You are on page 1of 24

This article was downloaded by: [Universita Studi la Sapienza]

On: 02 September 2014, At: 01:12


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ueqe20

Floor Response Spectra for Bare and


Infilled Reinforced Concrete Frames
a

Andrea Lucchini , Fabrizio Mollaioli & Paolo Bazzurro


a

Department of Structural and Geotechnical Engineering, Sapienza


University of Rome, Rome, Italy
b

University Institute for Advanced Studies (IUSS), Pavia, Italy


Accepted author version posted online: 22 Apr 2014.Published
online: 25 Aug 2014.

To cite this article: Andrea Lucchini, Fabrizio Mollaioli & Paolo Bazzurro (2014) Floor Response
Spectra for Bare and Infilled Reinforced Concrete Frames, Journal of Earthquake Engineering, 18:7,
1060-1082, DOI: 10.1080/13632469.2014.916633
To link to this article: http://dx.doi.org/10.1080/13632469.2014.916633

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of all the information (the
Content) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

Journal of Earthquake Engineering, 18:10601082, 2014


Copyright A. S. Elnashai
ISSN: 1363-2469 print / 1559-808X online
DOI: 10.1080/13632469.2014.916633

Floor Response Spectra for Bare and Infilled


Reinforced Concrete Frames
ANDREA LUCCHINI1 , FABRIZIO MOLLAIOLI1 ,
and PAOLO BAZZURRO2
Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Department of Structural and Geotechnical Engineering, Sapienza University of


Rome, Rome, Italy
2
University Institute for Advanced Studies (IUSS), Pavia, Italy
The objective of this article is to study the effects of structural nonlinear behavior on Floor Response
Spectra (FRS) of existing reinforced concrete frames. This study examines how the FRS vary with
the level of post-elastic behavior in buildings of different number of stories and masonry infill wall
configurations. The effect of damping modeling assumptions is also investigated. Differences and
similarities with findings from the literature are discussed. On the basis of the obtained results, a
commentary on the adequacy of basic assumptions used in predictive equations proposed by different
seismic codes is offered.
Keywords Non structural Elements; Floor Acceleration Response Spectra; Peak Floor
Accelerations; Reinforced Concrete Frame Structures; Post-Elastic Nonlinear Response; Infill Walls;
Damping Model Effects; Code Prediction Equations

1. Introduction
Non structural elements are those components of the building that are supported by the
structure and that are designed not to contribute to the carrying of gravity or lateral loads.
They are usually divided into the following main categories (e.g., see FEMA 74, 1994):
architectural elements (e.g., ceilings, partitions and panelling), building utility systems
(e.g., mechanical and electrical equipment such as elevators and solar panels), and furniture
and contents (e.g., library stacks and vending machines). These components make up for
a large portion of the replacement cost of a typical building. Past earthquakes have shown
how vulnerable these elements can be to earthquake-induced excitation. Frequently non
structural elements sustain more severe damage than the structure that supports them. Even
in those cases when earthquakes have caused minor or no damage to structural systems, non
structural elements have often suffered extensive damage mainly due to improper anchorage and bracing of components. Reducing seismic damage to these elements is fundamental
not only for economic reasons but also for the maintenance of the functionality of the building immediately after the earthquake. In addition, severe damage to non-structural elements
(e.g., falling chimneys and ceiling panels) can increase the risk of fatal injury to the occupants of the building. The photographs in Fig. 1, which were taken after the 2009 LAquila
earthquake in central Italy, show how non-structural damage can injure or kill occupants

Received 29 July 2013; accepted 16 April 2014.


Address correspondence to Andrea Lucchini, Department of Structural and Geotechnical Engineering,
Sapienza University of Rome, Via A. Gramsci 53, 00197 Rome, Italy. E-mail: andrea.lucchini@uniroma1.it

1060

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Floor Response Spectra for RC Frames

1061

FIGURE 1 Collapse of suspended ceilings: pictures (courtesy of Paolo Clemente) from


LAquila earthquake (Italy, 6th of April 2009). Note that these buildings suffered no visible
structural damage. Paolo Clemente and Fabrizio Mollaioli. Reproduced by permission
of Paolo Clemente and Fabrizio Mollaioli. Permission to reuse must be obtained from the
rightsholder.
and suggest how lengthy and expensive the repair process can be. Interestingly, both buildings in Fig. 1 did not experience any visible structural damage. Based on these premises,
it is evident how important it is in seismic assessment and design of building to accurately
evaluate the response of non structural elements to earthquake excitation.
The gold standard for the development of vulnerability relationships that link the
expected behavior of non structural elements to the level of acceleration they experience
during an earthquake is based on empirical data from past events. However, empirical,
good-quality data for many non structural components are scarce, and therefore, vulnerability relationships heavily rely on analytical response computations that are often based
on nonlinear dynamic analysis.
The seismic demand of elastic non-structural elements mounted on buildings experiencing linear elastic response to seismic excitation has been exhaustively investigated
[Villaverde, 1997] and many code-oriented predictive equations have been proposed (e.g.,
starting from those of Singh, 1975; to the most recent ones proposed in Reinoso and
Miranda, 2005; and Singh et al., 2006a,b). On the other hand, the number of studies on
the seismic response of non-structural elements attached to damaged buildings is limited.
Pioneering works on this topic of research are those of Lin and Mahin [1985] and Chen
and Soong [1988]. Despite the efforts spent, significant questions of how the floor response
spectra are affected by the nonlinear behavior of the supporting structure still remain unresolved. According to recent studies carried out by Chaudhuri and Villaverde [2008] and
Chaudhuri and Hutchinson [2011], structural damage may also cause amplifications of the
floor spectral ordinates at some oscillator periods when compared to those obtained when
the structure is artificially forced to remain linear during shaking. In particular, these studies state that significant amplifications can be observed when (a) the ground motion has
near-fault pulse-like characteristics with a dominant period close to the fundamental period
of the building; and (b) when the non structural component is located at the lower floors
and is in resonance with the higher modes of the building. In Sewell et al. [1986] it is
shown that amplifications of floor response spectral ordinates can also occur when buildings experience localized inelastic behavior. While studying the nonlinear seismic response
of particular precast concrete wall structures, Rodriguez et al. [2002] found that at the fundamental period of the structure the floor response spectral ordinates are always lower

