Professional Documents
Culture Documents
A Simplified Procedure For Nonlinear Static Analysis of Masonry Infilled RC Frames
A Simplified Procedure For Nonlinear Static Analysis of Masonry Infilled RC Frames
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a r t i c l e i n f o a b s t r a c t
Article history: Masonry walls are widely adopted in existing Reinforced Concrete (RC) structures, either as external infill
Received 24 October 2014 or as internal partitions. Although their mechanical contribution is usually neglected in structural anal-
Revised 23 May 2015 ysis and design, they significantly affect the seismic response of RC frames. This paper proposes a simpli-
Accepted 15 July 2015
fied procedure based on NonLinear Static (NLS) analysis for evaluating the seismic response of masonry
infilled RC frames. NonLinear Time History (NLTH) analyses are firstly carried out for understanding the
actual seismic response of such frames. To this end, an ‘‘equivalent-strut’’ model available in the scientific
Keywords:
literature is considered for simulating the nonlinear response of masonry walls under the cyclic actions
RC frames
Masonry infill
induced by seismic shakings. Then, based on the results of such NLTH analyses, the aforementioned sim-
Nonlinear analysis plified NLS procedure is formulated by unveiling a stable correlation between the observed dynamic
N2 Method response quantities and a simple scalar parameter that can be easily determined through two NLS anal-
yses carried out on the infilled frame and the corresponding bare one. A statistical description of the accu-
racy and reliability of the proposed method is finally proposed.
Ó 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2015.07.023
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
592 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
aforementioned CSM and the N2 Method was also presented [12]. assessment of its accuracy. The key findings of this study are finally
However, the effort towards formulating a simpler, accurate and summarised in Section 6.
reliable procedure for determining the seismic response of
masonry infilled RC structures by means of NLS analysis is still a
relevant challenge: it would be desirable that this procedure could 2. Modelling masonry infill walls in RC frames
be clearly based on an extension of the currently adopted (and val-
idated) relationships employed in NLS analyses of bare frames, Several modelling options are available in the scientific litera-
such as those assumed as part of the aforementioned N2 Method ture and can be nowadays considered for simulating the mechan-
[9,10]. ical behaviour of masonry infill walls: they range between the
Therefore, this paper presents a possible extension of the N2 simplest ‘‘equivalent strut’’ models (macro-models), whose first
Method to structures with masonry infill walls. Particularly, it conceptual definition dates back to the 1960s [13], to more recent
explains the conceptual genesis of the proposed procedure and and refined 2D/3D continuum models (Fig. 1), often formulated
describes the ‘‘inductive process’’ followed by the Authors in for- within the framework of the Finite Element Method (FEM). While
mulating the present proposal based on the results of a parametric the latter are not taken into account herein, the former are more
study. The monotonic and cyclic response of infill walls is simu- often employed in global analyses, as they are less computationally
lated through the attractive and computationally convenient demanding. However, defining their geometric parameters (i.e. the
approach based on the ‘‘equivalent strut’’ concept [13], widely width and depth of the diagonal strut and the equivalent nonlinear
accepted and validated in the scientific community. Particularly, behaviour of the masonry infill) is not straightforward, especially
a recent proposal is considered herein for determining the equiva- when openings, such as doors or windows, are present in the wall.
lent strut parameters depending on the actual geometric and Therefore, recent studies are available in the literature for describ-
mechanical properties of masonry walls, and taking into account ing the influence of relevant parameters, such as the vertical loads
the effect of openings within masonry walls. Various relevant acting on the frame [14], the elastic modulus and Poisson’s ratio of
parameters, such as the number of storeys, the distribution of walls infill masonry walls [15].
within the frame, the wall properties and the level of the seismic A thorough discussion about the aforementioned models is
excitation are taken into account in this analysis. The actual seis- beyond the scopes of this paper and can be found in another arti-
mic response of the frames under consideration was determined cle recently published by the same Authors [8]. Although alterna-
through NonLinear Time-History (NLTH) analyses intended as ‘‘nu- tive solutions based on adopting a variable number of struts for
merical experiments’’ and aimed at unveiling the role of infill walls simulating the mechanical response of walls are actually avail-
on the global structural response: the ratio between the displace- able in the literature [16], the analyses presented in this paper
ment demand obtained for the infilled frame and the one deter- are based on employing just two diagonal equivalent struts,
mined for the corresponding bare one is particularly scrutinised. which can carry loads only in compression as defined in a recent
Then, the results of NLTH analyses are considered for calibrating study [17]. This model defines the stiffness and the strength of
a simplified procedure based on the results of NonLinear Static the infill according to the horizontal direction and then the
(NLS) analyses carried out on both the infilled and the bare frames. force–displacement envelope of the diagonal placed between
These analyses were carried out for two levels of seismic intensity the beam–column joints have to be transformed to the direction
(PGA = 0.10g and PGA = 0.35g): on the one hand, it should be noted of the diagonal.
that the validation of the proposed method is bounded within the Hence, the infill walls are macro-modelled herein by means of a
aforementioned seismic intensity levels, albeit, on the other hand, tri-linear relationship representing the horizontal force–displace-
they are certainly representative of the actions to be considered for ment behaviour of masonry infill walls. The initial stiffness R1 of
serviceability and ultimate limit state checks in medium-to-high the curve represented in Fig. 2 is defined as follows [17]:
seismic hazard regions of Europe. Gw tw lw
The paper is organised as follows. Section 2 outlines the R1 ¼ ð1Þ
hw
State-of-the-Art on ‘‘equivalent-strut’’ models available in the lit-
erature and provides the details about the one adopted in in which Gw is the shear modulus of the masonry infill, while tw, lw
Section 3 for performing NLTH analyses of both bare and infilled and hw are the thickness, the length and the height of the masonry
frames. Then, Section 4 reports the formulation of the proposed wall, respectively. The maximum strength Fm is defined according
method based on NLS analyses and Section 5 outlines a statistical to a proposal available in the literature [17]:
beam beam
uts
l str
na
go
dia
column
column
column
column
masonry FE 2d elements
beam beam
ts
s tru
n al
go
column
column
column
column
dia
masonry FE 2d elements
Fig. 1. Modelling masonry infill with equivalent diagonal struts (on the left) and 2D continuum model (on the right).
