Engineering Structures 30 (2008) 31863192

Contents lists available at ScienceDirect

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

The effect of masonry infills on the seismic response of a four storey reinforced
concrete framea probabilistic assessment
Matja Dolek , Peter Fajfar
University of Ljubljana, Faculty of Civil and Geodetic Engineering, Jamova 2, SI-1000 Ljubljana, Slovenia

article

info

Article history:
Received 11 May 2007
Received in revised form
28 December 2007
Accepted 28 April 2008
Available online 5 June 2008
Keywords:
Probabilistic seismic assessment
Reliability analysis
Reinforced concrete frame
Masonry infill
Simplified non-linear analysis

a b s t r a c t
A relatively simple probabilistic approach for the seismic performance assessment of building structures
combines the SAC-FEMA method, which is part of the broader PEER probabilistic framework and permits
probability assessment in closed form, with the N2 method. In this approach, the most demanding part of
the PEER probabilistic framework, i.e. the Incremental Dynamic Analysis (IDA), is replaced by the much
simpler Incremental N2 (IN2) analysis. Predetermined default values for dispersion measures are needed
for the practical implementation of this approach. In the paper, the simplified approach is summarized
and applied to the seismic performance assessment of three variants of a four storey reinforced concrete
frame: a bare frame and two frames with masonry infill, one with openings and the other one without
them. The probabilities of exceedance of selected limit states are estimated and discussed. The results of
the analyses indicate that the probability of failure of the infilled frames with regularly distributed infill
is smaller than that of the bare frame. The beneficial effect of the infill is more evident in the probabilistic
analysis than in the deterministic analysis. Of the two infilled frames, the one with openings in the infill
has a higher probability of failure.
2008 Elsevier Ltd. All rights reserved.

1. Introduction
One of the methods developed for the seismic risk evaluation
of structures is the SAC-FEMA method (SAC is a joint venture
of the Structural Engineers Association of California, the Applied
Technology Council, and California Universities for Research in
Earthquake Engineering, FEMA is Federal Emergency Management
Agency), which permits probability assessment in closed form [1],
and represents a part of a broader PEER (Pacific Earthquake
Engineering Research Center) probabilistic framework [2]. Within
the framework of the SAC-FEMA method, the relationship between
the seismic intensity measure and the engineering demand
parameter is usually determined by Incremental Dynamic Analysis
(IDA) developed by Vamvatsikos and Cornell [3]. IDA is a powerful
tool for the estimation of seismic demand and capacity for
multiple levels of intensity. However, it requires a large number
of inelastic time-history analyses (and corresponding detailed data
on the ground motion time-histories and hysteretic behaviour
of structural elements), and is thus very time-consuming. It is
often possible to create summarized IDA curves with less input
data and less effort, but with still acceptable accuracy. One
possible approach is to determine the seismic demand for multiple
levels of seismic intensity using the N2 method [4] which is a

Corresponding author.
E-mail address: mdolsek@ikpir.fgg.uni-lj.si (M. Dolek).
0141-0296/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2008.04.031

practice-oriented nonlinear method based on pushover analysis

and inelastic response spectrum. Such an approach yields the
Incremental N2 (IN2) curve [5,6], which is intended to approximate
a summarized IDA curve. In this paper, this simplified approach
for probabilistic seismic performance assessment is summarized
and applied to three variants of a four storey reinforced concrete
(RC) frame: a bare frame and two infilled frames, one of them with
openings and another one without them. The example structures
are located at two different locations representing moderate and
high seismic hazard. For the determination of an IN2 curve for
an infilled frame, the extended N2 method, which is applicable to
infilled RC frames, was used [7,8]. The results are compared with
the results of the deterministic seismic assessment of the same
structures, presented in the companion paper [9].
Despite the relatively large number of seismic reliability studies
in the literature, few deal with infilled frames. One example of
the seismic reliability of reinforced concrete frames with masonry
infills is presented in [10].
2. Framework for probabilistic performance assessment
The simplified probabilistic performance assessment analysis
combines two procedures, i.e. the N2 method [4], which is usually
employed for a deterministic seismic performance assessment, and
a probabilistic assessment in closed form, upon which the SACFEMA steel moment frame guidelines [1] are based. In this section
the method used for the simplified probabilistic assessment