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1062

A. Lucchini, F. Mollaioli, and P. Bazzurro

than those found by imposing linear elastic structural behavior. At the higher periods of
vibration, on the other hand, they observed either a slight reduction or amplification. In contrast, results of a study by Medina et al. [2006] on moment-resisting frame structures with
distributed inelasticity showed that due to nonlinearity the floor response spectral ordinates
are always reduced, although they are more significantly reduced at periods close to the
fundamental period of the structure. Also, Taghavi and Miranda [2006] found reduction
of floor spectral ordinates at all periods, although they observed that the reductions produced by structural nonlinearity can be larger at periods close to those of the higher modes
of the structure than at the fundamental period of the structure. Sankaranarayanan and
Medina [2007] also studied the factors that most affect the floor response spectra in structures that undergo post-elastic nonlinear response. They showed that the most important
factors controlling the ordinates of the floor response spectra are the periods of vibration of
the structure, building height, floor level, and nonlinearity level of the structural response.
Few studies focused on floor response spectra in buildings where the presence of infill
walls was modeled (e.g., see Mollaioli et al., 2010). Some investigations of the capability
of several intensity measures to predict the acceleration response of non-structural elements
in infilled frames have been carried out (e.g., see Lucchini et al., 2013), but specific and
exhaustive studies on the influence of infill damage on floor response spectra have not yet
been undertaken.
Although these studies have led to interesting findings, general consensus about the
effects on the floor response spectra caused by the nonlinear response of the building is still
lacking. The objective of the present work is to contribute to this topic of research by reporting and discussing results from analyses carried out on bare and infilled reinforced concrete
frames, which, as will be shown later, respond differently to seismic excitation than the
buildings considered in the aforementioned studies. The investigated frames were selected
so as to be representative of strong beams-weak columns structures, designed according
to seismic codes that lack any consideration to capacity design. The rest of this article
is organized as follows. Section 2 describes the four structures considered in this study.
Section 3 presents the floor response spectra produced in these buildings by a large number of input ground motion records, and provides comments on the characteristics of the
spectra in relation to the post-elastic nonlinearity level of the building response. Section 4
offers comments on the limits of applicability of the prediction equations for FRS proposed
by different seismic codes.

2. Studied Buildings
The buildings selected as case studies are a 2-bay 4-story and a 2-bay 6-story structures
representative of existing constructions from Southern Europe. They are reinforced concrete frames designed according to the seismic code in force in Italy between 1996 and
2008. For each of the two frames, two infill wall configurations are considered: one bare
and one with brick infills at all stories representing internal partitions or weak perimeter
walls. The four frames will be referred to as 4b, 4w, 6b, and 6w. The number indicates the
number of stories and the letters b and w stands for bare and with walls, respectively.
The buildings have a stiffness and mass distributions that are regular in elevation, and, as
mentioned earlier, they are characterized by strong beams and weak columns (due to the
then-current code that lacked capacity design provisions). The length of each bay and the
height of each story are identical in all four models. Information on the span length and the
story height, as well as details on cross-sections dimensions and reinforcement of the structural members are given in the Appendix. Table 1 shows the results from modal analyses
carried out on these building models.

Floor Response Spectra for RC Frames

1063

TABLE 1 Periods (T) and participating mass ratios (PMR) of the first three modes of the
studied buildings. The modal analysis is run with stiffness matrix computed for structural
elements cracked after the application of the vertical loads
4b
Mode

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1
2
3

4w

6b

6w

PMR

PMR

PMR

PMR

0.89s
0.28s
0.15s

80%
11%
5%

0.30s
0.10s
0.06s

86%
10%
2%

1.22s
0.39s
0.21s

75%
11%
5%

0.44s
0.14s
0.08s

81%
11%
3%

Two sets of models were built for the 4b, 4w, 6b, and 6w frames: nonlinear models and equivalent-linear models. The seismic behavior of these models is evaluated
using OpenSees [McKenna et al., 2007]. In the first set, structural members are modelled
with force-based nonlinear elements characterized by distributed inelasticity. Second-order
effects due to geometric nonlinearity are accounted for through p-delta geometric coordinate transformations. The response of the infill walls is estimated using the equivalent
strut model proposed by Decanini and Fantin [1986] and Bertoldi et al. [1993]. The model
consists of a system of diagonal struts connected to the nodes of the frame that act only
in compression with a hysteretic behavior characterized by stiffness and strength degradation and loop pinching (Fig. 2). The strength capacity of the strut is evaluated considering
four possible in-plane failure modes of the wall: diagonal tension, diagonal compression,
shear failure at the horizontal wall-to-frame interface, and crushing of the wall corners in
contact with the frame. Out-of-plane failure modes are excluded. The parameter values of
these struts were calibrated after the work of Decanini et al. [2004] on typical infill walls
frequently used in Italy. These walls are made of 120 mm thick hollow bricks, and a mixture of cement, sand and lime mortar characterized by a compressive and shear strength
(the latter evaluated through diagonal compressive test) equal to 1.2 and 0.2 MPa, respectively, and an initial elastic modulus of 1,050 MPa. The second set of models comprises
equivalent-linear models of the 4b and 6b bare frames. These models are built using the
initial stiffness values of the structural elements of the corresponding nonlinear models,
with the initial stiffness calculated as the secant stiffness at the gravity load deformations.
A Rayleigh damping proportional to the mass and stiffness matrix is considered, with coefficients calibrated to provide a 5% damping at the first and third mode periods. In the
nonlinear models, the stiffness values used for computing damping coefficients and damping matrix are the initial one and the tangent one, respectively. As a consequence of this
modeling assumption, when the stiffness of the system drops to post-yield levels, the viscous damping forces greatly reduce and the energy is mainly dissipated through hysteretic

beam-column element

strut element
force
disp

FIGURE 2 Infill wall model: struts system and horizontal constitutive behavior.

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1064

A. Lucchini, F. Mollaioli, and P. Bazzurro

behavior. In the nonlinear models of the bare frames, two additional alternative damping
modeling approaches were also considered: a Rayleigh damping model proportional to the
mass and the initial stiffness matrix, and a tangent stiffness proportional damping with the
proportionality constant calibrated to provide 5% damping at the first mode period of the
system. The latter model, recently investigated by Petrini et al. [2008] among others, allows
the limitation of damping forces after the system yields.
The evaluation of the effects of the nonlinear behavior experienced by structural elements on the floor acceleration response spectra is obtained by comparing the responses
of the linear and nonlinear sets of models. The analyses results obtained adopting different
damping models are used to show the sensitivity of floor acceleration response spectra to
damping modeling assumptions. The explicit modeling of infill partitions allows to evaluate
the effects of damage of masonry infill walls (frames 4w and 6w) on the seismic behavior
of the frames, and thus, on the floor response spectra.
The dynamic response of the studied building models is evaluated via time-history
analyses by using in total an ensemble of 72 ground motions from 26 worldwide earthquakes. The used accelerograms, whose details can be found in the Appendix, are taken
from the Pacific Earthquake Engineering Research (PEER) Next Generation Attenuation
(NGA) database [PEER, 2011]. All of them have a usable frequency range that includes
the frequency range of the response of these buildings. The ensemble was not selected based
on a narrow range of magnitude values of the causative events nor of source-to-site distance
values. The suite of accelerograms consists in general of far-field standard records that
do not include any recognizable pulse in the velocity trace.
In each time-history analysis, the Maximum Inter-story Drift Ratio (MIDR) is also
recorded in addition to the floor accelerations. The MIDR, a parameter commonly considered as a simple global measure of structural damage, is used here to identify three
different ranges of the nonlinear response of the buildings, that is, a nearly linear (N-L)
range, a limited damage (L-D) range, and a significant damage (S-D) range. The MIDR values identifying the three different ranges, which are summarized in Table 2, were defined
based on the results of pushover analyses carried out on the bare frames (see Fig. 3). In the
N-L range the structure is expected to experience mild hinges on few elements without
a significant decrease in the lateral stiffness of the building. The lateral stiffness of the
building starts to significantly decrease in the response to loads that push the structures
into the L-D range. In the S-D range, the fully yielded lateral strength of the structure is
reached. Considerations on the effects of damage on FRS will only be made based on analyses conducted for records that cause these damage ranges in the structures. Note that at
deformations close to those expected at the onset of collapse, which in this study are associated to MIDR values greater than 4% for the 4-story and 5% for the 6-story buildings, the
seismic demand evaluation of non structural elements is, of course, not a major concern and
therefore, the floor accelerations recorded for this range of responses will not be discussed.