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 593
F
F δ
dia
Fm go
na
ls
tru
ts
R2
hw
hv
Fy
R3 lv
R1 Fu = 0
δy δm
δ
lw
Fig. 2. Force–displacement relationship of the diagonal strut measured in the
horizontal direction [17]. Fig. 3. Geometrical parameters involved in the definition of the homothetic
parameter a.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f ws t w lw that corresponds to an opening of around 67% of the solid walls
F m ¼ 0:818 1 þ C 2I þ 1 ; ð2Þ
CI above which the mechanical contribution of the infill wall can be
neglected due to its extremely weak expected response.
lw Finally, the definition of the cyclic response of the ‘‘equivalent
C I ¼ 1:925 ; ð3Þ
hw strut’’ model utilised in this paper is described into details in
Section 3.1 and validated with reference to experimental results
where fws is the shear strength of masonry, obtained from diagonal
available in the literature.
compressive tests.
The force Fy at the onset of cracking is assumed to be equal to
0.6 Fm, whereas the maximum horizontal displacement dm is 3. Parametric analysis
defined starting from different limitations of the storey drift. In
particular, the maximum force Fm is reached at a storey drift of A parametric study based on several NLTH analyses performed
0.2% for short walls, 0.15% for long walls with a window and on RC infilled and bare structures is presented in this section with
0.10% for walls with a door opening. the aim of investigating the influence of masonry infills on the glo-
The (negative) stiffness R3 of the post-peak softening branch is bal seismic response of RC frames. These NLTH analyses are pri-
derived by arbitrarily considering the displacement at collapse marily considered as a ‘‘numerical’’ experimental campaign
equal to 5 dm and the residual strength Fu equal to zero. aimed at ‘‘observing’’ the ‘‘actual’’ structural response of both bare
Furthermore, the role of openings is taken into account by intro- and masonry infilled structures.
ducing the following k0 parameter defined as follows [18]: Hence, the NLTH analyses lead at quantifying the f ratio defined
as follows, for each one of the frames under consideration
1:5 lv
k0 ¼ 1 ¼ 1 1:5 a P 0 ð4Þ
lw DNLTH
top;infill
f ¼ ; ð8Þ
where lv is the horizontal length of the opening in the wall under DNLTH
top;bare
consideration. For the sake of simplicity, and in line with the main
objectives of this paper, only window openings in homothetic trans- where DNLTH
top is the top displacement obtained through NLTH analy-
formation with the dimensions of walls are considered in this ses and the subscripts ‘‘bare’’ and ‘‘infill’’ refer to the ‘‘bare’’ and ‘‘in-
study: thus, their size is identified by the following dimensionless filled’’ frames, respectively.
parameter [18]:
a ¼ hv =hw ¼ lv =lw : ð5Þ 3.1. The structures selected for the parametric analysis
where hv and hw and the heights of opening and wall, respectively Two-, four- and six-storey 3D frames were defined by means of
(Fig. 3). a simulated design procedure based on codes and practices
It is worth highlighting that further proposals are available in adopted in Italy in the 1960s and 1970s of the past century
the literature for simulating the effect of more general sizes and [21,22]. Typical details for gravity load design only (i.e. smooth
positions of openings within the masonry walls [19,20]. bars, medium-to-low concrete compression strength, hooks and
However, the reduction factor k0 is considered herein for both largely spaced stirrups) were taken into account. In the design,
strength and initial stiffness R1, which can be easily quantified as infills were considered for their gravitational mass only. Fig. 4
follows [17]: reports the key dimensions in plan of the two structural scheme
under consideration in which three bays were along the y-direc-
G w t w lw
R1;0 ¼ k0 ; ð6Þ tion, whereas three and five bays were considered along the
hw
x-direction. The storey height was assumed equal to 3.50 m.
Moreover, the dimensions and the longitudinal reinforcement of
F m;0 ¼ k0 F m : ð7Þ
the columns are shown in Fig. 4.
According to Eq. (4), a can range between 0 for which k0 = 1 and cor- Table 1 reports the key information about geometry and rein-
responds to infill walls without openings, and 0.667 for which k0 = 0 forcement details of the beams. Transversal reinforcement
594 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
Fig. 4. Structural shape of the RC frames under consideration and column sections.
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 595
Table 2
Distribution of the columns within the structures.
four and two storeys were obtained by duly removing the lower Then, infilled frames were analysed and three different
floors from the six-storey ones. distributions of masonry walls were investigated (Fig. 5). Since
Concrete cubic strength Rcm = 25 MPa (fcm = 20.75 MPa) and the influence of infills on the global response of structures can be
reinforcement steel type AQ40 with fsm = 220 MPa (smooth bars) significantly affected by openings (e.g. windows or doors), opening
were adopted in order to simulate the behaviour of structural ratios as defined in Eq. (5) ranging between 0 and 0.60 were
materials in typical existing buildings designed in Europe in the considered [18].
past decades [21,22]. In the following, the configurations without Two Linear Elastic Design Spectra (LEDS) were considered for a
masonry infills are denoted as ‘‘bare structure’’. Soil Category A as required for ordinary buildings by EN 1998 [10]
Masonry infills were supposed to be made of artificial blocks of for spectra type A in order to simulate seismic actions. Although
expanded clay with a normal wall thickness tw = 30 cm. The very high values of peak acceleration (0.3–0.65g) were recorded
following mechanical properties were taken into account: in recent earthquakes [24], accelerations equal to 0.10g and
0.35g were considered. As a matter of fact, these two levels of shak-
– average compression strength fm = 4.00 MPa; ing intensity are intended at reproducing the seismic input which
– average shear strength fv0 = 0.30 MPa; has to be considered for serviceability and ultimate limit states in
– normal elastic modulus Ew = 3600 MPa; medium-to-strong seismic areas, according to the European and
– shear modulus Gw = 1080 MPa; Italian codes [10,25].
