Structures 10 (2017) 1–12

Contents lists available at ScienceDirect

Structures

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

Finite Element Study of the Effect of Interfacial Gaps on the in-Plane

Behaviour of Masonry Infills Bounded by Steel Frames
Xi Chen, Yi Liu ⁎
Dept. of Civil and Resource Engineering, Dalhousie Univ., Halifax, NS B3J 1Z1, Canada

a r t i c l e i n f o a b s t r a c t

Article history: A finite element study was conducted to investigate the in-plane behaviour and strength of concrete masonry
Received 23 March 2016 infills bounded by steel frames with the focus on the effect of interfacial gaps. Various gap locations and sizes
Received in revised form 28 September 2016 were studied. The results showed that the top beam-infill gap was more detrimental to the stiffness and strength
Accepted 10 November 2016
of infill than the column-infill gaps. Gap(s) had less effect on infills bounded by weak frames than strong frames.
Available online 11 November 2016
Different arrangements of gap locations at the column-infill interface affected the behaviour of infilled frames
Keywords:
only when the friction coefficient exceeded 1.0. The provisions contained in the American masonry standard
Masonry infills MSJC 2013 for design of infilled frames with gaps were examined. It was found that in most cases, the reduction
Steel frames factor for the gap effect specified in the MSJC 2013 produced conservative estimates for both the stiffness and
Interfacial gaps strength. Based on the finite element results, a set of lower bound expressions for stiffness and strength evalua-
Stiffness and strength tion over a wide range of gap size and locations were proposed.
Reduction factor © 2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
Finite element study

1. Introduction
Masonry walls built inside concrete or steel frames are commonly
known as masonry infills. When the infilled frame is subjected to in-
plane lateral loading, the infill with inherently large in-plane stiffness,
will attract large forces to the frame region and change the lateral load
distribution of the structure. The understanding of interaction between
the masonry infill and the surrounding frame is thus crucial in determi-
nation of infill contribution to the stiffness and strength of the infilled
system. Considerable amount of experimental research has been con-
ducted to investigate the interaction between the infill and its bounding
frame in an effort to quantify the infill contribution to the system
stiffness and strength [1–14]. These studies have contributed to the de-
velopment of the so-called “diagonal strut” method. This method re-
places the infill with an equivalent diagonal strut connecting two
loaded corners to resist the lateral loading. Once the strut width is
known, the stiffness of the system can be determined through a frame
analysis while the strength of the infill can also be formulated as a func-
tion of the width. Due to its simplicity, the “diagonal strut” concept has
been adopted in various international standards [15–17] for design of
infills, albeit with different forms of strut width equation. It should be
pointed out though, that the diagonal strut method contained in these
standards is only applicable to the so-called participating infills where
the interfacial gaps between the infill and the bounding frame are
zero or within a specified limit. For example, both the Canadian
⁎ Corresponding author.
E-mail address: yi.liu@dal.ca (Y. Liu).
masonry design standard S304-14 [15] and Eurocode 8 [17] require
that no interfacial gaps be existent for infills to be considered as partic-
ipating infills while no guidelines are provided for the treatment of in-
terfacial gaps in either document. The American masonry standard
MSJC 2013 [16] states that infills can be considered as participating pro-
vided that the top beam-infill gap is b 9.5 mm (3/8″), but in such a case, a
factor of 0.5 must be applied to the stiffness and strength of the infill.
However, there was no sufficient background information provided on
the given gap size limit or the reduction factor. In practice, initial gaps
between the infill and the bounding frame are common occurrence
due to shrinkage and settlement of the infill or defects in masonry work-
manship. These gaps could be located at either the beam-infill interface,
or column-infill interface, or both. Since the total elimination of gaps is
not realistic, it is thus important to evaluate the effect of gap on the infill
contribution to the system behaviour and strength.
Despite a large amount of research available to address the infill be-
haviour in general, studies focusing on the effect of gaps were limited.
Within the few available studies on gaps, the detrimental effect of
gaps on both the strength and stiffness of the infilled system was report-
ed [2,5,8,18]. However, the range of reduction observed in these studies
was scattered and due to limited data points, no correlation between
the gap and the infill stiffness and strength was defined. For example,
the experiment program of Yong [2], and Dawe and Seah [5] showed
that a 20 mm gap at the top beam-infill interface reduced both the
crack load and ultimate load of the infilled steel frame by 50 and 60% re-
spectively. Flanagan [6] reported that a 25.4 mm gap at the column-infill
interface of infilled steel frame resulted in nearly no reduction in the
ultimate capacity but a nonsymmetrical cracking pattern where more

