Engineering Structures 91 (2015) 125140

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Nonlinear seismic responses and lateral force transfer mechanisms of RC

frames with different inll congurations
Y.P. Yuen a,, J.S. Kuang b
a
b

Department of Civil Engineering, Bursa Orhangazi niversitesi, Bursa, Turkey

Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

a r t i c l e

i n f o

Article history:
Received 15 October 2014
Revised 23 February 2015
Accepted 24 February 2015
Available online 17 March 2015
Keywords:
Inlled frame
Nonlinear seismic response analysis
Discrete nite element
Unreinforced masonry wall

a b s t r a c t
Due to the architectural efciency of masonry-inlled reinforced concrete frames, the frames are highly
common structural forms for buildings. However, the inlls can signicantly modify the structural behaviour of these frames, which can be detrimental to the seismic performance of buildings. This study
investigates the seismic response and failure mechanisms of inlled RC-frame structures with ve different inll congurations: (1) full inlls, (2) 2/3-storey-height inlls, (3) a soft rst storey, (4) inlls with
window openings and (5) inlls with door openings. The nonlinear response history behaviour of the
masonry-inlled RC frames under four realistic earthquakes, namely, the 1979 El Centro, 1987
Superstition Hills, 1995 Kobe and 1999 Chi-Chi earthquakes, were simulated using discrete-nite element analysis with damage-based constitutive relations. The analysis indicated that the degrees of
continuity and regularity of the inll panels crucially affect the seismic performance of structures. As long
as out-of-plane collapse of inlls does not occur, full-height and continuous-inll panels can enhance the
overall stability and energy dissipation of frame structures. By contrast, discontinuous inlls can inict
serious damage localised at the points of discontinuity in the frame members. Furthermore, the analysis
revealed that the design concept of strong columnweak beam may not be always applicable to inlled
frames.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Inlled reinforce concrete (RC) frame structures are amongst
the most common structural forms for low- to medium-rise buildings in the world. For conventional inlled frame buildings, inll
panels are used as partitions or cladding, whereas the bounding
frame is designed as a structural skeleton to withstand external
loading. Despite their architectural efciency, catastrophic structural failures of and damage to inlled RC frame buildings have
been reported in nearly all destructive earthquake events, including the Turkey Kocaeli earthquake [1] in 1999, the Central Peru
earthquake in 2007 [2], the China Sichuan earthquake in 2008 [3]
and the Emilia earthquake in 2012 [4]. The actual seismic failure
modes and performances of inlled RC frames typically differ from
those anticipated based on the original structural analyses performed by design engineers.
The major cause of the mismatch between the observed and
theoretical seismic performance of inlled structures is engineers
neglect of the inuence of inlls on overall structural performance
Corresponding author. Tel.: +90 537 2643447.
E-mail address: [email protected] (Y.P. Yuen).
http://dx.doi.org/10.1016/j.engstruct.2015.02.031
0141-0296/ 2015 Elsevier Ltd. All rights reserved.

because inlls are normally considered non-structural components

[5,6]. However, the structural behaviour of inlled frames and bare
frames may differ because the presence of inlls, particularly
irregularly arranged inlls, can signicantly alter local stress distributions in the adjacent frame members and vary the overall
structural dynamic behaviour [7,8]. Furthermore, the brittle nature
of masonry inll materials often imposes more harm on the hysteretic behaviour of structures and causes severe localised damage
in the adjacent braced structural members under reversed cyclic
loading. These modications of structural action can be detrimental to the seismic performance of buildings, resulting in
non-reparable damage of the adjacent structural members or even
collapse of weak/soft storeys, jeopardising human lives and property [9].
Although the seismic vulnerability of inlled frame structures
has long been recognised and studied and different methods for
modelling inlls have been proposed [1013], provisions for their
design and modelling were not implemented in design standards
or guidelines until the last decade [14]. However, the various limitations of these models hinder their application in simulations and
studies of all possible failure modes of inlled frame structures
under seismic or general non-proportional loading. Such

126

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

difculties, including the simulation of complete evolutionary histories of damage and the associated eld of stress ow, are frequently encountered in inlls with complicated geometry, such
as inll walls with openings. Obviously, the introduction of openings in the inll walls would lead to discontinuous loading path
resulting drastic decrease in the strength and stiffness [15].
Thus, to simulate the highly nonlinear and loading-path dependent seismic responses of inlled frame structures, discrete models
that explicitly address crack propagation in inlls are the most
suitable approach. Accordingly, a discrete model for inlls with
damage-based cohesive crack representation that can simulate
the fracture and post-fracture behaviour of inlls has been developed [16].
The aim of this study is to investigate the seismic response and
vulnerability of multi-storey masonry-inlled RC frame structures
with typical regular or irregular inll congurations, including a
fully inlled frame, two structures with vertically discontinuous
inlls (a 2/3-storey-height inlled frame and an inlled frame with
a soft rst-storey), and two structures with opening inlls (an
inlled frame with window openings and an inlled frame with
door openings). The seismic behaviour, particularly the failure
modes, hysteresis behaviour and force transfer mechanisms, of
the inlled frame structures are thoroughly investigated via a series of nonlinear response history analyses using the discrete nite
element method.

and albeit complex and non-smooth boundary surfaces can be

dened for each element. In the combined discrete-nite element
method (DFEM), the models also consist of discrete objects, and
their interactions are dened as in the usual discrete element
method. The subtle difference between DEM and DFEM is the discrete objects, which are now no longer treated as rigid bodies but
discretised as nite continuum elements, for which deformation
and constitutive relations can be dened. DFEM, like DEM, is
superior for addressing problems involving strong discontinuous
kinematic elds, such as fracture, dislocation and shear band problems, while simultaneously simulating the deformation of elements under internal stress. DFEM is particularly suitable for this
study because cracks in mortar joints lead to discontinuities in displacement elds and masonry units are deformable and thus are
not suitable for treatment as rigid objects.
The concrete frame and masonry units are discretised into linear 8-node, 3-D solid elements, and their mechanical behaviour
is modelled according to the smeared isotropic damage-plasticity
law [20]. The embedded steel reinforcement is modelled with a
2-node, 3-D truss element, and the modied MenegottoPinto
model [21] is used to present the constitutive relationship. The
maximum meshing size for the frame and steel bars is not greater
than 200 mm, and the mesh is rened to 80 mm at the stress concentration zones; for the inlls, the meshing size is approximately
150 mm, as shown in Fig. 2.