– specific weight c = 8.00 kN/m3. NonLinear Time-History (NLTH) analyses were carried out on
the aforementioned structures; two sets of seven unscaled natural
The loads were determined according to Eurocode 1 [23] by accelerograms were selected from the European Strong Motion
taking into account the self-weight of the structural RC elements, Database [26] by considering the two target LEDS described above.
permanent loads on the floors equal to 5.16 kN/m2 and variable Fig. 6 shows the LEDS and the spectra of the natural accelerograms
loads equal to 3.50 kN/m2 for congregation areas. Moreover, a used in this research, while details about the natural accelero-
distributed permanent load equal to 7.20 kN/m was applied on grams are reported in Tables 3 and 4. Moreover, Fig. 7 represents
the external beams in order to simulate infills. According to a matrix of the analysed structures.
Eurocode 1 [23], the masses equal to 205.90 kN s2/m and The structures described above were modelled in OpenSEES
167.73 kN s2/m were assumed for the intermediate and the roof [27]. The nonlinear behaviour of beams and columns was simu-
floors of the 3-bays structure, respectively, while floor masses lated by employing force-based distributed plasticity elements by
equal to 348.43 kN s2/m and 295.59 kN s2/m were evaluated for using ‘‘nonlinear beam column’’ elements available in the software
intermediate and roof floors of the 5-bays structures, respectively. library [27]. Concrete01 model was employed for modelling both
6.0 20.0
000234ya EQ:108 000055xa EQ:34
18.0
5.0 006115ya EQ:2029 000055ya EQ:34
006326ya EQ:2142 16.0 000287xa EQ:146
4.0 006335xa EQ:2142 14.0 004674xa EQ:1635
Sa,e [m/s2]
Sa,e [m/s2]
1.0 4.0
2.0
0.0 0.0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
T [s] T [s]
Table 3
Key parameters of the natural accelerograms compatible with the LEDS 0.10g.
Waveform ID Station ID Earthquake name Date Mw Fault Epicentral PGA_X (m/s2) PGA_Y (m/s2) PGV_X (m/s) PGV_Y (m/s)
Mechanism distance
(km)
6342 ST2556 South Iceland 21/06/2000 6.4 Strike slip 20 1.0426 0.8446 0.0941 0.1217
(aftershock)
6336 ST2563 South Iceland 21/06/2000 6.4 Strike slip 24 1.1132 0.5434 0.1083 0.0409
(aftershock)
6335 ST2557 South Iceland 21/06/2000 6.4 Strike slip 15 1.2481 1.1322 0.1659 0.1083
(aftershock)
234 ST68 Montenegro 24/05/1979 6.2 Thrust 30 0.6669 0.7541 0.0445 0.0649
(aftershock)
6326 ST2496 South Iceland 21/06/2000 6.4 Strike slip 14 1.7476 1.1423 0.0966 0.1765
(aftershock)
6115 ST1320 Kozani 13/05/1995 6.5 Normal 17 2.0388 1.3962 0.0854 0.0656
6335 ST2557 South Iceland 21/06/2000 6.4 Strike slip 15 1.2481 1.1322 0.1659 0.1083
(aftershock)
Mean: 6.38 19.28 1.3007 0.9921 0.1087 0.0980
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 597
Table 4
Key parameters of the natural accelerograms compatible with the LEDS 0.35g.
Waveform ID Station ID Earthquake name Date Mw Fault Epicentral PGA_X (m/s2) PGA_Y (m/s2) PGV_X (m/s) PGV_Y (m/s)
mechanism distance
(km)
55 ST20 Friuli 06/05/1976 6.5 Thrust 23 3.4985 3.0968 0.2061 0.3262
4674 ST2486 South Iceland 17/06/2000 6.5 Strike slip 5 3.1176 3.3109 0.6122 0.2377
4674 ST2486 South Iceland 17/06/2000 6.5 Strike slip 5 3.1176 3.3109 0.6122 0.2377
55 ST20 Friuli 06/05/1976 6.5 Thrust 23 3.4985 3.0968 0.2061 0.3262
6349 ST2558 South Iceland 21/06/2000 6.4 Strike slip 5 7.2947 8.218 0.4557 0.9202
(aftershock)
287 ST93 Campano Lucano 23/11/1980 6.9 Normal 23 1.3633 1.7756 0.2056 0.3044
6349 ST2558 South Iceland 21/06/2000 6.4 Strike slip 5 7.2947 8.218 0.4557 0.9202
(aftershock)
Mean: 6.54 12.71 4.1693 4.4324 0.3934 0.4675
cover and core concrete (Fig. 8a), while Steel01 stress–strain law ‘‘Pinching4’’ material model and Fig. 10 highlights the significant
with 1% symmetric hardening was adopted for rebars [27] level of accuracy achieved by the adopted model in terms of
(Fig. 8b). Both mechanic and geometrical nonlinear effects were strength and stiffness values, and their degradation under cyclic
considered by means of the ‘‘P-Delta geometric transformation’’ actions.
command implemented in OpenSEES [27].
Truss elements were employed for representing the ‘‘equivalent
struts’’ which are intended to simulate the behaviour of masonry 3.2. Nonlinear Time History analysis (NLTH)
infills: they were supposed to sustain only compressive (negative)
axial forces according to the model proposed by Dolsek and Fajfar The two sets of seven natural accelerograms mentioned in the
[17]. Nonlinear cyclic behaviour was defined through the so-called previous section were used as input signals in NonLinear
‘‘Pinching 4’’ material model available in OpenSEES [27]. Fig. 9a Time-History (NLTH) analyses carried out on the aforementioned
depicts the static behaviour of the diagonal truss for increasing val- structures. Moreover, the investigation was extended to two-,
ues of the opening ratio defined by Eq. (5), whereas Fig. 9b shows a four- and six-storey frames with three and five bays with the dis-
possible cyclic response of the modelled equivalent strut. tribution of masonry infills reported in Fig. 5 and opening ratios a
The cyclic response of the truss element adopted in NLTH anal- equal to 0.00, 0.20, 0.40 and 0.60. However, only analysis in
yses was carefully calibrated by considering some experimental x-direction were performed for the sake of brevity. Fig. 11 depicts
results available in the scientific literature [28]. Particularly, the the cyclic response of a truss element intended at simulating an
cyclic force–displacement response observed in the experimental infill wall located at the first storey of a Type A
tests carried out on four specimens, originally referred to as I1, three-bay-four-storey frame with opening ratio a = 0.00. The cyclic
I2, I5 and I6, were considered for this calibration. Table 5 reports behaviour of the truss observed through NLTH analysis clearly
the numerical values of the parameters adopted for describing demonstrates that the truss element only responds in
the response of the truss elements via the aforementioned compression.
