Professional Documents
Culture Documents
Abstract: Reliable analytical and empirical models of the force-deformation parameters used to characterize the nonlinear behavior of
Downloaded from ascelibrary.org by CASA Institution Identity on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.
masonry panels are essential to simulating the seismic response of infilled reinforced concrete and steel frame systems. This paper presents
the development of empirical equations to predict the backbone curve parameters of infill panels modeled using equivalent struts. For this
purpose, a database of 264 infilled frame experiments is assembled from the existing literature. The experimental data from a subset of 113
specimens is used to calibrate the force-deformation parameters of the infill equivalent struts. Using the results from multivariate regression
analyses, empirical equations are proposed for the backbone curve parameters that define the axial response of the infill struts. Discussions
and recommendations for the cyclic degradation and pinching effect parameters are also presented. DOI: 10.1061/(ASCE)ST.1943-
541X.0002608. © 2020 American Society of Civil Engineers.
Author keywords: Masonry infilled frame; Database; Numerical modeling; Empirical model.
Number of Frame
and provide recommendations for the associated hysteretic param-
Reference specimens type Loading type eters. It is important to note that while the calibration was per-
formed using an implementation of the Lowes-Mitra-Altoontash
Abdul-Kadir (1974)a 12 Steel Monotonic pinching model in the Open System of Earthquake Engineering
Akhoundi et al. (2018)a 1 RC Quasi-static cyclic
(OpenSees) (Mckenna 1999), the backbone curve parameters can
Al-Chaar et al. (2002)a 4 RC Monotonic
Angel et al. (1994)a 7 RC Quasi-static cyclic also be used for nonpinching materials to represent the nonlinear
Anil and Altin (2007) 7 RC Quasi-static cyclic response of equivalent infill struts [e.g., the peak-oriented hyster-
Baran and Sevil (2010)a 3 RC Quasi-static cyclic etic model formulated by Burton and Deierlein (2014)]. Moreover,
Basha and Kaushik (2016)a 9 RC Quasi-static cyclic the cyclic degradation and pinching parameters can be used with
Bergami and Nuti (2015)a 2 RC Quasi-static cyclic the Lowes-Mitra-Altoontash model in non-OpenSees contexts.
Billington et al. (2009)a 1 RC Quasi-static cyclic
Blackard et al. (2009)a 4 RC Quasi-static cyclic
Bose and Rai (2014)a 1 RC Quasi-static cyclic Experimental Database
Calvi and Bolognini (2008)a 4 RC Quasi-static cyclic
Cavaleri and Di Trapani (2014)a 12 RC Quasi-static cyclic The entire experimental database consists of 257 one-bay one-story
Chiou and Hwang (2015)a 4 RC Quasi-static cyclic and 7 multi-bay one-story masonry infilled frames, collected from
Colangelo (2005)a 11 RC Pseudo-dynamic 49 journal publications, conference proceedings, and technical re-
Combescure et al. (1996)a 2 RC Quasi-static cyclic
ports. The sources of the database are listed in Table 1. The frame
Crisafulli (1997)a 2 RC Quasi-static cyclic
Da Porto et al. (2013) 6 RC Quasi-static cyclic types include both reinforced concrete and steel. The infill panels in
Dautaj et al. (2018)a 7 RC Quasi-static cyclic the database cover a range of different types of masonry units, re-
Dawe and Seah (1989)a 28 Steel Monotonic inforced and unreinforced masonry, panels with and without open-
Fiorato et al. (1970)a 7 RC Monotonic ings and with and without retrofit measures. A graphical summary
Flanagan and Bennett (1999) 8 Steel Quasi-static cyclic of the types of infilled frames in the database is provided in Fig. 1.
Gazic and Sigmund (2016)a 11 RC Quasi-static cyclic Since the focus of the current study is on RC infilled frames, the
Haider (1995)a 4 RC Quasi-static cyclic detailed characteristics of the infilled steel frames are omitted from
Kakaletsis and 6 RC Quasi-static cyclic Fig. 1. Table 2 presents the types of masonry units included in the
Karayannis (2008) database and the corresponding number of specimens. Approxi-
Khoshnoud and 2 RC Monotonic
mately 60% of the specimens are made from clay brick, slightly
Marsono (2016)a
Kumar et al. (2016)a 1 RC Quasi-static cyclic
Leuchars and Scrivener (1976)a 2 RC Quasi-static cyclic
Liu and Soon (2012) 10 Steel Monotonic
Mansouri et al. (2014)a 5 RC Quasi-static cyclic
Markulak et al. (2013) 6 Steel Quasi-static cyclic
Mehrabi et al. (1996)a 10 RC Monotonic/
Quasi-static cyclic
Misir et al. (2016)a 5 RC Quasi-static cyclic
Morandi et al. (2014)a 4 RC Quasi-static cyclic
Mosalam et al. (1997) 4 Steel Quasi-static cyclic
Pires et al. (1997)a 2 RC Quasi-static cyclic
Schwarz et al. (2015) 3 RC Quasi-static cyclic
Sigmund and Penava (2013)a 9 RC Quasi-static cyclic
Stylianidis (2012)a 5 RC Quasi-static cyclic
Tasnimi and Mohebkhah (2011) 5 Steel Quasi-static cyclic
Tawfik Essa et al. (2014)a 3 RC Quasi-static cyclic
Tizapa (2009)a 3 RC Quasi-static cyclic
Verderame et al. (2016)a 2 RC Quasi-static cyclic
Waly (2000)a 2 RC Quasi-static cyclic
Yorulmaz and Sozen (1968)a 7 RC Monotonic
Yuksel and Teymur (2011)a 2 RC Quasi-static cyclic
Zarnic and Tomazevic (1985) 3 RC Quasi-static cyclic
Zhai et al. (2016)a 3 RC Quasi-static cyclic
Zovkic et al. (2013) 3 RC Quasi-static cyclic
a Fig. 1. Summary of database structure.