http://dx.doi.org/10.1016/j.istruc.2016.11.001
2352-0124/© 2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
2 X. Chen, Y. Liu / Structures 10 (2017) 1–12
cracks developed in the lower portion of the panel. The study conducted
by Kadir [19] showed that a top beam-infill gap of 1.6 mm did not signif-
icantly affect the frame ultimate load but it reduced its cracking load. In
the experimental study conducted by Ng'andu [20] on infilled steel
frames, a 12 mm gap at the top beam-infill interface was found to re-
duce the system stiffness by about 20% but had no significant effect on
the cracking load or ultimate load. Nazief [21] conducted a finite ele-
ment study on infilled steel frames where the effect of beam-infill gap
and full separation gap (around the three infill-frame interfaces) was
studied. Results showed that a top beam-infill gap or a full separation
gap up to 5 mm did not have a significant impact on the infill ultimate
strength while the lateral stiffness was reduced by around 30%. A top
beam-infill gap of 10 to 15 mm resulted in a reduction in the ultimate
strength of 24% while a full separation gap of the same size resulted in
a reduction of 50%.
In the above mentioned studies, the gap scenarios were often fo-
cused on the top beam-infill [2,5,19,20] and with limited range of
sizes. The obtained information is not sufficient to define the effect of
gap location or size on the infill stiffness and strength in relation to
the infilled system. In light of this, this study was then motivated to fur-
ther investigate the effect of gaps on the behaviour of infills bounded by
steel frames. The numerical modeling technique was used where a finite
element model of the infilled frame subjected to in-plane lateral loading
was developed. The model was validated against experimental results
reported in the literature for infilled frames with or without gaps. The
objective of the study was to determine correlations between the gap
size and location and the lateral stiffness and strength of the infilled
frame and to assess the validity of design provisions on gapped infilled
frames contained in the American design standard MSJC 2013. To this
end, various gap locations, sizes and arrangement scenarios were con-
sidered as main parameters. The effects of bounding frame stiffness
and friction coefficient were also included in the study.
2. Finite element model
A simplified micro-modeling technique encoded in the software
ANSYS was used where the infill was modeled using homogeneous con-
tinuum elements and the mortar effect was modeled using a surface-
based cohesive contact model. The masonry unit was expanded in
each directions by the half mortar thickness and the expanded mortar
units interact with each other through the interface. The masonry
units were meshed into 2 × 2 plane-stress elements and placed in the
running bond. Details of the model is available in a previous paper pub-
lished by the authors [22]. For the interest of completeness and easy ref-
erence, the key components of the modeling are summarized in the
following.
2.1. Constitutive models for masonry and steel
The constitutive model for masonry in compression was based on
the experimentally calibrated model proposed by Priestley and Elder
[23] and it is expressed in Eqs. (1a)–(1c) where f′m is the compressive
strength of masonry. An elastic perfect plastic material model was
used for frame members where the elastic modulus and the yield
strength of steel were taken as 200 GPa and 350 MPa, respectively.
0   ε 2 
fm 2ε
ε ≤0:0015; σ¼ − ð1aÞ
0:9375 0:002 0:002
0 to compression failure while the factor β controls the coupling between
0:0015≤ε ≤0:0025; σ ¼ f m ð1−zm ðε−0:0015ÞÞ ð1bÞ
0:5
where zm ¼ 0 ð1cÞ
3 þ 0:29f m
0 −0:002
145 f m −1000
2.2. Interface model
The interface was modeled using surface-based cohesive contact
pairs to capture the cracking and sliding failure of the mortar joints.
Each contact pair consists of one zero-thickness surface-to-surface co-
hesive zone interface element, CONTA171, and one zero-thickness tar-
get element TARGE169 where CONTA171 was used to detect contact
or separation between “target” surfaces (TARGE169) and a deformable
surface defined by CONTA171 element. This interface element was
also used for the contact between the infill and the frame members. A
schematic view of the modeled infilled frame is shown in Fig. 1(a).
The geometry and configuration of inter-blocks contact pairs are
shown in Fig. 1(b). Cracking and sliding failure of mortar joints were
monitored using the linear elastic traction-separation laws as shown
in Fig. 2. Represented by the bilinear cohesive zone material model of
the CONTA171 element, separation initiates when the maximum nor-
mal or shear stress in a contact pair reaches the corresponding strength.
The shear stress and the relative slip behaviour is plotted in
Fig. 2(a) which shows separation begins at point A where the shear
strength (τt) is achieved and is completed at point C when the shear
stress reaches zero. The softening stage provides mortar joints with a re-
sidual shear strength to account for the frictional resistance of the joint
after shear crack occurred. For tension separation, a similar bilinear con-
stitutive relationship but with a cut-off was adopted. As Fig. 2(b) shows,
once the tension stress in the contact pair reaches the tensile strength
(ft) of mortar joint, the contact pair is considered separated to represent
the tensile cracking of mortar joint. The normal and shear stiffnesses, kn
and ks, were determined based on the following expressions suggested
by Lourenҫo [24]:
Eu Emortar
kn ¼ ð2Þ
t m ðEu −Emortar Þ
Gu Gmortar
ks ¼ ð3Þ
t m ðGu −Gmortar Þ
where Eu and Emortar, and Gu and Gmortar are the Young's moduli and
shear moduli of masonry and mortar respectively; and tm is the thick-
ness of mortar joints. In lieu of experimental data, Gu and Gmortar can
be taken as 0.4Eu and 0.4Emortar respectively.
2.3. Failure criteria
A Hill type yield criterion (f1) proposed by Lourenço and Rots [25]
was adopted to monitor the compression failure of masonry infill. This
failure surface was also shown by Dawe et al. [12] to produce results
in reasonably good agreement with experimental data.
σ 2x σ xσ y σ 2y τ 2xy
f1 ¼ 2
þβ þ 2 þγ −1 ≤0 ð4Þ
f mx f mx f my f f mx f my
my
where subscripts x, and y indicate the directions parallel and perpendic-
ular to the bed joint. Hence, σx and σy, and τxy are the nodal normal
stresses in the x- and y-direction, and the shear stress, respectively;
fmx and fmy are the uniaxial compressive strength of infills in the x-
and y-direction, respectively. As the masonry infill was assumed
orthotropic, the compressive strength of masonry in head joint direc-
tion (x-direction) was taken as 0.7 times that in bed joint direction (y-
direction) [16]. The factor γ accounts for the shear stress contribution
normal stresses in x- and y-directions. In lieu of experimental data, as
suggested by Dawe et al. [12] for masonry commonly used in practice,
the value of β and γ can be taken as −1.0 and 5.5 respectively.
The damage initiation criterion is defined using a Mohr-Coulomb
friction yield surface combined with a tension cut-off and elliptical com-
pression cap as shown in Fig. 3 where ft is the tensile strength of mortar
 X. Chen, Y. Liu / Structures 10 (2017) 1–12
a)
Beam element
Applied Load
Augment Spring
Plane stress element
b)
Fig. 1. a) Finite element model of the infilled frame, b) configuration of a joint.
joint and φ is the friction angle. Note that the compression cap coincides
with the failure criteria defined in Eq. (2). This failure surface has been
shown by Lourenço and Rots [25] and Al-Chaar and Mehrabi [26] to pro-
duce reasonably accurate estimates when compared with experimental
results.
2.4. Boundary conditions and loading procedure
The frame beam-to-column connection was modeled as rigid and
the columns were assumed to be fix-supported at ends. The bottom
beam of the bounding frame is vertically supported along the length.
A monotonically increased load with a defined load increment was ap-
plied at the frame beam level. During each iteration of analysis, tensile
and shear failure were checked at the interface and the compression
failure was checked at each masonry unit based on the failure surface
