buildings
Article
Numerical Model Calibration and a Parametric Study Based on
the Out-Of-Plane Drift Capacity of Stone Masonry Walls
Ibrahim Serkan Misir 1
Citation: Misir, I.S.; Yucel, G.
Numerical Model Calibration and a
Parametric Study Based on the
Out-Of-Plane Drift Capacity of Stone
Masonry Walls. Buildings 2023, 13,
437. https://doi.org/10.3390/
buildings13020437
Academic Editor: Nerio Tullini
Received: 22 December 2022
Revised: 23 January 2023
Accepted: 1 February 2023
Published: 4 February 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Buildings 2023, 13, 437. https://doi.org/10.3390/buildings13020437
and Gokhan Yucel 2, *
1 Civil Engineering Department, Dokuz Eylul University, Izmir 35390, Turkey
2 Civil Engineering Department, Osmaniye Korkut Ata University, Osmaniye 80010, Turkey
* Correspondence: gokhanyucel@osmaniye.edu.tr
Abstract: Failure under seismic action generally occurs in the form of out-of-plane collapses of walls
before reaching their in-plane strength in historical stone masonry buildings. Consistent finite element
(FE) macro modeling has emerged as a need for use in seismic assessments of these walls. This paper
presents the numerical model calibration of U-shaped multi-leaf stone masonry wall specimens tested
under ambient vibrations and out-of-plane (OOP) load reversals. The uncertain elastic parameters
were obtained by manual calibration of the numerical models based on ambient vibration test (AVT)
data of the specimens. To obtain nonlinear calibration parameters, static pushover analyses were
performed on FE models simulating quasi-static tests. The calibrated numerical models matched well
with the experimental results in terms of load–drift response and damage distribution. As a result,
the modulus of elasticity and tensile and compressive degrading strength parameters of masonry
walls were proposed. A parametric study was conducted to examine the effects of different materials
and geometric properties (tensile strength, aspect ratio, slenderness ratio, and geometric scale) on the
OOP behavior of stone masonry walls. A quite different strain distribution was obtained in the case
of a large aspect ratio, while it was determined that the geometric scale had no effect on the strain
distribution. Tensile strength was the dominant parameter affecting the load–drift response of the
models. Within the presented work, a practical tool for out-of-plane seismic assessment has been
proposed for the structures covered in this paper.
Keywords: historical stone masonry; out-of-plane performance; quasi-static test; ambient vibration
test; seismic performance; finite element analysis; parametric study
1. Introduction
The earthquake performance of historical masonry structures tends to deteriorate
day by day due to the seismic activities and other environmental effects they are exposed
to throughout their lifetime. During an earthquake, wall failure usually occurs in the
out-of-plane (OOP) direction before the walls reach their in-plane (IP) strength [1–6]. This
weakness leading to the OOP damage of walls has been revealed by damage observations
after earthquakes (Figure 1) [7–9]. Numerous OOP collapses have been reported, partic-
ularly in buildings without a diaphragm effect and with long unsupported wall lengths.
This weak behavior of masonry is also due to its inherent mechanical characteristics, het-
erogeneity, anisotropy, and assembly deficiencies [10]. However, there is a primary need
for developing cost-effective tools for use in assessing the performance levels of historical
masonry structures.
Research has focused on the OOP behavior of historical masonry structures in the
recent decade. OOP loading tests are performed under static or dynamic loads, i.e., using
an actuator [11–15], an airbag [16–18], or a shake table [19–21]. Specimens prepared to
represent the OOP behavior of historic masonry walls are often simplified due to laboratory
limitations, e.g., the production of large-scale specimens and the complexity and size of the
geometric properties of representative structures. For this cases, small-scale specimens were
https://www.mdpi.com/journal/buildings
Buildings 2023, 13, 437 2 of 18

tested [22–24]. However, since the response of brick-and-mortar phases forming the wall
under lateral loading varies with the size of the specimen and the loading rates, it becomes
more difficult to fulfill the requirements of similitude laws as the geometric scale becomes
smaller. Moreover, even if the brick size and mortar thickness are properly reduced, small-
Buildings 2023, 13, x FOR PEER REVIEW 2 of 18
scale masonry tends to be highly strong, but, also, highly flexible [22]. Therefore, generally,
subsystem tests were performed on full- or large-scale specimens [16,25–27].

(a) (b) (c)

Figure1.1.Recent
Figure RecentOOP
OOPcollapse
collapsemechanisms
mechanisms inin masonry
masonry structures
structures (a)(a) flexure
flexure damage
damage [7],[7](b,c)
, (b)OOP
and
(c) OOP rocking damage of unsupported walls [8,9].
rocking damage of unsupported walls [8,9].

Research
Finite elementhas(FE)
focused on the
modeling of OOP
masonry behavior
is still of historical masonry
in development. Threestructures
main modeling in the
recent decade.
approaches OOP loading
are being adopted, tests are performed
namely, macro, meso, underand static or dynamic
micro modeling loads, i.e., using
[1]. Even for
meso-scale modeling, where nonlinear behavior is represented by interfaces, its use into
an actuator [11–15], an airbag [16–18], or a shake table [19–21]. Specimens prepared
represent
the seismicthe OOP behavior
assessment of theofwhole
historic masonry
structure is walls are often
not feasible due simplified due to labora-
to its complexity and
tory limitations, e.g., the production of large-scale specimens
computational cost. Therefore, meso- and micro-scale models are preferred in research and the complexity and size
of thedetailed
where geometric properties
damage of representative
assessments are examined structures.
[28–30].For Thethis cases,
macro small-scale
modeling speci-
approach
is,mens
thus,were testedpreferred
generally [22–24]. However,
in such large since the response
analyses, and the of homogenization
brick-and-mortarmethods phases form-
[31]
ingemphasized.
are the wall under lateral loading
Although varies with
this approach the size ofinthe
is insufficient specimen and
determining thecollapse
local loading
rates, it becomes
mechanisms [32], itmore
has been difficult
shown tothat
fulfill the requirements
it simulates cracks that ofoccur
similitude
due tolaws as the
seismic geo-
actions
onmetric scale becomes
the structure properly. smaller. Moreover,
This explains whyeven theifmacro
the brick size and
modeling mortar thickness
approach is generally are
properlyto
preferred reduced,
obtain the small-scale masonryoftends
global behavior masonryto bestructures
highly strong, but, also,
[18,32–38]. highly flexible
However, when
[22]. Therefore,
seismic assessment generally,
of historical subsystem
masonrytests were modeled
buildings performed with onmacro
full- or large-scale
elements speci-
is needed,
itmens
is seen that displacement-based metrics in the out-of-plane direction to determine the
[16,25–27].
damage levelelement
Finite of the walls
(FE)are missing.ofRecently,
modeling masonryout-of-plane drift limits have
is still in development. Threebeen
mainproposed
model-
byingMisir et al. (2022)
approaches are beingas a adopted,
common namely,indicatormacro,of damagemeso,and andfor use modeling
micro in evaluating the
[1]. Even
performance levels of stone masonry walls during a non-linear
for meso-scale modeling, where nonlinear behavior is represented by interfaces, its use in analysis [15].
the This
seismic paper mainly presents
assessment of the wholea nonlinear
structure model
is notupdating
feasible due worktoon itsmacro-scale
complexity and FE
models calibrated
computational based
cost. on bothmeso-
Therefore, dynamic andand quasi-static
micro-scale modelstest data of large-scale
are preferred stone
in research
masonry wall specimens
where detailed damagetested in the OOP
assessments direction.[28–30].
are examined MaterialThe model
macroparameters,
modelingincluding
approach
the
is, degrading
thus, generally strengths
preferredat post-peak, wereanalyses,
in such large obtainedand in the
thefull range of responses.
homogenization methods More-
[31]
over, the level of damage in the wall specimens and the numerical
are emphasized. Although this approach is insufficient in determining local collapse results were compared.
Performance
mechanismslimits [32], itinhas
termsbeen ofshown
OOP driftthatratios were used
it simulates cracks to that
assess the damage
occur states of
due to seismic ac-
the
tions on the structure properly. This explains why the macro modeling approach istests
models. The paper first summarizes the in situ mechanical and ambient vibration gen-
(AVT)
erallyon Isabey Mosque
preferred to obtain(Figure 2), dated
the global 1375,oftaken
behavior masonryas the reference[18,32–38].
structures historical structure,
However,
and
when theseismic
dynamic and quasi-static
assessment cyclicmasonry
of historical OOP tests on threemodeled
buildings wall specimens
with macrorepresenting
elements
the
is needed, it is seen that displacement-based metrics in the out-of-plane directionofto
reference [39]. Test data have been successfully used in nonlinear simulations twode-
U-shaped stone masonry wall specimens with the same geometry
termine the damage level of the walls are missing. Recently, out-of-plane drift limits have but different morphology
(i.e.,
been double-leaf
proposed by DMW Misir andet three-leaf
al. (2022) as TMW mortared
a common walls).of
indicator Processing
damage and of AVT datainwith
for use eval-
Operational Modal Analysis (OMA) allowed the estimation
uating the performance levels of stone masonry walls during a non-linear analysis [15]. of elastic material parameters,
and quasi-static
This papertests mainlyallowed the determination
presents a nonlinear model of OOP load–drift
updating workresponses leading to
on macro-scale FE
calibration in the nonlinear range by a set of nonlinear static pushover analyses. Finally,
models calibrated based on both dynamic and quasi-static test data of large-scale stone
results of a parametric study aimed at generalizing the findings from these limited number
masonry wall specimens tested in the OOP direction. Material model parameters, includ-
of scaled tests were given.
ing the degrading strengths at post-peak, were obtained in the full range of responses.
Moreover, the level of damage in the wall specimens and the numerical results were com-
pared. Performance limits in terms of OOP drift ratios were used to assess the damage
states of the models. The paper first summarizes the in situ mechanical and ambient vi-
bration tests (AVT) on Isabey Mosque (Figure 2), dated 1375, taken as the reference his-
torical structure, and the dynamic and quasi-static cyclic OOP tests on three wall speci-
mens representing the reference [39]. Test data have been successfully used in nonlinear
simulations of two U-shaped stone masonry wall specimens with the same geometry but
 Buildings 2023, 13, x FOR PEER REVIEW 3 of 18