TABLE 2 Maximum Inter-story Drift Ratio (MIDR) values defining three different ranges
of nonlinear response of the studied buildings: the nearly linear (N-L) range, limited
damage (L-D) range, and significant damage (S-D) range
Range
N-L
L-D
S-D

4-story

6-story

MIDR<1%
1%<MIDR<2%
2%<MIDR<4%

MIDR<2%
2%<MIDR<3%
3%<MIDR<5%

Floor Response Spectra for RC Frames


6story bare

0.5

0.5

0.4

0.4
Vbase/W

Vbase/W

4story bare

0.3
0.2

0.3
0.2
0.1

0.1
0

MIDR [%]

2 3 4
MIDR [%]

FIGURE 3 Results of pushover analyses carried out on the 4b and 6b frame, with a lateral
force load vector proportional to the fundamental mode of the structure: lateral base shear
Vbase (normalized by the seismic weight W) vs. Maximum Inter-Story Drift Ratio (MIDR).

3. Floor Acceleration Response Spectra


This section discusses the results of the dynamic analyses on the bare and infilled frame
buildings carried out on both sets of equivalent-linear and nonlinear models. In particular, Peak Floor Accelerations (PFA) and Floor Response Spectra (FRS) obtained from
the selected input ground motions were monitored. PFA at a given story is defined as the
absolute value of the peak of the acceleration time history at that story. The period values considered for calculating the FRS are limited within the range of usable frequencies
of the selected input ground motion records as provided in the PEER NGA database flatfile. The (5%-damped) pseudo-acceleration response spectra of the ground motions records
used to analyze the 4-story and the 6-story buildings are displayed in Fig. 4. Three spectra
are shown for each building: the average spectrum of all the records causing a response of
the bare frames in the N-L range, the average spectrum of all the records that cause a L-D
response of such bare frames and, finally, the mean spectrum of all the strongest records,
namely those that induce a S-D response of the bare frames.
2

1.5

1.5
Sa [g]

Sa [g]

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1065

0.5

0.5
0

2
T [s]

T [s]

FIGURE 4 5%-damped mean response spectra of the records used to analyze the 4-story
building (left plot) and the 6-story building (right plot) in the N-L, L-D, and S-D ranges of
response.

1066

A. Lucchini, F. Mollaioli, and P. Bazzurro

4
Floor

0.2

0.5

0.8
1.1
PFAL [g]

0.2

1.4

4
Floor

Floor

The results obtained with both sets of equivalent-linear and nonlinear models are summarized and shown side by side in Fig. 5, which displays the distribution of peak floor
accelerations of the 4b and 6b frames (top and bottom plots, respectively) obtained by separately averaging the response of all the records that are in the N-L, L-D, and S-D ranges
of structural damage. (Note that no structural damage is modelled in the equivalent-linear
case but the attribution of records to a damage range is based only on the level of MIDR.)
By comparing the top and bottom plots of the figure, it can be noticed that the peak values
of the ground acceleration recorded at the base of the 4b and 6b frame models are different.
This is simply due to the fact that the intensity of the ground motions that produces the considered levels of structural damage in the two buildings is different. In the linear-equivalent
models, the PFA distributions along the buildings height (left panels) are characterized by
a similar profile in both frames: the accelerations initially tend to increase linearly with the
floor level, then remain almost constant and finally increase at the floors closer to the roof.
The sharp increase of the floor acceleration at the top of the building has been observed
also in other studies from the literature (e.g., Rodriguez et al., 2002; Taghavi and Miranda,
2005). In Singh et al. [2006a] this increase has been explained as a whiplashing effect
due to the higher modes. Also, the observed trend for the recorded PFA profiles can be
attributed to higher mode effects. In fact, in the case studies the contributions of higher
modes are non-negligible because of the large values of the ground motion spectral ordinates at these shorter periods of vibration. This can be observed, for example, by inspecting

Floor

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

3.1. Bare Frames (4b and 6b)

1
0.5

0.8
1.1
PFAL [g]

1.4

0.8
1.1
PFA [g]

1.4

0.5

0.8
1.1
PFA [g]

1.4

0.2

0.5

0.2

FIGURE 5 Peak Floor Accelerations (PFA) for the 4b frame (top row) and 6b frame
(bottom row) obtained from both the linear (left) and the nonlinear models of the buildings. The dotted, dashed, and solid curves represent the mean (over all the records of the
ensemble) PFA distribution along the height of the frame in the N-L, L-D, and S-D ranges,
respectively.

Floor Response Spectra for RC Frames


4

1067

3
2.5

3
Sa/PGA

Floor

2
2

1.5
1

1
0.5

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

0
s

T3 T2
0

T1
1

T [s]

FIGURE 6 First three modal shapes (first solid, second dashed, and third dotted) of the
4b frame (left) and 5%-damped pseudo-acceleration response spectrum of one record from
the ensemble (right). T1 , T2 , and T3 are the first three modal periods.
the plots in Fig. 6, obtained from one record in the ensemble. This figure depicts the first
three mode shapes on the left, and the pseudo-acceleration response spectrum with the corresponding periods of vibration identified on the right. From these plots it can be noted that
while the second mode contributes more to the response of the system at the roof level, the
third mode appears to be more significant at the first floor.
When the floor accelerations are evaluated using the nonlinear models (right panels
in Fig. 5), the average shape of the PFA distributions differs from the one obtained using
the equivalent-linear models. In particular, in the 6b frame building the average PFA profile at the upper floor levels becomes almost constant because of the significant reduction
of the floor acceleration at the roof. The reduction at the higher stories is even more significant in the 4b frame. In the nonlinear 4b case, the PFA distribution along the height
tends to a parabolic profile, with the maximum value occurring at the lower floor levels
of the building. It is worth noting that the observed differences between the PFA distributions obtained with the two types of models could have been reduced, especially in the N-L
range of response, if the equivalent-linear models were calibrated differently. For example,
an ad-hoc evaluation of the stiffness values of the cracked structural elements at the specific intensity level considered for the seismic loads would have definitely decreased the
differences between the responses of both models. In general, it is apparent that the effect
of structural damage tends to reduce the mean PFA along the height of the building and
particularly its maximum value. In the S-D range of response of both the 4b and 6b frames,
this reduction of the maximum value of the mean PFA is about 50%.
Moving from peak values to spectra, Figs. 7 and 8 show a comparison of the 2%damped acceleration FRS at the first and top floors obtained from both sets of models
for the 4b and 6b frames, respectively. At the roof level, the effect of structural damage
on the response both shifts the FRS peaks to longer periods and significantly reduces the
corresponding peak acceleration values. At the first floor, a similar but less dramatic trend
can be observed. Note, however, that the peaks of the FRS close to the third mode in the
nonlinear case are often slightly higher than those in the linear case especially for the 4b
frame.
Figure 9 shows the mean acceleration FRS obtained using only the set of nonlinear
models of the 4b and 6b frames. Again, the averaging is done by separately grouping the
results from records that caused damage in the three different specified ranges. In the N-L