defined above. If failure was detected, the tangent stiffness of the ele-
ment associated with the failure at that load increment was modified
according to the material constitutive model described above; and the
analysis was rerun with the updated system stiffness matrix until no
new failure was detected. The entire process was repeated for the sub-
sequent increment of load. A full Newton-Raphson iterative procedure
and sparse direct equation solver specified in ANSYS were adopted in
the solution control. To obtain the falling branch in the load vs. lateral
displacement response, an augmented spring was implemented as
shown in Fig. 1(a). The spring was implemented to regulate the load
when the infill suddenly cracked or crushed so that the load vs. lateral
3
Target element
Node K Node L
Node J Node I
Interface element
Bed Joint and Head Joint
Masonry Element
Contact element CONTA171
Target element TARGE169
Contact Normal
displacement curve can reflect this load drop. The stiffness of this spring
was set to be equal to the initial stiffness of the infilled frame.
2.5. Verification of the finite model
The finite element model was validated using experimental results
obtained from the available literature [1–4,18]. Nineteen specimens
were considered, of which twelve was without gaps and the remaining
seven was with gaps. Among seven gapped specimens, five specimens
had a top beam-infill gap (WB4, WB5, WD11, R3a and R3b) and two
specimens had both column-infill and beam-infill gap (R4a and R4b).
Detailed descriptions of test set-up and specimens are available in the
aforementioned references and the key characteristics are summarized
in Table 1. Table 2 shows the comparison results where kexp and Pexp are
the crack stiffness and ultimate load obtained from experimental results
whereas kFE and PFE are the corresponding results obtained from finite
element (FE) analysis. The table shows that for infills with or without
gap, the average test-to-FE ratios of ultimate load and stiffness are all
close to unity with the corresponding overall coefficient of variations
(COV) around 9% and 17% respectively. This shows that the finite ele-
ment model is capable of providing stiffness and strength values
which agree well with the test results. Fig. 4 compares the load vs. later-
al displacement curves for one infill without gap (specimen WB2) and
two gapped infills (specimen R3a and R4a). The crack pattern and fail-
ure mode comparison of these specimens are presented in Fig. 5. It is
noted that since the masonry unit was discretized into 2 × 2 elements,
4 X. Chen, Y. Liu / Structures 10 (2017) 1–12
a)
A C0). The infill was taken to be 2800 mm high by 2800 mm long. The infill
τt
Shear Stress
O umns and W360 × 79 beam; and 3) the weak frame (labeled as WF)
0 utA Slip
b)
ft A, cut-off point
Tensile Stress
O
0 unA
Gap
Fig. 2. Constitutive relationship for cohesive material, a) shear stress vs slip, b) normal
stress vs gap.
some cracking shown in the FE model was then through the masonry
units. These cracks are identified in circles for specimen R3a as an exam-
ple. Although these cracking was not predicted at the exact locations as
those in the experiments, the overall cracking pattern was accurately
predicted. Judging from comparisons presented in Table 2 and Figs. 4
and 5, it is believed that the finite element model can simulate the be-
haviour and failure mode reasonably well where the degradation of
stiffness due to cracking in the infill and development of cracking and
crushing as loading progressed is accurately captured.
3. Parametric study
The model specimens and associated parameters used in this para-
metric study are summarized in Table 3. Parameters considered includ-
ed the gap size and location; bounding frame stiffness; and friction
coefficient at the frame-infill interface. The gap size and location were
two main parameters where the total gap size ranging from 0 to
60 mm were considered in three location scenarios. These scenarios
were 1) a gap at the top beam-infill interface (labeled as BB); 2) gaps
at two column-infill interfaces (labeled as BC); and 3) a full separation
gap with equal width at three interfaces (labeled as AA). The model of
Coulomb friction mode
Cap mode
φ develop in the frame. As illustrated in Fig. 6, the following terms were
Fig. 3. Mohr-Coulomb friction yield surface.
the infill with no gaps was considered as the control model (labeled as
compressive strength was assumed to be 15 MPa and the tensile
strength was taken as the 1/10th of that value. The boundary conditions,
constitutive model and the analysis and loading procedures were as
previously described in Section 2.
Three types of bounding frames, representing normal, strong and
weak frames, were considered. They were 1) the normal frame (labeled
as NF) which consisted of W250 × 58 columns and W200 × 46 beam;
C 2) the strong frame (labeled as SF) which consisted of W310 × 129 col-
utC which consisted of W150 × 24 columns and W100 × 19 beam. In all
cases, the beams were oriented with their webs in the plane of the infill
while the columns were oriented with their webs perpendicular to the
plane of infill. Members of the strong and weak frame have an EI/L ap-
proximately 5 and 0.1 times respectively those of the normal frame.
The analysis for all gap location and sizes was performed for all three
types of frames.
The gap width and its arrangement are indicated in the specimen
label. For instance, specimen BB-10 stands for a model with a 10 mm
gap at the top beam-infill interface (total width = 10 mm); specimen
BC-15-30 stands for a model with a 15 mm gap at each of the column-
C infill interface with a total width of 30 mm; whereas specimen AA-10-
30 indicates a model with a separation gap of 10 mm between the infill
and three members of the frame with a total width of 30 mm. In the case
of column-infill gap scenario for specimen BC-10-20, two different ar-
rangements of gaps were also considered where specimens L0R20 and
L20R0 indicate that the total 20 mm gap was located between either
the right column or the left column and the infill whereas the other
column-infill interface was assumed in tight contact.
The default friction coefficient (μ) at the frame-infill interface was
chosen as 0.4 since it was recommended for concrete on steel surfaces
by King and Pandey [27] and it provided a reasonably good agreement
with experimental results. For specimens BC-10-20, L20R0 and L0R20,
two additional μ values of 0.7 and 1.0 were considered. The value of
0.7 is suggested by CSA S304-14 for sliding shear between masonry
and bare steel while the value of 1.0 was selected to represent sliding
on very rough surfaces.
A finite element analysis conducted on the bare frame showed that
for the normal frame (NF), yielding of columns occurred when the later-
al displacement reached 28 mm. For the weak (WF) and the strong
frame (SF), the yielding commenced at a lateral displacement of
62 mm and 23 mm, respectively. When the column-infill gap size ex-
ceeds the elastic limit for lateral displacement, yielding is expected to
initiate in the column base and beam-column joints before the gap is
closed. As can be seen, some gap sizes studied place the frame in the
yielding region. These cases are identified in Table 3.
4. Numerical study results
4.1. General discussion
The typical load vs. lateral displacement curves are presented in
Fig. 6 using specimen AA-10-30 bounded by the normal frame as an ex-
ample. It can be seen that before the gap was closed, the gapped infill
system behaved similarly to a bare frame and a sudden increase in stiff-
ness was observed upon the closure of the gap. The specimen reached
Tension failure at a larger displacement and thus higher member forces would
cut-off
used for evaluation of the infill behaviour. The initial stiffness, Ki, was
defined as the initial slope of the response curve of the gapped infill
frame; the secant stiffness at cracking load, Kcr, was defined as the
slope of the straight line connecting the origin and the first cracking
ft σ load point; and the tangent crack stiffness Kt, was defined as the slope
of the straight line connecting the point of the sudden stiffness increase
and the cracking load point. For the infilled frame without gaps (C0), the
 X. Chen, Y. Liu / Structures 10 (2017) 1–12 5