598 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
stress
f sm=220MPa
stress
ε cu=0.0036 ε co=0.0022
strain Es=210000MPa
strain
f cm,u=18.39MPa
f sm=220MPa
f cm=20.75MPa
(a) (b)
Fig. 8. Materials.
-600 -600
a=0.00 Cyclic behaviour
-500 a=0.20 -500
Static envelope
Axial Force [kN]
-200 -200
-100 -100
0 0
Table 5
Numerical values of the parameters adopted for describing the response of the truss via the ‘‘Pinching4’’ material model.
Calcarenite masonry Floating point values defining force (in N) and deformation points on the positive response envelope
infilled frame I1–I2 $ePf1 $ePf2 $ePf3 $ePf4 $ePd1 $ePd2 $ePd3 $ePd4
0.0 0.0 0.0 0.0 1.06e4 1.00e3 5.00e3 5.01e3
Floating point values defining force (in N) and deformation points on the negative response envelope
$eNf1 $eNf2 $eNf3 $eNf4 $eNd1 $eNd2 $eNd3 $eNd4
17,2469 28,7449 0.0 0.0 1.06e4 1.00e3 5.00e3 5.01e3
Clay tile masonry Floating point values defining force (in N) and deformation points on the positive response envelope
infilled frame I5–I6 $ePf1 $ePf2 $ePf3 $ePf4 $ePd1 $ePd2 $ePd3 $ePd4
0.0 0.0 0.0 0.0 1.65e4 1.00e3 5.00e3 5.01e3
Floating point values defining force (in N) and deformation points on the negative response envelope
$eNf1 $eNf2 $eNf3 $eNf4 $eNd1 $eNd2 $eNd3 $eNd4
14,4874 24,1457 0.0 0.0 1.65e4 1.00e3 5.00e3 5.01e3
Common to all Floating point value defining $rDispP $fFoceP $uForceP $rDispN $fFoceN $uForceN
specimens I1–I2–I5–I6 unloading and reloading 0.0 0.0 0.0 0.0 0.0 0.0
branches
Floating point values controlling cyclic $gK1 $gK2 $gK3 $gK4 $gKLim
degradation model for unloading stiffness 2.0 2.0 2.0 2.0 2.0
degradation
Floating point values controlling cyclic $gD1 $gD2 $gD3 $gD4 $gDLim
degradation model for reloading stiffness 0.1 0.1 0.1 0.1 1.0
degradation
Floating point values controlling cyclic $gF1 $gF2 $gF3 $gF4 $gFLim
degradation model for strength degradation 0.001 0.001 0.001 0.001 1.0
Floating point value used to define maximum energy dissipation under cyclic loading $gE 10e8
String to indicate type of damage (option: ‘‘cycle’’, ‘‘energy’’) $dmgType Cycle
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 599
Fig. 10. Simulated and experimental cyclic response of the infilled frames tested by Cavaleri et al. [28].
-100 – the masonry infill opening ratio a or, in a more general sense,
the actual strength/stiffness of the masonry walls;
0
100
15 10 5 0 -5 -10 -15
Table 6
deformation [mm] Top displacement demand (in cm) of the analysed structures by NLTH analysis.
Fig. 11. Cyclic response of the infill resulting from NLTH analysis. Storey Structure 3 bays 5 bays
0.10g 0.35g 0.10g 0.35g
DNLTH
top ðcmÞ DNLTH NLTH
top ðcmÞ Dtop ðcmÞ DNLTH
top ðcmÞ
The results in terms of top absolute displacement demands
2 Bare 2.2 11.3 2.2 11.9
DNLTH
top were evaluated as the average of the maximum response Infilled a = 0.60 1.3 10.2 1.4 10.0
obtained from the seven accelerograms [10]: particularly, Table 6 Type A a = 0.40 0.4 7.9 0.3 8.1
reports the numerical average values of the maximum response a = 0.20 0.3 4.1 0.2 4.3
a = 0.00 0.2 1.3 0.1 1.0
obtained for the two seismic intensities. Furthermore, the values
Infilled a = 0.60 1.6 10.9 1.9 10.3
of Table 6 are graphically represented in Fig. 12. Type B a = 0.40 0.7 9.3 0.7 9.7
It is worth highlighting that no brittle failure mechanism was a = 0.20 0.4 6.6 0.6 7.5
considered in these NLTH analyses. Therefore, they provide an esti- a = 0.00 0.3 4.8 0.4 3.9
mation of the seismic demand without considering the possibility Infilled a = 0.60 1.8 10.5 1.8 10.9
Type C a = 0.40 1.7 10.3 1.7 10.7
of shear failure of members and joints: the occurrence of brittle or
a = 0.20 1.7 10.3 1.7 10.7
ductile failure could be detected in post-processing the analysis a = 0.00 1.7 10.3 1.7 10.7
results by checking the force demand in members and joints and 4 Bare 4.2 16.9 4.1 15.9
comparing it to the corresponding capacity, as usual in Infilled a = 0.60 2.0 14.1 1.9 12.4
practice-oriented analyses. Type A a = 0.40 1.1 9.8 1.1 10.4
As expected, NLTH analyses pointed out that top displacement a = 0.20 0.9 9.7 0.7 8.6
a = 0.00 0.8 8.5 0.5 6.3
are generally (even significantly) smaller in infilled frames than
Infilled a = 0.60 2.2 14.6 2.2 13.8
