Engineering Structures 175 (2018) 861–878

Contents lists available at ScienceDirect

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

Effect of URM infills on inelastic floor response of RC frame buildings T

a b a,⁎ c
Mitesh Surana , Mayur Pisode , Yogendra Singh , Dominik H. Lang
a
Department of Earthquake Engineering, Indian Institute of Technology Roorkee, Roorkee, Uttarakhand 247667, India
b
Thornton Tomasetti, Mumbai 400013, India
c
Natural Hazards Division, Norwegian Geotechnical Institute, Sognsveien 70, 0855 Oslo, Norway

A R T I C LE I N FO A B S T R A C T

Keywords: The effect of unreinforced masonry infills on the floor acceleration response of inelastically responding mid-rise
URM infills RC frame buildings is studied using incremental dynamic analyses. Two structural building configurations,
RC frame buildings uniformly infilled, and with open ground storey are analyzed by using the FEMA P695 far-field ground-motion
Non-structural components suite. It is observed that the effect of structural nonlinearity is much more pronounced in case of infilled RC
Floor spectrum
frame buildings than bare RC frame buildings. Sequential failure of infills results in a significant elongation of
Spectral amplification
period of vibration resulting in the shifting of peaks in floor spectrum towards longer periods. Modified floor
spectral amplification functions are presented for both uniformly infilled and open ground storey RC frame
buildings. The proposed spectral amplification functions are validated using nonlinear analysis of typical
buildings for recorded as well as spectrum-compatible ground-motions and can be used to estimate the floor
response spectra directly from the code-based or site-specific design spectra.

1. Introduction
The assessment of floor acceleration demands is a crucial task in the
framework of performance-based seismic design of non-structural
components (NSCs). To this day, a significant number of attempts have
been made to study the floor acceleration demands in RC bare frame
buildings [1–8]. The crucial parameters affecting floor accelerations
have already been identified including the frequency content of the
ground motion [9–12], the dynamic characteristics of the supporting
structure (the “building structure” throughout this article is referred as
the “supporting structure”), the level of nonlinearity (inelasticity) of the
supporting structure [12–27], and both the period and damping of the
NSC [2–6,9,12,21]. Amplification of Peak Floor Acceleration (PFA)
along the height of the building has been identified to be governed by
the dynamic characteristics (i.e. frequency and mode shape) of the
supporting structure [4,5,9,11,18]. Some recent studies focused on ir-
regular supporting structures [25] and the development of probabilistic
models [26,27] for prediction of the floor accelerations.
Major findings of earlier studies on RC bare frame buildings include
that the amplification of the PFA along the height of the building re-
duces with increasing period of vibration as well as with increasing
inelasticity of the supporting structure [8,9,11,12]. The floor response
spectra (FRS) show peaks corresponding to the different modes of vi-
bration [1–6,9,12,18]. These peaks reduce with increasing inelasticity
of the supporting structure [1,12,13]. It has also been shown that the
floor spectra are amplified ground-motion spectra [9,10,12] and the
spectral amplification factors (defined as the ratio of the spectral or-
dinate at a given floor level to the spectral ordinate at ground level)
corresponding to the different modes of vibration follow the respective
elastic mode shapes along the height of the supporting structure
[12,21,25]. This observation was reasonably valid even in case of in-
elastically responding supporting structures [12].
Infills are generally treated as NSC and therefore generally ignored
in the seismic design of buildings. It is well accepted that the presence
of infills completely alters the dynamic behaviour of the buildings, as
they interact with the adjoining frame especially in the in-plane di-
rection. On the other hand, for out-of-plane action, the infills them-
selves are considered as decoupled secondary systems subjected to the
floor response as the input motion. Therefore, the estimation of floor
response of URM infilled RC frame buildings is not only crucial for the
safety of NSCs, but also for ensuring the safety of URM infills in out-of-
plane action.
Contrary to RC bare frame buildings, very limited studies
[18,28,29] have been conducted so far focusing on the effect of pre-
sence of Unreinforced Masonry (URM) infill walls on the floor accel-
eration demands. It was shown that the modelling of infills (even if
weak and distributed uniformly along the height) can significantly af-
fect both PFA and FRS [18]. Perrone and Filiatrault [28] highlighted

⁎
Corresponding author.
E-mail addresses: yogendra.eq@gmail.com (Y. Singh), dominik.lang@ngi.no (D.H. Lang).

https://doi.org/10.1016/j.engstruct.2018.08.078
Received 19 November 2017; Received in revised form 14 August 2018; Accepted 23 August 2018
Available online 01 September 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
M. Surana et al. Engineering Structures 175 (2018) 861–878

the inability of the current models available in literature to predict the

FRS of RC frame buildings with masonry infills subjected to frequent
earthquakes. Blasi et al. [29] showed that the variation of the elastic
modulus of the infill panels can significantly influence the dynamic
response of the structures, and hence can modify the distribution of the
PFA along the height of the supporting structure.
In general, buildings are designed to respond with different degrees
of inelasticity under earthquakes of different intensities. Therefore, the
floor response can be estimated accurately only by considering the ef-
fect of inelasticity of the building, especially in the framework of per-
formance-based design, in which the buildings are designed for dif-
ferent levels of ductility demand. However, current seismic design
codes [30–32] do not consider the effect of inelasticity of the supporting
structure and the level of shaking imparted to the NSCs. Therefore, the
present study attempts to include the effects of inelasticity of the sup-
porting structure on the floor response of RC frame structures, both
uniformly infilled (UI) with URM walls and with open ground storeys
(OGS; also known as “pilotis frames” and “buildings on stilts”). Build-
ings of the latter structural typology, although not considered to be
desirable for good seismic performance, are prevalent in India and
many other parts of the world [33–36]. For the present study, a total of
1232 nonlinear dynamic analyses are conducted using a suite of 22 far-
field ground-motion records that are applied on four different building
models (i.e. two uniformly infilled models and two models with open
ground storeys) having different dynamic characteristics in the two
orthogonal directions. Further, to study the effect of inelastic behaviour
of the supporting structure on floor response, seven different levels of
inelasticity (ductility demands) are considered. The spectral amplifi-
cation functions, which were developed for RC bare frame buildings in
an earlier study [12], are modified to predict the floor acceleration
demands for NSCs mounted on UI and OGS buildings. The developed
spectral amplification functions take into account the ground-motion
characteristics (acceleration response spectrum of the free-field ground-
motion), dynamic characteristics (periods and mode shapes) and level
of inelasticity of the supporting structure, as well as the frequency
tuning between the NSC and the supporting structure. The developed
spectral amplification functions are validated through nonlinear time-
history analyses using both recorded and spectrum-compatible time
histories.