Experiments used in calibration.
more than half of which are hollow. Concrete blocks, most of which
are solid, comprise approximately 30% of the specimens. Other less In addition, the relative stiffness parameter (λhcol ) proposed by
common types of brick (e.g., calcarenite, vitrified ceramic, cement Stafford Smith and Carter (1969) is also computed:
and pumice) are also included and there are nine autoclaved aerated
concrete specimens.
Em tw sinð2θÞ 0.25
For the purpose of the current study, the subset (134) of speci- λhcol ¼ · hcol ð1Þ
4Ef I col hw
mens that include only RC frames with unreinforced masonry panels
without openings or retrofit are first extracted. As one of the impor-
tant steps in this study is to calibrate the backbone modeling param- where θ ¼ tan−1 ðhw =lw Þ is the angle between the diagonal and hori-
eters of the axial response of the equivalent diagonal struts, explicit zontal of the panel, Ef = the modulus of elasticity of the frame, I col =
force-displacement curves from the tests are needed. However, some the moment of inertia of the column.
of the investigations did not report the force-displacement curves for The infilled frame failure mode classification scheme proposed
all tested specimens. After excluding those investigations that did not by Tempestti and Stavridis (2017) has been adopted for the current
include this information, 113 specimens remained and were used to study. In the original literature, the infilled frames are first classified
calibrate and develop empirical equations for the backbone curve into four categories based on two quantitative metrics, then each
parameters of the infill struts. category is assigned a distinct failure mode. For the current study,
Table 3 provides a statistical summary of the key structural var- because the failure observations in the experiments are available,
iables in the specimen subset used for subsequent calibrations. we did not use the quantitative metrics for classification, but instead
These variables are selected based on a review of prior literature assigned each specimen to a distinct failure mode based on
on the behavior of infilled frames. hcol is the column height from the available observations. Table 4 presents the adopted failure
the top of the base to the centerline of the beam, lbeam is the beam mode descriptions and the number of specimens corresponding
span to the centerline of the column, ld is the length of the diagonal to each one.
strut (measured based on the centerline dimension of the frame, as
is used in the structural models in subsequent sections), hw is the
height of the infill panel, lw is the length of the infill panel, hw =lw is Numerical Modeling and Calibration of the Infill
the aspect ratio of the infill panel, tw is the thickness of the infill Strut Parameters
panel, and P=Ag fc0 is the column axial load ratio. In most experi-
ments, the axial load was applied directly to the columns (to re- This section discusses the numerical modeling of infilled frames
present gravity loads). However, it is worth noting that in real using the Lowes-Mitra-Altoontash pinching model and details of
buildings some of the gravity load can be transmitted from the frame the process used to calibrate the parameters of the equivalent diago-
beams to the infill panels. ρt;col is the column transverse reinforce- nal struts. The single-strut model (in each direction) used for the
ment ratio, fm is the masonry prism strength, and Em is the masonry calibration process is consistent with the modeling framework pro-
modulus of elasticity. For consistency, the recorded values of fm and posed by Bose et al. (2018) and Stavridis et al. (2017), and adopted
Em are based on prism compression tests with loading applied in ASCE 41 (ASCE 2017). As discussed earlier, while this model-
perpendicular to the bed joint (some studies also performed the com- ing approach may not be able to precisely capture any localized
pression tests parallel to the bed-joints). In some cases where Em is failure modes, the total response of the masonry infill and the frame
not available, an empirical relationship, Em ¼ 550fm , as recom- can represent the global seismic response of the tested specimens.
mended by Kaushik et al. (2007) and FEMA 306, has been adopted. A schematic view of a one-bay one-story model is shown in Fig. 2.
Development of the Infilled Frame Model in OpenSees and cyclic degradations in the infill panel, and its effectiveness for
modeling the infill struts has been illustrated by Noh et al. (2017).
The numerical model of the infilled frame is developed in
OpenSees. However, it is worth noting that the overall approach A schematic response of the Lowes-Mitra-Altoontash model and
is applicable to other platforms (e.g., LS DYNA) that have similar associated parameters is shown in Fig. 3 (adapted from Lowes
implementations of the adopted material models. A concentrated et al. 2004). The parameters that are critical to the calibration pro-
plasticity model is used for the RC beams and columns, which con- cess described in the next subsection are presented in bold font. A
sists of elastic beam-column elements with two zero-length hinges response envelope, an unload-reload path and three damage rules
at the ends (Burton and Deierlein 2014; Haselton et al. 2008; define the hysteretic response in the Lowes-Mitra-Altoontash
Haselton et al. 2016; Noh et al. 2017), as shown in Fig. 2. The peak model. More specifically, eight points (16 parameters) are used
oriented hysteretic model developed by Ibarra et al. (2005) (also to construct a multilinear response envelope (shown as solid lines
denoted as the Ibarra-Medina-Krawinkler model) is used to model in Fig. 3). The unload-reload path (shown as dashed lines in Fig. 3)
the beam flexural hinges. As column failure is one of the primary is controlled by four points (six parameters). The unloading stiff-
concerns in the seismic performance of nonductile infilled frames, ness degradation, reloading stiffness degradation and strength deg-
it is critical that this mechanism is incorporated in the simulation. radation, are each controlled by five parameters, which predict the
Given this consideration, Burton and Deierlein (2014) used a flexu- damage index as functions of the displacement history and energy
ral spring with the Ibarra-Medina-Krawinkler model paired with a accumulation. More details of the model can be found in Lowes
shear spring with rigid-softening material to incorporate flexural et al. (2004).