Table 1
Details of specimens used in the validation of FE model.

H L f′m Em te Beam section Column section Gap scenario

(m) (m) (MPa) (MPa) (mm)

Infills with gaps

Yong [2] WB4 2.8 3.6 32.8 27,880a 66 W250 × 58 W250 × 58 20 mm top gap
WB5 2.8 3.6 31.0 26,350a 66
Richardson [4] WD11 2.8 3.6 24.6 12,400 66 W250 × 58 W250 × 58 20 mm top gap
Riddington [18] R3a 2.71 2.71 7.0 15,400 100 152 × 152 × 30 UC 152 × 152 × 30 UC 3 mm top gap
R4a 2.71 2.71 7.0 15,400 100 3 mm top gap + 1.5 mm gap on each side
R3b 2.48 2.48 7.0 15,400 100 406 × 140 × 39 UB 406 × 140 × 39 UB 3 mm top gap
R4b 2.48 2.48 7.0 15,400 100 3 mm top gap + 1.5 mm gap on each side

Infills without gaps

McBride [1] WA1 2.8 3.6 27.4 23,290 66 W250 × 58 W200 × 46
WA2 2.8 3.6 27.7 23,545 66
WA3 2.8 3.6 26.5 22,525 66
WA4 2.8 3.6 24.4 20,740 66
Yong [2] WB1 2.8 3.6 23.7 20,145a 66 W250 × 58 W200 × 46
WB2 2.8 3.6 33.3 28,305a 66
WB3 2.8 3.6 31.4 26,690a 66
Amos [3] WC1 2.8 3.6 31.7 26,945a 66 W250 × 58 W200 × 46
WC2 2.8 3.6 27.7 23,545a 66
WC7 2.8 3.6 33.4 28,390a 66
Riddington [18] R2a 2.71 2.71 7.0 15,400b 100 152 × 152 × 30 UCb 152 × 152 × 30 UC
R2b 2.48 2.48 7.0 15,400b 100 406 × 140 × 39 UBb 406 × 140 × 39 UB
a
Em was not reported in the paper, a value of 850f′m was assumed.
b
Universal steel columns and beams, British designation.

initial stiffness was denoted as Ks. Since this study focused on the behav-
iour and strength of the infill, the net lateral load Pnet (shown as the dot-
ted line) carried by the infill was used in the later discussion and it was
calculated as the ultimate load of the infilled frame (system load)
subtracted by the corresponding load for a bare frame at the same
displacement.
Fig. 7 compares the load vs. lateral displacement curves of specimens
with the same total gap size (AA-10-30, BB-30, and BC-15-30) bounded
by the normal frame. As shown in Fig. 7(a), the initial stiffnesses of
specimens AA-10-30 and BC-15-30 were similar and they were more
or less the same as that of the bare frame. This is expected since the
gap between the infill and the columns enabled some “free” deforma-
tion of the frame before the infill was engaged. In the case of specimen
BB-30, the initial stiffness was slightly higher than that of a bare frame.
In this case, although there was no bearing at the beam-infill interface,
the infill was engaged at the initial loading stage through bearing at
the two column-infill interfaces. Fig. 7(b) shows that before the gaps
were closed, only the infill of specimen BB-30 showed noticeable load

Table 2
Stiffness and strength comparison of the experimental and FE results.

kexp (kN/mm) Pexp (kN) kFE (kN/mm) PFE (kN) kexp P exp
kFE P FE

Infills with gaps

Yong [2] WB4 27 209 24 219 1.13 0.95
WB5 34 231 29 254 1.17 0.91
Richardson [4] WD11 14 167 13 175 1.08 0.95
Riddington [18] R3a 29 199 23 189 1.26 1.05
R3b 45 382 35 323 1.29 1.18
R4a 7 170 7 167 0.97 1.02
R4b 19 363 18 331 1.06 1.10
Avg 1.14 1.02
C.O.V. (%) 10 9

Infills without gaps

McBride [1] WA1 73 471 75 476 0.97 0.99
WA2 82 440 75 481 1.09 0.91
WA3 74 463 75 460 0.99 1.01
WA4 63 476 68 424 0.93 1.12
Yong [2] WB1 72 449 62 412 1.16 1.09
WB2 74 556 81 581 0.92 0.96
WB3 74 556 77 546 0.96 1.02
Amos [3] WC1 41 420 68 451 0.6 0.93
WC2 46 310 37 403 1.24 0.77
WC7 71 534 81 537 0.88 0.99
Riddington [18] R2a 30 209 36 226 0.83 0.92
R2b 60 411 67 423 0.90 0.97
Avg 0.96 0.97
C.O.V. (%) 17 9

Overall
Avg 1.02 0.99
C.O.V. (%) 17 9
6 X. Chen, Y. Liu / Structures 10 (2017) 1–12

a) 600 the resistance was mainly achieved through bearing at column-

infill interfaces. The high rigidity of beam-column connection of
500
strong frame yielded smaller rotation at a given lateral displacement
Load (kN)

400 and thus the contact area achieved between the top beam and the
300 infill was less than that in the normal or weak frame, resulting in
the majority of lateral load being transferred through the contact be-
200
tween columns-infill interfaces. The mode of failure of specimen BB-
100 10 was shown in Fig. 8(b) as an example. Load was transferred
0 through two alternative struts developed between the mid-height
0 5 10 15 20 25 30 of column and the two diagonal corners, which resulted in sliding
Lateral Displacement (mm) of the two portion of the infill with respect to each other along the
mortar joints. It should be pointed out that this failure mechanism
b) 250 was only observed in specimens with BB gap scenario and bounded
by strong frames.
200
4.3. Effect of gap size
150
Load (kN)

100 Fig. 9 shows the normalized stiffness (Kcr and Kt) and normalized
strength (Pnet) of infills plotted against the gap size for specimens with
50 the normal frame for all three gap scenarios. The stiffness and strength
values were normalized using corresponding values of control speci-
0
0 5 10 15 20 25 30 men C0 without gaps. Since specimen C0 behaved linearly to the first
Lateral Displacement (mm) cracking load, the Kcr and Kt for specimen C0 are identical and they are
essentially the initial stiffness Ks (Fig. 6). Fig. 9 shows that as gap size in-
c) 200 creased, both Kcr and Kt showed significant decreases and the reduction
in Kcr was more severe than in Kt. For example, for a given gap location
150 (BB for example), the normalized Kcr was reduced to 0.11 while Kt was
Load (kN)

about 0.56 with a beam-infill gap size of 30 mm (specimen BB-30).