2. Inlled RC frame prototype structures

3.2. Interfacial constitutive model for mortar joints

The prototype structure is a two-storey RC building frame

designed to resist earthquakes with a PGA of 0.15 g and detailed
to obtain an expected displacement ductility factor of 24 [17].
The uniaxial strengths of the concrete and longitudinal and transverse reinforcements are 20.1 MPa, 400 MPa and 235 MPa, respectively. The 125-mm-thick inll panels are composed of 600 
300  125-mm masonry units (compressive strength = 15 MPa)
and 10-mm-thick mortar joints (compressive strength = 5 MPa).
Five typical types of masonry-inll arrangement are considered
in this study: (1) full inlls, (2) 2/3-storey-height inlls, (3) soft
rst-storey, (4) inlls with window openings and (5) inlls with
door openings. The window size is 900 mm square, with two windows in each wall. The windows are located at 2/3 height (from
bottom to top) of the wall and 1/3 width (from left to right) of
the wall. The door size is 900  2100 mm, with one door in each
wall. The door is located in the middle-bottom of the wall. The
details of the prototype models are shown in Fig. 1. The types of
inlls selected for this study are rather widely adopted in Europe
[18] and Asia. In North America, typical constructional inlls
(CMU or clay bricks with type-N mortar) have similar and slightly
higher material strengths compared to the prototype inlls considered in this study.

As the strength of mortar joints is often much weaker than that

of masonry units in inll walls, damage and cracks will likely
propagate along the joints. To simulate the pre- and post-fracture
behaviour of mortar joints, a distinct approach is adopted in which
no mortar element is inserted between the masonry units but
damage-based cohesive interactions with nite sliding formulation
are enforced on the contact surfaces of the masonry units.
Cohesive elements are often used to model adhesives or interfacial zones between two components. Although most of the cohesive elements have zero thickness, continuum cohesive elements,
for which the direct membrane strains and in-plane membrane
shear strain are often assumed to be zero, are also quite popular
for cohesive zones of nite thickness. However, the major
disadvantage of the use of cohesive elements is that control of element distortion and hourglass deformation modes is necessary.
Damage or inelastic deformation will likely be concentrated in
the mortar joints, with severe distortion of the mortar elements,
and if reduced integration is employed, hourglass deformation
modes can be triggered under complicated loading conditions that
may impair the accuracy of the simulation. Thus the contact
formulation is a more effective tool to simulate extended crack
opening and propagation behaviour as well as component disintegration in inlls. The incremental form of the traction-separation
law for cohesive cracks considering the mixed-mode fracture behaviour of the mortar joints is given by

3. Discrete nite element models

3.1. DFE modeling
To adequately capture the local inll-frame interacting behaviour, detailed modelling and discretisation of the frame and inll
components are required. The discrete nite element models are
constructed with ABAQUS [19], while all components are discretely
modelled with appropriate elements, and mechanical interactions
are enforced amongst them. The discrete element method (DEM)
is commonly used for granular or particle contact and ow problems, in which element-to-element interactions are dened using
contact mechanics and/or potential functions. In contrast to typical
continuum FEM, the discrete elements are treated as rigid bodies,

8
9 2
9
38
knn 1  D  Ht n
0
0
>
>
< Dt n >
=
< Dun >
=
6
7
Dt s 4
0
ks 1  D
0
5 Dus
>
>
>
>
:
;
:
;
Dt t
Dut
0
0
kt 1  D
1
where t = {tn, ts, tt} and [u] = {un, us, ut} are the traction and displacement jump vector between two masonry unit surfaces, respectively,
and ke is an initial isotropic elastic stiffness tensor. The damage
variable D is a scalar parameter of value within [0,1], and normal
stiffness is assumed to be completely recovered under the compressive normal traction to account for the unilateral effect. H(tn) is a
step function: H(a) = 1 if a P 0, otherwise H(a) = 0, which is used

127

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

Fig. 1. Prototype structures: (a) bare frame; (b) full inlls; (c) 2/3-storey height inlls; (d) soft rst storey; (d) window opening inlls; (e) door opening inlls.

Fig. 2. Meshing and interaction of fully inlled frame model: (a) concrete frame and inlls; (b) steel reinforcement; (c) inlls-frame contact conditions.

to characterise the unilateral effect or stiffness recovery effect. Due

to the masonry units restraint, the macroscopic cracks, which
inict macroscopic stiffness degradation, in mortar joints are eventually aligned with the bed planes regardless of fracture modes. By
considering the major macroscopic damage mechanism and typical
structural actions experienced by inlls, it is sufcient to use a damage variable to model the stiffness degradation effect due to the

formation of cracks in mortar joints. The criterion of damage initiation is dened as

2

t2n b2s t 2s b2t t2t  f nt 0 tn P 0

2
2

b2s jts j tn tan us 2 b2t jt t j t n tan ut 2  f nt 0 tn < 0

128

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

where fnt is the tensile strength of the mortar joints; bs and bt are
the ratios of the tensile strength to the mode II shear cohesion
strength cs and mode III tear cohesion strength ct, respectively;
and /s and /t are the friction angles under mode II and mode III
deformation, respectively. The crack-driving force GT is dened as

DGT ft n gDsun  jt s j  ftn g tan /s Dus  jt t j  ftn g

 tan /t Dut 

where {} denotes the Macaulay bracket: {a} = a if a P 0, otherwise

{a} = 0.
It is assumed that the critical strain-energy release rate Gc under
mixed-mode fracture is represented [22] as

DGC mGIIC GIIIC  GIC


m
GIIC GIIIC
GT DGT
GT

where GIC, GIIC, and GIIIC are critical strain-energy release rates under
pure mode I, pure mode II, and mode III fracture, respectively; m is
an exponent that depends on the brittleness of the material; and G0
is the strain-energy release rate at damage initiation. The mixedmode fraction criterion BK law was developed and veried for ductile and brittle polymer materials, which have amorphous and polycrystalline solid structures, respectively. Hence, the fraction
criterion is suitable for a wide range of materials with different
molecular or grain structures. The evolution of damage is presented
in the form of

DD max

!
DGT
GT DGC

0
GC  G0 GC  G0 2

When the contact surfaces are under compression, the joints

behave similarly to MohrCoulomb materials, and crack propagation is governed by mode II fracture. However, when the surfaces are subjected to interacting tension and shear, a mixedmode fracture criterion of the BK form would control the crack
propagation. Once the mortar joints are fractured, D = 1, and the
post-cracking interaction property of the two contact surfaces follows Coulombs friction law; thus the slip criterion is dened as

f jt s j  ftn g tan /s 2 jtt j  ft n g tan /t 2 0

The friction force is treated as constant, i.e., the change is zero at

each time increment under the explicit Euler integration scheme,
and the possible change in friction force is accounted at the end
of integration. The model II or mode III fracture energy presented
here is the macroscopically averaged energy, minus the frictional
dissipation under normal compression, required for shearing
masonry composites to failure or fracture states. Thus frictions
are subtracted from the total shear tractions in Eq. (6), and when
a certain part of the strain energy is completely released due to
fracture D = 1, the frictional dissipation dictated by Coulombs
law (Eq. (7)) can continue. Hence, a smooth transition from the
traction-separation law to Coulombs law is attained.
The model has material parameters of 12, as calculated and
obtained from (3ke + 2b + 2u + 1fnt + 3G + 1m). The damage variable D is calculated using Eq. (5), and the interfacial shear and normal elastic stiffness are uncoupled, leaving only three elastic
constants. The presented traction-separation law simulates the
mortar joint mechanical behaviour at the macroscopic scale, and
thus the fracture energy is the macroscopically averaged energy,
minus the frictional dissipation under normal compression,
required for shearing masonry composites to fracture states.
Most of the model parameters, including the elastic constants, tensile bond strength, shear cohesion, friction angles, and mode I and
mode II fracture energy, are readily obtained from material tests,
such as joint shear tests and tension tests [23].