Buildings 2023, 13, 437 3 of 18

(a) (b)
Figure 2. (a) A general view of Isabey Mosque and (b) the inside view of the West courtyard wal
[38].

2. Reference OOP Tests

(a) test data of three stone masonry wall specimens,
The (b) namely, the double-leaf dr
jointed wall
Figure
Figure (DDW),
2.2.(a)
(a)AAgeneral
general theview
view double-leaf
ofofIsabey
Isabey Mosque
Mosque wall and
and with
(b)(b)
thethehydraulic
inside
inside viewview lime
of
of the the
West mortar
West (DMW),
courtyard
courtyard [38]. and t
wallwall
[38].
three-leaf wall specimen with hydraulic lime mortar (TMW), were used in the simulatio
2. Reference OOP Tests
as the 2.reference
Reference experimental
OOPof Tests data [39]. The latter wall includes two outer stone leav
The test data three stone masonry wall specimens, namely, the double-leaf dry-
and ajointed
rubbleThe
infill
walltest(DDW),
layerthree
data of the
asdouble-leaf
the third wall
stone masonry
leaf.withFigures
wall hydraulic
3a and
specimens,lime
3b
namely,
show
mortar
the geometries
the(DMW),
double-leafand dry-
the
of t
two- and three-leaf
jointed
three-leaf wall
wall(DDW), specimens,
specimen thewith respectively.
double-leaf
hydraulic wall
limewith The
mortar specimens
hydraulic
(TMW),lime werehad used1/6
mortar scale
in(DMW),
the U-shaped
and the geom
simulations
etry aimed
as
three-leafat the
the reference wallinteraction
experimental
specimen withofdata
the[39].
front
hydraulic The and
limelattertransverse
wall(TMW),
mortar includes walls interfering
two used
were outer instone at steep
leaves
the simulations and corne
aasrubble infill
the reference
The specimens layer
reflect as thegeometric
experimental
the thirddata leaf.[39].
Figures
and 3a andwall
Thematerial
latter 3bproperties
show
includesthe geometries
twoof outer of theleaves
stone
the reference two-historic
and
andthree-leaf
a rubble infillspecimens,
layer as respectively.
the third leaf. TheFigures
specimens 3a andhad 3b 1/6show
scalethe U-shaped
geometries geometry
of the
structure, Isabey Mosque, such as the wall morphology, stone material type, mortar co
aimed
two- and at the interaction
three-leaf of the front
specimens, and transverse
respectively. walls interfering
The specimens had 1/6 scaleat steep corners.geom-
U-shaped The
tent and strength,
specimens
etry aimedreflect
rubble
at thethe
infill-to-total
geometric
interaction ofand
the material
walltransverse
thickness
front andproperties
ratio,
ofwalls
the totalhistorical
reference
interfering
rows atstructure,
at steep
wall
corners.
height, a
pect ratio
Isabey andMosque,
The specimens slenderness
such asthe
reflect ratio.
the wallAdditionally,
geometric morphology,
and material theproperties
stone absencetype,
material ofthe
of lead
mortarclamps
reference content connecting
and
historical t
strength,
stones,structure,
overlapping rubble infill-to-total
featuressuch
Isabey Mosque, wall thickness
in wall-to-wall ratio,
as the wall morphology, total
connections, rows at
stone and wall
material height,
the type,
absenceaspect
mortar ratio
ofcon-
a rigid di
and
tent slenderness
and strength, ratio.
rubble Additionally,
infill-to-total the absence
wall
phragm effect are among the common features of the reference structure and the thicknessof lead clamps
ratio, total connecting
rows at wall the stones,
height, as-test spe
overlapping
pect ratio and features in wall-to-wall
slenderness connections,
ratio. Additionally, theand the absence
absence of a rigid
of lead clamps diaphragm
connecting the
imens.effect
Theare reference historicalfeatures
among the common
structure has wallsstructure
of the reference
of 1.8 m thick and up to 15The m high th
stones, overlapping features in wall-to-wall connections, andand thethe test specimens.
absence of a rigid dia-
surround the
reference
phragm effect
courtyard
historical from
structure
are among thehas
its three sides
walls offeatures
common
[39].
1.8 m thick
of the
The
and test specimens
up to 15structure
reference m high that reflect
andsurround
thethegeomet
the test spec-
and boundary
courtyard
imens. The conditions
from its three
reference of the[39].
sides
historical west Thecourtyard
structure test
has wallswall
specimens and
reflect
of 1.8 transverse
the
m thick and up towalls,
geometry and with
15 mboundary
high thattheir lo
conditions
unsupported
surroundlength of the west
(approx.
the courtyard courtyard
from 30itsm) wall and
andsides
three transverse
relatively
[39]. Thehigh walls, with their
structurereflect
test specimens long
as well unsupported
theas a lack of a sla
geometry
length (approx.conditions
30 m) and relatively high structure as transverse
well as a walls,lack ofwith a slab. The
The specimens have similar morphologies to that of the reference structure, long
and boundary of the west courtyard wall and their classified
specimens
unsupported havelength
similar morphologies
(approx. 30 m) and to that of the reference
relatively structure,
high structure classified
as well as a lackas “double-
of a slab.
“double-leaf
leaf wall
with with connection” up to medium height and wall” “three-leaf wall” after m
Thewallspecimens connection”
have similarup to medium
morphologies height
to that of andthe“three-leaf
reference structure, after medium
classified as
diumheight,
height, according
according
“double-leaf walltowith thetoclassification
the classification
connection” up proposed
to medium
proposed
by Binda
height etby Binda
al. [40]
and andet
“three-leaf
al.wall”
Binda [40] and
et al.after Binda
[41],me-
as et
demonstrated
[41], as demonstrated by endoscopic
by examinations
endoscopic throughout
examinations the reference
throughout
dium height, according to the classification proposed by Binda et al. [40] and Binda et al. structure.
the reference structure.
[41], as demonstrated by endoscopic examinations throughout the reference structure.