hinging and column shear failure. Sattar (2014) used a flexural
spring with the Ibarra-Medina-Krawinkler model, a shear spring Calibration
with the failure model developed by Elwood (2004), and an axial
spring with the failure model also developed by Elwood (in series) The flexural hinge parameters of the reinforced concrete beams and
to capture the complex flexure-shear-axial interactions. In this columns are determined using the semi-empirical equations pro-
study, flexural and shear springs are modeled in series at the col- posed by Haselton et al. (2016), whereas the column shear hinge
umn ends (top and bottom), as shown in Fig. 2. The flexural spring parameters are obtained using the equations by Elwood (2004).
is modeled using the Ibarra-Medina-Krawinkler model and the Once the parameters of the reinforced concrete frame are deter-
shear spring is modeled with Elwood’s shear failure model. It is mined, those governing the axial response of the infill struts are
worth noting that this frame model can also capture column shear calibrated through an iterative procedure, where the lateral hyster-
and flexural failure in the spans with no infill walls. Therefore, the etic curve of the entire one-bay one-story infilled frame from the
same column modeling approach can be used for all the columns numerical simulation is compared with the experimental hysteretic
when modeling a real building. curve, and the parameters of the infill struts are adjusted until an
The infill panel is modeled as two diagonal struts (one in each acceptable visual match is achieved (Burton and Deierlein 2014),
direction) using a truss element with the Lowes-Mitra-Altoontash i.e., close visual resemblance in the response envelope, the pinch-
pinching material model. Various material models have been used ing effects, and the unloading-reloading slopes.
by different researchers to simulate the nonlinear behavior of infill A good fit (compared to the experimental results) is achieved
struts. Some examples include an early version of the Ibarra- using the Lowes-Mitra-Altoontash material model because of the
Medina-Krawinkler model (Burton and Deierlein 2014), the con- flexibility to control a wide range of parameters. However, a
crete material model (Mohd-Yassin 1994; Bose and Stavridis trade-off between precision and complexity in parameter defini-
2018), the Kent-Scott-Park model (Scott et al. 1982; Noh et al. tions must be considered when applying the model. For the pur-
2017), the Bouc-Wen hysteretic model (Bouc 1967; Noh et al. pose of this study, several simplifying assumptions are made to
2017), and the Lowes-Mitra-Altoontash model (Furtado et al. 2015; determine the parameters for the Lowes-Mitra-Altoontash model
Noh et al. 2017). In this study, the Lowes-Mitra-Altoontash model applied to the equivalent infill struts, while seeking to maintain an
is adopted because of its flexibility in controlling pinching effects adequate level of calibration accuracy. Some of the modeling
Fig. 3. Schematic response of the Lowes-Mitra-Altoontash model. (Adapted from Lowes et al. 2004.)
assumptions discussed by Noh et al. (2017) are also adopted. The calibration process. A simplified approach is adopted in this
details of the calibration process are as follows: study, which relates the strength and stiffness degradation to
1. The points defining the multilinear backbone curve of the strut the displacement history (Lowes et al. 2004). More specifically,
axial response are calibrated such that there is a close match gK 1 and gK 3 are used to define unloading stiffness degradation,
between the simulated and experimental lateral response envel- with gKLim setting the limit for the damage index (Lowes et al.
ope of the infilled frame. In the Lowes-Mitra-Altoontash model, 2004). Similarly, ðgD1 ; gD3 ; gDLimÞ and ðgF1 ; gF3 ; gFLimÞ
eight points define this envelope. However, as the infill panel is are used to control the reloading stiffness and strength degrada-
modeled by a compression-only strut in each diagonal direction, tion, respectively. The other six parameters ðgK 2 ; gK 4 ; gD2 ;
only the negative (compression) branch [ðeNd1 ; eNf 1 Þ, ðeNd2 ; gD4 ; gF2 ; gF4 Þ are set to zero.
eNf2 Þ, ðeNd3 ; eNf 3 Þ, ðeNd4 ; eNf4 Þ] needs to be calibrated. 4. The dmgType parameter, which is set to “Cycle,” relates the
The force level of the positive (tension) branch is taken to be damage indices to the displacement history. gE is taken as
1% of that in the compression branch (Noh et al. 2017), result- 10 for all specimens (Noh et al. 2017).
ing in a response envelope similar to the constitutive model de- The calibration process follows the sequence of steps 1, 2, and 3
veloped by Cavaleri and Di Trapani (2014). After calibrating a discussed previously, i.e., first the response envelope is adjusted,
certain number of specimens, it was discovered that specifying then the pinching shape is addressed, and finally the cyclic degra-
another point between the yield and capping points does not dation parameters are determined. Table 5 lists the parameters in the
significantly influence the calibration accuracy (as measured Lowes-Mitra-Altoontash model that are used in the calibration, in-
by the dissipated hysteretic energy). Therefore, ðeNd2 ; eNf2 Þ cluding a description of how each parameter is used. Also note that
is assigned by linear interpolation after obtaining ðeNd1 ; eNf1 Þ for the calibration, the strut area and other infill parameters are not
and ðeNd3 ; eNf3 Þ. As a result, this step can be simplified to pre-determined using previously developed equations. Instead, a
calibrating three points: the yield point ðeNd1 ; eNf1 Þ, the cap- unit cross section area is used for the strut such that the calibrated
ping point ðeNd3 ; eNf3 Þ, and the point where the residual stress parameters (for example, eNf 1 ) are equal to the actual strut
strength is achieved ðeNd4 ; eNf 4 Þ. force values.