1068

A. Lucchini, F. Mollaioli, and P. Bazzurro


Limited Damage

4
2

0.5

1
1.5
Tc [s]

4
2

0.5

1
1.5
Tc [s]

4
2
0

Sa [g]

Sa [g]

Sa [g]

Sa [g]

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Significant Damage

Sa [g]

Sa [g]

Nearly Linear
8

4
2

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

4
2

0.5

1
1.5
Tc [s]

FIGURE 7 Comparison between the mean FRS at the roof (top) and first story (bottom)
obtained with the linear (dotted line) and nonlinear (solid line) models of the 4b frame. The
plots on the left column refer to the records in the N-L range, in the center to the L-D range,
and on the right to the S-D range.
case narrow-band peaks at periods close to the first three periods of vibration of the structures can be clearly identified. In the two more severe damage cases the peaks are still
pronounced but, as expected, shifted towards longer periods. In the 4b frame, ratios of the
periods of these three shifted peaks to the first three mode periods (see T1, T2 , and T3 in
Table 1) are equal to 1.3, 2.0, and 2.2, respectively. In the 6-story case, the period elongation is less significant, and these ratios are equal to 1.1, 1.5, and 1.7, respectively. Table 3
reports the results of modal analyses carried out on linear-models of the 4b and 6b frames
considering the secant-to-yielding stiffness instead of the initial stiffness (as in Table 1) for
the elements. The mode periods in Table 3 correspond closely to the period values of the
FRS peaks obtained in the L-D and S-D ranges. This implies that the shift of the FRS peaks
in the response is due to further cracking of the structural elements and to the yielding of
the rebars. By observing the plots in Fig. 9 it is also interesting to note that the maximum
floor spectral acceleration does not always occur at the roof level: in the 4b frame, in fact,
the contribution of the higher modes at lower floors produces the maximum peaks in the
FRS.
The period at which the FRS exhibit the highest amplification with respect to the corresponding PFA values is also not the same at all floors and for all damage levels. Figure 10,
which depicts the same FRS as in Fig. 9 but normalized by the mean PFA value, indicates that for the 4b frame in the S-D range of response the maximum amplification occurs
around T3 at the second story and around T1 at the fourth story. In the N-L and L-D range,
the maximum amplification at the fourth story occurs instead around T2 . The discrepancy
in the maximum amplification pattern can be explained by the different contribution to the
building response at the various stories from the different modes of vibration. It is also

Floor Response Spectra for RC Frames

4
2

0.5

1
1.5
Tc [s]

4
2

0.5

1
1.5
Tc [s]

0
0

4
2
0

Sa [g]

Sa [g]

Sa [g]

Sa [g]

Sa [g]

Sa [g]

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Significant Damage

Limited Damage

Nearly Linear
8

4
2

0.5

1
1.5
Tc [s]

1069

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

4
2

0.5

1
1.5
Tc [s]

0
0

FIGURE 8 Comparison between the mean FRS at the roof (top) and first story (bottom)
obtained with the linear (dotted line) and nonlinear (solid line) models of the 6b frame. The
plots on the left column refer to the records in the N-L range, in the center to the L-D range,
and on the right to the S-D range.
TABLE 3 Periods (T) and participating mass ratios (PMR) of the first three modes of
vibration. The modal analysis considered the effective stiffness of the yielded-elements of
the corresponding nonlinear models
4b
Mode
1
2
3

6b

PMR

PMR

1.30s
0.49s
0.29s

80%
11%
5%

1.50s
0.54s
0.32s

75%
11%
5%

important to note, from Fig. 10, that the maximum amplification of the normalized mean
FRS obtained using the nonlinear models never exceeds the value of 6 both in the 4b and 6b
frames. However, it is worthwhile remarking that the amplification level strongly depends
on the considered value of the component damping ratio. If the spectra were calculated
with a damping ratio different from 2%, the value of 5 would certainly change.
Figures 11 and 12 show PFA and FRS obtained with damping models alternative to
Rayleigh damping built with mass and tangent stiffness matrix. As shown in the plots
of Fig. 11, it is observed that the floor accelerations can be significantly underestimated
when a Rayleigh damping with an initial stiffness proportional term approach is adopted.
In general, the underestimation increases with the increase of structural damage. In the
investigated case studies, a maximum value of about 20% was observed (see the PFA value

1070

A. Lucchini, F. Mollaioli, and P. Bazzurro


Nearly Linear

Limited Damage

floor2
floor4

Sa [g]

2
1
0

0.5

1
1.5
Tc [s]

Sa [g]

Sa [g]

floor2
floor4
floor6

3
2
1

0.5

1
1.5
Tc [s]

2
1

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

2
1

0.5

1
1.5
Tc [s]

FIGURE 9 Mean 2%-damped FRS for the 4b frame (top) and 6b frame (bottom) obtained
by separately averaging the spectra for all records producing a given level of damage in the
nonlinear structural model.
Nearly Linear

Sa/PFA

4
3
2

3
2

0
0

0.5

1
1.5
Tc [s]

floor2
floor4
floor6

Sa/PFA

4
3
2
1
0

Significant Damage
6

Sa/PFA

floor2
floor4

5
Sa/PFA

Limited Damage
6

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

3
2
1

Sa/PFA

Sa/PFA

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Sa [g]

Sa [g]

Significant Damage

Sa [g]

0
0

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

3
2
1

0.5

1
1.5
Tc [s]

FIGURE 10 Mean 2%-damped FRS for the 4b frame (top) and 6b frame (bottom) normalized by the mean PFA of each set of records causing damage in the three different ranges
of response.

Floor Response Spectra for RC Frames

0.2

0.5

0.8
1.1
PFA [g]

1.4

0.2

0.5

0.8
1.1
PFA [g]

1.4

FIGURE 11 PFA for the bare frames in the N-L (dotted lines) and S-D range of response
(solid curves) obtained using a Rayleigh damping based on initial (black lines) and tangent
(grey lines) stiffness matrix.

4story

6story

3
Sa [g]

Sa [g]

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

6story

Floor

Floor

4story

1071

1
0

0.5

1
Tc [s]

1.5

0
0

0.5

1
Tc [s]

1.5

FIGURE 12 FRS at the roof of the bare frames in the N-L range of response obtained
using a Rayleigh damping model based on tangent stiffness matrix (black lines) and a
tangent stiffness proportional damping model (grey lines).

at the fifth story level of the 6b frame in the S-D range of response). Depending on the
adopted damping model, the frequency content of floor accelerations and the derived floor
response spectra can also change. This is shown in Fig. 12 that compares the FRS obtained
with Rayleigh damping with those from a tangent stiffness proportional model with proportionality constant calibrated to provide a 5% damping at the first mode period of the
structure. The damping modeling assumption affects only slightly the value of the FRS
peak at the first mode period but impacts greatly the FRS values in the vicinity of the higher
mode periods. In particular, in the studied buildings the peak values obtained with the tangent stiffness proportional model are up to 50% lower than those obtained with Rayleigh
damping. This is because the stiffness proportional model gives damping ratios that are
proportional to response frequency. If the proportionality constant is evaluated based on
the assumed value for the damping ratio at the fundamental period of the structure, the
damping values at higher mode periods can become unreasonably high. In the studied 4and 6-story frame building, for example, a damping ratio at the first mode period equal to
5% produces at the second mode period a damping ratio of about 15%.