2. Numerical study

For the present study, RC frame buildings with plan shape and
elevations as shown in Fig. 1 are considered. The building plan was
chosen from earlier studies [11,12] based on a field survey to consider
the variety of characteristics (e.g. typical storey height, bay width) of
the building stock in the National Capital Region (NCR) of India [37].
The heights of these buildings are considered as 4 and 8 storeys, re-
presenting two examples of the mid-rise building stock typical for the
NCR of India. The storey height is taken as 3.3 m, consistent with the
field observations. The thickness of URM infill walls is considered as
230 mm and 110 mm for exterior and interior walls, respectively. The
compressive strength of infill walls masonry is assumed to be 4.1 MPa
considering the fair quality of masonry, also consistent with typical
average compressive strength values for solid clay brick masonry in
Northern India [38,39]. A total of four building models are investigated
with two different heights (i.e. 4 and 8 storeys) and two different
configurations consisting of UI and OGS buildings.
The buildings are modelled in the building analysis and design
software ETABS [40,41]. Beams and columns are defined using 3D
frame elements and slabs are considered as rigid diaphragms. The
cracked section properties of beams and columns are derived following
ASCE 41 [42]. Dead and live loads on the buildings are assigned ac-
cording to IS 875 Part 1 [43] and IS 875 Part 2 [44], respectively. In
order to model the URM infill walls, the eccentric strut model of ASCE
41 [45] with some modifications as per Burton and Deierlein [46] is
Fig. 1. (a) Generic plan; (b) schematic elevation in longitudinal direction, for a
4 storey UI RC frame building; and (c) schematic elevation in longitudinal di-
rection, for a 4 storey OGS RC frame building. Elevations in the transverse
direction are not shown here for brevity. (The dashed lines in the floor plan
represent the floor slab boundaries, which are assumed to be rigid in plane. The
inclined lines in elevations represent the infills. All dimensions are in meter.)
used. The initial (uncracked) stiffness of the masonry infill wall is
considered as twice of the stiffness obtained from the equivalent strut
width model of ASCE 41, as recommended by Burton and Deierlein [46]
based on experimental investigations on URM infill walls. All the
buildings are designed as Special Moment Resisting Frames (SMRF),
following the most recent Indian Standards IS 1893 [47] and IS 13920
[48].
The buildings are designed for seismic actions corresponding to
Indian seismic zone IV (Effective Peak Ground Acceleration = 0.24 g),
and assumed to be situated on soil type I (hard soil/rock). The design
response spectrum corresponding to Maximum Considered Earthquake
(MCE) hazard as per the Indian Standard IS 1893 [47] is shown in
Fig. 2. All the considered building models are designed conforming to
the strong-column weak-beam (SCWB) design criteria with a SCWB
ratio of 1.40 [48]. P-delta effects are also considered both in the

862
M. Surana et al. Engineering Structures 175 (2018) 861–878

frame buildings [8,9,12]. This period ratio has its particular importance
in simplification of the floor amplification model in which the peak
corresponding to the second mode can be considered at a period ratio
close to 0.30.

2.1. Nonlinear modelling and analysis

For the nonlinear analysis, a lumped-plasticity model is used to

Fig. 2. Response spectrum at a rock site corresponding to the MCE hazard for
seismic zone IV.
analysis and design process. The material properties, typical section
sizes, and reinforcement ratios for beams and columns obtained from
design of the considered buildings, are summarized in Table 1.
Table 2 presents the dynamic characteristics of the considered
buildings (UI, OGS and uniformly infilled building with door and
window openings (UIWO) described later in this article) obtained from
the modal analysis. It is interesting to note that the ratio of the second
mode period (T2) to the first mode period (T1) obtained from Eigen-
value analysis is close to 0.30. This observation is found to be in good
agreement with the findings of earlier studies on both RC and steel bare
model the inelastic behaviour of both beams and columns. Flexural
(M3) hinges and interacting (P-M2-M3) hinges are assigned at both
ends of beams and columns (Fig. 3a), respectively, and the corre-
sponding backbone curve parameters are derived following ASCE 41
[42] guidelines. These parameters in ASCE 41 are obtained from a
cyclic envelope curve, and thereby include strength deterioration ef-
fects. To consider stiffness degradation under cyclic loading, the pivot
hysteresis model [49] is used. In this hysteretic model, the unloading
and reverse loadings tend to be directed towards specific points (called
the pivot points) in the force-deformation plane. The corresponding
hysteresis model parameters are adopted from CSI [40,49]. The shear
failure of columns due to the strut action of the infills is modelled as per
ASCE 41 [42].
A variety of micro- and macro-models are available in literature for
representing the nonlinear behaviour of infill walls [46,50–54]. Haldar
et al. [38] compared three different macro-models with the available
experimental results and concluded that an eccentric single strut model
is capable of predicting the governing failure modes in most of the
cases. Accordingly, in the present study, an eccentric single strut macro-
element model of URM infills has been adopted. Based on the experi-
ments conducted on infill panels, Burton and Deierlein [46] proposed
an experimentally calibrated macro-model for modelling of URM infill
panels. This model utilizes a trilinear force-deformation curve for URM
infill walls which also includes post-peak behaviour as shown in Fig. 3b.

Table 1
Typical sizes, material properties and reinforcement ratios for beams and columns.
Frame No. of Concrete compressive Rebar strength Beam size Column size (mm) Beam longitudinal Column longitudinal
configuration storeys strength (MPa) (MPa) (mm) reinforcement (in %) reinforcement (in %)

UI 4 30 500 250 × 350 350 × 350, 0.50–1.50 2.25–3.00

300 × 300
8 30 500 300 × 400 400 × 400, 0.40–1.20 1.50–3.50
350 × 350

OGS 4 30 500 250 × 350 350 × 350, 0.50–1.50 2.25–3.00

300 × 300
8 30 500 300 × 400 400 × 400, 0.40–1.20 1.50–3.50
350 × 350

Table 2
Dynamic characteristics of the considered building models.
Frame configuration No. of storeys Direction Period (s) Period ratio Modal mass participation factor (%)

T1 T2 T2/T1 αm1 αm2

UI 4 Longitudinal 0.40 0.14 0.35 85 11

Transverse 0.54 0.19 0.35 85 11
8 Longitudinal 0.71 0.24 0.34 82 11
Transverse 1.00 0.33 0.33 80 13

OGS 4 Longitudinal 0.77 0.18 0.23 99 01

Transverse 0.91 0.23 0.25 97 02
8 Longitudinal 0.94 0.29 0.31 94 05
Transverse 1.22 0.40 0.33 91 07

UIWO 4 Longitudinal 0.58 0.20 0.34 85 10

Transverse 0.59 0.22 0.37 83 10

T1 and T2 are the periods of vibration corresponding to the fundamental and the second modes of vibration, respectively, in the direction under consideration; and
αm1 and αm2 are the modal mass participation factors corresponding to the fundamental and the second modes of vibration, respectively, in the direction under
consideration.

863
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 3. Typical backbone curves for: (a) RC frame elements; and (b) URM infill walls. All the parameters are shown in the normalized coordinates. The subscript ‘n’
stands for the normalized deformation parameters with respect to the ultimate deformation.