in bare ones (Fig. 12). Particularly, Table 6 shows that the DNLTH
top val- Type B a = 0.40 1.8 11.6 1.9 11.7
ues are higher for the bare structures and reduces with the opening a = 0.20 1.4 9.6 1.6 10.9
a = 0.00 1.4 8.9 1.5 10.2
ratio a defined by Eq. (5). This is an expected effect due to the
Infilled a = 0.60 2.3 14.2 2.2 13.7
structural contribution of masonry infills to the lateral capacity Type C a = 0.40 1.9 12.5 1.8 11.3
of the structure. However, it should be clear that the results a = 0.20 1.9 12.5 1.8 11.3
reported in Table 6 cannot be merely summarised by claiming that a = 0.00 1.9 12.5 1.8 11.3
the masonry infill walls play a ‘‘beneficial’’ effect on the seismic 6 Bare 6.3 19.9 5.9 19.4
response of RC frames. This is generally true in terms of displace- Infilled a = 0.60 3.1 16.2 2.9 16.8
ment demand, but cannot be extended to the other aspects of Type A a = 0.40 1.9 11.7 1.8 13.6
a = 0.20 1.7 10.6 1.2 11.7
the structural response. In fact, the internal distribution of forces
a = 0.00 1.4 10.6 1.1 11.4
as well as the total force transferred to foundations can be signifi- Infilled a = 0.60 3.6 17.1 3.7 17.6
cantly different (and often higher) than the corresponding values Type B a = 0.40 2.2 13.7 2.3 15.0
determined for a bare RC frame. A thorough discussion about these a = 0.20 2.2 11.2 2.2 14.5
a = 0.00 2.3 12.7 2.2 11.9
effects and the general importance of considering masonry infills
Infilled a = 0.60 3.4 16.0 3.2 17.1
in simulating the seismic behaviour of RC frames is already avail- Type C a = 0.40 2.0 13.1 1.8 14.4
able in the recent scientific literature [8]. The influence of masonry a = 0.20 2.0 11.4 1.9 13.1
infill walls is especially relevant for structures designed for a = 0.00 2.0 10.8 1.9 10.9
600 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
3,00 16,00
bare Type A Type B Type C bare Type A Type B Type C
14,00
2,50
12,00
ΔtopNLTH [cm]
ΔtopNLTH [cm]
2,00
10,00
1,50 8,00
6,00
1,00
4,00
0,50
2,00
Two-storey (0,10g) Two-storey (0,35g)
0,00 0,00
a = 0.00 a = 0.20 a = 0.40 a = 0.60 a = 0.00 a = 0.20 a = 0.40 a = 0.60
(a) (b)
5,00 20,00
4,50 bare Type A Type B Type C bare Type A Type B Type C
4,00 16,00
3,50
ΔtopNLTH [cm]
ΔtopNLTH [cm]
3,00 12,00
2,50
2,00 8,00
1,50
1,00 4,00
0,50
Four-storey (0,10g) Four-storey (0,35g)
0,00 0,00
a = 0.00 a = 0.20 a = 0.40 a = 0.60 a = 0.00 a = 0.20 a = 0.40 a = 0.60
(c) (d)
8,00 24,00
bare Type A Type B Type C bare Type A Type B Type C
7,00
20,00
6,00
ΔtopNLTH [cm]
ΔtopNLTH [cm]
16,00
5,00
4,00 12,00
3,00
8,00
2,00
4,00
1,00
Six-storey (0,10g) Six-storey (0,35g)
0,00 0,00
a = 0.00 a = 0.20 a = 0.40 a = 0.60 a = 0.00 a = 0.20 a = 0.40 a = 0.60
(e) (f)
Fig. 12. Results of NLTH analyses in terms of top absolute displacement demands for the two seismic intensities.
– the seismic intensity level or, in a more general sense, the actual proposes a scalar ‘‘shape’’ parameter to simply describe the afore-
stress level achieved in the masonry wall described by its non- mentioned effects of masonry infills (see Section 4.3). The correla-
linear elastic-softening force–displacement curve (Fig. 11). tion emerging between this parameter and the infilled-to-bare
displacement ratio f is the main finding of this paper and repre-
Particularly, the values of f tend to the unit in the case of ‘‘weak’’ sents the pivot of the proposed simplified procedure for analysing
masonry walls and ‘‘high’’ seismic intensity (i.e. a = 0.60 at peak masonry infilled RC frames via NLS analyses: such a procedure is
ground acceleration 0.35g). On the contrary, the lowest values of finally summarised in Section 4.4.
the f ratio are achieved for ‘‘strong’’ masonry walls and ‘‘low’’ seis-
mic intensity (i.e. a = 0.00 at peak ground acceleration 0.10g). The 4.1. NonLinear Static analysis (NLS)
extreme values achieved by the f ratio are highlighted in bold in
Table 7. NLS (pushover) analyses are proposed with the aim of unveiling
possible correlations between the value of the f ratio and the
4. Formulation of the proposed procedure ‘‘shapes’’ of the two capacity curves obtained for the infilled and
the bare structures; they were performed on the structures
NonLinear Static (NLS) analyses can be carried out on the same described in Section 3.1 and led to obtaining their capacity curves.