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

is briefly explained. More detailed explanation can be found

elsewhere, e.g. in [6].
The SAC-FEMA method is based on some simplifying assumptions, which permit the formulation of a probabilistic assessment
in closed form [1]. An additional simplification is introduced by
employing the N2 method instead of the IDA analysis for the determination of the relation between the seismic intensity measure
and the engineering demand parameter. The curve which represents this relationship is usually called an IDA curve [3]. In the
simplified procedure it is substituted by an IN2 curve [5,6]. An IN2
curve is intended to approximate a summarized IDA curve, and is
not calculated for a single ground motion. The term summarized,
when related to IN2 curves, applies only to mean or median curves,
since the proposed simplified approach is not intended for the determination of dispersion. Default values for the dispersion measures for randomness and uncertainty in displacement demand
and capacity have therefore to be used in order to determine the
probability of exceedance of a given limit state.
The N2 method is a relatively simple nonlinear analysis
method for deterministic seismic assessment of buildings and
bridges, which combines pushover analysis of a multi degreeof-freedom (MDOF) model with the response spectrum analysis
of an equivalent single-degree-of-freedom model (SDOF). The
N2 method has been mainly used for the seismic assessment
of structures where the seismic demand (i.e. the engineering
demand parameter) for a given seismic intensity is compared
with the capacity corresponding to a given limit state. (Note
that the expression performance level used in FEMA 350 [11]
has basically the same meaning as limit state according to the
Eurocode terminology.) In a probabilistic performance assessment
the relationship between the seismic demand and the seismic
intensity has to be determined for different values of the seismic
intensity measure.
The IN2 curve represents the relation between an engineering
demand parameter and a seismic intensity measure. The top displacement is usually used as the engineering demand parameter,
and the spectral acceleration, i.e. the value in the elastic acceleration spectrum at the period of the idealized system, represents
the intensity measure. The engineering demand parameter and the
corresponding seismic intensity measure will be denoted as C and

sCa , respectively. The whole IN2 curve can be determined by repeating the N2 approach for increasing ground motion intensity
until failure occurs. In the simplest but very common case the
equal displacement rule applies, i.e. the inelastic displacement
is assumed to be equal to the elastic displacement of the system
with the same stiffness and mass, but with unlimited strength. In
such a case the IN2 curve is a straight line (with its origin at the
point (0, 0)) until failure occurs. It is necessary to determine only
the point corresponding to failure. In general, the shape of the
IN2 curve depends on the relation between the reduction factor,
ductility and period (the RT relation), which defines the inelastic spectra to be used in the N2 method for the determination of
seismic demand. For example, in the case of infilled RC frames, the
IN2 curve consists of straight lines, as presented in Fig. 1, for which
three points have to be evaluated by the N2 method [7,8] (in Fig. 1,
these are the points at the top displacements a, b and c). The idealized capacity diagram with indicated yield and near collapse points
is also shown. It is conservatively assumed that the structure fails
after the NC limit state is attained. Thus the IN2 line after the NC
limit state is horizontal. Knowing the IN2 curve, the engineering
demand parameter can be easily linked to the corresponding seismic intensity measures.

Once the seismic intensity sCa , which causes a selected limit

state, has been determined from the IN2 curve, the x confidence

3187

Fig. 1. A typical IN2 curve for an infilled RC frame building. Y and NC indicate the
yield and near collapse points.

level estimate of the annual probability (mean annual frequency) of

the exceedance of a given limit state PLS,x can be determined as [1]
 
sCa CH Cf Cx
PLS,x = H
(1)
#
"