100
50
0 impact on Kcr as the frame needs to deform a greater amount before the
0 5 10 15 20 25
Lateral Displacement (mm)
Experimental Finite element model
Fig. 4. Comparison of load vs. lateral displacement curves between FE and experiment
results: a) specimen WB2, b) specimen R3a (3 mm top gap) and c) specimen R4a
(3 mm top gap + 1.5 mm gap on each side).
carrying which further confirms the above observation of loading trans-
fer at the onset of loading. The general trend of all three specimens is
similar where an elastic linear behaviour was observed almost before
engagement of the infill indicating the frame deformed nearly elastical-
ly up to the closure of the gap(s) and after which point, the infill system
behaved similar to the control specimen C0. The sudden drops in the ris-
ing branch of the curve indicates the development of cracking and the
load increase immediately thereafter indicates the ability of the infill
to find the alternative loading paths for resisting additional load. After
the ultimate load, the system still showed marked ductility maintaining
about 80% of the ultimate load with increasing lateral deflection.
4.2. Failure mode
The finite element results showed that for all gapped specimens
bounded by the normal or weak frames, once the gap(s) was closed,
the diagonal strut developed in the infill, and failure was still pre-
dominated by crushing of masonry at loaded corners, which was
the same as the failure mode for the infill without gaps. One such fail-
ure mode is shown in Fig. 8(a). For specimens with gap sizes exceed-
ing the elastic lateral displacement of the frame, the frame began to
yield at column bases and beam-column joints before gap(s) was
closed. However, the FE model showed that the diagonal strut action
was still developed in the inelastic stage of the frame.
For specimens with BB gap scenario and bounded by strong
frames, however, the failure was initiated by sliding shear where
This is reasonable since Kcr is largely dependent on the lateral displace-
ment of the system before the infill engagement whereas Kt is a measure
of stiffness after the infill engagement. The increase in gap size has more
30 infill becomes engaged. The infill strength Pnet also showed significant
reduction as a result of an increase in the gap size where specimen
BB-30 had a Pnet of 54% of that of specimen C0. The general trend de-
scribed above also applies to gap scenarios AA and BC. The reduction
in infill strength is believed to be attributed to the reduction in contact
area developed between the infill and the frame. According to the diag-
onal strut concept, as the lateral loading increases, the flexural deforma-
tion of the frame causes the frame to separate from the infill and at
failure only two loaded corners remain in contact with the infill to pro-
vide the resistance. For a gapped specimen, as gap size increases, the
flexural deformation required to engage the infill is increasingly greater.
The larger deformation of frame members leads to less contact area be-
tween the infill and the frame, which ultimately reduces the effective
width of diagonal strut for the load transfer.
Referring to Fig. 9(a), while the yielding reduced Kcr significantly,
e.g. Kcr for specimen AA-20-60 was reduced below the stiffness of a
bare frame, the infill net load Pnet seems less affected by yielding of
the frame since there was no sudden change in the rate of reduction
as shown in Fig. 9(c). In practice, the gap size should be maintained
within the elastic limit of the frame deformation. In the case that
the gap size exceeds the elastic deformation of the frame, this
study suggests that the diagonal strut can still develop and a large
portion of the infill strength can still be achieved, provided that the
frame is allowed to develop its full plastic capacity.
4.4. Effect of gap location
It can also be observed in Fig. 9 that with the same total size of
gap, the gap scenario BB resulted in the greatest reduction in both
the stiffness and strength while the BC scenario showed the least re-
duction. It suggests that for a given total gap size, the beam-infill gap
has more detrimental effect on infill strength than the column-infill
gap. The load transferring mechanism for an infill without gaps is a
combination of direct bearing and friction between the infill and
 X. Chen, Y. Liu / Structures 10 (2017) 1–12
a)
b)
c)
Fig. 5. Comparison between crack patterns between FE and experiment results: a) specimen WB2, b) specimen R3a, and c) specimen R4a.
frame members. In the case of top beam-infill gap, the load transfer-
ring can only rely on the direct bearing between the infill and frame
columns; whereas in the case of column-infill gaps, the load is trans-
ferred initially through friction between the infill and the top beam
and after the gaps close, the bearing between the infill and the
frame members also partakes in loading transferring. Therefore, the
presence of the top gap is more disruptive in establishing an effective
loading path. Fig. 9(c) also shows that the reduction in stiffness or
strength for the case of full separation gap is not simply the addition
of the reduction due to the beam-infill gap (BB) and the column-infill
gap (BC). For example, specimen AA-5-15 scenario is equivalent to
the superposition of specimen BB-5 and specimen BC-5-10 scenarios.
The reduction in strength in specimen AA-5-15 (21%) is greater than
the summation of reduction in the latter two specimens (10% +
7% = 17%).
7
4.5. Effect of frame stiffness
The effect of gaps on infills bounded by different frames is presented
in Fig. 10 using gap scenario AA as an example. It can be seen that the
decreasing trend on Kcr, Kt and Pnet as a result of an increase in gap
size remained true for varying frame stiffness. However, as the stiffness
of the bounding frame increased, reductions in Kt and Pnet increased. In
other words, weak frames resulted in smaller reductions whereas
strong frames resulted in larger reductions in stiffness (Kt) and strength
(Pnet) than those bounded by the normal frame. This may be attributed
to the fact that a weak frame, especially a weak beam allows more rota-
tion to develop at beam-column joints which can accommodate the
presence of gap better than a stiffer frame. A stiff beam remains relative-
ly horizontal during loading with little end rotation at beam-column
joints. This leads to small contact regions at loaded corners. In the case
8 X. Chen, Y. Liu / Structures 10 (2017) 1–12

a)
Table 3 AA-10-30
FE model specimens and parameters. 300 BC-15-30
Gap location ID Gap size Frame type Fric. coef. 250 BB-30
Per side total

System load (kN)