To compute the strength of URM infills, three most prominent failure nonlinear static analysis of the building (using the method of ASCE 41
modes of infill walls (sliding shear, diagonal compression and diagonal [42]). In the present study, the strength ratio (R) is preferred over
tension) have been considered. Haldar et al. [38] compared the strength ductility demand (µ), as it can be obtained relatively easily. For the
of Indian URM infills with the aspect ratio (0.52 and 1.05) and com- design of new buildings, the strength ratio R can be estimated using the
pressive strength consistent with the present study in different modes of following correlation:
failure and showed that the sliding shear failure mode results in the
q
minimum strength of the URM infills and thereby governs the inelastic R=
Ω (2)
modelling of URM infills. The same observation is also made in the
present study and accordingly the yield strength is computed based on where q is the behaviour factor (also known as response reduction
the strength of infills in the sliding shear failure mode. The other factor), and Ω is the over-strength factor [30,31]. The over-strength
backbone curve parameters (e.g. the yield displacement, δy the capping factor may vary with the design code used and local construction
(ultimate) strength, Fc the plastic capping displacement, δp and the post- practices. In the absence of an available estimate or for accurate esti-
capping displacement, δpc) for URM infill walls are adopted from Burton mation of R, a nonlinear static analysis of the building is recommended.
and Deierlein [46]. It is particularly important to consider here that
these parameters were obtained for monotonic loading. Therefore,
3. Results and discussion
these parameters do not account for the cyclic deterioration of strength
and stiffness. To consider the cyclic deterioration effects in the infill
3.1. Amplification of floor acceleration along the height of the building
panels, these monotonic backbone curve parameters are adjusted (by
reducing strength and deformation capacity to account for cyclic de-
All the four considered buildings are analyzed using the 22 far-field
terioration effects implicitly) following the guidelines of PEER/ATC-72
ground-motion records applied in both directions separately, while the
[55]. As the objective of the present study is to estimate the global floor
records are scaled to seven levels of inelasticity. The different levels of
response of the infilled RC frame buildings, therefore simplified mod-
inelasticity are represented by an elastic case and different values for
elling methodology adopted in the present study is expected to be fairly
strength ratio varying from R = 1.00 to 3.50. Fig. 4 compares the ratio
reliable for the performance-based seismic design of NSCs.
of median PFA to peak ground acceleration (PGA) profiles with varying
To investigate the floor acceleration demand in the considered
levels of inelasticity (strength ratio) for the considered 4 storey building
structural models, Incremental Dynamic Analysis (IDA; [56]) is per-
models along the height of the building. The numerically obtained
formed. For the execution of the IDA, 22 far-field ground-motion re-
profiles are compared with the models prescribed in current codes
cords as identified in FEMA P695 [57] are used, and a Rayleigh
[30–32]. It is evident from Fig. 4 that the ratio of PFA/PGA at any
damping of 5% is assigned to the periods corresponding to the funda-
height of the building is higher in case of the UI frame as compared to
mental mode and the mode corresponding to a cumulative mass par-
the OGS building. This observation is due to the increase in the period
ticipation of 95%. For a given structural model, the major component
of vibration of the OGS building in comparison with the UI frame
(i.e. the horizontal component with higher PGA) of each ground-motion
(Table 2). Further, the ratio of PFA/PGA is increasing almost linearly in
record is applied along both principal directions of the building, sepa-
case of UI RC frame buildings, with a sudden increase at the roof level,
rately. The ground motion is scaled to multiple intensities based on the
whereas it is almost constant along the height of the building in case of
spectral acceleration at the fundamental period of the building, Sa (T1,
OGS buildings. This fact can be attributed to significant differences in
5%) in the considered direction. The structural inelasticity is considered
the dynamic characteristics of the two different structural systems (viz.
through a strength ratio (R) defined as:
UI and OGS) resulting in whiplashing effects of higher modes in case of
Sa (T1, 5%) UI RC frame buildings, and absence of such effects in case of OGS
R=
Say (1) buildings (Table 2).
The two phenomena of amplification of the motions along the
where Sa (T1, 5%) is the 5%-damped elastic spectral acceleration of the height of the building and inelastic energy dissipation go on simulta-
scaled ground motion at the building’s fundamental period T1, and Say neously in case of buildings subjected to severe shaking. As it was also
is the spectral acceleration at yielding, which has been obtained from a observed in previous studies [9,11,12], the floor acceleration