2. The unload-reload path parameters are adjusted to match the A calibration example for specimen TA2 of the experimental
pinching shape as exhibited in the hysteretic response measured investigation by Morandi et al. (2014) is illustrated in Fig. 4.
during the experiment. The Lowes-Mitra-Altoontash model uses The experimental hysteretic curve of the infilled frame is shown
six parameters to control pinching effects (through the points de- with the solid line in Fig. 4(a), while the numerical simulation result
fining the unload-reload path). In the current study, four para- is represented by the dashed line. A reasonable match between the
meters are calibrated: uForceN which determines the strength simulated and experimental hysteretic curve in terms of the overall
at the start of unloading from the tension branch, rDispN and response envelope, pinching effect and cyclic degradation, is ob-
rForce which define the point at which reloading begins, and served. A further comparison is made between the cumulative en-
uForceP, which is used to match the strength at which unloading ergy dissipation of the experimental and simulated response, as
from the compression branch begins. As the tension branch is illustrated in Fig. 4(b), where the horizontal axis is the displace-
negligible, rDispP and rForceP are not influential and values ment level of the test and the vertical axis is the cumulative energy.
of rDispP ¼ 0.2 and rForceP ¼ 0.5 are assumed. A reasonable visual match is achieved between the experimental
3. The Lowes-Mitra-Altoontash model uses 15 parameters to con- and simulated results.
trol the strength, unloading stiffness and reloading stiffness Fig. 5 further presents four calibration results based on experi-
degradation, which significantly adds to the complexity of the ments by Cavaleri and Di Trapani (2014). Specimen S1B-2 is used
Fig. 4. Calibration result for specimen TA2 in Morandi et al. (2014): (a) hysteretic response; and (b) cumulative energy dissipation.
as an example to illustrate the process for determining the transition For experiments with available digital force-deformation data, a
point between the negative post-peak stiffness and the residual cumulative energy dissipation comparison is also conducted for
strength ðeNd4; eNf4Þ [the circled point in Fig. 5(a)]. This point further validation.
is estimated by observing the start of the stabilizing portion of the
hysteretic response, where the load remains approximately constant
with increased displacement. Once the residual strength transition Empirical Model for the Backbone Curve Parameters
point is identified, the corresponding deformation is converted to of the Equivalent Infill Struts
the diagonal direction and used as eNd4 (needs to be divided by ld
to obtain strain) in the Lowes-Mitra-Altoontash model. eNf4, This section presents the statistical analysis results for the cali-
which specifies the strut residual strength, is then adjusted until the brated model parameters that define the axial force-displacement
simulated hysteretic response of the entire infilled frame reasonably relationship of the infill struts. The main focus of this paper is
matches the experimental response, as shown in the calibration re- calibrating and predicting the force-displacement parameters that
sult in Fig. 5(a). define the backbone curve, which can be used to model the infill
The calibration process is performed on the data set of struts using pinching materials such as the one by Lowes-Mitra-
113 specimens until an adequate visual match between experimen- Altoontash and others (e.g., peak-oriented models like the Ibarra-
tal and simulated response is achieved for each experiment. Medina-Krawinkler model). These parameters are presented here as
Fig. 5. Calibration results for specimens: (a) S1B-2; (b) S1A-1; (c) S1C-1; and (d) S1C-3 of the Cavaleri and Di Trapani (2014) experiments.
Fig. 6. Axial force-deformation response of the infill strut model. these backbone parameters and those that define the Lowes-Mitra-
Altoontash model. An empirical equation is established to relate each
backbone parameter to several geometric and material properties of
the initial stiffness (K e ), the yield strength (Fy ), the capping strength the infilled frame. No regression analysis is performed for the pinch-
(Fc ), the deformation corresponding to the capping strength (dc ), ing and cyclic degradation parameters. However, recommendations
the post-capping stiffness (K pc ), and the residual strength (Fres ), for the appropriate range of the parameter values are presented later
as shown in Fig. 6. Table 6 provides the relationship between in this section.
Fig. 7. Trends between K e and (a) Em ; (b) fm ; (c) hw =lw ; (d) tw ; (e) ld ; and (f) λhcol .
Table 7. Summary of regression the results for the proposed empirical Table 9. Stepwise elimination based on the partial F-test (round 2)
model including all six parameters
Predictors F statistic P value
Standard
lnðλhcol Þ 1.44 0.23
Predictors Coefficient error t value P value
lnðld Þ 3.87 0.05
lnðtw Þ 1.13 0.24 4.63 1 × 10−5 lnðtw Þ 31.03 1.91 × 10−7
lnðEm Þ 0.72 0.15 4.90 3.44 × 10−6 lnðEm Þ 42.52 2.37 × 10−9
lnðhw =lw Þ −1.16 0.30 −3.86 1.98 × 10−4 lnðhw =lw Þ 15.64 1.38 × 10−4
lnðfm Þ −0.06 0.14 −0.46 0.65
lnðld Þ −0.53 0.31 −1.70 0.09
lnðλhcol Þ −0.34 0.28 −1.23 0.22 Table 10. Stepwise elimination based on the partial F-test (round 3)
Intercept −2.49 2.05 −1.22 0.23
Predictors F statistic P value
lnðld Þ 3.29 0.07
Table 8. Stepwise elimination based on the partial F-test (round 1) lnðtw Þ 30.19 2.64 × 10−7
lnðEm Þ 59.67 6.11 × 10−12
Predictors F statistic P value
lnðhw =lw Þ 18.25 4.19 × 10−5
ln ðfm Þ 0.21 0.65
lnðλhcol Þ 1.52 0.22
lnðld Þ 2.89 0.09 version of Eq. (7) based on the current data set. The much lower R2
lnðtw Þ 21.48 1.03 × 10−5 value obtained for we =ld (0.25) compared to K e (0.54) [based on
lnðEm Þ 24.03 3.44 × 10−6 Eq. (5)] is worth noting and may influence the decision regarding
lnðhw =lw Þ 14.87 1.98 × 10−4
which of the two formulations [Eq. (5) or (6)] to use.