However, parameters related to mode III (GIII, bIII and /III) and
mixed-modes fracture (m) are generally not available. Following
usual practice, mode III parameters are taken to be the same as
those of mode II, given that mode III action is not signicant in
the studies considered in this paper or in most practical situations.
The exponent m takes a value of 2.2, as recommended for brittle
materials [22]. The cohesive cracking behaviour of mortar joints
under single mode and mixed modes of loading is illustrated in
Fig. 3.
The model is implemented in ABAQUS through the user-subroutine VUINTERACTION for traction-separation behaviour in general contact simulations. The integration scheme used for the
implementation of the model is based on a standard modied
explicit Euler scheme with substepping [24]. To illustrate the
numerical algorithms of the model implementation, the solution
steps for the mortar joints stressed from the elastic state to the
damaged state, which are similar for other states, are outlined in
Fig. 4. Various verication simulations were performed with the
proposed modelling method. The results of the simulations are in
very good agreement with the experimental results, including
material shear and tension tests (Fig. 5) and structure loading tests
(Fig. 6). Details of the simulations and comparisons between the
experimental and numerical results have been presented previously [16].
Furthermore, the interfacial interaction between the frame and
the inlls is modelled as a frictional contact problem enforced with
the penalty method. The cohesion at the interface between the
inll panel and the frame is neglected because the cohesion would
be destroyed by shrinkage of inlls, leaving a gap at the interface.
Moreover, the overall responses would not be signicantly affected
by the presence of cohesion because the major interacting forces
between the bounding frame and the inll panel are transmitted
through compressive struts running along the panel and braced
at the frame regions within the contact lengths. Tension ties are
weaker than compressive struts, and the ties cannot be formed
effectively in the inll panels due to the weak tensile strength of
the inlls, which can be easily destroyed at the early stage of loading. Hence, the interface is modelled to transmit compression and
frictional shear without tension by the contact interaction method.
3.3. Computational scheme
Due to the highly nonlinear nature of the simulation problem
involving nonlinear material responses, complicated and evolving
constraints and contacts amongst components and nonlinear geometric effects, the response history analysis employs an explicit
central-difference integration scheme with a time increment of
17 s, while double-precision numbers are used in the analysis to
retain sufcient precision. Explicit integration is chosen to avoid
convergence problems, and a very small time increment is used
to satisfy the stability limit of the explicit integration.
A rough estimation of the number of integration steps reveals
that nearly a billion increments are required for a complete timehistory analysis of earthquakes with a typical duration of 30
50 s. Due to the large amount of data generated, it is virtually
impractical to save all integration steps to output les. In this
study, following usual practices, the structural responses are saved
at certain time intervals or time frames. Field variables such as
stress, strain, and damage are recorded at time intervals of 0.1 s,
while nodal outputs at certain positions such as reactions and displacements are recorded at intervals of 0.01 s. Thus, the amount of
stored data can be reduced to a manipulable level without signicant loss of important structural information for post-analysis.
This type of modelling and analysis with a considerably long
computation time seems slightly impractical for the purposes of
engineering design but is necessary if detailed and accurate

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

129

Fig. 3. Fracture behaviour of mortar joints: (a) mode II fracture; (b) mode I fracture; (c) mixed mode I and mode II fracture.

nonlinear structural responses of inlled frames are indeed

needed. For design purposes, a preliminary analysis can be performed with simplied models, such as strut-and-tie models. If
serious damage and collapse of inlls that would signicantly jeopardise structural stability are suspected or observed, a detailed
nonlinear analysis with advanced modelling techniques is required
for rigorous studies. Sub-modelling techniques in which only some
important parts of the structure are detailed, modelled and tied to
the simplied global structure can be used to reduce computational effort.
3.4. Validation for dynamic behaviour of large assemblage structure
In addition to the verication at the material and component
levels as shown in Figs. 5 and 6, veracity of the modelling at
large-scale assemblage level is also checked by simulation of a
large scale shake table experiment by Hashemi and Mosalam,
2007 [29]. The raw experimental data can be downloaded from
can be found on the Network for Earthquake Engineering
Simulation (NEES) central data repository at https://central.nees.
org/. The prototype structure is an assemblage of three-interior
bays (length = 4115 mm) in the rst-storey (height = 3429 mm)
of a hypothetical ve storey RC building containing unreinforced
masonry inll walls. The inll wall is composed of clay bricks
(102 mm  203 mm  68 mm) and ASTM C270Type N mortar. To
reect the mass and loading effect of the upper storey of the prototype structure on the test sub-structure, a dead weight of
32.12 tons is placed on the RC slab and post-tensioning loads
(145 kN for exterior columns; 290 for interior columns) are applied
on the columns. The average concrete compressive strengths on
the
experiment
day
are
38.3 MPa
for
columns

(305 mm  305 mm),

37.2 MPa
for
beams
(long
span:
267 mm  343 mm; short span: 305 mm  229 mm) and slab
(95 mm thick), and 34.3 MPa for foundation. The masonry prisms
average compressive strength is 17.0 MPa. The yield stresses of the
reinforcing steel and post-tensioning steel are 458 MPa and
1034 MPa respectively. Following the same modelling method, a
discrete nite element model for the test structures is developed
as shown in Fig. 7.
The test structure was subjected to eight sequential base excitations of which time histories are scaled from two real earthquake
ground motions: 1994 Northridge Tarzana record (designated
TAR) and 1999 Dzce Lamont record (designated DUZ). Visible
damage such as cracks at the wallcolumn interfaces and in the
mortar head-joints started to occur during the excitation of TAR
level 4 (PGA = 0.625 g) and level 6 (PGA = 1.039 g). Furthermore,
major cracking and early signs of corner crushing occurs in the
inll wall during the DUZ level 7 excitation (PGA = 1.242 g).
Cracking and initiation of concrete spalling at the bases of the
braced columns were also observed after the DUZ 7 excitation.
Since the behaviour of the test structure is nearly complete
elastic before the TAR 4 excitation, the computational analysis here
concerns only the responses of the structure due to base acceleration time history from TAR 4 to DUZ 7. The computed base shear
and oor displacement time histories are plotted against the
experimental results in Fig. 8 and the simulated damage patterns
in the structure are shown in Fig. 9. It can be seen that the comparison between the computational and the experimental results
show a good agreement in terms of both global structural
responses and local damage patterns. Thus, the proposed modelling can accurately simulate the dynamic behaviour of real
inlled frame structures.