864
M. Surana et al.
Fig. 4. Variation of median PFA/PGA along the height of the 4 storey RC frame buildings: (a) UI – longitudinal direction; (b) UI – transverse direction; (c) OGS –
longitudinal direction; and (d) OGS – transverse direction.
amplification decreases with increasing period of vibration. With in-
creased levels of inelasticity, the PFA/PGA reduces further. In case of
OGS buildings, the period is much longer than the UI frame and when
subjected to large ductility demand, these buildings show a net de-
amplification (Fig. 4c and d).
A comparison of the PFA/PGA models of the current codes (EN 1998
and FEMA P750) with the numerically obtained PFA/PGA profiles
shows that the current code models neither consider the effect of period
of vibration nor of inelasticity of the supporting structure. Thereby,
both models produce significantly conservative estimates of PFA/PGA
for mid-rise supporting structures responding inelastically. The ob-
served trends are consistent in case of the 8 storey buildings as well,
while the conservatism of the code models [30–32] in prediction of
PFA/PGA increases further in comparison with the 4 storey buildings.
However, this conservatism (in case of both UI and OGS buildings) is
observed to be much smaller than those observed in RC bare frame
buildings [9,11]. This observation can be attributed to a relatively in-
creased rigidity of the infilled frames due to the presence of URM infill
walls.
3.2. Floor spectrum profiles along the height of the building
Fig. 5 presents the coefficient of variation (COV) in floor spectra
normalized using three different schemes (viz. normalization by PFA,
PGA, and ground response spectrum) for a 4 storey UI RC frame
building. It can be observed that normalizing the floor spectra with the
corresponding ground spectrum leads to the least COV in elastic as well
865
Engineering Structures 175 (2018) 861–878
as inelastic cases, irrespective of the floor level under consideration. A
similar observation was already made in case of RC bare frame build-
ings in earlier studies [12,25]. A closer observation of the COV shows
that for periods close to zero, the normalization with respect to PFA
actually leads to the least COV. However, in this case, to construct the
FRS, the PFA is also to be obtained from PGA (PFA is usually expressed
as a function of PGA and the floor level under consideration). There-
fore, the prediction of FRS using PFA also leads to the total variability
close to that in case of normalization of the floor spectrum using PGA,
even at periods close to zero. These observations are made in case of all
buildings considered in the present study, but results are not presented
here for brevity. Accordingly, in the present study, the floor accelera-
tion demands are presented using normalization with respect to the
corresponding input ground-motion spectrum. This normalization di-
rectly provides the spectral amplification at different periods.
Figs. 6 and 7 present the median 5%-damped normalized FRS ob-
tained at the center of mass (coinciding with the geometric center in the
present case of symmetric buildings) for the 4 and 8 storey buildings
considered in the present study. The FRS (SAF (T)) are normalized by the
respective ground response spectrum, SAG (T) and shown at three dif-
ferent relative heights for both UI and OGS buildings in the longitudinal
direction. Similar results are also obtained in the transverse direction,
but, for the sake of brevity, are not presented here. These curves di-
rectly represent the spectral amplification factors as a function of nor-
malized period (T/T1). From Figs. 6 and 7 (and also from the normal-
ized FRS obtained for the transverse direction of the considered
buildings), it has been observed that the UI buildings exhibit peaks
M. Surana et al.
Fig. 5. COV of the floor spectra normalized by PGA, PFA and ground response spectrum, for a 4 storey UI RC frame building subjected to a suite of 22 far-field
ground-motion records: (a) Z/H = 0.50 (Elastic); (b) Z/H = 1.00 (Elastic); (c) Z/H = 0.50 (R = 2.00); and (d) Z/H = 1.00 (R = 2.00).
corresponding to the fundamental as well as the second mode of vi-
bration. In case of OGS buildings, the peak in the spectral amplification,
corresponding to the second mode almost disappears, and the peak
spectral amplification factors corresponding to the second mode of vi-
bration are observed to be close to unity in this case (irrespective of the
level of inelasticity and the floor level under consideration). The re-
duced amplification corresponding to the second mode of vibration, in
case of OGS buildings, can be attributed to an insignificant contribution
of the second mode of vibration in the dynamic response of the
building, as also evidenced from the dynamic characteristics of the OGS
buildings (Table 2). Further, these peaks in the spectral amplification
reduce with the increase in the inelasticity of the supporting structure.
This observation is in good agreement with the earlier study on RC bare
frame buildings [12].
A comparison of Figs. 6a and 7a with Figs. 6d and 7d, respectively,
exhibits that the spectral amplification at the first floor level in case of
the OGS building is much higher (approximately 4 times of the ground
spectral ordinate, for elastic response) as compared to the UI frame
(approximately 2 times of the ground spectral ordinate, for elastic re-
sponse). The increased amplification in case of the OGS buildings can
be attributed to the presence of stiffness irregularity, leading to a sig-
nificantly increased mode shape coefficient at first floor level when
866
Engineering Structures 175 (2018) 861–878
compared with the corresponding UI frame buildings. This increased
spectral amplification, in case of OGS buildings, cannot be captured
using the current code models [30–32] which completely ignore the
role of mode shape in estimating the FRS. This observation further
highlights the need to include the dynamic characteristics of the sup-
porting structure in the floor amplification model.
Two very prominent effects are observed in the normalized FRS of
infilled RC frame buildings with increased levels of inelasticity (Figs. 6
and 7). These are: (i) a reduction in the spectral amplification corre-
sponding to not only the fundamental mode but also corresponding to
the second mode of vibration; and (ii) shifting of the peaks in floor
spectrum profiles towards longer periods (due to period elongation of
the inelastically responding supporting structure). The former was al-
ready observed in a previous study on bare RC frames, by Surana et al.
[12], while the latter was not observed, so prominently, in previous
studies [12,58] on bare RC frame buildings, and therefore, in those
studies, the effect of inelasticity on period elongation was neglected.
Fig. 8 presents the peak spectral amplification factors corresponding
to the fundamental mode of vibration, at the roof level for the con-
sidered UI and OGS buildings. The obtained COV (in the peak spectral
amplification factor for different buildings having different periods of
vibration) for a given strength ratio, varies between 7 and 11%,
M. Surana et al.
Fig. 6. Median normalized floor spectra (median of the 22, 5%-damped floor response spectra, with ordinate normalized by the corresponding ground response
spectrum and abscissa normalized by the fundamental mode period) for the 4 storey RC frame buildings in longitudinal direction: (a) Z/H = 0.25 (UI); (b) Z/
H = 0.50 (UI); (c) Z/H = 1.00 (UI); (d) Z/H = 0.25 (OGS); (e) Z/H = 0.50 (OGS); and (f) Z/H = 1.00 (OGS).
implying that the peak spectral amplification corresponding to the
fundamental mode can be considered to be independent of the period of
vibration. The same observation has also been made for the spectral
amplification corresponding to the second mode of vibration (only in
case of UI RC frame buildings, since spectral amplification close to unity
is observed in case of OGS buildings), but, again, the results are not
reported herein for brevity.
Fig. 9 compares the variation of the effective period of vibration
(Teff) with the strength ratio (R) for inelastic UI (Fig. 9a), OGS (Fig. 9b)
(both shown by circles) and RC bare frame buildings (shown by
squares, and investigated in the earlier study by Surana et al. [12]). The
effective (elongated) period of the inelastic structure is considered as
the period corresponding to the peak spectral amplification obtained
from the median normalized FRS. It is observed that, in case of RC bare
frame buildings (investigated in the earlier study by Surana et al. [12]),
the period elongation is insignificant and the peaks in the floor spec-
trum can be reasonably located using the period of vibration obtained
from an elastic analysis. On the other hand, in case of UI and OGS
buildings, this period elongation effect is significant. The reason for a
significant period elongation in case of UI and OGS buildings can be
attributed to the sequential failure of URM infill walls starting from the
lower storeys and progressing upward with the increasing levels of
inelasticity. The reduction in stiffness due to failure of URM infill walls
is much more pronounced than that due to yielding of beams in case of
bare frame buildings; and it results in a significant change in the dy-
namic characteristics of the supporting structure, and hence in the floor
response.
Fig. 9 also presents a comparison of the numerically obtained
median Teff with the period obtained from Eqs. (3) and (4), for UI and
OGS buildings, respectively.
867
Engineering Structures 175 (2018) 861–878
R2 + 1
Teff = T1
2 (3)
Teff = T1 R (4)
The closeness of the Eqs. (3) and (4) with the median Teff indicates
that these can be used for predicting the Teff of the inelastic URM in-
filled RC frame buildings, for the considered levels of inelasticity, with
reasonable accuracy. These expressions have been later used in the
present study, for predicting Teff of buildings with different levels of
inelasticity.
Figs. 6 and 7, are reproduced in Figs. 10 and 11, respectively, by
using Teff in place of T1 for normalization of the period on the abscissa.
The figures show that normalizing the FRS using Teff of the inelastic
structure, leads to the convergence of peaks for different inelasticity
levels, close to the normalized period equal to unity. However, the ef-
fect of period elongation due to infill wall failure is not as significant for
the second mode as it is for the fundamental mode. As a result, this
normalization of the FRS with effective period leads to a slight diver-
gence of peaks corresponding to the second mode.
Fig. 12 presents a comparison of the peak spectral amplification
factors corresponding to the fundamental and the second mode along
the height, normalized by the respective mode shape for different
buildings (except in the case of second mode for OGS buildings) con-
sidered in the present study. The solid and dotted lines, in the figure,
show the mode shape normalized amplification factors in the long-
itudinal and in the transverse directions, respectively, for the same
building. The lines of same colour show a particular strength ratio.
It can be observed that almost vertical lines are obtained when the
peak spectral amplification factors are normalized by the corresponding
M. Surana et al.
Fig. 7. Median normalized floor spectra (median of the 22, 5%-damped floor response spectra, with ordinate normalized by the corresponding ground response
spectrum and abscissa normalized by the fundamental mode period) for the 8 storey buildings in longitudinal direction: (a) Z/H = 0.125 (UI); (b) Z/H = 0.50 (UI);
(c) Z/H = 1.00 (UI); (d) Z/H = 0.125 (OGS); (e) Z/H = 0.50 (OGS); and (f) Z/H = 1.00 (OGS).
Fig. 8. Spectral amplification factors corresponding to the fundamental mode
of vibration at the level of roof for different buildings investigated in the present
study.
mode shape. This observation indicates that the spectral amplification
factors along the building height more or less follow the respective
elastic mode shape. This observation holds good even in case of the
second mode of vibration (Fig. 12c), and at different levels of in-
elasticity. This observation has been found out to be in agreement with
the earlier studies on inelastically responding RC bare frame buildings
[12,21]. The presented results suggest that the elastic mode shape of
the UI and OGS buildings can be used to capture the variation of the
spectral amplification along the height of both, elastic as well as in-
elastic, buildings. Furthermore, the envelope values of the peak spectral
amplification factors for elastically responding UI and OGS buildings,
corresponding to the fundamental and the second modes of vibration
can be roughly estimated as 7.50 and 3.00, respectively (unity
868
Engineering Structures 175 (2018) 861–878
corresponding to the second mode in case of OGS buildings). The ob-
served values of the peak spectral amplification corresponding to the
fundamental and the second modes of vibration are consistent with the
earlier studies [9,12] on steel and RC bare frame buildings.
3.3. Modified floor amplification model accounting for the influence of
infills
In an earlier study [12] spectral amplification functions were de-
veloped for inelastic RC bare frame buildings. These functions estimate
the peak spectral amplifications corresponding to the first two modes of
vibration. Parabolic curves were fitted to obtain the spectral amplifi-
cation within impact zones of the fundamental and the second modes of
vibration. For periods outside the impact zones of the first two modes,
the model amplifies the ground response spectrum by constant factors
[12]. Based on the observations of the present study, the spectral am-
plification functions are modified to account for the influence of URM
infills. The developed (modified) spectral amplification functions are
shown in Fig. 13. In the modified model for spectral amplification
functions, two separate parabolic functions are used within the impact
zones of the fundamental and the second modes, in order to replicate
the two peaks in case of UI frame buildings (Fig. 13a); whereas a single
parabolic function is used to represent the sole peak corresponding to
the fundamental mode in case of the OGS buildings (Fig. 13b). The
impact zone of the second mode is considered between the normalized
periods 0 and 0.5, whereas the impact zone of the fundamental mode
starts from the normalized period 0.5 and continues up to the point of
intersection with the long-period amplification factor (AL), which is
assumed to be constant.
The spectral amplification, A0 corresponding to zero period (which
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 9. Variation of the effective period of vibration with strength ratio: (a) UI buildings; and (b) OGS buildings. (Different colours show different heights of buildings,
circular-shaped points show infilled RC frame buildings, whereas square-shaped points show bare RC frame buildings.)

Fig. 10. Median normalized floor spectra (median of the 22, 5%-damped floor response spectra, with ordinate normalized by the corresponding ground response
spectrum and abscissa normalized by the effective period) for the 4 storey buildings in longitudinal direction: (a) Z/H = 0.25 (UI); (b) Z/H = 0.50 (UI); (c) Z/
H = 1.00 (UI); (d) Z/H = 0.25 (OGS); (e) Z/H = 0.50 (OGS); and (f) Z/H = 1.00 (OGS).

869
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 11. Median normalized floor spectra (median of the 22, 5%-damped floor response spectra, with ordinate normalized by the corresponding ground response
spectrum and abscissa normalized by the effective period) for the 8 storey buildings in longitudinal direction: (a) Z/H = 0.125 (UI); (b) Z/H = 0.50 (UI); (c) Z/
H = 1.00 (UI); (d) Z/H = 0.125 (OGS); (e) Z/H = 0.50 (OGS); and (f) Z/H = 1.00 (OGS). Some of the curves (for buildings having longer effective periods) ended
before the effective period ratio of 3, since the floor spectra were studied for periods up to 5 s.