(a) (b)
(a) (b)
Figure 3. Stone unit layout for (a) double-leaf wall specimen (dimensions in mm) and (b) three-leaf
wall
FigureFigurespecimen
3. Stone (the
unitunit middle
layout forof thedouble-leaf
wall was filled withspecimen
rubble). (dimensions in mm) and (b) three-le
3. Stone layout for(a) wall
(a) double-leaf wall specimen (dimensions in mm) and (b) three-leaf
wall specimen (the(the
wall specimen middle
middleofofthe
the wall wasfilled
wall was filled with
with rubble).
rubble).
Buildings 2023, 13, x FOR PEER REVIEW 4 of 18
Buildings 2023, 13, 437 4 of 18

Using
Using AVT AVT data datataken
takenfrom
fromthethe
west andand
west north courtyard
north wallswalls
courtyard of theof reference struc-
the reference
ture, it was found that the dominant free vibration modes of
structure, it was found that the dominant free vibration modes of the walls were inthe walls were in the OOP
the
direction and the frequency corresponding to the first mode was
OOP direction and the frequency corresponding to the first mode was 3.91 Hz [38]. Ref- 3.91 Hz [38]. Reference
specimens reflect the
erence specimens mean
reflect thevalues
meanofvalues
the modulus of elasticity
of the modulus and the uniaxial
of elasticity and thecompres-
uniaxial
sive strength of historic walls (4530 MPa and 2.30 MPa, respectively)
compressive strength of historic walls (4530 MPa and 2.30 MPa, respectively) obtained obtained by semi-by
destructive
semi-destructiveflat-jack tests performed
flat-jack on theon
tests performed reference structure
the reference [39]. The
structure [39].lime
The mortar used
lime mortar
in theinmortar
used jointed
the mortar specimens
jointed specimensreflects thethe
reflects content of of
content the
thehistoric
historicmortar
mortarand andthe
the mean
mean
compressive
compressive strength (2.30 MPa) of the reference structure. The initial shear strengths and
strength (2.30 MPa) of the reference structure. The initial shear strengths and
the characteristic value
the characteristic value of of the
the internal
internal friction
friction angle
angleofofmasonry
masonryunitsunitswere
werefvoifvoi =
= 0.35
0.35 MPa
MPa
and α αkk ==0.33
0.33for
for14-day-old
14-day-oldsamples,
samples,respectively.
respectively. TheThe specimens
specimens contained
contained marble
marble blocks
average compressive
with average compressiveand andtensile
tensilestrengths
strengthsofof94.2
94.2 and
and 14.6
14.6 MPaMPaandand
anan average
average den-
density
sity of 2710
of 2710 kg/mkg/m3 . The
3. The results of AVT (Figure 4) on the wall specimens were also
results of AVT (Figure 4) on the wall specimens were also used as
reference data.

First mode f = 29.38 Hz

Second mode f = 39.76 Hz

(a) (b)
Figure
Figure 4.
4. (a)
(a) AVT setup (1:
AVT setup (1: front
front wall
wall accelerometers shown inside
accelerometers shown inside the
the red
red circles,
circles, 2:
2: left
left transversal
transversal
wall accelerometers shown inside the blue circles, 3: data acquisition system), and (b) first
wall accelerometers shown inside the blue circles, 3: data acquisition system), and (b) first two
two free
free
vibration modes of the DMW specimen.
vibration modes of the DMW specimen.

The specimens were

The specimens weretested
testedunder
underOOPOOPloads
loadsrepresenting
representing thethe seismic
seismic inertial
inertial forces
forces by
by applying the loads at four specific points on the front wall, as shown
applying the loads at four specific points on the front wall, as shown in Figure 5a. Lateral in Figure 5a.
Lateral loads generated
loads generated by a hydraulic
by a hydraulic actuator actuator were transferred
were transferred to theatwall
to the wall at the inter-
the intersection
sectionofpoints
points a 2 × of a 2 × 2 grid
2 regular regular gridalong
placed placed along one-third
one-third and two-thirds
and two-thirds of the height of the
andheight
width
and width
of the wall.ofThetheloads
wall. were
The loads were distributed
distributed on the wallon the the
with wallhelp
with ofthe help of abeam
a spreader spreader
and
beam and four semi-rigid plywood plates (each with an area of 2500
four semi-rigid plywood plates (each with an area of 500 × 800 mm ). Target displacements × 800 mm 2). Target

displacements
were increasedwere on theincreased
basis ofon thethe basis of response
structural the structural response
of the specimens of thebyspecimens
a semi-cyclic by
a semi-cyclic procedure that resulted in the strength envelopes shown
procedure that resulted in the strength envelopes shown in Figure 5b. No vertical load in Figure 5b. No
vertical
was load to
applied wastheapplied
top of tothethe top of theasspecimens,
specimens, as in the
in the reference reference
structure. The structure. The
vertical and
vertical
horizontalandaxeshorizontal axes
represent, represent, respectively,
respectively, the resultant
the resultant force applied force
to theapplied
center of tothe
thefront
cen-
ter ofand
wall thethefront
OOPwall and the OOP
displacement displacement
measured at the measured
top mid-span at the topwall.
of the mid-spanDMWofand theTMW
wall.
DMW
specimensand behaved
TMW specimens behaved
quite similarly quite similarly
regarding regarding
the strength the strength
envelopes, with a moreenvelopes,
stable
with
damagea morestatestable
of DMWdamage state drifts.
in higher of DMW in higher drifts.
 Buildings
Buildings
Buildings 13,13,
2023,
2023,
2023, 13, x FOR
437
x FOR PEER
PEER REVIEW
REVIEW 5 of55 18
of 18
of 18

(a)(a) (b)(b)
Figure
Figure
Figure 5. (a)
5. (a)
5. (a)
OOPOOP test
testtest
OOP setup,
setup, and
setup, and (b)
(b)(b)
and load–drift
load–drift response
response
load–drift of wall
of of
response wall
wall specimens.
specimens.
specimens.

TheThe
The
wall wall
wall specimens
specimens
specimens couldcould
could exhibit
exhibit
exhibit a certain
aa certain
certain amount
amount
amount of of
ofbending bending
bending deformation
deformation
deformation under under
under out-
out-of-plane
of-plane displacement
displacement due todue
their to their
slender slender
structure structure
and
out-of-plane displacement due to their slender structure and long unsupported spans. long and long
unsupported unsupported
spans. They spans.
also
They
They also
continued
also continued
to
continuedcarry loads to carry
to carryup loads loads upup
to relatively to relatively
large drift large
to relatively large
ratios drift
thanks
drift ratios
to
ratios the thanks
cantilever
thanks to behavior
to the the cantile-
cantile- of
ver
verthe behavior of the
wall andofquasi-static
behavior wall
the wall and and
loading quasi-static
conditions.
quasi-static loading conditions.
Bothconditions.
loading specimens showed Both specimens
a linear elastic
Both specimens showed
showed a a
branch
linear
up
linear to elastic
the
elastic peak branch
branchloadup up
and to
thethe
to passed peakpeak
into
loadaload
smooth
and and passed
softening
passed into into a smooth
behavior.
a smooth softening
Considering
softening behavior.
the OOP
behavior.
Considering
response
Considering ofthe the
theOOP OOP
specimens,response
response DDW of is
of the thefarspecimens,
from the
specimens, DDW DDW
mortar isjointed
is far far
from from the
walls
the mortar
in terms
mortar jointed walls
of stiffness
jointed walls
in
andterms of
strength, stiffness
probably and strength,
due to the probably
lack of due
tensile to the
strength lack of
that
in terms of stiffness and strength, probably due to the lack of tensile strength that governs tensile
governs strength
the that
whole governs
behavior.
the
theThe whole
whole behavior.
testsbehavior.
were stopped
The The attests
tests the
werewere
onset stopped
of theat
stopped at the
collapse
the onset
onsetof each of specimen.
of the the collapse
collapse of each
ofFigure
each specimen.
6 shows
specimen. the
Figure 6 shows the global and local mechanisms in the
Figure 6 shows the global and local mechanisms in the wall and Figure 7 shows the rela-the
global and local mechanisms in the wall and Figure 7 wall
shows and
the Figure
relationship7 shows the
between rela-
load–drift
tionship
tionship response
between
between and
thethe the crack
load–drift
load–drift patternand
response
response obtained
and
thethe for
crack
crack the DMWobtained
pattern
pattern specimen
obtained for[15].
for The
thethe DMW OOP
DMW
limit
specimen states
specimen [15]. ofThe
[15]. theThespecimens
OOP OOP limit could
limit be
states
states ofidentified
of
thethe (Figure
specimens
specimens 7a)
could
could inbeterms
be of top (Figure
identified
identified drift ratio
(Figure 7a)based
7a)
in in
on these
terms of observations,
top drift ratio and
based proposed
on these as 0.4%,
observations, 1.2%, and and
terms of top drift ratio based on these observations, and proposed as 0.4%, 1.2%, and 3.2% 3.2%
proposed for DMW
as 0.4%, corresponding
1.2%, and 3.2%
forto
forDamage
DMWDMW Limitation to
corresponding
corresponding (DL),
Damage Severe
to Damage Damage
Limitation
Limitation (SD),
(DL),
(DL), and Near
Severe
Severe Collapse
Damage
Damage (NC)
(SD),
(SD), and andlimit
NearNear states,
Col- Col-
respectively.
lapse (NC) It
limitshould
states,be noted
respectively. that theIt recommended
should be noted
lapse (NC) limit states, respectively. It should be noted that the recommended NC limit NC thatlimitthe state of 3.2%
recommended [15] NCis taken
limit
into
state ofaccount,
state of
3.2%3.2% although
[15][15]
is taken the
is taken intomaximum
into account,
account, drift reached
although
although theduring
the the experiments
maximum
maximum driftdrift shown
reached
reached wasthe
during
during 3.8%
the
in Figure
experiments
experiments 7a.
shownshown was was
3.8% 3.8% in Figure
in Figure 7a.7a.