130

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

Fig. 4. Solution steps for the mortar joints stressed from the elastic state to the damaged state.

Fig. 5. Verication simulations: (a) cyclic shear test [25] and (b) tension test [26].

4. Characteristics of vibration modes and ground motions

The effects of inlls on the undamaged modal characteristics of
the structures are rst studied. The rst lateral vibration modes of
the inlled frame and the bare frame models are extracted as
shown in Fig. 10. Due to the strong bracing action provided by

the inlls to the bare frames, the natural vibration periods of the
rst lateral deection modes T1 are signicantly reduced from
0.426 s to 0.088 s, 0.181 s and 0.243 s for the fully inlled frame,
2/3-storey-height inlled frame and inlled frame with a soft rst
storey, respectively. The openings slightly weaken the bracing
action of full inlls, and the lateral vibration periods of the inlled

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

131

Fig. 6. Verication simulations of structure tests: (a) masonry pier shear test [27]; (b) out-of-plane loading test on inlled frame [28].

Fig. 7. DFE model of the large-scale test structure [29] (a) mesh of concrete frame and inll wall; (b) reinforcing steel bars embedded inside the concrete frame.

frame with window-openings and inlled frame with door-openings are 0.117 s and 0.147 s, respectively.
As shown in Fig. 10, the structures with vertically continuous
inlls, i.e., the fully inlled frame and the two inlled frames with
openings, have smooth lateral mode shapes, but the inlled frame
with a soft rst storey and the 2/3-storey-height inlled frame
have uneven mode shapes in which deformation is localised in
the soft parts of the structures, i.e., unbraced lengths of columns.
Furthermore, the total lumped masses on the rst and second
oors of the local bare frame are 215.76 tons and 191.30 tons,
respectively. The masonry inlls introduced an additional mass

of 5.45 tons, which is only approximately 1% of the total mass of

the whole structure, to the inlled frames with full inlls. Thus,
unlike the bracing action, the inll mass has relatively little effect
on the overall dynamic properties of the structures.
The nonlinear response histories of the inlled frames are analysed in the context of four realistic earthquakes. The earthquake
records adopted in this analysis are 1979 El Centro 1140-component at USGS-station 5056 (PGA = 0.14 g), 1987 Superstition Hills
225-component at USGS-station 5051 (PGA = 0.46 g), 1995 Kobe
000-component at station Takatori (PGA = 0.61 g) and 1999 ChiChi EW-component at station CHY080 (PGA = 0.96 g). The 1979 El

132

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

5. Failure modes and force transfer mechanisms

Subjected to the four ground motions, the resulting storey drift
response time histories and hysteresis loops (showing base-shear
against top drift) of the prototype structures are plotted in
Figs. 12 and 13, respectively. The frames, except those with openings, behave elastically, with the development of minor tensile
cracks on inll panels and frames near the corners for the El
Centro earthquake. However, at the stronger excitation levels of
the Superstition Hills, Kobe and Chi-Chi earthquakes, signicant
damage and inelastic strain development, such as propagation of
major cracks across entire inll panels and cracking and even
crushing of concrete, occurs in the structures.
Under the three stronger earthquakes, Superstition Hills, Kobe
and Chi-Chi, the incurred deformation shapes and damage distributions are similar for a specic type of inlled frame. This similarity is not in terms of absolute magnitudes of deformation and
scales of damage but rather generic deformation shapes and locations of major damage. Undoubtedly, the exact incurred damage
patterns and deformation shapes of a specic type of inlled frame
greatly depend on the loading history and thus can never be identical. Nevertheless, the structural behaviours of a particular inlled
frame under different earthquakes should at least bear some
resemblance to each other, provided that the loading planes are
the same. Hence, as an example, contour plots of the developed
inelastic strain on the structures and the associated minor principal stress distribution under the Chi-Chi earthquake are plotted
in Figs. 14 and 15 to illustrate the general deformation shapes
and damage distributions.

5.1. Effect of inlls on the seismic performance of frame structures

Fig. 8. Comparisons of numerical analysis and shake-table experimental results: (a)
oor displacement time histories; (b) base-shear time histories.

Centro and the 1999 Chi-Chi ground motions can be considered

fortication and rare earthquakes with 475-year and 2436-year
return periods, respectively, which would be experienced by the
prototype RC frame structure.
The frequency contents of the four ground motions are characterised by the elastic response spectra, as shown in Fig. 11, and the
mean period Tm. The mean period is a measure of the frequency
content of a ground motion and is the mean of the periods of the
Fourier amplitude spectrum, where the weight factor is the square
of the Fourier amplitudes. It is dened by

P 2
C Ti
T m Pi i 2
iCi

where Ci is the Fourier amplitude coefcient corresponding to period Ti.

Compared with other spectral parameters, such as the predominant period, the mean period is generally a more objective measure
of the frequency content inuencing structures. The mean periods
of the El Centro, Superstition Hills, Kobe and Chi-Chi earthquakes
are 0.49 s, 1.19 s, 1.14 s and 0.83 s, respectively. The vibration
strength of the El Centro earthquake is concentrated in the highfrequency range (T < 0.4 s). In contrast to the El Centro motion,
the Kobe and Superstition Hills earthquakes have rather evenly
distributed vibration strengths over a high to medium frequency
range (T < 2 s), while the spectrum of the NS component of the
Chi-Chi earthquake exhibits a sharp peak at approximately
T = 0.9 s. Because the Chi-Chi motion was recorded near the source,
its history of ground velocity reveals the pulse behaviour.