200
None C0 0 NF, SF, WF 0.4, 0.7, 1.0
Full separation AA-1-3 1 3 NF, SF, WF 0.4 150
AA-5-15 5 15
AA-10-30a,b 10 30
100
AA-15-45a,b 15 45
50
AA-20-60a,b 20 60
Beam-infill BB-1 1 1 NF, SF, WF 0.4 0
BB-5 5 5 0 20 40 60 80
BB-10 10 10 Lateral Displacement (mm)
BB-15 15 15
BB-20 20 20 b)
BB-30a,b 30 30 AA-10-30
Column-infill BC-1-2 1 2 NF, SF, WF 0.4 160
BC-15-30
BC-5-10 5 10 0.4 140
BC-10-20 10 20 0.4, 0.7, 1.0 BB-30
120
L0R20 0, 20 20 NF 0.4, 0.7, 1.0

Net load (kN)

L20R0 20, 0 20 0.4, 0.7, 1.0 100
BC-15-30a,b 15 30 NF, SF, WF 0.4 80
BC-20-40a,b 20 40
60
a
Total gap size exceeds the maximum elastic lateral displacement for the normal frame 40
(NF).
b
Total gap size exceeds the maximum elastic lateral displacement for the strong frame 20
(SF). 0
0 20 40 60 80
Lateral Displacement (mm)

of Kcr, it is greatly influenced by the bare frame stiffness. As a result, a
weak frame has a low Kcr while a strong frame has a high Kcr.
The above findings are also valid for BB and BC gap scenarios in gen-
eral. However, a noted difference is that the normalized strength Pnet of
infills bounded by strong frames for BB gap scenario showed a more
drastic decrease than those bounded by normal or weak frames. This
is shown in Fig. 11. This is believed to be attributed to the failure mech-
anism change in specimens with BB gap scenario and bounded by
strong frames. As previously discussed, for these specimens, the failure
was initiated by sliding shear where the top left portion of the infill
tended to slide with respect to the bottom right portion of the infill.
This load transfer mechanism was much less efficient than the original
diagonal strut which ultimately led to a lower strength.
4.6. Effect of column-infill gaps
In practice, the gap may be present at either side of the column-infill
interface. The effect of location of column-infill gap along with friction
coefficient is discussed in this section. Fig. 12 compares the net load
vs. lateral displacement curves of specimens L20R0, BC-10-20 and
L0R20 for different μ values. It shows that increasing μ from 0.4 to 0.7
has little impact on the behaviour of the specimen. The initial infill
Ks Kcr Kt
250
200
Ki
Load (kN)
Fig. 7. Load vs. lateral displacement curves of infill with gaps: a) total system load, b) net
load of infill.
behaviour of infills with μ = 0.4 and 0.7 was non-participating and
after the lateral displacement of frame reached about 20 mm, the infill
became engaged and the system had a sudden increase in stiffness.
When μ was 1.0, however, different column-infill gap locations re-
sulted in different infill behaviour. Fig. 12(c) shows that when μ = 1.0,
specimen L0R20 was participating from the onset of loading with a Ki
of 44.6 kN/mm, just slightly lower than that of specimen C0 (62.6 kN/
mm). Since in this case the gap on the loaded column side was non-
existent, the diagonal strut began to develop at the onset of loading.
For both specimens BC-10-20 and L20R0, due to the gap at the loaded
column side, the initial infill behaviour was non-participating until the
displacement reached 10 and 20 mm respectively.
In terms of strength, when μ = 0.4 and 0.7, there was little variation
in the net strength among the three gapped specimens. When μ = 1.0,
the lowest net strength (166 kN) occurred in specimen L0R20 while the
highest total strength (189 kN) occurred in specimen L20R0 and speci-
men BC-10-20 had a strength (172 kN) in between. The finding that the
infill (specimen L0R20) with the highest initial stiffness but lowest
strength may be attributed to the fact that the right side of column-
infill gap in specimen L0R20 was never fully closed. This can be seen
No gap (C0) in the finite element failure mode comparison shown in Fig. 13. The di-
bare frame (NF) rect bearing of right column at the bottom right corner was absent for
AA-10-30 specimen L0R20, and the transferring of lateral load relied solely on
AA-10-30 (Net load) the beam-infill interface. For specimen L20R0, the left gap was closed
through frame deformation and thus the lateral load was transferred

150 through both the column-infill interface and beam-infill interface at

both loaded corners. As a result, specimen L20R0 had a better-
100 developed diagonal strut than specimens L0R20 and BC-10-20 and
thus achieved higher ultimate load.
50 Pnet In summary, when the friction coefficient is b 1.0 which is commonly
the case for masonry on steel surfaces, the effect of friction coefficient to
0
0 20 40 60 80
the infill stiffness and strength for gapped frames is negligible. When
Lateral Displacement (mm) the friction coefficient is N 1.0 which may be considered for very rough
surface contact, the infill stiffness and strength are noticeably affected
Fig. 6. Definitions of Ki, Kcr, Kt and Pnet in this study. by the column-infill gap location.
 X. Chen, Y. Liu / Structures 10 (2017) 1–12 9

a) b)

Fig. 8. Compressive stress distribution in FE models a) specimenAA-5-15, b) specimen BB-10bounded by the strong frame (grey areas indicate crushing failure of masonry).

4.7. Discussion of gap treatment in MSJC 2013
As mentioned earlier, the guidelines contained in the MSJC 2013 on
the treatment of interfacial gaps in the design of masonry infills specify a
9.5 mm limit for the beam-infill gap along with a 0.5 reduction factor for
the infill stiffness and strength. In practice, most steel frames have an
elastic deformation N 10 mm. Therefore, it is safe to assume that for
most practical frames, the compressive diagonal strut can be developed
without resulting in plastic strains at column bases or beam-column
joints if the gap is within 10 mm. The finite element results of the entire

a) 1.2 a) 1

1.0 0.8
Normalized Kcr

Normalized Kcr

0.8
0.6
0.6
0.4 0.4

0.2 0.2
0.0
0 10 20 30 40 50 60 0
Gap Size (mm) 0 10 20 30 40 50 60
Gap Size (mm)
b) 1.2
b)
1.0 1
Normalized Kt

0.8 0.8
Normalized Kt

0.6 0.6
0.4 0.4
0.2
0.2
0.0
0 10 20 30 40 50 60 0
Gap Size (mm) 0 10 20 30 40 50 60
Gap Size (mm)

c) 1.2 c)
1
BC-5-10
1.0
Normalized Pnet

0.8
Normalized Pnet

0.8
0.6
0.6
0.4
0.4 BB-5 AA-5-15
0.2
0.2
0
0.0
0 10 20 30 40 50 60
0 10 20 30 40 50 60
Gap Size (mm) Gap Size (mm)
AA (weak frame) AA (normal frame) AA (strong frame)