(a)(a) (b)(b)
Figure
Figure 6. Observed
6. Observed damage
damage in the
in the DMWDMW specimen
specimen (a) (a) inner
inner andand outer
outer faceface bending
bending cracks
cracks on on
thethe
Figure 6. Observed damage in the DMW specimen (a) inner and outer face bending cracks on the
front wall (dashed and continuous lines, respectively) and (b) tensile cracks (within the rectangle).
front wall (dashed and continuous lines, respectively) and (b) tensile cracks (within the rectangle).
front wall (dashed and continuous lines, respectively) and (b) tensile cracks (within the rectangle).
 Buildings
Buildings 13,13,
2023,
2023, 437 PEER REVIEW
x FOR 6 of6 18
of 18

up to 6th cycle

beyond the 6th cycle

(a) (b)
Figure 7. 7.
Figure (a)(a)
Load–drift response,
Load–drift and
response, (b)(b)
and crack pattern
crack in in
pattern thethe
DMW specimen
DMW [15].
specimen [15].

3. 3.
Numerical Simulations
Numerical Simulations
TheThe seismic
seismic assessment
assessment of of historical
historical masonry
masonry structures,
structures, each
each with
with unique
unique structural
structural
features,isisa achallenging
features, challengingtasktask with
with complex material
materialbehavior,
behavior,wall
wallconnection
connection details, and
details,
andboundary
boundary conditions.
conditions.ForForthis
thisreason,
reason, it necessaryto
it is necessary toadopt
adoptananadequate
adequate constitutive
constitutive
material
material model
model andandupdate
updateitsitsproperties
propertieswith withexperimental
experimentaldata.data.Modeling
Modeling masonry
masonry
structures with the FE macro modeling approach is widely preferred because of itsits
structures with the FE macro modeling approach is widely preferred because of good
good
consistency
consistency and and low
low computational
computational cost.
cost. Therefore,
Therefore, thethe numerical
numerical models
models in in this
this paper
paper
have
have been
been setsetupuponon macro
macro scale
scale toto
bebe comparable
comparable with
with practical
practical applications.
applications. Firstly,
Firstly, inin
this
this section,
section, thethe adopted
adopted material
material model
model is isintroduced.
introduced. Secondly,
Secondly, the the numerical
numerical model
model
was
was updated
updated using
using thethe dynamic
dynamic and and
staticstatic test obtained
test data data obtained
in the in the laboratory.
laboratory. Then, non- Then,
nonlinear static pushover analyses were performed on the updated
linear static pushover analyses were performed on the updated numerical models until numerical models until
the same ultimate top drifts were achieved in the quasi-static OOP tests.
the same ultimate top drifts were achieved in the quasi-static OOP tests. In the last section, In the last section,
parametric
parametric studies
studies were
were conducted
conducted onon the
the validated
validated numerical
numerical model
model ofofDMWDMW to to examine
examine
thethe effect
effect ofof material
material and
and geometric
geometric properties
properties ononthethe
OOP OOP behavior
behavior of of a masonry
a masonry wall.
wall.
For frequency analysis, static test simulation, and parametric analysis,
For frequency analysis, static test simulation, and parametric analysis, a total of 3, 2, and a total of 3, 2, and
10 simulations were run, respectively.
10 simulations were run, respectively.
3.1. Adopted Material Model
3.1. Adopted Material Model
Nonlinear FE analyses were performed on 3D numerical models in the ABAQUS
Nonlinear FE analyses were performed on 3D numerical models in the ABAQUS
software [42] to simulate the OOP response of stone masonry wall specimens. The combined
software [42] to simulate the OOP response of stone masonry wall specimens. The com-
behavior of stone units and mortar joints in the wall specimens was represented by solid
bined behavior of stone units and mortar joints in the wall specimens was represented by
elements using the macro-scale modeling approach, which is computationally more efficient
solid elements using the macro-scale modeling approach, which is computationally more
for constructing larger building models compared to micro and meso modeling [1]. Some
efficient formodel
material constructing largerhave
parameters building
beenmodels
updatedcompared to microfor
and proposed and
themeso
use modeling
in practical
[1]. Some material model parameters have been updated and proposed
engineering applications, based on the simulations of the OOP response and thefor the uselimit
in prac-
states
tical engineering applications, based on the simulations of the OOP response
of the specimens. The Concrete Damage Plasticity model (CDP), defined in the ABAQUS and the limit
states of the
material specimens.
library, was usedTheto Concrete Damage
simulate the Plasticity
nonlinear model
behavior of the(CDP), defined
specimens. Thisinmodel,
the
ABAQUS material
originally library,
formulated to was used to
represent simulate
the theofnonlinear
behavior behavior
quasi-brittle of the(i.e.,
materials specimens.
concrete)
This model, originally formulated to represent the behavior of quasi-brittle materials
under cyclic and dynamic loads [43,44], has been used in masonry modeling [32,33]. Under (i.e.,
concrete) under cyclic and dynamic loads [43,44], has been used in masonry
uniaxial tensile stresses, the material follows a linear branch up to tensile strength inmodeling
[32,33]. Under uniaxial tensile stresses, the material follows a linear branch up to tensile
Buildings 2023, 13, x FOR PEER REVIEW 7 of 18
Buildings 2023, 13, 437 7 of 18

strength in accordance with the defined modulus of elasticity. Tensile strength corre-
sponds to the
accordance point
with theatdefined
which microcracks
modulus of begin to form.
elasticity. Thestrength
Tensile materialcorresponds
follows the prede-
to the
point
fined atstress–strain
which microcracks
curve inbegin
thetopost-peak
form. Therange.
material follows
When thethe predefined stress–strain
compressive and tensile
curve inare
stresses theexceeded,
post-peakarange. When
reduction the
rule incompressive and tensile
the elastic stiffness stresses
governs are exceeded,
the material a
(Figure
reduction
8). rule in the elastic stiffness governs the material (Figure 8).

(a) (b) (c)

Figure 8.
Figure 8. Tension (a), compression
Tension (a), compression (b),
(b), and
and uniaxial
uniaxial cyclic
cyclic (c)
(c) behavior
behavior of
of the
the CDP.
CDP.

compressive stresses
Uniaxial tensile and compressive ((t𝜎and
stresses (σ andσc )𝜎are
) are defined
defined as as follows:
follows:

𝜎 = 1 − 𝑑 𝐸 𝜀 −pl𝜀 (1)
 
σt = (1 − dt ) E0 ε t − ε t (1)