Fc ¼ 0.003766f0.196
m t0.867
w l0.792
d
Fig. 9. Trends between Fc and (a) Em ; (b) fm ; (c) hw =lw ; (d) tw ; (e) ld ; and (f) λhcol .
Fig. 10. Relationship between Fy and Fc . increases, the entire frame tends to have a more ductile response,
and the infill panel often crushes because of compression at large
drifts instead of sliding shear, which tends to occur at smaller drift
demands. As a result, the displacement at the point where the
strength starts to degrade tends to be larger.
K pc
Fig. 11. Relationship between Fres and Fc . ¼ −1.278f −0.357
m tw −0.517
Ke
R2 ¼ 0.32; σ̂ ¼ 0.46 ð14Þ
Empirical Equation for the “Yield” Strength F y where fm and tw are in units of MPa and mm, respectively.
It is worth noting that the R2 values for Eqs. (13) and (14)
Fig. 10 shows a scatterplot of Fy against Fc . A strong association are relatively low. This means that the variance of the response var-
between Fy and Fc is observed. Therefore, Fy can be simply com- iable is not fully captured by the adopted predictors. Nevertheless,
puted as a proportion of Fc after the latter is determined. After the statistical significance indicates that the predictors provide
regressing Fy against Fc , a simple equation is obtained valuable information about the response variable. More impor-
tantly, the computed coefficients can reasonably capture the rela-
Fy ¼ 0.72Fc R2 ¼ 0.98; σ̂ ¼ 0.13 ð11Þ
tionships between the response variable and the adopted predictors.
Eq. (11) provides a simplified approach to estimating the yield Fig. 14 presents comparisons between the parameter values pre-
point in the backbone curve using Fc , which is consistent with pre- dicted using the developed empirical equations and the original
viously developed models (e.g., Stavridis et al. 2017; Dolšek and calibrated parameters after removing the outlier datapoints. The
Fajfar 2008). mean and median of the ratios of the predicted to calibrated param-
eter values are listed in Table 11. The ranges of the mean and
median ratios are 0.95–1.14 and 0.84–1.08, respectively, indicating
Empirical Equation for the Residual Strength F r es that the empirical models predict the backbone parameters with
The analysis for Fres is based on 75 experimental observations reasonable accuracy.
where the residual strength is reached as indicated by the load-
displacement curves. Fig. 11 shows a strong trend between the cali- Cyclic Degradation and Pinching Parameters
brated Fres and Fc . Therefore, similar to Fy , Eq. (12) provides a
simplified approach to estimating Fres using Fc . This section discusses the calibration results for the parameters that
The obtained empirical equation is given by control the cyclic degradation and pinching effects which are not
included in the regression analysis. Although the material models
Fres ¼ 0.4Fc R2 ¼ 0.81; σ^ ¼ 0.41 ð12Þ provide well-defined rules to simulate the response incorporating
the degradation effects, the influence of the structural characteris-
tics on the degradation parameters could not be extracted from the
Empirical Equation for Deformation at
experimental data and is still not well understood. More in-depth
Peak Strength d c
studies are necessary in this respect. Nevertheless, the statistical
This subsection presents the analysis of the axial deformation of values of these degradation parameters are used to recommend
the infill strut at peak strength (dc ) normalized by the length of the appropriate ranges for use in numerical modeling.
Fig. 12. Trends between dc =ld and (a) Em ; (b) fm ; (c) hw =lw ; (d) tw ; and (e) λhcol .
As discussed earlier, this study assumes gK 2 , gK 4 , gD2 , gD4 , as discussed in the calibration process. The ranges of rDispN,
gF2 , and gF4 to be zero. Three sets of parameters: ðgK 1 ; gK 3 ; rForceN, and uForceN are 0.02–0.3, 0.02–0.4, and 0–0.3,
gKLimÞ, ðgD1 ; gD3 ; gDLimÞ, and ðgF1 ; gF3 ; gFLimÞ, are used respectively.
to relate cyclic unloading stiffness degradation, reloading stiffness
degradation, and strength degradation to the displacement history.