Fig. 9. Normalized stiffness and strength of infills with varying gap sizes and locations: Fig. 10. Normalized stiffness and strength of infills bounded by different frames (AA
a) Kcr, b) Kt, and c) Pnet. scenario): a) Kcr, b) Kt and c) Pnet.
10 X. Chen, Y. Liu / Structures 10 (2017) 1–12

1 a)
0.8
Normalized Pnet

0.6

0.4

0.2

0
0 10 20 30
Gap Size (mm)
BB (weak frame) BB (normal frame) BB (strong frame)

Fig. 11. Normalized Pnet of infills bounded by different frames with BB scenario gaps.

parametric study in terms of normalized stiffness and strength are plot-

ted against the gap size (total width) in Fig. 14. The lower bound of all
results is also shown in the figure together with the 0.5 reduction factor
line. It can be seen that, in the case of Kcr, the 0.5 reduction factor be-
comes un-conservative when gap size reaches 3 mm. In the case of Kt, b)
the reduction factor is adequate up to a gap size of 20 mm and beyond
that, the reduction factor is only un-conservative for a few infills bound-
ed by the strong frame, as identified in Fig. 14(b). This suggests that the
stiffness referred to in the MSJC 2013 should be intended for the tangent
stiffness after the infill engagement. In the case of infill strength, up to a

a)
250

200
Load (kN)

150

100

50

0
0 10 20 30 40 50
Lateral Displacement (mm)

b) 300

250
Load (kN)

200
Fig. 13. Distribution of compressive stress prior to failure in a) specimen L20R0 and
150 b) specimen L0R20.

100
total gap size of 10 mm, the 0.5 reduction factor provides a markedly
50 conservative estimate for infills with normal and weak bounding
0 frames. The average reduction factor for a gap size of 10 mm is around
0 10 20 30 40 50 0.83. For strong bounding frames, however, the 0.5 reduction factor is
Lateral Displacement (mm)
closer to the FE results for a gap size of 10 mm but on the un-
c) conservative side for a few cases beyond 10 mm. Note that these speci-
300 mens were infills with BB gap scenario and bounded by strong frames, it
250 suggests that the reduction factor should be used with caution with
strong bounding frames. In the case of normal and weak frame, the
200
Load (kN)

0.5 reduction factor was conservative up to 30 mm.

150 To obtain reduction factor expressions that may cover all parameters
as discussed previously, a linear regression analysis on the finite ele-
100
ment results was conducted. A set of analytical equations was then pro-
50 posed as follows for the lower bound curves of normalized Kcr, Kt
0
and Pnet.
0 10 20 30 40 50
Lateral displacement (mm) 0:0324x þ 1:42 2
R F;cr ¼ ; R ¼ 0:996 ð5aÞ
0:817x þ 1:42
C0 L0R20 BC-10-20 L20R0

Fig. 12. Net load vs. lateral displacement curves of infill with different μ: a) μ = 0.4, b) μ =
0:0103x þ 2:97 2
R F;t ¼ ; R ¼ 0:981 ð5bÞ
0.7, and c) μ = 1.0. 0:157x þ 2:97
 X. Chen, Y. Liu / Structures 10 (2017) 1–12 11

a) showed that the reduction effect of gap(s) on Kcr, Kt and Pnet are almost
1.2 identical to the case of a square infill. Hence, the proposed reduction fac-
1.0 tor equations are also valid for other aspect ratios. It is noted that these
equations were developed for infills bounded by steel frames. For other
Normalized Kcr

0.8
bounding frame materials, it is believed that the general trend present-
0.6 ed is still valid while the set of equations may not be directly applicable.
0.4 Since the model used a single frame configuration, its applicability to
multi-storey, multi-bay infilled frames and infills with openings needs
0.2
further investigation.
0.0
0 10 20 30 40 50 60 70
Gap size (mm) 5. Conclusions
b) 1.2 A finite element model was developed to study the in-plane stiffness
1.0 and strength of masonry infills bounded by steel frames with the focus
on the effect of frame-to-infill interfacial gaps. A parametric study on
Normalized Kt

0.8 the effect of gap size and location, bounding frame stiffness, and friction
0.6 coefficient on the lateral stiffness and strength of infilled frames was
BC-20-40
AA-15-45 conducted. The validity of design guidelines on gapped infills contained
0.4 BB-20 AA-20-60
BB-30
in the MSJC 2013 was examined. Some conclusions stemmed from this
0.2 study are as follows:
0.0 For all specimens considered, an increase in gap size resulted in a re-
0 10 20 30 40 50 60 70 duction in infill secant crack stiffness and tangent crack stiffness, as well
Gap size (mm) as in the infill strength. The degree of reduction is dependent on the gap
c) location and the bounding frame stiffness. When comparing beam-infill
1.2
and column-infill gap scenarios for a given total gap size, the beam-infill
1.0 gap resulted in the greater reduction in both the stiffness and strength.
The reduction in stiffness or strength for the full separation gap is not
Normalized Pnet

0.8
simply the addition of the reduction of the cases of beam-infill and
0.6 column-infill gaps. It is greater than the summation of reduction in
SF-BB-10
SF-BB-20 the latter two cases.
0.4 SF-BB-5 SF-BB-15 SF-BB-30 SF-AA-15-45
For varying bounding frame stiffness, results showed that gap(s) has
0.2 less effect on infills bounded by weak frames than on those bounded by
0.0 stronger frames. In the case of beam-infill gap on infills bounded by
0 10 20 30 40 50 60 70 strong frames, sliding shear failure was observed which resulted in a
Gap size (mm) significant reduction in infill strength.
FE data points For the friction coefficient, results showed that increasing μ from 0.4
MSJC 2013 reduction factor line to 0.7 has no significant effect on the infill behaviour as related to the
Lower bound envelope column-infill gap location. However, when μ = 1.0, the infill stiffness
Analytical equation of lower bound envelope and strength are noticeably affected by the column-infill gap location.
The examination of MSJC 2013 shows that in the case of infill
Fig. 14. Comparison of the lower bound envelope from FE results and the MSJC reduction strength, up to a total gap size of 10 mm, the 0.5 reduction factor pro-
factor: a) Kcr, b) Kt, and c) Pnet. vides a markedly conservative estimate for infills with normal and
weak bounding frames. For strong bounding frames, however, the 0.5
reduction factor is adequate and even slightly on the high side. In the
case of tangent crack stiffness, the reduction factor provides marked un-
0:202x þ 0:37 2 derestimates for a gap size up to 10 mm for all bounding frame cases.
R F;net ¼ ; R ¼ 0:996 ð5cÞ
0:501x þ 0:37 A set of equations were proposed for calculating reduction factors for
stiffness and strength over a wide range of gap size and locations and
where x is the total gap size, “RF,cr”, “RF,t” and “RF,net” are reduction fac- they were shown to achieve a good agreement with FE results.
tors for Kcr, Kt and Pnet, respectively, and R2 is the coefficient of determi-
nation of the regression analysis.
Acknowledgements
The comparison of these equations with FE results is illustrated in
Fig. 14. It can be seen that all proposed analytical equations obtained a
The authors wish to recognize the contribution of financial assis-
R2 N 0.98, indicating a good performance of equations against FE results.
tance by the Canadian Concrete Masonry Producers Association and
Natural Sciences and Engineering Research Council of Canada.
4.8. Application of reduction factors and limitations