1072

A. Lucchini, F. Mollaioli, and P. Bazzurro

4
Floor

Floor

Figures 13 and 14 show the mean PFA and the mean FRS obtained from the analyses
performed on nonlinear models of the 4w and 6w frames. To allow a direct comparison of
the results from nonlinear models of the bare and infilled frames, in this section the mean
PFA and acceleration FRS for the N-L, L-D, and S-D ranges are computed using the same
set of records that caused those damage ranges in the analysis of the 4b and 6b frames.
Because of the infill walls influence, the absolute values and distribution of PFA along
the height of the building differ from those obtained for the corresponding bare frames. This
can be observed in the plots of Fig. 13, which display the ratios of the PFA from the 4w and
6w frames to those from the 4b and 6b frames discussed earlier. It can be noted that, despite
the uniform distribution of the infill walls, the values of the PFA ratio differ from one along
the height of the building in each of the three nonlinear ranges of response. In the cases
under study the infill walls do not modify the stiffness distribution of the building along
the height, but they decrease the periods of vibration. This decrease causes a change in the
relative contribution that is given by each mode to the global response of the system and,
therefore, also a change in the distribution of the acceleration response along the height.
In addition, differences between the PFA of the bare and infilled frames can also be due to
the additional contribution to the global energy dissipation capacity of the system offered
by the infill walls. From an inspection of Fig. 13, both reductions and amplifications of the
PFA values can be observed (with values ranging between 0.8 and 1.4) without a clearly
identifiable trend. However, for both the 4- and 6-story structures the higher values of the

3
2

2
1

0.2

0
0.6

0.5

0.8
1.1
PFAW [g]

1.4

4
Floor

Floor

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

3.2. Infill Frames (4w and 6w)

3
2

0.8
1
1.2
PFAW/PFA

1.4

0.8
1
1.2
PFAW/PFA

1.4

3
2

0.2

0
0.6

0.5

0.8
1.1
PFAW [g]

1.4

FIGURE 13 Distribution of mean PFA obtained for the 4w frame (top) and 6w frame
(bottom) for the records causing N-L (dotted line), L-D (dashed line), and S-D (solid line)
ranges of response in the corresponding bare frames. The plots on the left show the absolute
values of PFA for the walled frames (PFAW ) and those on the right show the ratios of the
PFAW to the corresponding values of PFA for the bare frames.

Floor Response Spectra for RC Frames


Nearly Linear

Limited Damage

Sa/PFA

Sa/PFA

4
3
2

3
2
1

0
0

0.5

1.5
1
Tc [s]

floor2
floor4
floor6

5
4

Sa/PFA

Sa/PFA

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

3
2
1
0

Sa/PFA

floor2
floor4

Significant Damage

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

3
2
1

Sa/PFA

6
5

1073

0
0

0.5

1
1.5
Tc [s]

0.5

1
1.5
Tc [s]

3
2
1

0.5

1
1.5
Tc [s]

FIGURE 14 Mean 2%-damped FRS for the 4w frame (top) and 6w frame (bottom) normalized by the mean PFA of each set of record causing damage in the three different ranges
of response in the corresponding bare frames.
PFA ratio are observed in the N-L range of response. On the basis of these results, it is clear
that even when the infill walls are weak and uniformly distributed within the structure, their
contribution to the mean PFA along the height of the building can be significant and should
not be neglected.
By comparing the plots of Figs. 12 and 14, it can be noted that the peaks of the FRS
for the 4w and 6w frames are both broader and generally less pronounced than those of
the bare frames. This is to be expected given the effect of the damaged partitions on the
periods of vibration of the structure. The reduction in the absolute values of the peaks of
the FRS for the infilled frames is prevalent but some exceptions were found (e.g., the peak
at the first mode period of the 4-story FRS at second floor in the N-L range of response).
Note also that the difference between FRS for bare and infilled frames is less significant for
the S-D level since the completely cracked partitions do not affect as much the structural
response, which begins to increasingly resemble that of a bare frame.

4. Comparisons with Findings from the Literature


As mentioned earlier, the literature on this subject is plentiful. Chaudhuri and Hutchinson
[2011] studied the effect of structural damage on PFA values in steel moment-resisting
frames subjected to earthquakes of increasing intensities. Taghavi and Miranda [2006]
investigated the effect of structural nonlinearity on both PFA and FRS in generic inelastic
regular moment-resisting structures. Although the building types are somewhat different,
some similarities between the results presented here and those of these earlier studies can
be found. According to Chaudhuri and Hutchinson [2011] the nonlinear response of the
structure, in general, produces a reduction of the PFA values. Taghavi and Miranda [2006]

A. Lucchini, F. Mollaioli, and P. Bazzurro


6

4
Floor

Floor

1074

3
2

1
0

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

0.5

1
1.5
PFA/PFAL

0.5

1
1.5
PFA/PFAL

FIGURE 15 Average ratio of PFA from the analyses carried out on the nonlinear models
(PFA) of the 4b (left) and 6b (right) frame to the PFA from the equivalent-linear models
(PFAL ). The dotted, dashed, and solid curves represent the distribution along the height of
the frames of the average PFA/PFAL ratio obtained separately from the records in the N-L,
L-D, and S-D response ranges, respectively.
also observed a decrease in the PFA value due to structural damage that is more significant around the mid height and at the top of the structure. They consequently found, in
accordance with the results of the present study, that due to structural damage the highest acceleration value is recorded at a floor level lower than the roof level. According to
Chaudhuri and Hutchinson [2011], an amplification of PFA values can also occur in stiff
structures whose acceleration response is significantly affected by the higher modes. The
amplification may occur prevalently at lower stories and, in rarer cases, at upper stories as
well. As stated earlier, the same behavior was also confirmed by the results of this study
for the 4b and 6b frame analyses. Figure 15 summarizes how the average ratio of the PFA
from the nonlinear model to the PFA, again, from the linear model exceed one at the lower
stories in both frames, at the fifth story for the 6b frame.
For the structures under study, the amplifications of PFA in the nonlinear cases cannot be explained as the product of localized damage, which can produce a change in the
stiffness distribution and, thus, in the dynamic properties of the structure during the shaking, as stated by Sewell et al. [1986]. Modal analyses run for both the 4b and 6b frames at
each time step of the nonlinear dynamic analyses show that the mode shapes of the frames
do not significantly change with increased damage. The amplification of PFA values can
instead be explained by an increased contribution of the higher modes to the response of
the damaged structure that, unlike the first mode, become less damped as the damage progresses when the considered Rayleigh damping model (proportional to the mass and the
tangent stiffness matrix) is adopted. This explanation can be appreciated more clearly by
examining Figs. 16 and 17. Figure 16 shows the variation of the first three mode periods
and of the corresponding damping ratios evaluated during a time-history analysis of the 4b
frame. The damping values decrease at the higher modes while increases at the first mode
with period lengthening (Fig. 17). Hence, the amplification of PFA for the nonlinear cases
may be an artificial by-product of the modeling approach of damping adopted, rather than
being a characteristic behavior.
The findings regarding the effects of structural damage on floor response spectra are
substantially in line with those of other researchers (e.g., Chaudhuri and Villaverde, 2008).
Structural damage may produce amplifications of FRS and, therefore, increased demand on
non-structural components when: the supporting structure is short and stiff; the components
are located on the lower floor levels; the period of vibration of the component is in tune with