𝜎 = 1 − 𝑑 𝐸 𝜀 −pl𝜀 (2)
σc = (1 − dc ) E0 ε c − ε c (2)
where 𝐸 is the initial modulus of elasticity, 𝑑 and 𝑑 are the scalar damage variables,
𝜀 andE0𝜀 is are
where the total
initialdeformations, and 𝜀 dtand
modulus of elasticity, and𝜀dc are
arethe
thescalar damageplastic
equivalent variables, εt
defor-
pl pl
and
mations
ε c are total deformations, and
under tension and compression,
ε t and are the
respectively.
ε t equivalent plastic deformations
The post-peak behavior under ten-under
tension and compression,
sion is defined respectively.
as a function The post-peak
of the cracking behavior
strain, which is under
obtainedtension is defined
by the followingas
aequation:
function of the cracking strain, which is obtained by the following equation:

t𝜀 ==
εck ε t 𝜀−−
εel0t𝜀 (3)
where 𝜀 is the tensile elastic deformation. The equivalent tensile plastic deformation pl
where εel0t is the tensile elastic deformation. The equivalent tensile plastic deformation ε t
𝜀 and the elastic stiffness reduction can be obtained as follows:
and the elastic stiffness reduction can be obtained as follows:
𝑑 𝜎
pl𝜀 =ck𝜀 − dt σt (4)
εt = εt − 1−𝑑 𝐸 (4)
(1 − dt ) E0
1−
(1 1−−d)𝑑 ==(1 1−−st𝑠dc𝑑)(1 − sc𝑠dt𝑑
(5)
) (5)
where 𝑠 and 𝑠 are the function of stress and the elastic stiffness recovery factors, which
where st and sc are the function of stress and the elastic stiffness recovery factors, which
can be calculated
can be calculated as
as follows:
follows:
1−
st 𝑠= = 1− (σ)𝜎
wt𝑤H𝐻 (6)
(6)
sc = 1 − wc (1 − H (σ )) (7)
𝑠 =1−𝑤 1−𝐻 𝜎 (7)
In these equations, wt and wc are the material parameters expressing the stiffness
In these
recovery equations,
coefficients, and𝑤they 𝑤 arebetween
andremain the material parameters
the values expressing
of ‘0’ denoting thethe stiffness
absence of
recovery coefficients, and they remain between the values of ‘0’ denoting the absence
stiffness recovery and ‘1’ denoting the complete recovery. H (σ ) is the Heaviside function of
stiffnessas
defined recovery
0 for σ <and ‘1’ denoting
0 and theThe
1 for σ > 0. complete
defaultrecovery. 𝐻 𝜎 isrecovery
values of stiffness the Heaviside function
coefficients for
defined as
uniaxial 0 for
cyclic 𝜎 < 0 and
behavior 1 for 𝜎 > 0. The default values
(tension-compression-tension) are wt of
= 0stiffness
and wc recovery
= 1 (Figurecoeffi-
8c).
cients for uniaxial cyclic behavior (tension-compression-tension) are 𝑤 = 0 and 𝑤 = 1
3.2. Construction
(Figure 8c). and Calibration of Numerical Models
FE macro models of the specimens were constructed using eight-node hexahedral 3D
solid elements (C3D8). FE mesh is shown with the OOP loading scheme in Figure 9a. Fixed
support was assumed at the base of the walls, as displacements were prevented at the
bottom of the specimens confirmed by LVDT measurements. A mesh sensitivity analysis
 3.2. Construction and calibration of Numerical Models
FE macro models of the specimens were constructed using eight-node hexahedral 3D
Buildings 2023, 13, 437 8 of 18
solid elements (C3D8). FE mesh is shown with the OOP loading scheme in Figure 9a. Fixed
support was assumed at the base of the walls, as displacements were prevented at the
bottom of the specimens confirmed by LVDT measurements. A mesh sensitivity analysis
basedon
based onnine
ninemeshing
meshingoptions
optionswith
withdifferent
differentnumbers
numbersofofFEs FEswaswasperformed,
performed,and andthethe
resultswere
results werecompared
comparedin in terms
terms of
of the
the first
first mode
mode frequencies
frequenciesof ofthese
thesemodels,
models,asasshown
shownin
inFigure
Figure9b.
9b.AsAsaaresult,
result, the
the model
model was
was generated with 1762
generated with 1762 FEs,
FEs,resulting
resultingin inaamesh
meshsizesize
of 0.15 m (which is half the wall thickness) to achieve a fine demonstration
of 0.15 m (which is half the wall thickness) to achieve a fine demonstration of the strain of the strain
andstress
and stresscontour
contourplots
plotsusing
usingthe
theleast
leastnumber
numberofofelements
elementsforforthe
thesake
sakeofofcomputational
computational
efficiency. By using four interconnected surfaces on the model, the displacement
efficiency. By using four interconnected surfaces on the model, the displacement loading loading
appliedover
applied overfour
fourpoints
pointson onthe
thefront
frontwall
wallofofthe
thespecimens
specimenscould
couldbe besimulated.
simulated.These
These
surfaces are connected to a reference point located at the midpoint of the
surfaces are connected to a reference point located at the midpoint of the front wall usingfront wall using
a structural distribution couple following a quadratic weighting rule. This method is usedis
a structural distribution couple following a quadratic weighting rule. This method
toused to couple
couple both
both the the displacement
displacement and rotation
and rotation of eachofattached
each attached
point point
to theto the average
average dis-
displacement and rotation of the surface nodes within the influence
placement and rotation of the surface nodes within the influence radius [45]. radius [45].

(a) (b)
Figure 9. (a) FE mesh and the coupled surfaces (purple areas) for four-point loading in the models,
Figure 9. (a) FE mesh and the coupled surfaces (purple areas) for four-point loading in the models,
(b) mesh convergence plot based on the natural frequency of the first mode.
(b) mesh convergence plot based on the natural frequency of the first mode.