2 

k
1 2
CH = exp
Cf = exp
,
2DR + 2CR ,
2 H
2b2
s

(2)


k2
Cx = exp Kx
2DU + 2CU .
b2

 
sCa is the median value of the hazard function at the
H

seismic intensity sCa . It provides a median estimate of the annual

probability that the seismic intensity will be equal to or exceed

the level sCa , i.e. the seismic intensity corresponding to the

median displacement capacity C . k is a parameter of the hazard
(sa ) = ko (sa )k , and b is
function, idealized in the form H
a parameter of the function relating the engineering demand
parameter to the intensity measure, i.e. of the so-called IDA curve,
or, in the case of the simplified method, of the IN2 curve. IDA
or IN2 curve is idealized in the form D (sa ) = a (sa )b . Kx is
the standardized normal variate associated with the probability
x of not being exceeded. For example, the values Kx = 0, 1
and 1.28 are associated with 50%, 84% and 90% confidence levels,
respectively.
H is the dispersion measure for hazard. The product
 
 
sCa CH represents the mean value of the hazard function H sCa .
H
Other parameters represent the dispersion of the engineering
demand parameter (i.e. the top displacement) due to ground
motion variability (randomness) and due to variability related to
structural modeling and analysis (uncertainty). DR and CR are
the dispersion measures for randomness in the top displacement
demand and capacity, and DU and CU are the dispersion measures
for uncertainty in the top displacement demand and capacity. For
practical applications, predetermined default values for dispersion
measures, based on statistical studies of typical structural systems,
will be needed. In the example shown in this paper, some rough
preliminary estimates were used.
3. Probabilistic performance assessment of example structures
3.1. Description of the example structures and the mathematical
modeling
The example structures are the same as in the companion
paper [9]. A bare and two infilled four-storey plane RC frames
have been studied (Fig. 2). The buildings had been designed to
reproduce the design practice in European and Mediterranean
countries about forty to fifty years ago [12]. The elements of the
RC frame were modelled by one-component lumped plasticity
elements, consisting of an elastic beam and two inelastic rotational
hinges. The infills were modelled by means of two diagonal strut

3188

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

Fig. 2. The elevation view of the studied structures.

elements which can carry loads only in compression. The diagonals

were placed between the beamcolumn joints. More details about
the structures and their models are described in the companion
paper.
3.2. Limit states and the target probability of failure
In the study, the limit states of damage limitation (DL),
of significant damage (SD), and of near collapse (NC) were
investigated. Their definitions are given in the companion
paper [9].
The probabilities of exceedance of these limit states were
calculated with a 50% confidence level (i.e. without considering
uncertainties in demand and capacity, Kx = 0), and with a 90%
confidence level. A 90% confidence level was used, for example, for
the global collapse criterion for moment resisting frames in FEMA350 [11] and by Yun et al. [13].
In earthquake engineering, no general consensus on the target
probabilities for different limit states has been reached so far.
Some values for the target probabilities of failure are provided
in the Probabilistic model code [14]. In this document, the value
Po = 105 has been recommended as a target probability of
exceedance for the ultimate limit state for a common design
situation (moderate consequences of failure and normal cost of
safety measures). A much larger value Po = 5 104 has been
recommended for a large relative cost of safety measures typical
for existing buildings. These values can be used as a rough measure
in order to better understand the results presented in this paper.
3.3. Earthquake motion and hazard
One of the advantages of the N2 method, which was used for
the determination of the IN2 curves of the example structures, is
the presentation of earthquake motion by an elastic acceleration
response spectrum and not by a set of ground motion records,
which is required in the IDA analysis. In our example, the EC 8
spectrum for ground type A was used (Fig. 5(a)). For the sake of
simplicity it was assumed that the shape of the spectrum does not
change with the level of peak ground acceleration. Note, however,
that the IN2 curve can be determined also by using uniform hazard
spectra, by assuming different shapes for different levels of seismic
intensity.
In addition to the elastic response spectrum, hazard curves for
the intensity measures are needed for a probabilistic performance
assessment. In the study it was assumed that the example
structures are located in the SE part of Slovenia (moderate hazard
MH) and in the south of Italy (high hazardHH). The hazard curves
for the peak ground acceleration are shown in Fig. 5(b). The HH