Floor Response Spectra for RC Frames


2.5

10
[%]

T [s]

1.5
1
0.5
0

12

mode1
mode2
mode3

8
6
4

10

15

10

t [s]

15

t [s]

FIGURE 16 Periods (left) and corresponding damping ratios (right plot) of the first three
modes of the 4b frame evaluated at each step of the nonlinear dynamic analysis.
10
8
[%]

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1075

6
4
1
2

2
20

3
40
[Hz]

60

FIGURE 17 Damping applied to the first three modes of the 4b frame evaluated using the
initial (undamaged) stiffness matrix according to the Rayleigh damping model used in the
nonlinear dynamic analyses.
the higher modes of the structure. However, the maximum amplification observed in this
study, which was equal to 1.18, is much lower than the value of 1.8 found by Chaudhuri
and Villaverde [2008]. The reasons explaining the amplifications in the FRS that are caused
by structural nonlinearity are the same of those provided earlier for the amplification of the
PFA.
In most cases, again, the effect of the nonlinear response is the reduction of the FRS
peak values and its amount depends, in general, on the floor level, the period of vibration,
and the level of damage experienced by the structure. For example, in the 4b frame the
reduction is more accentuated in the S-D than the N-L range, and in the S-D case is more
pronounced at the first mode period while at the N-L case in second mode period. Thus,
structural nonlinearity affects floor response spectra both near the fundamental period of
the structure and near the higher periods of vibration as well. This finding is in accordance
with the results of Taghavi and Miranda [2006]. On the contrary, Medina et al. [2006], who
worked on steel moment-resisting frames, reported reductions that were always more pronounced at the fundamental mode period of the structure than at the higher mode periods.
This discrepancy in the identification of the periods where the reductions can occur is likely
due to the different dynamic characteristics of the buildings analyzed in the two studies.
Finally, as a consequence of cracking of the structural elements and progressive yielding of rebars, the FRS peaks shift towards longer period values. This behavior is in line
with the results obtained by Rodriguez et al. [2002], but not with those of Taghavi and
Miranda [2006]. In this latter study, in fact, while the response of the structure enters into

1076

A. Lucchini, F. Mollaioli, and P. Bazzurro

the severe nonlinear range, the amplifications of the FRS peaks reduce without any shift of
the corresponding periods. The difference between the effects on the floor response spectra
produced by the nonlinear structural behavior observed in the two studies is due to the different constitutive models used for evaluating the response of the structure. In this study,
cracking of concrete is explicitly accounted for in the fiber sections models of the forcebased nonlinear elements. Whereas, in the moment-resisting frames structures studied in
Taghavi and Miranda [2006], the hysteretic behavior of the member elements is modeled
using simple back-bone bilinear constitutive laws.

The findings of the present study are not fully consistent with the prediction equations
proposed for the evaluation of the floor response spectra by three different seismic codes
and guidelines: the Eurocode 8 [EN 1998-1, 2004], New Zealand Standard 1170.5 [NZS,
2004], and NEHRP Recommended Seismic Provisions [FEMA P-750, 2009]. These three
codes were chosen because they are often recognized as representing the state-of-the-art
in seismic design. As such, they are often considered or adopted outright in the national
standards used in many countries. A comparison between the elevation distribution of the
PFA and FRS at the roof level of the 6b frame suggested by these three codes is displayed
in Fig. 18.

5.1. Eurocode 8 (EC8)


The EC8 profile simply assumes a linear distribution of PFA along the height of the building. This behavior fails to recognize the significant effects on PFA of both higher modes,
which can occur even if the participation mass excited by these modes is small, and of structural damage. As observed in this study (Fig. 19), the PFA profiles may differ significantly
from a simple linear one, as in the case of the investigated buildings. The values suggested
by the EC8 appear, on average, to overestimate PFA at higher stories and underestimate
it, albeit slightly, at lower stories the average values computed for both the 4- and 6-story
bare and infilled frames. As noted earlier, however, the underestimation can be partially
due to the Rayleigh damping modeling approach (proportional to the mass and the tangent
stiffness matrix) adopted here.

3
EC8
FEMA
NZS

2.5
FRS/PFA

0.8
z/H

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

5. Comments on Code Prediction Equations

0.6
0.4
0.2

2
1.5
1
0.5

2
PFA/PGA

Tc/T1

FIGURE 18 Comparison of estimates of PFA (normalized by PGA) and floor response


spectra (normalized by PFA) at the roof level of the 6b frame obtained using the three
codes.

Floor Response Spectra for RC Frames


infilled

0.8

0.8

0.6

0.6

6story
z/H

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

z/H

0.4

0.4

0.2

0.2
1

1.5
2
PFA/PGA

0
0.5

2.5

0.8

0.8

0.6

0.6

z/H

4story
z/H

bare
1

0
0.5

0.4

0.4

0.2

0.2

0
0.5

1.5
2
PFA/PGA

1077

2.5

0
0.5

1.5
2
PFA/PGA

2.5

1.5
2
PFA/PGA

2.5

FIGURE 19 Average PFA distribution for the three ranges of response (N-L dotted, L-D
dashed, and S-D solid) normalized by the Peak Ground Acceleration (PGA) of the input
motion from this study, compared with the PFA/PGA distribution proposed by the EC8
(dashed grey line).
Another discrepancy between the results of this study and the EC8 relates to the
predicted maximum amplification of the FRS ordinates with respect to the PFA at the corresponding floor. If the equation proposed by the EC8 is applied, it can be observed that
the FRS to the PFA at the same story do not exceed a value of about 2.5 (which is equal to
2.2 for z/H=1) while the results shown here (Figs, Figure 10 and 14) have mean estimates
of the maximum values as high as five (and for some records as high as almost 10). Values
of 2.5 are exceeded at periods close to those of the first three modes of the structures analyzed here. The FRS suggested by the EC8 (Fig. 18) also has a single peak located at the
fundamental period T1 of the supporting structure and does not recognize that large spectral
amplification may occur when the component resonates with higher periods of vibration
as well. By observing Fig. 18, it can also be noted that EC8 does not allow for floor spectral accelerations higher than PFA for periods longer than twice T1 . Larger floor spectral
ordinates for periods greater than 2T1 are possible, as shown in Fig. 20. In addition, the prediction equation in EC8 suggests that the spectral amplification factors at different floors
are very similar to one another while this study found that the shape of the FRS strongly
depends on the floor level (Fig. 9). Finally, the EC8 floor response spectra do not depend
on dynamic properties of the non-structural component besides its fundamental period.
However, the FRS clearly also depend on the damping of the non structural component.
This dependence, which has been shown by others (e.g., Sullivan et al., 2013), was also
confirmed in the present study by findings (not shown here because of space limitations)
from analyses characterized by different damping ratios.