set of frames, in both bare and infilled configurations, to under- According to EN 1998 [10] provisions, two different horizontal
stand the influence of masonry walls on the resulting shape of force distributions were applied and increased for performing
the capacity curves. This section summarises the results of such pushover analyses. Particularly, a distribution proportional to the
analyses (see Section 4.1), compares them with the ones obtained shape of the first mode (called ‘‘modal’’ in the following) and one
from the NLTH analyses reported in Section 3 (see Section 4.2) and proportional to the masses (denoted ‘‘uniform’’ in the following)
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 601
Table 7 40%
Values of f ratio obtained from NLTH analysis. Infill A a=0.00
35%
Storey Structure 3 bays 5 bays Infill A a=0.20
30%
Infill A a=0.40
0.10g f 0.35g f 0.10g f 0.35g f
Vb / W [%]
25% Infill A a=0.60
2 Infilled Type A a = 0.60 0.61 0.90 0.64 0.84
20% Bare structure
a = 0.40 0.17 0.70 0.13 0.68
a = 0.20 0.13 0.36 0.08 0.36 15%
a = 0.00 0.11 0.12 0.06 0.08
Infilled Type B a = 0.60 0.74 0.96 0.84 0.86 10%
a = 0.40 0.31 0.82 0.34 0.81 5%
a = 0.20 0.20 0.59 0.26 0.63
a = 0.00 0.16 0.43 0.18 0.33 0%
Infilled Type C a = 0.60 0.83 0.93 0.81 0.92 0.00% 0.50% 1.00% 1.50% 2.00%
a = 0.40 0.79 0.91 0.78 0.90 Δtop / H [%]
a = 0.20 0.79 0.91 0.78 0.89
a = 0.00 0.79 0.91 0.78 0.89 Fig. 14. Capacity curves of the four-storey five-bay structure derived for the modal
4 Infilled Type A a = 0.60 0.47 0.84 0.46 0.78 distribution.
a = 0.40 0.27 0.58 0.26 0.65
a = 0.20 0.21 0.58 0.17 0.54
a = 0.00 0.18 0.51 0.13 0.40
Infilled Type B a = 0.60 0.53 0.86 0.55 0.87 30%
a = 0.40 0.43 0.69 0.47 0.73 Infill A a=0.0
a = 0.20 0.33 0.57 0.40 0.68 25% Infill A a=0.2
a = 0.00 0.33 0.53 0.37 0.64 Infill A a=0.4
Infilled Type C a = 0.60 0.54 0.84 0.53 0.86 20%
Vb / W [%]
Infill A a=0.6
a = 0.40 0.46 0.74 0.45 0.71
a = 0.20 0.45 0.74 0.45 0.71 15% Bare
a = 0.00 0.45 0.74 0.45 0.71
10%
6 Infilled Type A a = 0.60 0.49 0.82 0.48 0.87
a = 0.40 0.30 0.59 0.30 0.70
5%
a = 0.20 0.26 0.53 0.20 0.60
a = 0.00 0.23 0.53 0.19 0.59
0%
Infilled Type B a = 0.60 0.57 0.86 0.62 0.91
0.00% 0.50% 1.00% 1.50% 2.00%
a = 0.40 0.35 0.69 0.39 0.78
a = 0.20 0.35 0.56 0.37 0.75 Δtop / H [%]
a = 0.00 0.36 0.64 0.38 0.61
Infilled Type C a = 0.60 0.53 0.81 0.53 0.88 Fig. 15. Capacity curves of the six-storey five-bay structure derived for the modal
a = 0.40 0.32 0.66 0.31 0.74 distribution.
a = 0.20 0.31 0.57 0.32 0.68
a = 0.00 0.32 0.54 0.32 0.56
Infill A a=0.60 ment demands evaluated through NLS and NLTH analyses are in
40%
Bare structure good agreement for the bare structures and the accuracy of the
30% N2 Method for RC bare structures was demonstrated. Moreover,
the number of bays of the structures, generally, does not affect
20%
the displacement demand. Hence, the following identity, generally
10% accepted in the common practice (at least for regular frames) is
0% assumed in the following of this study:
0.00% 0.50% 1.00% 1.50% 2.00%
Δtop / H [%] DNLTH NLS
top;bare ¼ Dtop;bare : ð9Þ
Fig. 13. Capacity curves of the two-storey five-bay structure derived for the modal In other words, it is once again confirmed that the N2 Method leads
distribution. to an accurate estimation of the displacement demand of bare RC
602 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
Table 8
Top displacement demand (in cm) of the bare structures by N2 Method.
0.10g DNLS
top;bare ðcmÞ 0.35g DNLS
top;bare ðcmÞ 0.10g DNLS
top;bare ðcmÞ 0.35g DNLS
top;bare ðcmÞ
frames (and, then, of their ‘‘performance point’’ on the capacity correlation between f and some relevant parameters related to the
curves determined by NLS analyses). seismic response of the structure as simulated by NLS analyses.
Therefore, a direct estimation of the corresponding displace- Particularly, the following parameter a can be defined by consider-
ment that would be obtained on the corresponding infilled struc- ing the ‘‘shape’’ of the two capacity curves corresponding to the
ture can be, in principle, determined as follows: infilled frame and the corresponding bare one (Fig. 17):
The observations highlighted at the end of Section 3.2 about the Method [9].
general trends of the f ratio can be now turned into a quantitative Actually, the parameter a ranges between 0 and 1: values of a
close to 0 can be achieved for ‘‘strong’’ masonry walls and ‘‘low’’
seismic actions, while values close to the unit correspond to
‘‘weak’’ masonry infills and ‘‘high’’ seismic actions. Moreover, dif-
25
N2-Method 2-storeys ferent values of a can be determined for a given structure (whose
Bare Structures behaviour with and without infills is captured by the two different
20 3-storeys capacity curves schematically depicted in Fig. 17) depending on
the expected seismic demand level. Fig. 18 shows that two differ-
4-storeys
ent values of a can be determined for the two values of displace-
ΔtopNLS [cm]
15
3-bays ment demand, ideally identified with the values DNLS top;bare
5-bays
determined through the NLS analysis (and the N2 Method) on
10 the bare frame for the two design spectra corresponding to the
Earthquake two intensity levels of seismic shaking considered in this study.
intensity
5 Table 9 reports the a values determined for all the structures
0.10g
analysed in Section 3.2 by assuming the displacement demands
0.35g DNLS
top;bare summarised in Table 8.
0
0 5 10 15 20 25
The values of a reported in Table 9 are in a one-to-one corre-
spondence with the ones determined for the f ratio in Table 7.