The nonlinear time history results of the bare frame are

regarded as the baseline data for investigating the effect of different inll congurations on the structural behaviour of RC frames.
The bare frame has stable hysteretic behaviour, as shown in
Fig. 13, similar to that of typical ductile RC frames, due to the successful development of a beam-sway mechanism in which plastic
hinges are formed at the beam ends and column bases, which
are referred to as the fuse regions, as shown in Fig. 14.
Based on the capacity design approach for frame structures subjected to seismic effects, the beam-sway mechanism is the most
desirable failure mechanism because the earthquake induced
vibration energy can be stably dissipated in designated and wellconned regions, i.e., beam ends and column bases. Thus, the columns, which serve as the most important structural components to
support buildings, are protected from severe damage during an
earthquake.
Furthermore, the regions of plastic hinges, in which the concrete is well conned by stirrups, have high rotational ductility
and do not undergo serious strength and stiffness degradation
under cyclic loading, facilitating stable cyclic loaddeformation
behaviour of the frame structures. Therefore, even when the bare
frame is subjected to the strong Chi-Chi earthquake excitation,
for which the PGA is 0.82 g, approximately 5.4-fold larger than
the original design level of 0.15 g, the structural performance
remains excellent, without fatal damage.
In contrast to the bare frame, the fully inlled frame exhibits
localised damage pattern and different levels of pinching and
slip-lock phenomena. Although the fully inlled frames have a
symmetric and regular inll arrangement, the rst-storey columns
still unavoidably suffer severe damage, as shown in Fig. 14, compared with those in the second storey. This damage is due to the
brittle failure of the inll panels and the inability of uniform
non-designed inll panels to guarantee that the new strength

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

133

Fig. 9. Simulated damage patterns of the inlled frame: (a) crack strain in the concrete frame; (b) plastic strain in the reinforcement; (c) damage in the inll wall.

Fig. 10. First lateral vibration modes of inlled and bare frame structures: (a) bare frame T1 = 0.426 s; (b) fully inlled frame T1 = 0.088 s; (c) 2/3-storey height inlled frame
T1 = 0.181 s; (d) inlled frame with soft rst storey T1 = 0.243 s; (e) inlled frame with window openings T1 = 0.117 s; (f) inlled frame with door openings T1 = 0.147 s.

134

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

Fig. 11. Elastic response spectra of ground-motion histories.

distribution in the structure can properly meet the new seismic

demand for each storey, which is affected by the altered stiffness
distribution.
Because lower storeys generally bear stronger seismic actions,
the brittle inlls in these storeys are often cracked rst. Inll cracking immediately causes a decrease in the overall stiffness and
strength of the corresponding storeys due to the damage of the
compressive struts running along the inll panels. The resulting
stress redistribution and possible soft-storey effect incur larger
deformation to the storey, with seriously cracked inll panels, as
observed for the rst storey in this study. Such localised deformation in those members can scarcely be accommodated by the original frame design method, which does not consider the effect of the
inll on the structure, resulting in serious and localised damage.
Signicant shear forces are transmitted through the struts
formed in inll panels to the bounding frame members in small
braced areas. These locally stressed regions suffer signicant shear
damage. The damage distributions, as shown in Fig. 14, can be
readily related to the principal stress (compressive) distribution
in the structure after major cracking of inlls, as shown in
Fig. 15. In addition to the shear action, columns in the fully inlled
frames can bend exurally only after the cracked inll panels are
fully mobilised in lateral sliding. However, the exural deformation of columns is still greatly restrained by the joining of inll
panels, such that the effective bending lengths of columns is signicantly reduced, as shown in Fig. 15; in other words, the short
column effect is incurred. Hence, the major damage suffered by
the frame members is shear damage, while exural damage also
occurs to the columns, although to a smaller extent.
Furthermore, gravity loading imposed on the structure is shared
by RC frames and inll panels, thus incurring axial damage on
the columns that is very minor compared to the shear and exural
damage. However, this argument is true only if the inll panels do
not collapse. As demonstrated in a biaxial excitation analysis [30],
if the inll panels collapse in the out-of-plane directions, the lateral-force transfer mechanisms and vertical loads supporting the
action of the structure can be seriously jeopardised, and severe
axial damage can be imposed to the columns, as commonly
observed in post-earthquake eld investigations.
Furthermore, while plastic hinges cannot be formed in some of
beam ends, some occur in unexpected locations of the beams, such
as near the middle of the top beams, also resulting in the bracing
actions provided by inlls, as shown in Fig. 15. Because the frame
is designed as a typical ductile frame and only the member ends
are conned with stirrups, the formation of plastic hinges at those
unconned regions would impair the ductility of the structure.

Although the fully inlled frame exhibits localised damage patterns where the concrete in the rst storey column bases is completely crushed, the global stability of the structures is enhanced
by the introduction of fully integrated inlls. The overall structural
stability is mainly contributed by the vertical supporting and
robust lateral force bracing actions provided by the fully integrated
inlls, as shown in Fig. 15. The robust lateral force transfer mechanisms impart stable slip-lock hysteresis behaviour to the fully
inlled frames, with some pinching phenomena. A benecial effect
of inlls has been reported previously [31]. However, again, this
stability can be retained only if the inll panels do not collapse.
As shown in Fig. 15, the lateral-force transfer mechanisms are
quite different before and after major cracking occurs in the inll
panels because the sliding and opening of major cracks alters the
struts locations and orientations. While initial cornercorner
diagonal struts are destroyed, new struts form after the sliding of
inll panels is halted by the bounding frame. The post-cracking
struts are more vertically inclined and originate from column bases
because the horizontally sliding cracks obstruct the transfer of
horizontal force throughout the entire inll panel. The formation
of post-cracking compressive struts may differ amongst other
types of inlls, and thus this observation may be restricted to the
case of inlls comprising a strong masonry unit with weak mortar.
However, as mentioned previously, the initial fundamental frequency of a frame structure can be signicantly amplied by inll
walls. As a result, the inlled frame structures experience much
greater seismic forces than the bare frame in the elastic states.
Hence, the incurred maximum base shear of the fully inlled
frames is 2.8-fold higher than that of the bare frame under the El
Centro earthquake. Because the structures are deformed to inelastic range under the other three stronger earthquakes, the maximum base shear is dictated by the base shear capacity of the
structures. Nevertheless, as shown in Fig. 12, the inlled frames
with higher stiffness tend to attract higher seismic demands at
an early loading stage before exceeding the capacity, after which
strength and stiffness degradations occur. The additional strength
introduced by the brittle inlls cannot compensate for the elevated
forces, and thus more severe damage is inicted on the frame
members.
5.2. Effect of vertically discontinuous inlls
Undesirable localised damage and failure modes are observed in
the structures with vertically discontinuous inlls, as shown in
Fig. 14. Under strong seismic excitations, the damage incurred in
the inlled frame with a soft rst storey is even more seriously
localised in the rst storey column ends, and similarly, some levels
of pinching are observed in the hysteresis loops, as shown in
Fig. 13.
Under the El Centro earthquake, the incurred maximum base
shear of the inlled frames with a soft storey is 1.1-fold higher than
that of the bare frame due to the stiffening effect of the inlls on
the structure. Due to the abrupt change in the vertical distribution
of the lateral stiffness and the strength of the structure, the non-inlled rst storey becomes a soft storey. In fact, the rst in-plane
lateral mode shape, as shown in Fig. 10, already suggests that the
earthquake-induced deformation will be localised in the rst
storey. As a result, an undesirable column-sway failure mechanism
occurs in the structure subjected to earthquake excitations, in
which plastic hinges, with some concrete crushing, are mainly
formed in the column ends in the rst storey. By contrast, the inlls
restrain the deformation of the upper storey, and thus little damage is incurred in the upper storey.
The column-sway failure mechanism is very dangerous for
structural stability because the vertical bracing strength will be
signicantly reduced following the localised damage of the

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

135

Fig. 12. Storey-drift time histories of inlled frame structures [vertical axis: drift (mm); horizontal axis: time (s)].

columns supporting the weight of the upper storeys. This response

is evident in the total collapse of a particular storey, i.e., the pancaking phenomenon, in structures with soft or weak storeys.
Given the serious damage incurred in the rst storey columns of
the inlled frame structure, the pancaking phenomenon will
likely occur if the structure is subjected to larger seismic demands
or larger vertical loading.