Table 12 presents the statistical summary of the calibrated cyclic Conclusion
degradation parameters. In general, using values close to the mean
The main purpose of this study is to characterize the nonlinear
in Table 12 provides an approximate match between the simulated
force-deformation parameters of infill panels modeled using equiv-
and experimental hysteretic curves. A more precise match requires
alent diagonal struts, which is essential to performing reliable sim-
further adjustments.
ulations of the nonlinear response of infilled frame systems subject
Table 13 presents the statistical summary for the parameters that
to seismic loads and performance-based assessments.
control the pinching effect. rDispP and rForceP are assumed to An experimental database of infilled frames was assembled as
be 0.2 and 0.5, respectively, as discussed earlier. The pinching part of the study. The entire database consists of 264 infilled frame
parameters vary significantly across different specimens. Therefore, specimens subjected to monotonic, quasi-static, and pseudo-
a generalized set of values might not provide a reliable approxima- dynamic loading, among which 191 specimens have reinforced
tion. However, the ranges of values presented in Table 13 can provide concrete frames and 73 specimens have steel frames. To achieve
some suggestions for setting these parameters. uForceP is usually consistency in the generalized modeling assumptions, 113 rein-
assigned a negative value. The larger magnitude of this parameter is forced concrete infilled frame specimens are used to calibrate
because the actual pinching point is adjusted by uForceP · ePf3, and develop empirical equations for the infill strut backbone curve
while ePf 3 is a tension branch parameter that is set to be small, parameters.
Fig. 13. Trends between K pc =K e and (a) Em ; (b) f m ; (c) hw =lw ; (d) tw (e) ld ; and (f) λhcol .
The calibration of the strut model parameters is based on the in the nonlinear analysis models of multi-bay multi-story masonry
pinching material model proposed by Lowes et al. (2004). An iter- infilled reinforced concrete frame systems for limit state (e.g., col-
ative procedure is used such that the simulated lateral hysteretic re- lapse, demolition) assessments.
sponse curve of the entire one-bay one-story infilled frame achieves a There are several unavoidable limitations associated with the
good visual match with the values reported in the experiment. current study. Although the collected database covers a variety
Based on the calibration results, multivariate regression analyses of masonry-infilled frame tests, the primary scope of this study
are performed to develop empirical predictive equations for the is on reinforced concrete frames with unreinforced solid infill pan-
backbone curve parameters of equivalent diagonal struts, including els, which limits the number of specimens that can be utilized in the
the initial stiffness (K e ), the yield strength (Fy ), the capping calibration and the applicability of the empirical equations. Further
strength (Fc ), the deformation at capping strength normalized by research efforts incorporating specimens with other characteristics,
the strut length (dc =ld ), the post-capping stiffness (K pc ), and the such as masonry panels with openings and masonry infilled steel
residual strength (Fres ). The predictor variables include several frames, are needed to support the development of more generalized
geometric and material properties of the infill panels, which are and comprehensive empirical models. In addition, because of the
easily obtained from design or experimental documentation. The differences in the objectives and details provided among the pub-
prediction errors are quantified to inform the uncertainties associ- lications used to assemble the database, several structural properties
ated with using the empirical equations. Statistical information for having potential influence on the nonlinear behavior of the infill
the parameters controlling pinching effects and cyclic degradation panels could not be collected for many of the test specimens, which
is also presented and discussed. Despite using a pinching material reduces the range of available predictors. The calibration process is
as the basis of the calibration, the backbone curve parameters ob- based on a simplified numerical model that is able to capture the
tained from the empirical equations can be used for modeling infill global seismic response of the infilled frame. The accuracy of the
struts with non-pinching materials (e.g., the peak-oriented hysteretic calibration could be further improved by using more complex
material model). Subsequently, the strut model can be incorporated numerical models for the frame and infill. Also, the developed
Fig. 14. Calibrated versus predicted response parameters: (a) K e ; (b) Fc ; (c) Fy ; (d) Fres ; (e) dc =ld ; and (f) K pc =K e .
Table 11. Mean and median of the ratio of the predicted to the calibrated Table 12. Statistical summary of the cyclic degradation parameters
parameters
Summary gK 1 gK 3 gKLim gD1 gD3 gDLim gF1 gF3 gFLim
Statistics Ke Fc Fy Fres dc =ld K pc =K e
Min 0.2 0.05 0.10 0.18 0.10 0.06 0.00 0.00 0.00
Mean 1.14 1.06 0.99 0.95 1.11 1.11 Max 0.9 0.50 0.95 1.50 1.50 0.70 1.00 1.50 0.30
Median 1.08 1.03 0.97 0.84 1.00 0.97 Median 0.75 0.20 0.95 0.50 0.20 0.25 0.80 0.90 0.20
Mean 0.69 0.22 0.92 0.57 0.36 0.26 0.75 0.98 0.18
COVa 0.26 0.49 0.13 0.34 0.87 0.36 0.19 0.28 0.26
a
Coefficient of variation.
empirical equations are for the backbone curve parameters of the
infill struts. The parameters associated with pinching effects and
cyclic degradation have not been examined in detail. More in-depth models was evaluated using the training data (i.e., the same data
studies are needed to improve the characterization of these param- used to develop the model). Ideally, the model performance should
eters. Finally, because of the small number (relatively speaking) of be evaluated using a data set that is different from the one used in its
available experiments, the predictive capability of the empirical development (i.e., testing data).
.1007/s40091-015-0086-5. under earthquake loading.” J. Struct. Eng. 137 (11): 1291–1302. https://
Gazic, G., and V. Sigmund. 2016. “Cyclic testing of single-span weak doi.org/10.1061/(ASCE)ST.1943-541X.0000376.
frames with masonry infill.” Grad evinar 68 (8): 617–633. Liu, Y., and S. Soon. 2012. “Experimental study of concrete masonry infills
Haider, S. 1995. “In-plane cyclic response of reinforced concrete frames bounded by steel frames.” Can. J. Civ. Eng. 39 (2): 180–190. https://doi
with unreinforced masonry infills.” Ph.D. dissertation, Dept. of Civil .org/10.1139/l11-122.
and Environmental Engineering, Rice Univ. Lowes, L., N. Mitra, and A. Altoontash. 2004. A beam-column joint model
Haselton, C. B., A. B. Liel, S. L. Lange, and G. Deierlein. 2008. Beam- for simulating the earthquake response of reinforced concrete frames.
column element model calibrated for predicting flexural response lead- PEER Rep. No. 2003/10. Berkeley, CA: Pacific Earthquake Engineer-
ing to global collapse of RC frame buildings. Berkeley, CA: Pacific ing Research Center, Univ. of California.