3.2.1.
3.2.1.Linear
LinearModel
ModelCalibration
Calibration
Calibration
Calibration of theinitial
of the initialnumerical
numericalmodel modelwas wascarried
carriedout outinintwo
twostages.
stages.InInthe
thefirst
first
stage,
stage,the
theelastic
elasticmaterial
materialparameters
parametersofofthe themodel
modelwere weremanually
manuallycalibrated
calibratedusing
usingOMA
OMA
results
resultsprocessed
processedwith with AVTAVT test data
test data[15]. TheThe
[15]. modal estimation
modal estimation was was
performed
performedusingusing
the
ARTeMIS
the ARTeMIS software
software[46] [46]
withwith
the the
EFDD EFDD (Enhanced
(Enhanced Frequency
Frequency Domain
DomainDecomposition)
Decomposition)
analysis
analysistechnique
technique [47]. In In
[47]. thethe
calibration,
calibration, the the
modulus
modulus of elasticity was was
of elasticity chosen as theasun-
chosen the
certain material
uncertain parameter
material parameterand was
andupdated.
was updated. Using Using
the widely accepted
the widely expression
accepted in the
expression
literature [48] E =[48]
in the literature [(200~1000) α fc] (where
E = [(200~1000) α fc] fc is thefcuniaxial
(where compressive
is the uniaxial strengthstrength
compressive of the
of the
wall and wall and a multiplier
a multiplier α is addedα isto
added to the expression
the expression to consider to consider
the scalethe scale the
effect), effect),
upperthe
upper
limit limit of
of [1000 [1000
α fc] fc] was
wasα taken as taken as an
an initial initial
value forvalue for the modulus
the modulus of elasticity
of elasticity for mortar- for
mortar-jointed
jointed wall (DMW walland (DMW
TMW)and TMW)
models. Themodels. The mean compressive
mean compressive strength of thestrength
mortarof(2.1the
mortar
MPa) for(2.1
DMW MPa) andfor DMW
the averageandofthe
theaverage of the meanstrengths
mean compressive compressive of thestrengths
mortar of the
(1.87
mortar (1.87 MPa) and rubble infill (2.96 MPa) for TMW were
MPa) and rubble infill (2.96 MPa) for TMW were used as fc, and the initial value of the used as fc, and the initial
value of of
modulus the modulus
elasticity wasof calculated.
elasticity was For calculated.
only the dryFor only wall
jointed the dry
(DDW)jointed walla (DDW)
model, value
model, a value
consistent consistent
with the with
static test the static
results test results
was assigned aswas assigned
the initial as the initial
modulus modulus
of elasticity. Theof
elasticity. The Poisson’s ratio was set to 0.25 for each specimen. Table 1 summarizes the
modulus of elasticity in both the initial and the updated numerical models, as well as the
mass density. In the DMW and TMW models, the calibrated module successfully converged
close to the upper limit (with 14% and 0.6% errors, respectively) of the expression above.
 upper
upper
upper limit limit
limit(with (with
(with 14%
14%and
14% and
and0.6% 0.6%
0.6% errors,
errors,
errors, respectively)
respectively)
respectively) ofoftheofexpression
the the expression
expression above.
above.
above.
upper
upper
upper
upper
upper
upperlimit
limit
limit
limit
upper limit
limit
(with
(with
(with
(with
limit (with
(with
14%
14%
14%
14%
(with 14%
14%
and
and
and
and
14% and
and
0.6%
0.6%
0.6%
and 0.6%
0.6%
errors,
0.6% errors,
errors,
errors,
errors,
errors,
0.6% respectively)
respectively)
respectively)
respectively)
respectively)
respectively)
errors, of the
ofof
the
of
respectively) ofexpression
of
the
the
of the
the expression
expression
expression
expression
expression
the above.
above.
above.
above.
above.
above.
expression above.
TableTable
Table1.1. 1. Elastic
Elastic
Elastic material
material
material properties
properties
properties ofof of initial
initial
initial FEFE FE models
models
models and
and and
the
the the results
results
results ofof of model
model
model updating.
updating.
updating.
Table
Table
Table
TableTable
Table
1.
1.1.
Table 1. Elastic
1.
Elastic
Table 1. Elastic
Elastic
Elastic
1. Elastic
1. material
material
material
Elastic
material
material
material
Elastic properties
properties
properties
material
properties
properties of
ofof
properties
material of
properties ofinitial
of
initial
properties of initial
initial
initial
initial
of FE
FEFE
FE
initial FE models
FE
models
initial FE models
and
models
models
models
models
FE and
and
and
models and
and
the
and
the
the
the
and the results
the
results
the results
results
results
results
the of
ofof
of
results of model
of
model
results of model
model
model
model
of updating.
updating.
updating.
model updating.
updating.
updating.
updating.
model updating.
DDW
DDW DDW
Model Model DMW
Model DMW DMW Model TMW
Model
Model TMW TMWModel
Model Model
DDW
DDW
DDW
DDW DDW
DDW
Model
Model
DDW Model DMW
Model
Model
Model
Model DMW
DMW
DMW DMW
DMW
DMWModel
Model Model TMW
Model
Model
Model
Model TMW
TMW
TMW TMW
TMW
Model
TMWModel
Model Model
Model
Model
Model
Buildings 2023, 13, 437 9 of 18
Ini-Ini-
Ini- Ini- Ini-Ini-
Ini- Ini- Ini-Ini-
Ini- Ini-
Material
Material
Material
Material
Material
Material
Material
PropertiesIni-
Properties
Properties
Properties PropertiesIni-
Properties
Properties Ini-Ini-
Ini-
tial
tial tial
Updated
Updated
Updated
Ini-
UpdatedUpdated
Updated
Updated Ini-Ini-
Ini-
Ini-
Ini-
tial
tial tial
Updated
Updated
Updated
Ini- Updated
Updated
Updated
Updated Ini-Ini-
Ini-
Ini-
Ini-
tial
tial Ini- Updated
Updated
Updated
tialUpdated
Updated
Updated
Updated
Material
Material Properties
Material Properties
Properties tial
tial Updated
tial
tial Updated
Updated tial
tial Updated
tial
tial Updated
Updated tial
tial tial
tial Updated
Updated
Updated
tial
tialtial tial
tialtial tial
tialtial
Table 1. Elastic material properties of initial FE models and the results of model updating.
Modulus
Modulus
Modulus of elasticity,
ofofelasticity,
elasticity,
Modulus
Modulus
Modulus
Modulus
Modulus
Modulus of
of
Modulus of
of of elasticity,
of elasticity,
elasticity,
elasticity,
elasticity,
elasticity,
of elasticity, 200200 233
200 23323312,600 12,600
12,600 10,87310,873
10,873 14,40014,485
14,400
14,400 14,48514,485
DDW E (MPa)
E EModel
(MPa)
(MPa) 200200
200
200 200DMW
200
200 233
233
233
233 23312,600
233
Model
233 12,600 12,600
12,600
12,600
12,600
12,60010,873
10,87310,873
10,873
10,873
10,873
10,873 14,400 14,400
14,400
14,400
TMW
14,400
14,400 14,485
Model
14,400 14,485
14,485 14,485
14,485
14,485
14,485
E
EE E EE (MPa)
E
(MPa)
(MPa)
(MPa)
(MPa) (MPa)
(MPa)
Material Properties Initial Updated Initial Updated Initial Updated
Mass
Mass Mass
density
density density
Mass
Mass Mass
Mass
density
density density
density 2456
2456 2456 - - 2508 2508 - - 2446 2446 -- -
Modulus of elasticity,
200 MassMass density
density
Mass
(kg/m
(kg/m 3) 3density
(kg/m )
233 33) 2456
2456
2456 2456 --- -- --- 10,873
2456
12,600
24562456
2508
2508
2508
2508
25082508 --- -- 14,400
2508
2508
2446
--- 2446
2446
2446
24462446 14,485
2446
2446 -- -- ---
E (MPa) (kg/m
(kg/m
(kg/m
(kg/m (kg/m
(kg/m
3
(kg/m
3
))
3))33)) 33)
3