ΔtopNLTH [cm]
Moreover, it is easy to understand that a and f are clearly corre-
Fig. 16. Top displacement demand of the bare structures: NLTH vs. NLS analysis.
lated as shown by Fig. 19 for the two-storey frames and
NLS
Abare
NLS NLS
top,bare Δ top,bare Δ top
(a) (b)
Fig. 17. Defining the areas under the capacity curves of both infilled (a) and bare (b) structure.
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 603
40% Pn 2
ðf f Þ ðai aÞ
35% Δtop,bare
NLS,0.10g
Δtop,bare
NLS,0.35g
Infilled R2 ¼ P i¼1 i 2 P ð12Þ
n n ða a Þ2
30% Bare i¼1 f i f i¼1 i
Vb / W [%]
25%
in which f and a
are the average values of the f and a parameters,
20% respectively, evaluated on the basis of the n frames under investiga-
15% tion. On the basis of its definition, the coefficient of determination
10% can assume values ranging between 0 and 1, denoting values of
R2 close to the unit if a linear correlation can be defined between
5%
f and a with good agreement.
0% Fig. 22 depicts the f a comparison charts obtained for the
0.00% 0.20% 0.40% 0.60% 0.80%
structures under consideration. The coefficient of determination
Δtop / H [%] is also shown therein. Particularly, Fig. 22a shows the bunch of
points of coordinates (a, f) grouped according the number of stor-
Fig. 18. Definition of the areas under the capacity curves for the two-storey
structure. eys of the analysed structures. Values of R2 equal to 0.892, 0.924
and 0.918 were obtained for the two-, four- and six-storey struc-
tures, respectively. In Fig. 22b the comparison between f and a is
discriminated in respect to the distribution of infill within the RC
Figs. 20 and 21 for four- and six-storey structures, respectively. frames and values of R2 ranging between 0.854 for structures
Indeed, the bunch of (a, f) points represented within the aforemen- Type A and 0.937 for the configuration Type C can be observed.
tioned figures are clearly grouped around the bisector segment. Fig. 22c shows the bunch of points of coordinates (a, f) grouped
Furthermore, quantitative parameters can be introduced for according the opening. Finally, the comparison reported in
quantifying the correlation between f and a. Such a parameter is Fig. 22d highlights the high linear correlation between f and a for
the so-called coefficient of determination R2 providing a measure both low and strong earthquakes.
of the linear correlation that characterises the two key parameters Fig. 23 shows the comparison between f and a obtained for all
under investigation and is defined as follows: the structures and a coefficient of determination R2 equal to
Table 9
Values of the a parameter obtained from NLS analysis.
1.00 1.00
Two-storey Two-storey
frames: f-α frames: f-α
0.80 relationship in 0.80 relationship in
respect to the respect to the
distribution of opening ratios
infills 0.60
0.60 a=0.00 3-bays
Type A 3-bays a=0.00 5-bays
f
f
1.00 1,00
Two-storey Two-storey
frames: f-α frames: f-α
0.80 relationship in 0,80 relationship in
respect to the respect to the
seismic horizontal
intensity pattern
0.60 0,60
modal
f
f
0.10g 3-bays
3-bays
0.40 0,40
modal
0.10g 5-bays
5-bays
uniform
0.20 0.35g 3-bays 0,20
3-bays
uniform
0.35g 5-bays
5-bays
0.00 0,00
0.00 0.20 0.40 0.60 0.80 1.00 0,00 0,20 0,40 0,60 0,80 1,00
α α
(c) (d)
Fig. 19. f a relationship for the two-storey structures.
1.00 1.00
Four-storey Four-storey
frames: f-α frames: f-α
0.80 relationship in 0.80 relationship in
respect to the respect to the
distribution of opening ratios
infills
0.60 0.60 a=0.00 3-bays
Type A 3-bays a=0.00 5-bays
f
f
1.00 1,00
Four-storey Four-storey
frames: f-α frames: f-α
0.80 relationship in 0,80 relationship in
respect to the respect to the
seismic horizontal
intensity pattern
0.60 0,60
modal
f
0.10g 3-bays
f
3-bays
0.40 0,40
0.10g 5-bays modal
5-bays
0.20 0.35g 3-bays uniform
0,20
3-bays
0.35g 5-bays uniform
0.00 5-bays
0,00
0.00 0.20 0.40 0.60 0.80 1.00 0,00 0,20 0,40 0,60 0,80 1,00
α α
(c) (d)
Fig. 20. f a relationship for the four-storey structures.
E. Martinelli et al. / Engineering Structures 101 (2015) 591–608 605
1.00 1.00
Six-storey Six-storey
frames: f-α frames: f-α
0.80 relationship in 0.80 relationship in
respect to the respect to the
distribution of opening ratios
infills
0.60 0.60 a=0.00 3-bays
Type A 3-bays a=0.00 5-bays
f
f
0.10g 3-bays
3-bays
0.40 0,40
modal
0.10g 5-bays
5-bays
uniform
0.20 0.35g 3-bays 0,20
3-bays
uniform
0.35g 5-bays
5-bays
0.00 0,00
0.00 0.20 0.40 0.60 0.80 1.00 0,00 0,20 0,40 0,60 0,80 1,00
α α
(c) (d)
Fig. 21. f a relationship for the six-storey structures.
Dtop;infill ¼ a DNLS
top;bare ð14Þ
0.884 was observed. On the basis of the qualitative and quantita-
tive observations explained above, the f a relationship was
approximated through the following linear law: obtained by introducing the analytical correlation (13) within
the Eq. (10).
f ðaÞ ¼ a: ð13Þ
5. Further considerations about the accuracy of the proposed
4.4. Outline of the proposed procedure simplified procedure
The correlation between the parameters f and a emerged at the The results obtained by applying the proposed procedure on the
structures under investigation are presented herein and compared
end of Section 3 and the other observations pointed out by the
results of the parametric analysis reported therein directly lead with the ones obtained through NLTH analyses.