Earthquake Engineering Research Center. Mainstone, R. J. 1971. “On the stiffness and strengths of infilled frame.”
Haselton, C. B., A. B. Liel, S. C. Taylor-Lange, and G. G. Deierlein. 2016. Supplement, Proc. Inst. Civ. Eng. (S4): 57–90.
“Calibration of model to simulate response of reinforced concrete Mainstone, R. J., and G. A. Weeks. 1974. “The influence of a bounding
beam-columns to collapse.” ACI Struct. J. 113 (6): 1141–1152. frame on the racking stiffness and strengths of brick walls.” In Proc.,
https://doi.org/10.14359/51689245. 2nd Int. Brick Masonry Conf. Stoke-on-Trent, UK: British Ceramic
Hocking, R. R. 1976. “A biometrics invited paper. The analysis and selec- Research Association.
tion of variables in linear regression.” Biometrics 32 (1): 1–49. https:// Mansouri, A., M. S. Marefat, and M. Khanmohammadi. 2014. “Experimen-
doi.org/10.2307/2529336. tal evaluation of seismic performance of low-shear strength masonry in-
Holmes, M. 1961. “Steel frames with brickwork and concrete infilling.” fills with openings in reinforced concrete frames with deficient seismic
Proc. Inst. Civ. Eng. 19 (4): 473–478. https://doi.org/10.1680/iicep details.” Struct. Des. Tall Special Build. 23 (15): 1190–1210. https://doi
.1961.11305. .org/10.1002/tal.1115.
Ibarra, L. F., R. A. Medina, and H. Krawinkler. 2005. “Hysteretic models Markulak, D., I. Radić, and V. Sigmund. 2013. “Cyclic testing of single bay
that incorporate strength and stiffness deterioration.” Earthquake Eng. steel frames with various types of masonry infill.” Eng. Struct. 51 (Jun):
Struct. Dyn. 34 (12): 1489–1511. https://doi.org/10.1002/eqe.495. 267–277. https://doi.org/10.1016/j.engstruct.2013.01.026.
Jeon, J.-S., A. Shafieezadeh, and R. Desroches. 2014. “Statistical models Mckenna, F. T. 1999. “Object-oriented finite element programming: Frame-
for shear strength of RC beam-column joints using machine-learning works for analysis, algorithms and parallel computing.” Ph.D. dissertation,
techniques.” Earthquake Eng. Struct. Dyn. 43 (14): 2075–2095. Dept. of Civil and Environmental Engineering, Univ. of California,
https://doi.org/10.1002/eqe.2437. Berkerley.
Kakaletsis, D. J., and C. G. Karayannis. 2008. “Influence of masonry Mehrabi, A. B., P. Benson Shing, M. P. Schuller, and J. L. Noland. 1996.
strength and openings on infilled R/C frames under cycling loading.” “Experimental evaluation of masonry-infilled RC frames.” J. Struct.
J. Earthquake Eng. 12 (2): 197–221. https://doi.org/10.1080/13632 Eng. 122 (3): 228–237. https://doi.org/10.1061/(ASCE)0733-9445
460701299138. (1996)122:3(228).
Kaushik, H. B., D. C. Rai, and S. K. Jain. 2007. “Stress-strain character- Misir, I. S., O. Ozcelik, S. C. Girgin, and U. Yucel. 2016. “The behavior
istics of clay brick masonry under uniaxial compression.” J. Mater. Civ. of infill walls in RC frames under combined bidirectional loading.”
Eng. 19 (9): 728–739. https://doi.org/10.1061/(ASCE)0899-1561 J. Earthquake Eng. 20 (4): 559–586. https://doi.org/10.1080/13632469
(2007)19:9(728). .2015.1104748.
Khoshnoud, H. R., and K. Marsono. 2016. “Experimental study of masonry Mohd-Yassin, M. H. 1994. “Nonlinear analysis of prestressed concrete
infill reinforced concrete frames with and without corner openings.” structures under monotonic and cycling loads.” Ph.D. dissertation,
Struct. Eng. Mech. 57 (4): 641–656. https://doi.org/10.12989/sem Dept. of Civil and Environmental Engineering, Univ. of California,
.2016.57.4.641. Berkerley.
Klingner, R. E., and V. V. Bertero. 1978. “Earthquake resistance of infilled Morandi, P., S. Hak, and G. Magenes. 2014. “In-plane experimental re-
frames.” J. Struct. Div. 104 (6): 973–989. sponse of strong masonry infills.” In Proc., 9th Int. Masonry Conf.