curve exceeds the MH curve by about 100%. For example, the peak
ground acceleration for the 2475 year return period amounts to
0.67, and 0.32 g, for the high and the moderate seismic hazard
regions, respectively. Discrete values obtained from the same
hazard curves were used in the deterministic analyses described
in the companion paper.
3.4. IN2 curves
IN2 curves are determined based on the results of pushover
analyses and on the ground motion data presented in Section 3.3.
A single point on the IN2 curve can be obtained by N2 analysis. In
the companion paper the N2 approach has been summarized and
applied to the example structures. For convenience, the capacity
diagrams for all three test examples determined in the companion
paper have been re-plotted in Fig. 4. The elastic spectra in Fig. 4
represent the seismic demand at which the NC limit states are
attained in the three example structures. Note that Fig. 4 applies
to equivalent SDOF systems.
The IN2 curves for the equivalent SDOF systems are defined by
the elastic spectral acceleration and the corresponding inelastic
displacement. As explained in Section 2, only a few points of the
IN2 curve have to be determined by the N2 method in order to
construct the IN2 curve. For the bare frame, only the point at
the NC limit state has to be determined, since the period of the
idealized system T = 0.83 s exceeds the corner period TC (Fig. 3(a)),
and the equal displacement rule can thus be applied to this
particular example. The displacement at the NC limit state was
already determined as Sd,NC = 9.4 cm (see Section 4.3 in the
companion paper). Considering the equal displacement rule, the
elastic displacement is equal to the inelastic displacement and
the corresponding elastic spectral acceleration Sae,NC is obtained
as Sae,NC = 42 Sd,NC /T 2 = 0.54 g. The IN2 curve for the
equivalent SDOF system representing the bare frame (Fig. 4) is then
constructed as a straight line from the origin to the point at the NC
limit state defined by Sd,NC and Sae,NC . After this point it is assumed
that the IN2 curve is horizontal (see Section 2).
The IN2 curve for the MDOF model of the bare frame is
presented in Fig. 5. The top displacement of the MDOF system,
DNC , and the displacement of the equivalent SDOF system, Sd,NC , are
related by the formula DNC = Sd,NC = 9.4/1.29 = 12.1 cm, where
is the transformation factor from the MDOF to the equivalent
SDOF system and vice versa. In Fig. 5, the intensity measure is
presented not only in terms of the spectral accelerations (on the
right hand side), but also in terms of the peak ground accelerations
(on the left hand side). The ratio between the spectral acceleration
Sae (T = 0.83 s) and the peak ground acceleration ag is equal to 1.2
(Fig. 3). Consequently, the peak ground acceleration at which the
NC limit state is attained amounts to ag,NC = 0.54 g/1.2 = 0.45 g.

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

3189

Fig. 3. (a) The normalized EC8 spectrum (Soil type A) with indicated periods of idealized systems for the bare frame and the infilled frames; (b) Hazard curves for the
moderate (MH) and high hazard (HH) regions, with indicated return periods.

A comparison of demand and capacity (Fig. 5) shows that, in

the case of the infilled frame, at the MH location, the capacities are
larger than demands for all three limit states, with exception of
DL limit state for PI frame. At the HH location, demand exceeds
capacity at the NC limit state, whereas the opposite applies to
the SD limit state. At the DL limit state, capacity exceeds demand
in the case of the fully infilled frame, whereas demand exceeds
capacity in the case of the partially infilled frame. In the case of the
bare frame, the NC limit state is attained at a smaller peak ground
acceleration than in the case of the infilled frames (0.45 g against
0.55 g and 0.59 g).
3.5. Results of probabilistic performance assessment

Fig. 4. Capacity diagrams, IN2 curves, and elastic spectra defining the seismic
demand at the NC limit state for the equivalent SDOF systems representing the bare
frame and the infilled frames.