These reduction factors (Eqs. (5a)–(5c)), developed based on nor- References

malized stiffness and strength trend, are intended to be used to modify
the stiffness and strength of the infills without gaps. Although the study
suggests that some yielding in frame members does not adversely im-
pact the infill strength, it is recommended that the gap size be kept
within the elastic limit of the frame deformation to ensure the frame in-
tegrity in the event of infill failure. It is also noted that while the previ-
ous discussion was presented for aspect ratio equal to 1.0, other aspect
ratios (h/l = 1.4 and 0.78) were also investigated and the results
[1] McBride R. The Behaviour of Masonry Infilled Steel Frames Subjected to Racking.
MScE Thesis; New Brunswick, Canada: University of New Brunswick; 1984.
[2] Yong TC. Shear Strength of Masonry Infilled Panels in Steel Frames. MScE Thesis;
New Brunswick, Canada: University of New Brunswick; 1984.
[3] Amos K. The Shear Strength of Masonry Infilled Steel Frames. MScE Thesis; New
Brunswick, Canada: University of New Brunswick; 1985.
[4] Richardson J. The Behaviour of Masonry Infilled Steel Frames. MScE Thesis; New
Brunswick, Canada: University of New Brunswick; 1986.
[5] Dawe JL, Seah CK. Behaviour of masonry infilled steel frames. Can J Civ Eng 1989;
16(6):865–76.
12 X. Chen, Y. Liu / Structures 10 (2017) 1–12
[6] Flanagan RD. Behavior of Structural Clay Tile Infilled Frames. Ph.D Dissertation;
Knoxville: The University of Tennessee; 1994.
[7] El-Dakhakhni WW. Nonlinear Finite-element Modelling of Concrete Masonry-
infilled Steel Frame. MScE thesis; Philadelphia: Drexel University; 2002.
[8] Liu Y, Soon S. Experimental study of concrete masonry infills bounded by steel
frames. Can J Civ Eng 2012;39(2):180–90.
[9] Liu Y, Manesh P. Concrete masonry infilled steel frames subjected to combined in-
plane lateral and axial loading — an experimental study. Eng Struct 2013;52:331–9.
[10] Dhanasekar M, Page AW, Kleeman PW. A finite element model for the in-plane be-
havior of brick masonry. Proceedings of the 9th Australasian Conference on Mecha-
nisms of Structures; 1984. p. 262–7.
[11] El-Dakhakhni WW, Elgaaly M, Hamid AA. Three-strut model for concrete masonry-
infilled steel frames. J Struct Eng 2003;129(2):177–85.
[12] Dawe JL, Seah CK, Liu Y. A computer model for predicting infilled frame behaviour.
Can J Civ Eng 2001;28:133–48.
[13] Stafford-Smith B. Behaviour of square infilled frames. J Struct Div 1966;92(1):
381–404.
[14] Mainstone R. On the stiffnesses and strengths of infilled frames. Proc Inst Civ Eng
1971(Supplement IV):57–90.
[15] CAN/CSA S304-14. Design of Masonry Structures. ON, Canada: Canadian Standard
Association; 2014.
[16] ACI 530-13/ASCE 5-13/TMS 402-13. Building Code Requirements for Masonry Struc-
tures. United States: Masonry Standards Joint Committee; 2013.
[17] European committee for standardization. Eurocode 8: Design of Structures for Earth-
quake Resistance — Part 1: General Rules, Seismic Actions and Rules for Buildings;
2004.
[18] Riddington JR. The influence of initial gaps on infilled frame behaviour. Proc Inst Civ
Eng 1984;77:295–310.
[19] Kadir MRA. The structural behavior of masonry infill panels in framed structures.
Ph.D. thesis; University of Edinburgh; 1974.
[20] Ng'andu BM. Bracing Steel Frames With Calcium Silicate Element Walls. Ph.D Disser-
tation; Eindhoven, Netherlands: Eindhoven University of Technology; 2006.
[21] Nazief MA. Finite Element Characterization of the Behaviour of Masonry Infill Shear
Walls With and Without Openings. Ph.D Dissertation; Alberta, Canada: University of
Alberta; 2014.
[22] Chen X, Liu Y. A finite element study of the effect of vertical loading on the in-plane
behavior of concrete masonry infills bounded by steel frames; 2016. Eng Struct 3:
118–29. http://dx.doi.org/10.1016/j.engstruct.2016.03.010.
[23] Priestley MJN, Elder DM. Stress-strain curves for unconfined and confined concrete
masonry. ACI J 1983;80(3):192–201.
[24] Lourenço PB. Computational Strategies for Masonry Structures. PhD thesis; Delft,
Netherlands: Civil Engineering Department, Delft University of Technology; 1996.
[25] Lourenço PB, Rots JG. A solution for the macro-modelling of masonry structures. Pro-
ceedings of the 11th International Brick/Block Masonry Conference. Shanghai,
China: Tongji University; 1997. p. 1239–49.
[26] Al-Chaar B, Mehrabi A. Constitutive models for nonlinear finite element analysis of
masonry prisms and infill walls. Report ERDC/CERL TR-08-19, US Army Corps of En-
gineers; 2008.
[27] King GJW, Pandey PC. The analysis of infilled frames using finite elements. Proc Inst
Civ Eng, Part 2 1978;65:749–60.