The frames with 2/3-storey-height inlls have a lower stiffness

than the fully inlled frames and thus incur a maximum base shear
that is only 1.5-fold higher than that of the bare frame under the El
Centro earthquake. When the structures deform beyond elastic
ranges under stronger earthquakes, the frame with 2/3-storeyheight inlls has the poorest seismic performance amongst the
six prototype structures. As shown in Fig. 14, the damage is

136

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

Fig. 13. Hysteresis loops of inlled frame structures [vertical axis: base shear (kN); horizontal axis: top drift (mm)].

seriously localised in the central columns, and a very severe pinching phenomenon occurs in the hysteresis loops, as shown in Fig. 13,
indicating that the structure loses most of its lateral stability during the earthquake excitations.
The damage is highly localised in the central column in the rst
storey, and even total failure, i.e., crushing of the concrete and
buckling of longitudinal reinforcements across the entire column
section, occurs in the rst storey for the Kobe and Chi-Chi excitations. This failure occurs because the central column is restrained
on both sides by the inlls, while the edge columns are only
restrained on one side, as shown in Fig. 15. Thus, the central column attracts a lateral seismic force that is 1.72.6-fold higher than
that attracted by the edge columns.

The effective length of the columns is signicantly shortened

under the deection restraints imposed by inlls, which lead to
severe and brittle damage of those columns due to shear-controlling action rather than the exural-controlling action targeted by
the original design. This response is commonly referred to as the
short column or captive column effect. Shear damage is highly
undesirable for conventionally designed and detailed RC columns,
particularly under cyclic loadings, because it would result in serious strength and stiffness degradation of the columns, inducing
acute pinching phenomena in the hysteresis loops.
Moreover, unlike the frame with fully integrated inll walls, the
2/3-storey-height inll walls do not support the building against
the vertical burdening weight, which is solely supported by the

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

137

Fig. 14. Damage patterns of inlled frames under the Chi-Chi earthquake with (a) crack strain at peak response in frame; (b) plastic strain at peak response in reinforcement;
(c) ultimate compressive damage in frame.

captive columns of the structure. After signicant shear damage

accumulates in the central short column in the rst storey, the
degraded axial force capacity is no longer sufcient to withstand
the axial force imposed on the column. Consequently, severe concrete crushing followed by longitudinal bar buckling is apparent in
the rightmost plot of Fig. 14. The structure also loses its lateral stability, characterised by a nearly at curve in the hysteresis loops
shown in Fig. 13.
5.3. Effect of openings
Comparing with the frames with full-height solid inlls, the
introduction of openings in the inll panels signicantly reduces

the incurred bracing action against the bounding frame. The maximum base shears experienced by the inlled frames with window
openings are reduced to approximately two-thirds to as much as
one-half of that experienced by the fully inlled frames; surprisingly, the damage to the bounding RC frame and the inll panel
is the most severe amongst the six structures, as shown in
Figs. 14 and 15. The structure is able to attain higher lateral stability than the frame with captive columns, as shown in Fig. 13.
Although the incurred seismic forces are quite signicantly
reduced by the introduction of openings, very serious damage that
is much more severe than that inicted upon the fully inlled
frames subjected to higher seismic forces is inicted on the bounding RC frame under seismic loading. This apparently abnormal

138

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

Fig. 15. Force transfer mechanisms: plots of the principal compressive stress distribution of inlled frames (a) before major cracking and (b) after major cracking.

behaviour is due to the partial collapse of the holed inll panels, as

shown in Fig. 14. Stored strain energy in the collapsing inll components is suddenly released, resulting in severe, localised damage
to the inll. In other words, following the toppling of inlls, most of
the seismic vibration energy is absorbed by the bounding frames
rather than the inlls in inlled frame structures with window
openings, in contrast to the fully inlled frames, in which most

of the vibration energy is apparently absorbed by the non-toppled

inlls.
Nevertheless, the inlled frames with window openings can
attain quite reasonable post-damage lateral stability, although
the structures suffer more extensive damage than other inlled
frames. Again, the vertical and lateral bracing actions provided
by the full-height inll panels contribute this stability to the

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

structure. This result demonstrates the importance of the continuity of the inll distribution in a bounding frame to the global stability of the structure. A discontinuous inll distribution, such as
the inlled frames with captive columns, tends to have lower
post-damage lateral stability because any force-ow discontinuity
can prohibit the development of other force transfer mechanisms,
decreasing the potential for force redistribution. Therefore, even
with such severe damage being inicted on the structure, the continuous non-toppled inlls in the inlled frames with window
openings are still able to transmit the lateral and vertical forces
through multiple strut bracing mechanisms, as shown in Fig. 15.
By contrast, while other frames behave elastically under the
1979 El Centro earthquake, the inlled frames with openings exhibit a ratcheting phenomenon in which the irrecoverable inelastic
drift develops progressively with each loading cyclic, as shown in
Figs. 12 and 13, quite similar to the fatigue effect. This phenomenon occurs because the cracking strength of the inll panels
with openings is lower than that of solid inlls due to stress concentration in the corners, as shown in Fig. 15. As a result, sliding
cracks are initiated at those corners and progressively propagate
along the bed joints under cyclic loadings, and the inelastically
deformed inll panels restrain the undamaged bounding frame
from deecting back to the original conguration, resulting in
the development of residual drift, even if the structure is unloaded.
However, under other ground excitations, the ratcheting phenomenon for the inlled structures with window-openings is not
obvious because the development of the residual drift incurred
by sliding of the inlls along crack planes initiated from the window corners is limited. This limitation occurs because the continuous inlls beneath the window-openings cannot slide as freely as
those above and restrain the deformation of the adjacent columns.
Further increases in residual drift require inelastic deformation of
the RC frame structures, which often do not exhibit a strong ratcheting phenomenon. This halting of the ratcheting effect is commonly referred to as the shakedown effect.
Similar to window openings, door openings weaken the bracing
action of inlls to the bounding frame. The incurred base shear is
reduced to approximately 80%, as much as 50% of that experienced
by the fully inlled frames. As shown in Fig. 14, the structures exhibit rather asymmetrical damage patterns; the rightmost columns
suffer more severe damage than the leftmost columns, while opposite damage distributions are evident in the beams, which more
damage is inicted in the left beams. Although the inlled frames
with door openings exhibit lower initial stiffness and thus experience less seismic forces than the structures with window openings
under the excitation of the El Centro earthquake, the post-damage
stiffness and base shear of the former structures become considerably larger than that of the later structures under the other
three strong earthquake excitations. This increase can be understood by observing the initial and post-damaged force transfer
mechanisms, as shown in Fig. 15.
The door structures have weaker initial bracing action than
the window structures, but after the inlls suffer severe damage
with the formation of major cracks or even partially collapse, the
post-damaged lateral bracing mechanisms, as characterised by
the inclined compressive struts, of the door structures are much
stronger and more robust than the window structures. As a
result, the incurred base shears of the door structures are larger
in the post-damage states and can reach magnitudes as high as 80%
of that experienced by the fully inlled frames. The structural damage of the frames with door openings is less than that incurred on
the frames with window openings due to reduced disintegration of
inll components.
However, the damage is apparently more severe than that to
the fully inlled frames, or more precisely speaking, the damage
distribution is considerably asymmetric, as indicated previously.