Fig. 12. Variation of mode shape-normalized peak spectral amplification factors along the height of the building for varying levels of strength ratio for: (a) fun-
damental mode (UI buildings); (b) fundamental mode, (OGS buildings); and (c) second mode, (UI buildings). The solid lines represent the longitudinal direction,
whereas the dotted lines represent the transverse direction. In case of OGS buildings, mode shape normalized peak spectral amplification factors corresponding to the
second mode are close unity and independent of level of inelasticity, therefore have not been shown herein.

is also equal to PFA/PGA), is expressed as: and for inelastic response (R ≥ 1),
For elastic response,

T −2.5 ⎞ 8.75−3.5T1 ⎞ ⎞ Z 7−2R Z

1.0 ⩽ A0 = 1.0 + ⎜⎛ ⎛ 1
⎜ ⎟ R+⎛⎜ ⎟⎟ ⩽ 1.0 + ⎛ ⎞
2.5−T1 ⎞ Z ⎝⎝ 3T1 ⎠ ⎝ 3T ⎠⎠ H ⎝ 3 ⎠H
1 ⩽ A0 = 1 + ⎛ ⎜ ⎟ ⩽ 3.0 1

⎝ T1 ⎠ H (5) (6)

870
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 13. Developed modified spectral amplification functions for: (a) UI frame buildings; and (b) OGS buildings. (*Normalization of the periods is carried out with
respect to the effective period of vibration, corresponding to a given level of inelasticity.)

Fig. 14. Comparison of median PFA/PGA profiles obtained in the present study with code models and the model developed earlier [12] for RC bare frame buildings:
(a) 4 storey UI; (b) (b) 8 storey UI; (c) 4 storey OGS; and (d) 8 storey OGS. Please note that in Figures (a) and (b) the proposed model (Eq. (5)) overlaps with the FEMA
model, and in Figures (c) and (d) the proposed model is independent of Z/H, T1, and R.

where T1 is the fundamental period of vibration of the building for the
direction under consideration, R is the strength ratio as defined earlier
in Eq. (1), Z is the height from the base to the floor at which the re-
sponse spectrum is being constructed, and H is the total height of the
building from its base. Eqs. (5) and (6) are applicable for the UI frames
and in case of OGS buildings (as evidenced in Fig. 4c and d), a constant
value of A0 equal to unity and independent of the period of vibration,
floor level under consideration, and level of inelasticity is re-
commended.
The peak values of the spectral amplification function, corre-
sponding to the fundamental and the second modes, at ith floor, are
proposed as:

871
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 15. Comparison of median normalized floor spectra obtained in the present study with the proposed model for UI and UIWO buildings: (a) Z/H = 0.50 (Elastic);
(b) Z/H = 1.00 (Elastic); (c) Z/H = 0.50 (R = 2.00); and (d) Z/H = 1.00 (R = 2.00). (The number in the legend represents the number of storeys, whereas ‘L’ and ‘T’
represent the ‘Longitudinal’ and ‘Transverse’ directions, respectively, of the building.)

For elastic response, are the first mode shape coefficients at ith floor and roof, respectively;
φ2,i and φ2,roof are the second mode shape coefficients at ith floor and
⎛ ϕ1, i ⎞ roof, respectively; and R is the strength ratio as defined in Eq. (1). It
A1 = 7.50 ⎜ ⎟ ⩾ 1.00
ϕ (7) should be noted that in case of the OGS buildings, the proposed model
⎝ 1, roof ⎠
considers spectral amplification only for periods larger than 0.5Teff. For
⎛ ϕ2, i ⎞ periods shorter than 0.5Teff, a constant spectral amplification equal to
A0 ≤ A2 = 3.00 ⎜
ϕ ⎟ unity is recommended, irrespective of the floor level under considera-
⎝ 2, roof ⎠ (8)
tion. Eqs. (7)–(10) are reproduced from the earlier study on RC bare
and for inelastic response (R ≥ 1), frame buildings [12] whereas Eqs. (11)–(13) are modified to account
for the effective period of vibration considering the effect of inelasticity.
⎛ ϕ1, i ⎞ Knowing the values of A0, A1, and A2, the median amplification
A1 = 7.50 ⎜ (0.95R−0.65) ⩾ 1.00
ϕ1, roof ⎟ (9) function can be obtained using Eqs. (11)–(13) as:
⎝ ⎠
For 0 ≤ T/Teff ≤ 0.5
⎛ ϕ2, i ⎞ C1
Ao ⩽ A2 = 3.00 ⎜ (0.90R−0.25) A (T ) = −C2
ϕ2, roof ⎟ (10) 2
⎝ ⎠
where A1 and A2 are the spectral amplification factors corresponding to
1+ ( T2
−
T
Teff Teff ) (11)

the first and the second modes of vibration, respectively; φ1,i and φ1,roof for 0.5 ≤ T/Teff ≤ TL/Teff,

872
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 16. Comparison of median normalized floor spectra obtained in the present study with the proposed model for OGS and UIWO RC frame buildings: (a) Z/
H = 0.50 (Elastic); (b) Z/H = 1.00 (Elastic); (c) Z/H = 0.50 (R = 2.00); and (d) Z/H = 1.00 (R = 2.00). (The number in the legend represents the number of storeys,
whereas “L’ and ‘T’ represent the ‘Longitudinal’ and ‘Transverse’ directions, respectively, of the building.)

A (T ) =
C3
−C4
2
1 + 1− T( T
eff ) (12)
for T/Teff ≥ TL/Teff,
Z the supporting structure (whichever is smaller) is recommended for the
A (T ) = AL = 1 + 2 ⩽ A1
H (13)
where T1, T2 and Teff are the periods of vibration of the building cor-
responding to the fundamental and second modes, and the effective
(elongated) period obtained from Eq. (3) or (4), respectively; T is the
period of vibration of the NSC; AL is the long-period amplification
factor; TL is the normalized period beyond which the spectral amplifi-
cation function is assumed to be constant. The constants C1, C2, C3, and
C4 can be obtained using the known values of the amplification function
A(T) at specific values of period, T (i.e. at T = 0, T2, 0.5Teff, and T1).
The period TL can be obtained from the intersection of the parabola (Eq.
(12)) with the long-period amplification factor (Eq. (13)). In case of
NSCs having periods much longer than the fundamental period of the
supporting structure, the floor spectral acceleration tends to converge
to the ground spectral acceleration, irrespective of the floor level [58].
Therefore, Eq. (13) produces conservative estimates of the peak floor
spectral acceleration demand for very long-period NSCs. An upper limit
on the periods of NSCs up to 5 s or 4 times of the fundamental period of
applicability of the developed model. For NSCs with even longer per-
iods, the proposed function may yield too conservative estimates, par-
ticularly at upper floors.
3.4. Comparison of the proposed spectral amplification functions with
numerical results
The PFA/PGA and spectral amplification functions obtained from
Eqs. (5)–(13) are compared with the numerically obtained PFA/PGA
and spectral amplification profiles. Fig. 14 presents a comparison of the
predicted A0 profiles with the median A0 values obtained in this study,
for the investigated building models, exemplarily for the transverse
direction and for two different levels of inelasticity (the elastic case and

873
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 17. Comparison of floor spectra obtained from nonlinear dynamic analyses using the spectrum-compatible time histories, with the current code models, and the
proposed model, for the 4 storey UI building: (a) Z/H = 0.50 (Elastic); (b) Z/H = 1.00 (Elastic); (c) Z/H = 0.50 (R = 2.00); and (d) Z/H = 1.00 (R = 2.00). B-
ICC000, CHY101 and H-E11140 correspond to floor response obtained from nonlinear dynamic analyses using the spectrum-compatible time histories.