1078

A. Lucchini, F. Mollaioli, and P. Bazzurro


6story
5

Sa/PFA

Sa/PFA

4story
5

2
1
0

1
0

Tc/T1

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

Tc/T1

FIGURE 20 Floor response spectra normalized by the PFA at the roof of the studied bare
and infilled frame structures (denoted with solid and dotted lines, respectively) evaluated in
the S-D range of response. The abscissa of the FRS curves is normalized by the fundamental
mode period values of the bare frame structures.
5.2. New Zealand Standard 1170.5 (NZS)
In NZS the PFA/PGA profile along the height of the building is defined through the socalled floor height coefficient, CHi , which depends on the total height of the building, hn .
In particular, if hn is lower than 12m the PFA/PGA profile is linear with a maximum value
at the roof equal to 1+hn /6. For taller buildings, the profile is bilinear with a constant
value equal to 3 for floor levels higher than 12m (or higher than 0.2hn , when hn > 60 m).
The results obtained in the present study, however, demonstrate that the PFA/PGA profile
depends on dynamic properties of the building that are not captured by its total height only
(e.g., compare Figs.10 and 14 for the bare and infilled frames). As in the case of EC8, the
maximum value of PFA/PGA is predicted at the roof level, which is in contrast with the
results found in this study. Hence, the same comment on this matter previously offered for
EC8 applies here as well.
At each floor level the spectral ordinates of the FRS are evaluated by multiplying the
PFA value by the spectral shape coefficient Ci . This coefficient, however, is considered to
be dependent only on the absolute value of the fundamental period, Tc , of the non structural
component rather than on the ratio between Tc and the fundamental period of vibration of
the building. The results of the analyses shown here do not support such a simplification.
NZS states a minimum Ci value of 0.5 for spectral ordinates related to periods Tc equal
to or longer than 1.5s. This value appears to be non-conservative. For example, in Fig. 20
the Sa /PFA ratio of the 6b frame at Tc equal to 1.5s (i.e., at Tc /T1 =1.5s/1.22s=1.23) is
considerably higher than 0.5. For Tc values lower than 0,75s, on the other hand, the Ci coefficient is assumed to reach its maximum value of 2. This study has shown that depending
on both the level of damage, building dynamic properties and spectral period of the FRS
the maximum component amplification can exceed the value of 2. Again, as for the EC8,
the ordinates of the FRS do not change with the floor level. This assumption, which does
not account for the different contributions of the modes of vibration in the FRS at different
floors, is not supported by the results presented here.

Floor Response Spectra for RC Frames

1079

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

5.3. FEMA P-750 (FEMA)


Similarly to EC8, the FEMA document proposes a linear PFA/PGA profile along the height
of the building but its maximum value at the roof is equal to 3 rather than to 2.5 as in the
EC8 case. As explained in the FEMA document, that value was established after the examination of recorded acceleration data for short and moderate height structures in response
to large California earthquakes. If the dynamic properties of the structure are available, the
FEMA document also provides an approach to the computation of floor accelerations that
utilizes modal analysis.
FEMA defines the amplification factors for the ordinates of the FRS with respect to the
PFA at the same floor via the component amplification coefficient, ap . As the coefficient Ci
in the NZS, ap controls the shape of the FRS. In the first of the two alternative formulations
proposed by the guidelines, the value of ap depends on the component stiffness. For spectral
ordinates of the FRS with period shorter than 0.06s, ap is set equal to 1. This means that
according to FEMA, non-structural components that stiff can be considered rigid or rigidly
attached to the floor. For longer periods, namely, for components that are flexible (or not
rigidly attached) ap is instead set equal to 2.5. The second formulation, which was derived
from the NCEER work of Bachman et al. [1993], provides a more accurate estimate of ap
as a function of the Tc /T1 ratio. According to this formulation, the maximum value of ap ,
applicable to Tc /T1 values between 0.7 and 1.4, is again equal to 2.5 (Fig. 18). As stated
earlier, a maximum amplification value of 2.5 appears to be non-conservative in some cases.
The upper bound of the period range of applicability proposed by the second formulation
in FEMA is high enough to cover possible large amplifications of the FRS ordinates at
periods slightly longer than T1 that may occur when the response of the supporting structure
becomes nonlinear. However, as observed earlier, the large amplification that may occur
for the spectral ordinates of the FRS at periods close to those of the higher modes of the
supporting structure is not adequately accounted for by the lower bound of the range of 0.7.
The FEMA guidelines are more conservative than EC8 and NZS for spectral ordinates of
the FRS longer than twice T1 where the ap value is equal to 1. The value of 1 is in better
agreement with the results of the analyses presented here.

6. Conclusions
The results from this study shed some additional light on the effects of post-elastic nonlinear behavior of the supporting building on Peak Floor Accelerations (PFA) and Floor
acceleration Response Spectra (FRS). The floor acceleration response was investigated for
strong beam-weak column old vintage reinforced concrete frames with and without infill
walls using a large set of far-field ground motion recordings. The findings of this study
were compared and contrasted with those of others and with the prescriptions of three stateof-the-art codes and guidelines. The results from the present study suggest the following
conclusions.

Effects of post-elastic nonlinear behavior:

The post-elastic nonlinear behavior of the supporting building significantly affects


the profile of the PFA along the height of the building and the ordinates of the FRS
at all stories.
In general, disregarding the post-elastic nonlinearity of the building response
produces the following.
 Conservative estimates of the PFA at mid height of the structure and at the roof
level.

1080

A. Lucchini, F. Mollaioli, and P. Bazzurro


Conservative estimates of the spectral acceleration of the FRS at all periods
besides those slightly longer than the periods of the higher modes of the supporting building. The lengthening of the oscillation periods due to cracking
and progressive yielding of rebars in structural elements causes the peaks of
the FRS in the nonlinear case to slightly exceed the ordinates of the FRS in the
linear case.
The amount of overestimation of PFA and FRS ordinates depends, in general,
on the level of structural damage, on the dynamic properties of the supporting
building, and on the story level.
Effects of damping modeling assumptions:
When FRS and PFA are derived from floor accelerations calculated via nonlinear dynamic analyses of the supporting structure, attention must be paid in the
selection and calibration of the adopted damping model.
If a Rayleigh damping proportional to the mass and tangent stiffness matrix of
the structure is adopted, the PFA at lower and upper stories can slightly increase
due to structural nonlinearity. When this damping model is adopted, the higher
modes become less damped as the damage progresses and their contribution to
the response of the structure can increase.
If a Rayleigh damping model is built based on the initial stiffness matrix of the
structure instead of the tangent one, the PFA can be significantly underestimated.
If a stiffness proportional damping model calibrated at the first mode period of
the structure is adopted instead of Rayleigh damping, higher mode contribution
to FRS are underestimated.
Effects of modeling infill walls:
Modeling the infill walls, even if they are weak and uniformly distributed such
as those in the case studies analyzed here, can significantly affect the profile
of the PFA along the height of the building and the ordinates of the FRS at all
stories. Modeling the infill walls may cause both increase and decrease of floor
accelerations. In the infilled case, the FRS peaks become less pronounced, and
the absolute values are frequently reduced when compared to those of the bare
frames.
Comparison of results from this study and predictive equations prescribed by codes
and guidelines.
The Eurocode 8, New Zealand Standard 1170.5, and FEMA P-750 documents,
which represent the state-of-the-art in seismic design codes and guidelines, predict PFA profiles and FRS ordinates that are somewhat inconsistent with the
findings of the present study. These documents do not seem to fully consider the
influence on PFA and FRS caused by the level and type of damage experienced by
the supporting building. In general, these documents prescribe PFA values at the
roof level that are overestimated and peaks of the FRS that are underestimated.


Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

To conclude, it is important to remark that these conclusions are not general but hold for
structures similar to those considered in the present study.

Funding
The financial support of both the Italian Ministry of the Instruction, University and
Research (MIUR), and the Italian Network of University Laboratories of Seismic
Engineering (ReLUIS) is gratefully acknowledged.

Floor Response Spectra for RC Frames

1081

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

References
Bachman, R. E., Drake, R. M., and Richter, P. J. [1993] 1994 Update to 1991 NEHRP Provisions
for Architectural, Mechanical, and Electrical Components and Systems, National Center for
Earthquake Engineering Research, Buffalo, New York.
Chaudhuri, S. R. and Villaverde, R. [2008] Effect of building nonlinearity on seismic response of
nonstructural components: a parametric study, Journal of Structural Engineering ASCE 134(4),
661670.
Chaudhuri, S. R. and Hutchinson, T. C. [2011] Effect of nonlinearity of frame buildings on peak
horizontal floor acceleration, Journal of Earthquake Engineering 15(1), 124142.
Chen, Y. Q. and Soong, T. T. [1988] State-of-the-art review: seismic response of secondary systems,
Engineering Structures 10(4), 218228.
Decanini, L. D., Mollaioli, F., Mura, A., and Saragoni, R. [2004] Seismic performance of masonry
infilled R/C frames, Proc. of 13th WCEE, Paper 165, Vancouver.
Decanini, L. D. and Fantin, G. E. [1986] Modelos simplificados de la mamposteria incluida en
porticos. Caracteristicas de rigidez y resistencia lateral en estado limite, Jornadas Argentinas de
Ingenieria Estructural, Buenos Aires, Argentina, 2, 817836 (in Spanish).
EN 1998-1 [2004] Eurocode 8: Design of Structures for Earthquake Resistance - Part 1: General
Rules, Seismic Actions and Rules for Buildings, CEN, Brussels.
Lin, J. and Mahin, S. A. [1985] Seismic response of light subsystems on inelastic structures,
Journal of Structural Engineering ASCE 111(2), 400417.
Lucchini, A., Cheng, Y., Mollaioli, F., and Liberatore, L. [2013] Predicting floor response spectra
for RC frame structures, Proc. of the 4th ECCOMAS Thematic Conference on Computational
Methods in Structural Dynamics and Earthquake Engineering (COMPDYN 2013), Paper
No. 1334, eds. M. Papadrakakis, N. D. Lagaros, and V. Plevris, Kos Island, Greece, June 1214.
McKenna, F., Fenves, G. L., and Scott, M. H. [2007] OpenSees: Open System for Earthquake
Engineering Simulation, http://opensees.berkeley.edu, PEER, University of California, Berkeley,
CA.
Medina, R. A., Sankaranarayanan, R., and Kingstonb, K. M. [2006] Floor response spectra for
light components mounted on regular moment-resisting frame structures, Engineering Structures
28(14), 19271940.
Mollaioli, F., Lucchini, A., Bruno, S., De Sortis, A., and Bazzurro, P. [2010] Floor acceleration
demand in reinforced concrete frame structures with masonry infill walls,Proc. of the 9th US
National and 10th Canadian Conference on Earthquake Engineering, 7, 58455854, Toronto,
Canada, July 2529.
FEMA 74 [1994] Reducing the Risks of Nonstructural Earthquake Damage: A Practical Guide,
Federal Emergency Management Agency, Washington, D.C.
FEMA P-750 [2009] NEHRP Recommended Seismic Provisions for New Buildings and other
Structures: Part 2, Commentary to ASCE/SEI 7-05, Federal Emergency Management Agency,
Washington, D.C.
NZS 1170.5 [2004] Structural Design Actions, Part 5: Earthquake actions New Zealand, Draft DR
00902.
PEER [2011] PEER Ground Motion Database Beta Version, 2011, http://peer.berkeley.edu/
peer_ground_motion_database/site, Pacific Earthquake Engineering Research Center.
Petrini, L., Maggi, C., Priestley, M. J. N., and Calvi G. M. [2008] experimental verification of viscous damping modeling for inelastic time history analyses, Journal of Earthquake Engineering
12(1), 125145.
Reinoso, E. and Miranda, E. [2005] Estimation of floor acceleration demands in high-rise buildings
during earthquakes, The Structural Design of Tall and Special Buildings 14(2), 107130.
Rodriguez, M. E., Restrepo, J. I., and Carr, A. J. [2002] Earthquake-induced floor horizontal
accelerations in buildings, Earthquake Engineering and Structural Dynamics 31(3), 693718.
Sankaranarayanan, R. and Medina, R. A. [2007] Acceleration response modification factors for
non-structural components attached to inelastic moment-resisting frame structures, Earthquake
Engineering and Structural Dynamics 36(14), 21892210.

Downloaded by [Universita Studi la Sapienza] at 01:12 02 September 2014

1082

A. Lucchini, F. Mollaioli, and P. Bazzurro

Sewell, R. T., Cornell, C. A., Toro, G. R. and McGuire, R. K. [1986] A study of factors influencing
floor response spectra in nonlinear multi-degree-of-freedom structures, JABEEC Report no. 82,
Palo California, Dept. of Civil and Environmental Engineering, Stanford University.
Singh, M. P. [1975] Generation of seismic floor spectra, Journal of the Engineering Mechanics
Division ASCE 101(5), 593607.
Singh, M. P., Moreschi, L. M., Surez, L. E., and Matheu, E. E. [2006a] Seismic design forces. I:
Rigid nonstructural components, Journal of Structural Engineering ASCE 132(10), 15241532.
Singh, M. P., Moreschi, L. M., Surez, L. E., and Matheu, E. E. [2006b] Seismic design
forces. II: Flexible nonstructural components, Journal of Structural Engineering ASCE 132(10),
15331543.
Sullivan, T. J., Calvi, P. M., and Nascimbene, R. [2013] Towards improved floor spectra estimates
for seismic design, Earthquakes and Structures 4(1), 109132.
Taghavi, S. and Miranda, E. [2005] Approximate floor acceleration demands in multi-storey
buildings. II: Applications, Journal of Structural Engineering ASCE 131(2), 212220.
Taghavi, A. and Miranda, E. [2006] Probabilistic seismic assessment of floor acceleration demands
in multi-story buildings, Report No. 162, June 2006, The John A. Blume Earthquake Engineering
Center, Dept. of Civil and Environmental, Engineering, Stanford University.
Villaverde, R. [1997] Seismic design of secondary structures: state of the art, Journal of Structural
Engineering ASCE 123(8), 10111019.

You might also like