Mass density
2456 - 2508 - 2446 -
(kg/m3 ) The The
Thefirst first
firstmode
mode mode of vibration
ofofvibration
vibration ofofeach of each
each specimen
specimen
specimen was,
was, was,
as as expected,
asexpected,
expected, ananOOP an OOP
OOP mode.
mode. mode. Among
Among Among
The
TheThe
The The
The
first
first
first
The first
first
mode
mode
mode
first mode
first mode
mode
of
of
mode of ofvibration
of
vibration
vibration
of vibration
vibration
vibration
of vibrationof
ofof
ofeach
each ofeach
of
each
each
of each
specimen
specimen
specimen
each specimen
specimen
specimen
specimen was,
was,
was,
was, was,
was,
as
as as
was,as asexpected,
as
expected, expected,
expected,
expected,
expected,
as expected,an
an an
anOOP anOOP
an
OOP
OOP
OOP
an OOP
mode.
OOPmode.
mode.
mode. mode.
mode.
Among
mode. Among
Among
Among Among
Among
Among
these
thesespecimens,
these specimens,
specimens,the the
thehighest highest
highestfrequency frequency
frequencyvalue value
valuewas wasforwasforthe for
theTMWthe TMW
TMWspecimen. specimen.
specimen.This This
Thisresult result result
these
these
these
these these
these specimens,
specimens,
specimens,
specimens,
specimens,
thesespecimens,
specimens, the
thethe
the the
the
highest
highest
highest highest
highest
highest
the frequency
frequency
highest frequency
frequency
frequency
frequency
frequency value
value
value
value value
value
was
value was
waswas was
was
for
for
wasfor
for the
thefor
for
the
the
for the
the
TMW
TMW
TMWTMW
the TMW
TMW specimen. specimen.
specimen.
specimen.
specimen.
specimen.
specimen. This
This
This
This This
This
result
Thisresult
result
result result
result
result
agrees
The agrees
agrees with
with
first with
the
mode the
thegreater
greatergreatervalues
values
of vibration values
ofeach
ofof of
ofinitial initial
initial stiffness
specimen stiffness
stiffness obtained
was, obtained
obtained
asobtained ininTMW
expected, in
TMW TMW
an OOP OOP
OOPtest,
OOP test,
test,which
mode. which
Among which
isisde- is de-
de-
agrees
agrees
agrees agrees
agrees
with
with with
with
the
the the
the
greater
greater greater
greater
values
valuesvalues
values initial
of of
of
initialinitial
initial
stiffness stiffness
stiffness
stiffness obtained obtained
obtained in inTMWin
TMWin TMW
TMWOOPOOP OOP
OOP
test,
test, test,
test,
which
which which
which
is isde-
de-is de-
is de-
agrees
tailed
tailed
these inwith
agrees
tailed with
inthe
inSection
specimens,
tailed
thegreater
with
Section
in the
Section
greater
the
Section 3.3.
3.3.
highestThe
The values
greater
3.3.
3.3.
values
The
first
first two
frequency
The
ofofinitial
values
first
two
first
initial
two
modes
modes
twovalue
stiffness
of initial
modes
modes
stiffness
obtained
obtained
was for
obtained
stiffness
obtainedobtained
by
the
obtained
obtained
bythe
TMWthe
by
in
bydynamic
theinTMW
dynamic TMW
in TMW
dynamic
specimen.
the dynamic
OOP OOP
tests
tests test,
OOP
tests
and
This and
tests
test,
thewhich
the
result
and
which
test,
and the is
which
numerical
numerical
agrees
the
de-
is de-
is
numerical
numerical
de-
tailed
tailed tailed
tailed
tailed in
inin
in in
Section
SectionSection
Section
Section 3.3.
3.3.
3.3.
3.3. 3.3.
TheThe
TheThe The
first
first
first first
two
two
first two
two two
modes
modes
modes modesobtained obtained
obtained
obtained by
by bythe
the by
the the
dynamic
dynamic
dynamic dynamictests
tests
tests tests
and
andand and
the
thethe the
numerical numerical
numerical
numerical
with
tailed
models
models
models
the
inhave
have
have
greater
Section
been
been
values been 3.3.
compared
compared The
ofcompared
first
in
initialin in inmodes
two
Table
Table
stiffness 2.modes
Table2.The obtained
The
obtained
obtained
2.Modal
The
Modal Modal
in
by the
by
Assurance
Assurance
TMW
dynamic
the
Assurance dynamic tests
Criterion
Criterion
OOP Criterion
test, which
and
tests
Criterion (MAC)
(MAC) the
and
(MAC)
is detailed
numerical
isthe numerical
isusedisinused
used toto to
models
models
models
modelsmodels
models
have
have
have
models have have
have
been
been
been
havebeen been
been
compared
compared
been compared
compared
compared
compared
compared in in
inTable
Table in Table
in
Table
Table
in Table
2.
2.
Table The
2.
The
2. The2.Modal
2.
The
2. The
The
Modal
Modal
Modal
The Modal
Modal
Assurance
Modal Assurance
Assurance
Assurance
Assurance
Assurance
Assurance Criterion
Criterion
Criterion
Criterion
Criterion
Criterion(MAC)
(MAC)
(MAC)
(MAC) (MAC)
(MAC)
(MAC)is
is used
is
is is used
is
used
used
used
is used
to to to
toto
used to
to
check
check
Section check
thethe
3.3. the compliance
compliance
compliance
The first two of of of
mode
mode
modes mode shapes
shapes
obtained shapes
obtained
by obtained
obtained
the using
using
dynamic using
OMAOMA OMA
tests andand
and and
FEA
FEA
the FEA
results. results.
results.
numerical TheThe The
updated
updated
models updated
check
check
check check
check
the
thethe the
the compliance
compliance
compliance
compliance
compliance of
of ofmodeofmode
of
mode
mode mode
shapes
shapes
shapes shapes
shapes
obtained obtained
obtained
obtained using
using using
using
OMAOMA OMA
OMAandand and
and
FEA
FEA FEA
FEA
results.
results.results.
results.
TheThe The
The
updated
updated updated
updated
have
checkcheck
models
models
been
the
models
of of compliance
the
DMW ofcompliance
DMW
compared DMWandand
in
of
and
TMW
TMW
Table
mode
of
TMW mode
had
2. had
The
shapes
ahad
avery
Modal aobtained
shapes
very obtained
very
low lowobtained
low
mean
mean
Assurance
using
using
mean OMA
using OMA
frequency
frequency
Criterion
OMA
frequency andand FEA
and
error
error
(MAC)
FEA
error results.
(<1%)
(<1%) results.
FEA
is used inthe
to
The
inresults.
(<1%) The
incheck
the theupdated
The
first
first updated
firstupdated
mode,
mode,
the mode,
models
models
models
modelsmodels
models
of
of
models DMW
of
DMW
of of
of
DMW
DMW
of DMW
DMW
DMWand
andand
and and
and
TMW
TMW
TMWTMW
and TMW
TMW
had
had
TMW had
had a
ahad
had
a
hadvery
very
a a
verya
very
a very
very
low
low
verylow
low low
low
mean
mean
mean
lowmean mean
mean
frequency
mean frequency
frequency
frequency
frequency
frequency
frequency error
error
error
error error
error
(<1%)
(<1%)
(<1%)
(<1%)
error (<1%)
(<1%)
in
in
(<1%) in
inthe
the
the in
in
the
in the
the
first
first
first
first
the first
first
mode,
mode,
mode,
mode,
first mode,
mode,
mode,
and
complianceand
andmeanmean mean
MAC
of MAC
mode MAC values
values values
shapes ofof0.852of
obtained 0.852
0.852and and and
0.865
using 0.865
0.865for forthe
OMA for
theandthe
first
first first
and
FEA and
andsecond second
secondmodes
results. modes
The modes compared
compared
compared
updated models totothe to the
the
andand
and
and and
and
mean
mean
mean
mean
and mean
mean
MAC
MAC
meanMAC
MAC MAC
MAC
values
values
values
MAC values values
values
of
of
values 0.852
of
0.852
of of
of
0.852
0.852
of 0.852
0.852
and
andand
0.852and and
and
0.865
0.865
0.865
0.865
and 0.865
0.865
for
forfor
for
0.865 the
the for
for
the
the
for the
the
first
first
first
first
the first
first
and
andand
and
first and
and
second
second
second
andsecondsecond
second
modes
second modes
modes
modes modes
modes
compared
modes compared
compared
compared
compared
compared
compared to
to to
tothe
the to the
to
the
the
to the
the
tests.
oftests.
DMW tests.
tests.TheThe
and The
small
small
TMW small
difference
had difference
difference ininthe
a very in
theMAC
low the
MAC
mean MAC values
values values
frequencyover over
overthe theunit
error the
unit unit
value
value
(<1%) value
isisthe
in most is
most most
first likely
likely likely
mode, duedue due
andtothe
todue to the
the
tests.
tests.
tests.tests.
tests.
The
TheThe
The The
The
small
small
small
small small
small
differencedifference
difference
difference
difference
difference in
ininthe
the
in thein
in
the MAC
MACthe
the
MAC
MAC MAC
MAC
values
values
values
values values
values
over
over
overover over
over
the
thethe
the the
the
unit
unit
unitunit unit
unit
value
value
value
value value
value
is most
is
isthe
most
is is
is
most most
most
likely
likely
likely likely
likely
due
duedue due
to
to to the
the
the to
to the
the
tests.
heterogeneity
heterogeneity
mean MAC The
heterogeneity
values small
ininthe indifference
ofthe the
specimens,
specimens, in
specimens, thewhich
which MAC
which values
cannot
cannot cannot
bebe over the
besecond
considered unit
considered
considered value
inin inmost
the isthemostlikelylikely
numerical
numerical
numerical due dueto
models.
models. the
to the
models.
heterogeneity
heterogeneity
heterogeneity
heterogeneity
heterogeneity
heterogeneity
heterogeneity in
inin
inthe
the in0.852
in
the
the
in the
the and
specimens,
specimens,
specimens,
the
0.865
specimens,
specimens,
specimens,
specimens, which for
which
which
which which
which
which
the
cannot
cannot
cannot
first
cannot cannot
cannot
be
be
cannot
and
be
be be considered
be
consideredconsidered
considered
considered
considered
be considered
modes
in
inin
inthe
the incompared
in
the
the
in the
the
numerical numerical
numerical
numerical
numerical
numerical
the numerical
to the
models.
models.
models.
models. models.
models.
models.
These
These
tests. These
The highly highly
highly
small acceptable
acceptable
acceptable
difference inMAC
MAC theMAC values
values
MAC values show
show
values show
that
over that
thatthe the
thecalibrated
unit calibrated
calibratedvalue models
models
is mostmodelscancan can
simulate
likely simulate
simulate
due the
tothe the the
These
These
These
These These
These
highly
highly
highly
highlyhighly
highly
acceptableacceptable
acceptable
acceptable
acceptable
acceptable MAC
MAC
MACMAC MAC
MACvalues
values
values
values values
values
showshow
show
show show
show
that
that
that
that that
that
the
thethe
the the
the
calibrated calibrated
calibrated
calibrated
calibrated
calibrated models
models
models models
models
cancan
can can
can
simulate
simulate
simulate simulate
simulate the
the the
the
These
theexperimental
heterogeneity highly
experimental
experimental
experimental
modes
modes acceptable
modes
in themodes
quite MAC
quitesatisfactorily
quite
specimens,satisfactorily
quite
values
satisfactorily
whichand
satisfactorily
show
and
andvalidate
cannot that
validate
validate
and the
bevalidate the
themodels.
considered models.
the in the numerical models. thethe
calibrated
the models.models
models.
models can simulate
can simulate
experimental
experimental
experimental
experimental
experimental
experimental modes
modes
modes
modesmodes
modesquite
quite
quite
quitequite satisfactorily
satisfactorily
satisfactorily
satisfactorily
satisfactorily
quite satisfactorily and
andand
and and
validate
validate validate
validate
validate
and the
the
validate the
the the
models.
models. models.
models.
models.
the models.
These highly acceptable MAC values show that the calibrated models can simulate the
Table
Table Table
experimental2.2. 2. Comparison
Comparison
Comparison
modes ofof
quite of experimental
experimental
experimental
satisfactorily andand and
numerical
and numerical
numerical
validate mode
mode themode shapes.
shapes.
models. shapes.
TableTable
TableTable
Table
Table 2.
Table 2. Comparison
2. Comparison
2. Comparison
Table
2.2. 2. Comparison
Comparison
Comparison ofof
Comparison of
2. Comparison experimental
experimental
experimentalofexperimental
of
and
and
and
of experimentalexperimental
of experimental
of experimental
numerical
numerical
numerical
and mode
mode
mode
numerical shapes.
shapes.
shapes.
mode and numerical
and
and numerical
and
shapes. numerical
numericalmodemode mode
shapes.
mode shapes.
shapes.
shapes.
Mode
1 1 11 of experimental and numerical mode shapes.Mode
2.Mode
Mode
Comparison
Table Mode Mode
Mode2 2 22
Mode
Mode
Mode
Mode
Mode Mode
Mode
1 11 111 1 Mode
ModeMode
2 2 Mode
Mode
Mode
Mode2 22 22
Mode
Spec.
Spec.
Spec. Experimental Numerical
Experimental
Experimental Numerical
Numerical
Mode 1 Results Experimental
Results
Results Experimental Numerical
Experimental Numerical
Numerical
Mode 2 Results
Results
Results
Spec.
Spec.Spec.
Spec.
Spec.
Spec. ExperimentalNumerical
Experimental
Experimental
Experimental
Experimental
Spec. Experimental
Experimental Numerical
Numerical
Numerical
Numerical
Numerical
Numerical Results Experimental
Results
Results
Results
Results
Results
Results Experimental Numerical
Experimental
Experimental
Experimental
Experimental
Experimental Numerical
Numerical
Numerical
Numerical
Numerical
Numerical Results
Results
Results
Results
Results
Results
Results
Spec. Experimental Numerical Results Experimental Numerical Results