for R = 2.00). These profiles are also studied in the longitudinal di-
rection and other strength ratios, but are not presented here for the sake
of brevity.
It can be observed that the earlier developed model for PFA/PGA
[12] for RC bare frame buildings, is reasonably accurate in predicting
the floor acceleration demands, even in case of UI buildings, taking into
account both the effect of period of vibration and inelasticity of the
supporting structure. Further, in case of OGS buildings, PFA/PGA is
close to unity and almost independent of the inelasticity of the sup-
porting structure, which is also consistent with the recommendations of
the present study. On the other hand, the PFA/PGA recommended by
the codes (EN 1998 and FEMA P750) are too conservative for the OGS
buildings.
In order to further test the applicability of the developed spectral
amplification functions, an additional 4 storey URM infilled RC frame
building with door and window openings (UIWO), which was not used
to derive the set of spectral amplification functions, is considered. The
plan of the building UIWO building has been considered to be the same,
as shown in Fig. 1. Window openings of sizes 2.0 m × 1.2 m and
3.0 m × 1.2 m are considered in the exterior infills along the long-
itudinal and transverse directions, respectively, and door openings of
size of 1.2 m × 2.0 m are considered in the interior infills along the
longitudinal direction. For simplicity, the doors and windows are
placed centrally in the infill. These door and window openings resulted
in a reduction of 18–23% of the area of individual infill panels.
A number of studies [59,60] are available in literature to predict the
stiffness and strength of the URM infill walls with openings. In the
present study, the guidelines recommended by Asteris [59] have been
used to model the effect of openings, which resulted in stiffness of the
perforated URM infill panels reduced to 35–40% of the corresponding
solid infill panel. The other modelling parameters have been chosen as
discussed earlier in Section 2.1. The dynamic characteristics of the
UIWO RC frame building are reported in Table 2. It has been observed
that the consideration of door and window openings resulted in 45%
and 9% elongation of the period of vibration of the UIWO building as
compared to UI building, in the longitudinal and the transverse direc-
tions, respectively. A larger elongation of the period in the longitudinal
direction, as compared to the transverse direction, can be attributed to

874
M. Surana et al.
Fig. 18. Comparison of floor spectra obtained from nonlinear dynamic analyses using the spectrum-compatible time histories, with the current code models, and the
proposed model, for the 4 storey OGS building: (a) Z/H = 0.25 (Elastic); (b) Z/H = 0.50 (Elastic); (c) Z/H = 1.00 (Elastic); (d) Z/H = 0.25 (R = 2.00); (e) Z/
H = 0.50 (R = 2.00); and (f) Z/H = 1.00 (R = 2.00). B-ICC000, CHY101 and H-E11140 correspond to floor response obtained from nonlinear dynamic analyses
using the spectrum-compatible time histories.
the fact that the longitudinal direction frames have openings (door or
window) in all the panels of URM infills, whereas the transverse di-
rection frames have openings only in exterior panels.
IDA has been performed on the 4 storey UIWO building using the
same suite of 22 ground-motion records applied in both the directions
separately and spectral amplification functions have been obtained.
Figs. 15 and 16 present a comparison of the median floor spectral
amplification profiles, obtained in this study, for the UI, OGS, and
UIWO buildings, with the proposed modified spectral amplification
functions, developed in Section 3.3. The floor spectral amplification
profiles are presented for all the investigated building models, at mid-
height (0.5H) and roof level (H), for two different strength ratios, re-
presenting the elastic response and moderate inelasticity (R = 2.00).
It is to be noted that only A0 (in case of UI frame buildings) in Eqs.
(5) and (6) is dependent of the supporting structure’s period of vibra-
tion, whereas A1 and A2 are independent of the period of vibration of
the supporting structure. The effect of period of vibration of the sup-
porting structure on A0 will lead to slightly different values of spectral
amplification at periods close to zero (only in transverse direction of the
8 storey UI RC frame building, since in other cases the A0 is governed by
the upper bound value in Eqs. (5) and (6)). In constructing the floor
spectral amplification profiles in Fig. 15 (shown by gray-shaded area),
A0 corresponding to T1 = 0.40 s is used for comparison. Further, as can
be seen from Eqs. (9) and (10), the peaks corresponding to the different
modes of vibration remain unaffected at the roof level, whereas a slight
difference is expected at the mid-height due to difference in mode
shapes of different buildings.
It can be observed that the proposed model captures well the peaks
corresponding to the two modes of vibration as well as the effect of
inelasticity of the supporting structure, at both the floor levels. It is
875
Engineering Structures 175 (2018) 861–878
interesting to note that the elastic response of the UIWO building is
similar to that of an UI RC frame building with softer infills (infills with
reduced stiffness due to presence of openings). However, as the in-
elasticity in the supporting structure is induced, the failure of URM
infills in the lower storeys alters the dynamic behaviour of the UIWO
building, dramatically, and renders it closer to the OGS building. As can
be observed from Figs. 15 and 16, the floor amplification in the UIWO
building is closer to the UI buildings in the elastic case and to the OGS
buildings in the inelastic case. Accordingly, the spectral amplifications
for the elastic case of UIWO are shown in Fig. 15 and for the inelastic
case (R = 2.00) are shown in Fig. 16. Further, the representation of the
peaks in the spectral amplification function through parabolic functions
is observed to be advantageous since it caters to the variability in the
estimation of the effective period of vibration of the inelastic supporting
structure (Figs. 15(c, d) and 16(c, d)).
3.5. Validation of the proposed model
To validate the proposed spectral amplification functions, linear and
nonlinear dynamic analyses using spectrum-compatible time histories
are performed on the 4 storey UI, OGS and UIWO buildings as described
earlier in this article. The details of the ground-motion records used to
generate spectrum-compatible time histories can be found in Surana
et al. [12].
Fig. 17 presents a comparison of the floor spectra obtained from the
spectrum-compatible time history analyses with those obtained from
current code models and the proposed model, for a 4 storey UI RC frame
building. It can be observed that the current code models are non-
conservative and unable to capture the peaks in the floor spectra, cor-
responding to the fundamental and the second modes of vibration, for
M. Surana et al. Engineering Structures 175 (2018) 861–878

Fig. 19. Comparison of floor spectra obtained from nonlinear dynamic analyses using the spectrum-compatible time histories, with the current code models, and the
proposed model, for the 4 storey UIWO building: (a) Z/H = 0.50 (Elastic); (b) Z/H = 1.00 (Elastic); (c) Z/H = 0.50 (R = 2.00); and (d) Z/H = 1.00 (R = 2.00). B-
ICC000, CHY101 and H-E11140 correspond to floor response obtained from nonlinear dynamic analyses using the spectrum-compatible time histories.

both, elastic as well as inelastic supporting structures. This observation
is contrary to the earlier study on bare RC frames [12] where the code
models are observed to be conservative for inelastically responding
structures. This non-conservativeness of the code models, in case of
inelastically (and obviously in case of elastically) responding UI RC
frame building lies in the fact that the current code models for pre-
diction of floor accelerations are based on PGA (PFA) and thereby ig-
nore the spectral shape of the ground motions, which have their peaks
in the short-period range (close to the period of vibration of the con-
sidered UI RC frame buildings). The proposed model being based on the
ground response spectrum and considering peaks corresponding to the
first two modes of vibration, predicts the floor accelerations in rea-
sonable agreement with the spectrum-compatible time history analyses,
for both, elastic as well as inelastic buildings, over the entire building
height.
Fig. 18 presents a comparison of the floor spectra for the 4 storey
OGS building obtained from the spectrum-compatible time history
analyses with those obtained from current code models and the pro-
posed model. In this case, the spectra have also been compared at a
normalized height (Z/H) of 0.25, which represents the first floor level
where a large discontinuity in stiffness and strength is present. It can be
observed that, in this case, the current code models are slightly non-
conservative in predicting the peak floor spectral acceleration corre-
sponding to the fundamental mode of vibration, for elastic response of
the supporting structure. This non-conservatism reduces with in-
creasing inelasticity of the supporting structure. On the other hand, the
code models (particularly FEMA P750) become significantly con-
servative at periods longer than the fundamental period of vibration, in
case of inelastically responding OGS building, whereas, the developed
model is reasonably accurate in the influence zone of the fundamental
period and slightly conservative in the short-period range. This con-
servativeness of the developed model in the short-period range (influ-
ence zone of higher modes) is due to the decision to ignore deamplifi-
cation in this spectral range for OGS buildings (Fig. 13(b)).
Fig. 19 presents a comparison of the floor spectra obtained from the
spectrum-compatible time history analyses with those obtained from
current code models and the proposed model, for the 4 storey UIWO
building. It is to be noted that for this building, the FRS for elastic case
are obtained from the developed model for UI RC frame buildings,
whereas the FRS for inelastic case are obtained from the developed

