, Sophianopoulos D, Chrysostomou C. Mathematical macro- Manual. Berkeley: Pacific Earthquake Engineering Research Center; 2007〈http://
modeling of infilled frames: State of the art. J Struct Eng 2011;137:1508–17. opensees.berkeley.edu/OpenSees/manuals/usermanual/
[28] Asteris PG, Cavaleri L, Trapani FD, Sarhosis V. A macro-modelling approach for the OpenSeesCommandLanguageManual.pdf〉.
analysis of infilled frame structures considering the effects of openings and vertical [35] Mander JB, Priestley MJN, Park R. Theoretical stress-strain model for confined
loads. Struct Infrastruct Eng 2016;12:551–66. concrete. J Struct Eng 1988;114:1804–26.
[29] Asteris PG, Cotsovos DM, Chrysostomou CZ, Mohebkhah A, Al-Chaar GK. [36] Li ZX, Zhou XJ, Xia DT, Xu XB, Liu ZF. Experimental study on mechanical properties
Mathematical micromodeling of infilled frames: state of the art. Eng Struct of non-autoclaved and non-sintered fly-ash perforated brick masonry. Earthq Eng
2013;56:1905–21. Eng Vib 2012;32:92–8. [in Chinese].
[30] Ministry of Housing and Urban-Rural Developmenet of the People's Republic of [37] Akhaveissy AH, Desai CS. Unreinforced masonry walls: Nonlinear finite element
China. Technical specification for concrete small-sized hollow block masonry analysis with a unified constitutive model. Arch Comput Method E
buildings (JGJ/T 14-2011). Beijing, China: China Architecture and Building Press; 2011;18:485–502.
2011. [38] Ministry of Housing and Urban-Rural Developmenet of the People's Republic of
[31] Ministry of Housing and Urban-Rural Developmenet of the People's Republic of China. Code for design of concrete structure (GB50010-2010) Beijing, China. China
China. Code for seismic design of buildings (GB50011-2010). Beijing, China: China Architecture and Building Press; 2010.

110