According to EN 1998 [10] provisions, the displacement
to the proposed procedure for determining the seismic response
of masonry infilled RC frames. Particularly, this procedure is based demand of the structures evaluated through NLS analysis was
taken equal to the maximum value between the ones obtained
on the following operational steps:
applying the two different horizontal force distributions ‘‘modal’’
and ‘‘uniform’’. Hence, Fig. 24 shows the comparison between the
1. performing two NLS analyses on both the masonry infilled
structure and the corresponding bare one according to the cur- (average of seven) displacement demand values determined by
rent code provisions [10,25]; means of the top displacement Dtop;infill ¼ DNLTH
top;infill obtained through
2. determining the performance point of the bare structure NLTH analyses (x-axis) and the corresponding value determined
DNLS for the earthquake of interest through the N2 via the proposed procedure based on NLS analyses and described
top;bare
in Section 4. The four charts reported in Fig. 24, in which the results
Method [9];
were grouped for highlighting the influence of the parameters
3. evaluating the areas under the capacity curves of both the
affecting the response of the masonry infilled structures under
infilled ANLS NLS
infill and the bare Abare structures, up to a displacement consideration, clearly confirm the high level of correlation between
equal to the demand DNLS
top;bare determined at the point 2; the numerical results estimated via NLTH analysis and the corre-
4. evaluating the parameter a by applying Eq. (11); sponding ones derived by Eq. (14). Moreover, the significant level
5. evaluating the displacement capacity Dtop;infill of the infilled of accuracy of the proposed procedure is demonstrated. Fig. 25 out-
structure by means of the simple equation: lines this comparison and reports the value R2 = 0.934 obtained for
606 E. Martinelli et al. / Engineering Structures 101 (2015) 591–608
1.00 1.00
2-storeys Type A
f
2 2
R =0.892 R =0.854
0.40 (2-storeys) 0.40 (Type A)
2 2
R =0.924 R =0.866
0.20 (4-storeys) 0.20 (Type B)
2 2
R =0.918 R =0.937
(6-storeys) (TypeC)
0.00 0.00
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
α α
(a) (b)
1.00 1.00
a=0.00
0.10g
a=0.20
0.80 0.80
a=0.40
0.35g
0.60 a=0.60 f 0.60
f
2
R =0.786 2
0.40 (a=0.00) 0.40 R =0.887
2
R =0.875 (0.10g)
(a=0.20)
0.20 R2=0.921 0.20 2
(a=0.40) R =0.763
R2=0.809 (0.35g)
0.00 (a=0.60) 0.00
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
α α
(c) (d)
the coefficient of determination. Particularly, the comparison chart Finally, with the aim of providing Readers with a measure, that
in Fig. 25 demonstrates the very good agreement between the could be directly employed in simplified reliability analyses to take
numerical results estimated via NLTH analysis and the correspond- into account the so-called ‘‘epistemic uncertainty’’ [29], the follow-
ing ones derived by the proposed NLS procedure. ing reliability factor bD intended at covering the aforementioned
Furthermore, in order to provide a quantitative measure of the uncertainties was evaluated as the standard deviation of the natu-
accuracy of the proposed procedure, the distribution of the ratio x ral logarithm of the ratio x [30]:
between the displacement demand obtained by the proposed pro-
bD ¼ rðln xÞ: ð16Þ
cedure and the NLTH analyses was investigated:
DNLS
top;infill 1.00
x¼ : ð15Þ
DNLTH
top;infill
structures
0.80 as a whole
Moreover, the relevant statistical parameters (i.e., average value l
and standard deviation r) of the x ratio were determined and 0.60
Fig. 26 shows its cumulative distribution of frequency.
f
25.00 25.00
2-storeys Type A
Δtop,infillNLS [cm]
Δtop,infillNLS [cm]
2 2
R =0.953 R =0.940
10.00 (2-storeys) 10.00 (Type A)
2 2
R =0.979 R =0.932
5.00 (4-storeys) 5.00 (Type B)
2 2
R =0.957 R =0.935
(6-storeys) (TypeC)
0.00 0.00
0.00 5.00 10.00 15.00 20.00 25.00 0.00 5.00 10.00 15.00 20.00 25.00
Δtop,infillNLTH [cm] Δtop,infillNLTH [cm]
(a) (b)
25.00 25.00
a=0.00
0.10g
a=0.20
20.00 20.00
a=0.40
Δtop,infillNLS [cm]
Δtop,infillNLS [cm]
0.35g
15.00 a=0.60 15.00
2
R =0.961
2
10.00 (a=0.00) 10.00 R =0.854
2
R =0.967 (0.10g)
(a=0.20)
5.00 R2=0.923 5.00 2
(a=0.40) R =0.796
R2=0.935 (0.35g)
0.00 (a=0.60) 0.00
0.00 5.00 10.00 15.00 20.00 25.00 0.00 5.00 10.00 15.00 20.00 25.00
Δtop,infillNLTH [cm] Δtop,infillNLTH [cm]
(c) (d)
Fig. 24. Accuracy of the proposed method and coefficients of determination R2.
25.00
1.00
0.90
structures 0.80
20.00 as a whole
0.70
Δtop,infillNLS [cm]
0.60
Pr(xi)
15.00 0.50
xˆ = 0.926
0.40
95%=1.286
5%=0.652
10.00 0.30
2
R =0.934 0.20
(structures 0.10 σ =0.323
5.00 as a
0.00
whole) 0.00 0.50 1.00 1.50 2.00
x
0.00
0.00 5.00 10.00 15.00 20.00 25.00
Fig. 26. Cumulative distribution of the actual-to-predicted top displacement ratio.
Δtop,infill NLTH [cm]
Fig. 25. Accuracy of the proposed method and coefficients of determination R2 for
all the structures under consideration. 6. Conclusions
25.00 References
Δtop,infillNLS eβD
structures [1] Penelis GG, Kappos AJ. Earthquake-resistant concrete structures. 1st
20.00 as a whole ed. Abington, Oxon (UK): Taylor & Francis; 1997.
[2] Kakaletsis DJ, Karayannis CG. Influence of masonry strength and openings on
Δtop,infillNLS [cm]