876
M. Surana et al.
model for OGS buildings. Similar to the UI RC frame buildings, the code
models are observed to be non-conservative for the elastically re-
sponding supporting structure with openings in infill walls. The ob-
served non-conservatism for code models vanishes, as the building re-
sponds inelastically. However, it is to be noted that the FRS shown in
Fig. 19(c, d) correspond to a moderate level on inelasticity (R = 2.00).
For higher levels of inelasticity, the code models are expected to be
quite conservative. On the other hand, the FRS obtained from the
proposed model are reasonably close to the numerically obtained
spectra in elastic as well as inelastic cases.
4. Conclusions
Incremental dynamic analyses are performed on mid-rise UI and
OGS RC frame building models in order to obtain the floor acceleration
response considering different levels of inelasticity of the supporting
structure. In total, 1232 nonlinear dynamic analyses are conducted and
the results are compared with an earlier study on bare RC frames to
identify the effect of URM infills on floor response. As found out in the
earlier study, the floor spectrum is observed to be better correlated with
the ground response spectrum, in comparison with the PFA and PGA, in
case of infilled RC frames. Due to the combined effect of sequential
failure of infills in successive storeys along the height of the building
and yielding of RC members, the increased levels of inelasticity lead to
a more significant elongation in the period of vibration of UI and OGS
buildings, as compared to bare frame buildings. This effect causes
shifting of peaks in floor spectrum profiles, towards longer periods. For
the investigated buildings, the effective (elongated) period is observed
to be up to 3.25 times of the elastic period. Further, the peak spectral
amplification corresponding to the fundamental and the second modes
of vibration, in case of UI and OGS buildings, is close to that observed in
RC bare frame buildings. These peak spectral amplification factors
follow the corresponding elastic mode shapes, reasonably closely, along
the height of the building, for both elastic and inelastic buildings, even
in the presence of infills.
Based on these observations and identification of the effects of infills
on floor amplification, the earlier developed spectral amplification
functions for RC bare frame buildings, are modified to account for the
influence of infills, in case of inelastically responding UI and OGS
buildings. The developed spectral amplification functions are validated
using spectrum-compatible time-history analyses on UI, OGS and UIWO
buildings. The developed floor spectral amplification functions over-
come limitations of current code models, since these take into account
the ground-motion characteristics, effects of dynamic characteristics
(periods and mode shapes) and level of inelasticity and corresponding
effective period of the supporting structure, and frequency tuning of the
NSC and the supporting structure. The developed spectral amplification
functions can be used to construct the FRS either from a code-based
design response spectrum or from a site-specific response spectrum. The
developed amplification functions are particularly useful in the fra-
mework of the performance-based design, as they will allow to develop
design FRS that are consistent with the targeted performance of the
supporting structure.
The floor spectral amplification functions developed in the present
study are based on UI and OGS buildings subjected to uni-directional
earthquake ground-motions, recorded in the far-field. Due to the pe-
culiar characteristics (e.g. fling effect, velocity pulse) of near-field
ground-motion records, a separate study is recommended for these
motions. The present study was conducted by modelling in-plane be-
haviour of URM infills, which means that a further study considering
the in- and out-of-plane interaction is also required. As noted by
Vukobratović and Fajfar [58], the floor spectral accelerations tend to
converge to the ground spectral accelerations in the very long-period
range. Therefore, the developed floor spectral amplification functions
may yield to too conservative results for NSCs having periods longer
than those considered in this study.
877
Engineering Structures 175 (2018) 861–878
Acknowledgements
The first and second author’s fellowships granted by the Ministry of
Human Resource Development, Government of India allowed the con-
duct of the present study. The support received from Ministry of Human
Resource Development, Government of India, is gratefully acknowl-
edged. The ground-motion records used in this study were obtained
from the PEER NGA WEST 2 database. The authors would also like to
express their thanks to three anonymous reviewers for very valuable
comments, resulting in the significant enhancement of the manuscript.
References
[1] Rodriguez ME, Restrepo JI, Carr AJ. Earthquake-induced floor horizontal accel-
erations in buildings. Earthquake Eng Struct Dyn 2002:693–718.
[2] Taghavi A, Miranda E. Approximate floor acceleration demands in multi-storey
buildings II: applications. J Struct Eng 2005;131(2):212–20.
[3] Medina RA, Sankarnarayanan R, Kingston KM. Floor response spectra for light
components mounted on regular moment-resisting frame structures. Eng Struct
2006;28:1927–40.
[4] Singh MP, Moreschi LM, Suarez LE, Matheu EE. Seismic design forces I: Rigid non-
structural components. J Struct Eng 2006;132(10):1524–32.
[5] Singh MP, Moreschi LM, Suarez LE, Matheu EE. Seismic design forces I: flexible non-
structural components. J Struct Eng 2006;132(10):1533–43.
[6] Sankaranarayanan R, Medina RA. Acceleration response modification factors for
non-structural components attached to inelastic moment-resisting frame structures.
Earthquake Eng Struct Dyn 2007;36:2189–210.
[7] Chaudhuri SR, Villaverde R. Effect of building nonlinearity on seismic response of
non-structural components: a parametric study. J Struct Eng 2008;134(4):661–70.
[8] Chaudhuri SR, Hutchinson TC. Effect of nonlinearity of frame buildings on peak
horizontal floor accelerations. J Earthquake Eng 2011;15(1):124–42.
[9] Weiser JD, Pekcan G, Zaghi AE, Itani AM, Maragakis EM. Floor accelerations in
yielding special moment resisting frame structures. Earthquake Spectra
2013;29(3):987–1002.
[10] Jiang W, Li B, Xie WC, Pandey MD. Generate floor response spectra: Part 1 Direct
spectra-to-spectra method. Nucl Eng Des 2015;293:525–46.
[11] Surana M, Singh Y, Lang DH. Effect of response reduction factor on peak floor
acceleration demand in mid-rise RC frame buildings. J Institut Eng – Series A, India
2017;98(1–2):53–65.
[12] Surana M, Singh Y, Lang DH. Floor spectrum of inelastic RC buildings considering
ground-motion characteristics. J Earthquake Eng 2018;22(3):488–519.
[13] Lin J, Mahin SA. Seismic response of light subsystems on inelastic structures. J
Struct Eng 1985;111(2):400–17.
[14] Politopoulos I, Feau C. Some aspects of floor spectra of 1 DOF nonlinear primary
structures. Earthquake Eng Struct Dyn 2007;36:975–93.
[15] Flores FX, Garcia DL, Charney FA. Assessment of floor accelerations in special steel
moment frames. J Constr Steel Res 2015;106:154–65.
[16] Vukobratović V, Fajfar P. A method for the direct determination of approximate