Koliou, M., and A. Filiatrault. 2017. “Development of wood and steel Whyteleafe, UK: International Masonry Society.
diaphragm hysteretic connector database for performance-based earth- Mosalam, K., R. N. White, and P. Gergely. 1997. “Static response of infilled
quake engineering.” Bull. Earthquake Eng. 15 (10): 4319–4347. https:// frames using quasi-static experimentation.” J. Struct. Eng. 123 (11):
doi.org/10.1007/s10518-017-0141-7. 1462–4169. https://doi.org/10.1061/(ASCE)0733-9445(1997)123:11
Kumar, M., M. Haider, and S. H. Lodi. 2016. “Response of low-quality (1462).
solid concrete block infilled frames.” Proc. Inst. Civ. Eng. Struct. Build. Noh, N. M., L. Liberatore, F. Mollaioli, and S. Tesfamariam. 2017.
169 (9): 669–687. https://doi.org/10.1680/jstbu.15.00068. “Modelling of masonry infilled RC frames subjected to cyclic loads:
Leuchars, J. M., and J. C. Scrivener. 1976. “Masonry infill panels subjected State of the art review and modelling with OpenSees.” Eng. Struct.
to cyclic in-plane loading.” Bull. N. Z. Nat. Soc. Earthquake Eng. 9 (2): 150: 599–621. https://doi.org/10.1016/j.engstruct.2017.07.002.
122–131. Panagiotakos, T. B., and M. N. Fardis. 1996. “Seismic response of infilled
Liberatore, L. 2001. “Approcci innovativi in termini di energia e di spos- RC frames structures.” In Proc., 11th World Conf. on Earthquake
tamento per la valutazione della risposta sismica di strutture a più gradi Engineering. Oxford, UK: Pergamon.
di libertà.” [In Italian.] Ph.D. Dissertation, Dept. of Structural and Paulay, T., and M. J. N. Priestley. 2009. Seismic design of reinforced
Geotechnical Engineering, Univ. of Rome. concrete and masonry buildings. New York: Wiley.
characteristics on seismic collapse of concrete frame buildings.” Ph.D. Environnement-Structures, Université Paris-Est.
dissertation, Dept. of Civil, Environmental, and Architectural Engineer- Verderame, G., P. Ricci, C. Del Gaudio, and M. De Risi. 2016. “Experi-
ing, Univ. of Colorado at Boulder. mental tests on masonry infilled gravityand seismic-load designed RC
Sattar, S., and A. B. Liel. 2016. “Seismic performance of nonductile rein- frames.” In Proc., 16th Int. Brick and Block Masonry Conf. London:
forced concrete frames with masonry infill walls. I: Development of a CRC Press.
strut model enhanced by finite element models.” Earthquake Spectra Waly, A. A. 2000. “Experimental and analytical work on the seismic per-
32 (2): 795–818. https://doi.org/10.1193/90914EQS139M. formance of different types of masonry infilled reinforced concrete
Schwarz, S., A. Hanaor, and D. Z. Yankelevsky. 2015. “Experimental re- frames under cyclic loading.” Master’s thesis, Graduate School of
sponse of reinforced concrete frames with AAC masonry infill walls to Natural and Applied Sciences, Dokuz Eylül Univ.
in-plane cyclic loading.” Structures 3 (Aug): 306–319. https://doi.org Yorulmaz, M., and M. Sozen. 1968. Behavior of single-story reinforced
/10.1016/j.istruc.2015.06.005. concrete frames with filler walls. Civil Engineering Studies SRS-337.
Scott, B. D., R. Park, and M. J. Priestley. 1982. “Stress-strain behavior of Champaign, IL: Univ. of Illinois Engineering Experiment Station,
concrete confined by overlapping hoops at low and high strain rates.” College of Engineering, Univ. of Illinois at Urbana-Champaign.
Ame. Concr. Inst. J. 79 (1): 13–27. Yuksel, E., and P. Teymur. 2011. “Earthquake performance improvement of
Sigmund, V., and D. Penava. 2013. “Influence of openings, with and with- low rise RC buildings using high strength clay brick walls.” Bull. Earth-
out confinement, on cyclic response of infilled R-C frames: An exper- quake Eng. 9 (4): 1157–1181. https://doi.org/10.1007/s10518-010-9242-2.
imental study.” J. Earthquake Eng. 18 (1): 113–146. https://doi.org/10 Zarnic, R., and M. Tomazevic. 1985. “Study of the behaviour of masonry
.1080/13632469.2013.817362. infilled reinforced concrete frames subjected to seismic loading.” In
Stafford Smith, B., and C. Carter. 1969. “A method of analysis for infilled Proc., 7th Int. Conf. on Brick Masonry. Melbourne: Brick Development
frames.” Proc. Inst. Civ. Eng. 44 (1): 31–48. https://doi.org/10.1680 Research Institute.
/iicep.1969.7290. Zhai, C., J. Kong, X. Wang, and Z. Chen. 2016. “Experimental and finite
Stavridis, A., J. Martin, and S. Bose. 2017. “Updating the ASCE 41 pro- element nalytical Investigation of seismic behavior of full-scale ma-
visions for infilled RC frames.” In Proc., 2017 Structural Engineers sonry infilled RC frames.” J. Earthquake Eng. 20 (7): 1171–1198.
Association of California (SEAOC) Convention. Sacramento, CA: https://doi.org/10.1080/13632469.2016.1138171.
Structural Engineers Association of California. Zovkic, J., V. Sigmund, and I. Guljas. 2013. “Cyclic testing of a single bay
Stylianidis, K. 2012. “Experimental investigation of masonry infilled R/C reinforced concrete frames with various types of masonry infill.” Earth-
frames.” Open Constr. Build. Technol. J. 6 (1): 194–212. https://doi.org quake Eng. Struct. Dyn. 42 (8): 1131–1149. https://doi.org/10.1002/eqe
/10.2174/1874836801206010194. .2263.