139

This increase in severity occurs because the initially symmetric lateral force transfer mechanisms are destroyed by the ratcheting
effect. In contrast to the window structures, the inlls founded
on the oors in the door structures are not horizontally continuous due to the introduction of the door openings, which results in
unrestrained sliding of the inll panels, as shown in Fig. 15. Sliding
of the inll panels leads to strong variation of the lateral force
transfer mechanisms, as well as the resulting drifts and hysteresis
loops, of the structures under loading acting in two opposite
directions.
Although the lateral stability remains quite satisfactory for the
prototype structure with door openings adopted in this study, if
the structures were taller, the stability would decrease because a
signicant PD effect would be triggered under large vertical loading and lateral drift.
5.4. Summary
Inclusion of inlls would signicantly distort general convexshaped (exural-controlled) hysteresis loops of ductile RC frames
without inlls to concave-shaped (shear-controlled) or, more
specically, slip-lock hysteretic loops. As shown in Fig. 13, the hysteretic loops of inlled frames simulated by the proposed modelling also replicate this slip-lock hysteresis phenomenon under
dynamic excitations. This slip-lock phenomenon is attributed to
the opening and closure of major cracks in the inlls under
reversed cyclic loading. The slip stage upon load reversal denotes
opening of major cracks in inlls, and thus the bracing action cannot be effectively developed in the inll panels. However, immediately after complete closure of the major cracks, i.e., the lock stage,
lateral forces can be continuously transmitted through the inll
panels, allowing the inlled structure to gain stiffness and strength
again upon further deformation.
In addition to the signature hysteresis phenomenon, another
commonly observed phenomenon is localised damage of inlled
structures, even with a uniform and regular distribution of inlls
[16,32]. Two major effects of inlls incur this localised damage.
First, signicant shear forces are transmitted through the struts
formed in the inlls to the bounding frame members in small contact lengths of approximately 0.10.35 clear lengths of braced columns [33]. The locally stressed regions suffer signicant shear
damage. Second, brittle cracking of inlls often occurs rst at lower
storeys, where larger seismic actions are usually induced to the
structural members. Moreover, the frame members must brace
the strain energy released from the cracking inlls. Consequently,
seismic damage in multi-storey inlled frames, regardless of regular or irregular inll conguration, is often more severe in lower
storeys than upper storeys. Irregular inll congurations and
arrangements of the panels in frame could exacerbate the damage
localisation due to discontinuity in stiffness and strength distributions. This damage localisation phenomenon of inlled frames
is also reected in the simulations, as shown in Fig. 14.
The numerical simulations of inlled frames with openings
revealed an interesting phenomenon in which signicant ratcheting structural deformation occurred under cyclic loading, as shown
in Figs. 12 and 13. Analogous to the fatigue effect, this ratcheting
phenomenon is caused by cracking and unrestrained sliding of
inlls with openings, which in turn destroys the symmetry of lateral-load transfer mechanisms and hysteresis loops. This phenomenon has seldom been reported in the literature because
most of the quasi-static reverse cyclic tests are displacement controlled with symmetric displacement cycles. Nevertheless, signicant asymmetric lateral-load transfer mechanisms of inlled
frames with openings have been reported [34,35].
In summary, the numerical simulations with the proposed
modelling technique are capable of replicating the generally

140

Y.P. Yuen, J.S. Kuang / Engineering Structures 91 (2015) 125140

observed nonlinear dynamic behaviour of inlled frame structures

under cyclic loading. In addition to general failure modes, the distribution of internal forces or force ows in the inlled structures
and the relationship between the incurred damage and lateral-load
transfer mechanisms can be more closely investigated based on the
numerical results.

[2]

[3]

[4]

6. Conclusions
[5]

The seismic behaviour of ve typical types of inlled RC frame

structures, namely, a fully inlled frame, an inlled frame with
2/3-storey-height inlls, an inlled frame with a soft rst storey,
an inlled frame with window openings and an inlled frame with
door openings under seismic excitations, were investigated indepth. The results of the analysis support the following
conclusions.

[6]
[7]
[8]
[9]
[10]

(1) Under seismic excitations, as long as out-of-plane collapse of

inlls does not occur, full height inll panels in full inlls,
window and door structures can enhance the overall
stability and energy dissipation of frame structures.
(2) Central short columns in 2/3-storey-height inlled frames
experience more severe damage due to the inll effect compared to the edge short columns. This difference occurs
because the central columns are restrained on both sides
by inlls, while the edge columns are only restrained on
one side, thus leading to approximately 1.72.6-fold higher
lateral seismic forces.
(3) Irregular arrangements of inll walls incur serious stress and
damage localisation. Abrupt changes in the vertical stiffness
of the inlled frame with a soft rst storey lead to concentrated damage of the unbraced soft-storey columns, reducing the vertical load carrying capacity. As reected by the
modelling, the axial-load carrying capacity of the columns
can be signicantly degraded after shear and/or exural
damage.
(4) Inlled frames with window or door openings exhibit shakedown or ratcheting phenomena, respectively, in hysteresis
behaviour due to the propagation of sliding cracks initiated
from the stress concentration regions around the corners
of the openings. Ratcheting deformation and, consequently,
shakedown, is halted by the kinematic restraint imposed
by continuous inlls beneath the window openings after full
slippage of the sliding cracks. However, unrestrained ratcheting deformation in the inlled frames with door openings
incurs serious asymmetric damage patterns and lateral force
transfer mechanisms.
(5) Columns of inlled frame structures suffer much greater
damage than the adjacent connecting beam members.
Thus, the capacity design concept of strong columnweak
beam may not always be applicable to inlled frames due
to the effect of inlls on the bare frame.