floor response spectra for inelastic structures. Bull Earthq Eng 2015;13(5):1405–24.
[17] Calvi PM, Sullivan TJ. Estimating floor spectra in multiple degree of freedom sys-
tems. Earthquakes Struct 2014;7(1):17–38.
[18] Lucchini A, Mollaioli F, Bazzurro P. Floor response spectra for bare and infilled
reinforced concrete frames. J Earthquake Eng 2014;18(7):1060–82.
[19] Petrone C, Magliulo G, Manfredi G. Floor response spectra in RC frame structures
designed according to Eurocode 8. Bull Earthq Eng 2015. https://doi.org/10.1007/
s10518-015-9846-7.
[20] Petrone C, Magliulo G, Manfredi G. Seismic demand on light-acceleration-sensitive
non-structural components in European reinforced concrete buildings. Earthquake
Eng Struct Dyn 2015;44(8):1203–17.
[21] Lucchini A, Franchin P, Mollaioli F. Probabilistic seismic demand model for non-
structural components. Earthquake Eng Struct Dyn 2016;45(4):599–617.
[22] Pan X, Zheng Z, Wang Z. Estimation of floor response spectra using modified modal
pushover analysis. Soil Dyn Earthquake Eng 2017;92:472–87.
[23] Pan X, Zheng Z, Wang Z. A multimode method for estimation of floor response
spectra. J Earthquake Eng 2017. https://doi.org/10.1080/13632469.2016.
1277440.
[24] Vukobratović V, Fajfar P. Code-oriented floor acceleration spectra for building
structures. Bull Earthq Eng 2017;15(7):3013–26.
[25] Surana M, Singh Y, Lang DH. Effect of irregular structural configuration on floor
acceleration demand in hill-side buildings. Earthquake Eng Struct Dyn
2018;47(10):2032–54.
[26] Lucchini A, Franchin P, Mollaioli F. Uniform hazard floor acceleration spectra for
linear structures. Earthquake Eng Struct Dyn 2017;46(7):1121–40.
[27] Lucchini A, Franchin P, Mollaioli F. Median floor spectra of linear structures with
uncertain properties. Earthquake Eng Struct Dyn 2017.
[28] Perrone D, Filiatrault A. Seismic demand on non-structural elements: influence of
masonry infills on floor response spectra. In: 16th European Conference on
Earthquake Engineering, Thessaloniki, Greece; 2018.
[29] Blasi G, Perrone D, Aiello MA. Fragility functions and floor spectra of RC masonry
infilled frames: influence of mechanical properties of masonry infills. Bull Earthq
Eng 2018;1–26. https://doi.org/10.1007/s10518-018-0435-4.
[30] American Society of Civil Engineers (ASCE). Minimum Design Loads for Buildings
M. Surana et al.
and Other Structures, ASCE Standard, ASCE 7-10, Reston, Virginia, United States;
2010.
[31] European Committee for Standardization EN 1998-1 (EC-8). Design of structures for
earthquake resistance – part 1: general rules, seismic actions and rules for buildings,
Brussels; 2004.
[32] Federal Emergency Management Agency (FEMA). NEHRP recommended seismic
provisions for new buildings and other structures, FEMA P750, Washington, D.C;
2009.
[33] Agarwal P, Thakkar SK, Dubey RN. Seismic performance of reinforced concrete
buildings during Bhuj Earthquake of January 26, 2001. ISET J Earthquake Technol
2002;39(3):195–217.
[34] Ozcebe G, Ramirez J, Wasti ST, Yakut A. 1 May 2003 Bingol Earthquake,
Engineering Report; 2003, p. 1–125.
[35] Kappos AJ, Panagopoulos G. Fragility curves for reinforced concrete buildings in
Greece. J Struct Infrastruct Eng 2010;1–2:39–53.
[36] Jain SK. Post-Earthquake handling of buildings. In: Jain SK, Lettis WR, Murty CVR,
Bardet JP, editor. 2001 Bhuj, India Earthquake Reconnaissance Report. Earthquake
Spectra, Supplement A to Volume 18; 2002, p. 297–317.
[37] DEQ. Seismic vulnerability of multistory buildings in Noida. Tech. Rep. Department
of earthquake engineering, Indian Institute of Technology Roorkee, Roorkee, India;
2009, p. 1–118.
[38] Haldar P, Singh Y, Paul DK. Identification of seismic failure modes of URM infilled
RC frame buildings. Eng Fail Anal 2013;33:97–118.
[39] Kaushik HB, Rai DC, Jain SK. Stress-strain characteristics of clay brick masonry
under uniaxial compression. J Mater Civ Eng 2007;19(9):728–39.
[40] CSI (Computers and Structures Inc.). CSI analysis reference manual. Computers and
Structure Inc., Berkeley, California, USA; 2016.
[41] CSI (Computers and Structures Inc.). Integrated building analysis and design soft-
ware (ETABS 2016). Computers and Structures Inc. Berkeley, California, USA; 2016.
[42] American Society of Civil Engineers (ASCE). seismic evaluation and retrofit of ex-
isting buildings, ASCE/SEI 41-13, Reston, Virginia; 2013.
[43] Bureau of Indian Standards (IS 875 Part 1). Indian Standard – Code of Practice for
Design loads (other than earthquake) for Buildings and Structures (Dead Loads),
New Delhi, India; 1987.
[44] Bureau of Indian Standards (IS 875 Part 2). Indian Standard – code of Practice for
Design Loads (other than earthquake) for Buildings and Structures (Live Loads),
New Delhi, India; 1987.
[45] American Society of Civil Engineers (ASCE). ASCE/SEI 41-06: Seismic
Rehabilitation of Existing Buildings. American Society of Civil Engineers, Reston,
VA, USA; 2007.
878
Engineering Structures 175 (2018) 861–878
[46] Burton HV, Deierlein GG. Simulation of seismic collapse in nonductile reinforced
concrete frame buildings with masonry infills. J Struct Eng 2014;140(8):A
401–4016.
[47] Bureau of Indian Standards (IS 1893 Part 1). Indian Standard – Criteria for
Earthquake Resistant Design of Structures, Part1 – General Provisions and
Buildings, New Delhi, India; 2016.
[48] Bureau of Indian Standards (IS 13920). Indian Standard – Ductile Design and
Detailing of Reinforced Concrete Structures Subjected to Seismic Forces – Code of
Practice, New Delhi, India; 2016.
[49] Dowell RK, Seible F, Wilson WL. Pivot hysteresis model for reinforced concrete
members. Struct J 1998;95(5):607–17.
[50] Klinger RE, Bertero V. Earthquake resistance of infilled frames. J Struct Division
1978;104(ST6):973–89.
[51] Zarnic R, Gostic S. Masonry infilled frames as an effective structural sub-assem-
blage. ed. Seismic Design Methodologies for The Next Generation of Codes; 1997, p.
335–346.
[52] Dolsek M, Fajfar P. The effect of masonry infills on the seismic response of a four
storey reinforced concrete frame – A probabilistic assessment. Eng Struct
2008;30:1991–2001.
[53] Asteris PG, Antoniou ST, Sophianopouloas DS, Chrysostomou CZ. Mathematical
macromodelling of infilled frames: State of the art. J Struct Eng
2011;137(12):1508–17.
[54] Asteris PG, Cotsovos DM, Chrysostomou CZ, Mohebkah A, Al-Chaar GK.
Mathematical micromodelling of infilled frames: State of the art. Eng Struct
2013;56:1905–21.
[55] Applied Technology Council (PEER/ATC-72-1). Modelling and acceptance criteria
for seismic design and analysis of tall buildings, applied technology council,
Redwood City, California; 2010.
[56] Vamvatsikos D, Cornell CA. Incremental dynamic analysis. Earthquake Eng Struct
Dyn 2002;31:491–514.
[57] Federal Emergency Management Agency (FEMA). Quantification of seismic per-
formance factors, FEMA P695, Washington, D.C; 2009.
[58] Vukobratović V, Fajfar P. A method for the direct estimation of floor acceleration
spectra for elastic and inelastic MDOF structures. Earthquake Eng Struct Dyn 2016.
https://doi.org/10.1002/eqe.2779.
[59] Asteris PG. Lateral stiffness of brick masonry infilled plane frames. J Struct Eng
2003;129(8):1071–9.
[60] Asteris PG, Giannopoulos IP, Chrysostomou CZ. Modelling of infilled frames with
openings. Open Construct Build Technol J 2012;6:81–91.