ff4.30
fE f=Ef=4.30 =Hz
EE = 4.30 Hz
Hz Hz
4.30 fE = 5.08 Hz
fEfE=f=E5.08
5.08
E = Hz
=f5.08 Hz Hz
Hz5.08
E = 4.30 Hz
DDW =ff=
fNf=NfN4.30 N=
N =4.30
4.30 4.30
Hz Hz
4.30 Hz
HzHz fN f= f
5.54 =
fN = 5.54 Hz Hz Hz
N =f 5.54
N
N = 5.54
Hz
5.54 Hz
DDW
DDW
DDW
DDW % Error: 0.00 % Error: 9.06
%% %Error:
Error:%Error: 0.00
Error: 0.00
0.000.00 %%Error:
% % 9.06
Error:
Error: 9.06
9.069.06
Error:
MAC:
MAC:MAC:
MAC:
MAC: 0.9540.954
0.9540.954
0.954 MAC:
MAC:
MAC: 0.904
MAC:
MAC:0.904 0.904
0.904
0.904

fEE== 29.38
f29.38
fE f=E f=29.38 29.38
Hz Hz
Hz Hz fEE ==Hz
f39.76
fEfE=f=E39.76
=39.76 39.76
39.76
Hz Hz
Hz Hz
E = 29.38 Hz
fNfN=fN ff
29.09
= 29.09
N =
= 29.09
Hz
= 29.09Hz
N HzHz
29.09 Hz f
fN =
N = ff
37.49
fN37.49 =
= 37.49
N N =Hz Hz Hz
37.49
37.49
Hz Hz
DMWDMW
DMW
DMW
DMW %
%%Error: %Error:
% Error:
Error: 0.99
Error: 0.99
0.990.99
0.99 %%Error:
%Error:%
Error:5.71
% Error:
Error:
5.71 5.71
5.715.71
MAC: MAC:MAC:
MAC: 0.813
0.813 0.813
0.813 MAC:
MAC: 0.905
MAC:
MAC: 0.905
0.9050.905
MAC: 0.813 MAC: 0.905

f33.91
E = 33.91
fE f=E f=33.91 Hz Hz fE = 51.86 Hz
f51.86
E = 51.86
Hz Hz
E Hz
E = 33.91 Hz fEfE=f=E51.86
=51.86 HzHz
fNfN=fN f
=34.01
N =
34.0134.01
Hz
= 34.01
N HzHz
Hz f
fNfN=fN=43.81 =
=43.81 43.81
43.81
N
N HzHzHz Hz
TMWTMW
TMW
TMW
%
%%Error: % Error:
Error:
Error: 0.29
0.29 0.29
0.29 %%Error:
%Error: % 15.52
Error: Error: 15.52
15.52
15.52
MAC:
MAC: MAC:
MAC: 0.789
0.789
0.789
0.789 MAC: MAC:
MAC: 0.786
MAC: 0.786
0.786
0.786

3.2.2. Nonlinear Model Calibration

The nonlinear behavior of the models was described by the Concrete Damaged Plas-
ticity (CDP) material model proposed by Lubliner et al. (1989) [43]. This material model,
which was originally developed to represent concrete behavior under cyclic or dynamic
loads, is also frequently used in studies that model the seismic behavior of historical ma-
sonry structures [32,33]. Material parameters assigned as defaults are given in Table 3. In
this table, ‘dilation angle’ is the angle that expresses the change in the volume of the element
under shear stresses; ‘σbo/σco’ refers to the ratio of the initial biaxial compressive strength
to the initial uniaxial compressive strength, and ‘K’ refers to the ratio between the secondary
stress constants in the tension and compression regions [45]. The mortar strengths were
Buildings 2023, 13, 437 10 of 18

directly taken from the tests by Misir et al. (2022) [15]. In the DMW specimen, the average
tensile (from the two-point bending test) and compressive strengths of the mortar on the
wall on test day were found to be 0.43 MPa and 2.10 MPa, respectively. For the TMW
specimen, these mortar strengths were found to be 0.31 MPa and 1.87 MPa, respectively.
Assigned compressive and tensile strengths are given in Table 4. The tensile strength was
taken as 5% of the compressive strength [18], while the compressive strength was taken
as the mean value of the mortar samples tested on the wall test day. The strain sequence
in the table was adopted from Valente and Milani (2019) [32] and the degrading values of
strengths were calibrated by a series of nonlinear static pushover analyses detailed in the
following section.

Table 3. The default parameters of CDP Material Model.

Dilation Angle Eccentricity σ bo /σ co K

10◦ 0.1 1.16 0.666

Table 4. Data points assigned for the nonlinear behavior of the CDP model.

Compressive Tensile
Yield Stress Inelastic Yield Stress Inelastic
(MPa) Strain (MPa) Strain
DMW TMW DMW TMW
2.0 1.8 0 0.105 0.095 0
2.1 1.9 0.005 0.075 0.063 0.00025
1.0 0.9 0.010 0.040 0.045 0.00057
0.5 0.5 0.031 0.012 0.036 0.00121

3.3. Nonlinear Static Pushover Analysis and the Comparison of Results with the OOP Tests
A series of nonlinear static pushover analyses in the OOP direction was performed
on the updated wall models. Analyses were performed using the implicit solver in the
ABAQUS software [42]. Two static analysis steps were defined based on the default fully
automatic time incrementation. Each analysis was conducted by first applying the self-
weight of the wall, followed by applying a four-point displacement load as in the tests.
The nonlinear equilibrium equations were solved by the full Newton technique. The
loading surfaces shown in Figure 9, representing the semi-rigid load distribution plates
as used in the OOP testing, were defined in four corresponding areas in the model to
enable the simulation of quasi-static loading over four synchronized points. Figure 10
shows the numerical load–drift response compared to the experimental envelopes of the
DMW and TMW specimens. The models showed results that were quite consistent with
the experimental findings. The narrow linear rising region in the elastic range shows that
both the specimens and models exhibited a rather brittle behavior. The initial stiffnesses
estimated by the updating of the DMW and TMW models using dynamic data were slightly
higher (16% and 4%, respectively) than the initial stiffnesses determined by the static tests
for this region. Since this is related to the elastic parameters of the models, the analysis
up to this stage allowed us to evaluate the compatibility of ambient vibration and the
elastic part of the static test results. It was determined in the nonlinear analysis that the
dominant model parameter affecting the load capacity was the tensile strength. For both
structural models, the load capacity was successfully achieved with an assigned tensile
strength of 5% of the compressive strength of the mortar. Numerical and experimental
results agree in the post-peak interval, but there are examples in the literature that the
numerical post-peak curve drops sharply than in the experiment [49,50]. However, the
lateral forces are calculated to be slightly larger for the calibrated models in this paper.
This result is appropriate since the strength degradation due to cyclic loads applied during
the tests lowers the envelope curve slightly below the monotonic pushover curve. Due to