[11]
[12]
[13]

[14]
[15]
[16]
[17]
[18]

[19]
[20]
[21]

[22]

[23]
[24]
[25]

[26]

[27]
[28]

[29]

[30]

Acknowledgments
[31]

The support of Hong Kong Research Grant Council under Grant

No. 614011 and Scientic and Technological Research Council of
_
Turkey (TBITAK)
under project number 214M236 are gratefully
acknowledged.
References
[1] Sezen H, Whittaker AS, Elwood KJ, Mosalam KM. Performance of reinforced
concrete buildings during the August 17, 1999 Kocaeli, Turkey earthquake, and

[32]

[33]
[34]
[35]

seismic design and construction practise in Turkey. Eng Struct

2003;25:10314.
Earthquake Engineering Field Investigation Team (EEFIT). The 2007 August 15
magnitude 7.9 earthquake near the coast of Central Peru. UK: The Institution of
Structural Engineers; 2007.
Earthquake Engineering Field Investigation Team (EEFIT). The Wenchuan,
China earthquake of 12 May 2008. UK: The Institution of Structural Engineers;
2008.
Manfredi M, Prota A, Verderame GM, De Luca F, Ricci P. 2012 Emilia
earthquake, Italy: reinforced concrete buildings responses. Bull Earthq Eng
2014;12:227598.
De Luca F, Verderame GM, Gmez-Martnez F, Prez-Garca A. The structural
role played by masonry inlls on RC building performances after the 2011
Lorca, Spain, earthquake. Bull Earthq Eng 2014;12:19992026.
Mehrabi AB, Shing PB. Behaviour and analysis of masonry-inlled frames. Prog
Struct Eng Mater 2002;4(3):32031.
Baloevic G, Radnic J, Harapin A. Numerical dynamic tests of masonry-inlled
RC frames. Eng Struct 2013;50:4355.
Su L, Shi J. Displacement-based earthquake loss assessment methodology for
RC frames inlled with masonry panels. Eng Struct 2013;48:43041.
Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry
buildings. New York: Wiley; 1992.
Saneinejad A, Hobbs B. Inelastic design of inlled frames. J Struct Eng ASCE
1995;121(4):63450.
Cavaleri L, Fossetti M, Papia M. Inlled frames: developments in the evaluation
of cyclic behaviour under lateral loads. Struct Eng Mech 2005;21:46994.
Puglisi M, Uzcategui M, Flrez-Lpez J. Modeling of masonry of inlled frames,
Part I. The plastic concentrator. Eng Struct 2009;31:1138.
Asteris PG, Cotsovos DM, Chrysostomou CZ, Mohebkhah A, Al-Chaar GK.
Mathematical micromodeling of inlled frames: State of the art. Eng Struct
2013;56:190521.
American Society of Civil Engineers. Seismic Rehabilitation of existing
buildings. ASCE/SEI 41. Reston, Virginia; 2006.
Chen Z, Liu Y. Numerical study of in-plane behaviour and strength of concrete
masonry inlls with openings. Eng Struct 2015;82:22635.
Kuang JS, Yuen YP. Simulations of masonry-inlled RC frames failure. P I Civil
Eng Eng Comput Mech 2013;166(4):17993.
SAC. Code for seismic design of buildings. GB 50011-2001. Beijing: China
Architecture & Building Press; 2001.
Negro P, Verzeletti G. Effect of inlls on the global behaviour of R/C frames:
energy considerations from pseudodynamic tests. Earthq Eng Struct D
1996;25:75373.
ABAQUS, Inc. ABAQUS theory manual; 2010.
Lubliner J, Oliver J, Oller S, Oate E. A plastic-damage model for concrete. Int J
Solids Struct 1989;25(3):229326.
Sakai J, Kawashima K. Modication of the Giuffre, Menegotto and Pinto Model
for unloading and reloading paths with small strain variations. J Struct Mech
Earthq Eng 2003;738(I-64):15969 (in Japanese).
Benzeggagh ML, Kenanem M. Measurement of mixed-mode delamination
fracture toughness of unidirectional glass/epoxy composites with mixed-mode
bending apparatus. Compos Sci Technol 1996;56(4):43949.
Rots JG. Structural masonry: an experimental/numerical basis for practical
design rules. Rotterdam: A.A. Balkema; 1997.
Sloan SW, Abbo AJ, Sheng D. Rened explicit integration of elastoplastic
models with automatic error control. Eng Computat 2001;18(1/2):12154.
Atkinson RH, Kingsley GR., Saeb S, Amadei B, Sture, S. A laboratory and in situ
study of the shear strength of masonry bed joints. In: Proceedings of the Eighth
International Brick and Block Masonry Conference, 1921 September 1988,
Trinity College, Dublin, Republic of Ireland; 1988.
Vermltfoort ATh, Van der Pluijm R. Deformation Tensile and compression
experiments on bricks, mortar and masonry. Report no. B-91-0561. TNOBOUW/TU Eindhoven, Building and Construction Research; 1991.
Raijmakers TMJ, Vermltfoort A Th. Deformation Controlled Meso Shear Tests
on Masonry Piers. Report no. B-92-1156, TNO-BOUW/TU Eindhoven; 1992.
Angel R, Abrams D, Shapiro D, Uzaeski J, Webster M. Behaviour of reinforced
concrete frames with masonry inlls. UILU-ENG-94-2005, Department of Civil
Engineering, University of Illinois at Urbana-Champaign; 1994.
Hashemi A, Mosalam KM, Seismic evaluation of reinforced concrete buildings
including effects of masonry inll walls. PEER Technical Report 2007/100.
Berkeley: University of California; 2007.
Yuen YP, Kuang JS. Nonlinear responses and failure of inlled RC frame
structures under biaxial seismic excitation. In: 15th world conference on
earthquake engineering. Lisbon, Portugal; 2012.
Ricci P, De Risi MT, Verderame GM, Manfredi G. Inuence of inll distribution
and design typology on seismic performance of low- and mid-rise RC
buildings. Bull Earthq Eng 2013;11:1585616.
Zarnic R, Gostic S, Crewe AJ, Taylor CA. Shaking table tests of 1:4 reduced-scale
models of masonry inlled reinforced concrete frame buildings. Earthq Eng
Struct D 2001;30:81934.
Crisafulli FJ, Carr AJ. Proposed macro-model for the analysis of inlled frame
structures. Bull New Zealand Soc Earthq 2007;40(2):6977.
Kakaletsis DJ, Karayannis CG. Inuence of masonry strength and openings on
inlled R/C frames under cycling loading. J Earthq Eng 2008;12(2):197221.
Surendran S, Kaushik HB. Masonry inll RC frames with openings review of inplane lateral. Open Construct Build Tech J 2012;6:12654 (Suppl 1-M9).