The DL and SD limit states are also indicated in the IN2 curve
in Fig. 5. The top displacements (capacities), corresponding to
these limit states, were determined by pushover analysis (Fig. 10
in the companion paper). The demands, representing the ground
motions with return periods of 225, 475 and 2475 years for high
and moderate hazard locations, are also presented in Fig. 5. The
comparison between demands and capacities suggests that, for the
high hazard location, the demand exceeds the capacity for all limit
states. The opposite situation can be observed in the case of the
moderate hazard location. Note that such a comparison is based
on the results of deterministic analyses, and is also presented in
the companion paper in a different form.
In the case of the infilled frames the equal displacement rule is
not applicable, and the NC point is not sufficient to construct the
IN2 curve. In addition to this point, two other points are needed
(Section 2), i.e. points which correspond to the yielding of the
idealized system and to the displacement of the idealized system
at the beginning of the degradation of the infills (points a and b in
Fig. 1). All three points can be determined by using the N2 method
applicable to infilled frames [7]. After the NC point, a horizontal
line is assumed (Section 2). For example, in the case of the partially
infilled frame, the displacement capacity corresponding to the NC
limit state amounts to 8.09 cm. The spectral acceleration at the
period (T = 0.22) of the equivalent SDOF system, Sae,NC , and the
corresponding peak ground acceleration, ag,NC , amount to 1.37g
and 0.55g, respectively (Section 4.3 in the companion paper). These
coordinates determine the NC point in Fig. 5. Similarly, the other
points defining the IN2 curve can be obtained from the results of
N2 analyses. For both of the infilled frames, the IN2 curves for the
equivalent SDOF system and for the MDOF system are presented in
Figs. 4 and 5, respectively.

For a probabilistic assessment, the approach summarized in

Section 2 will be used. In order to apply Eqs. (1) and (2), the hazard
curve (Section 3.3) and the IN2 curve (Section 3.4) have to be
known, as well as the values for the dispersion measures.
In the simplified approach, it is convenient to use default
values for the dispersion measures. Such values will have to be
determined in the future, based on extensive parametric studies.
For the example structures it was assumed that the dispersion
measure DR amounts to 0.4 for the bare frame, and to 0.7 for
the infilled frames. These values may be typical for, respectively,
buildings with moderate to long periods, and buildings with short
predominant periods [8]. A value of 0.25 was adopted for CR , DU
and CU . Note that the adopted values of the dispersion measures
for uncertainty are in agreement with the total uncertainty
dispersion measure UT = (2DU + 2CU )0.5 = 0.35 used in the
SAC-FEMA seismic performance evaluation [11]. The dispersion
measure for the hazard H was arbitrarily assumed to be 0.3.
For the determination of parameters k and b, the hazard curve
and the IN2 curve have to be idealized as explained in Section 2. In
order to facilitate comparisons of results for the three investigated
structures, the hazard curve and the IN2 curve have been expressed
in terms of the peak ground acceleration ag , rather than that of
the spectral acceleration at the natural period of the structure,
which is different for each of the three example structures. Note
that the fixed shape of the acceleration spectrum defines the
relation between the peak ground acceleration and the spectral
acceleration at any period.
In the idealization procedure the intervals have to be selected
in such a way that a close agreement between the actual and
approximated idealized curves is achieved. Based on results
of a previous study, intervals between 0.25 ag,LS and 1.0 ag,LS
were chosen for the approximation of the IN2 curve, and the
interval from 0.25 ag,LS to 1.25 ag,LS for the approximation of
the hazard curve, where ag,LS is the peak ground acceleration
corresponding to the selected limit state (LS). Since the intervals
for the approximation of the hazard curve vary for different limit

3190

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

Table 1
The parameter k corresponding to the hazard curve for moderate (MH) and high
hazard (HH) locations, for all three structures and limit states
Hazard

Structure

Limit state
DL

SD

NC

MH

BF
PI
FI

1.50
2.27
2.94

2.79
3.12
3.38

3.03
3.37
3.52

HH

BF
PI
FI

1.60
2.18
2.54

2.47
2.64
2.76

2.59
2.76
2.83

Table 2
The parameter b of the approximated IN2 curve, for all three structures and limit
states
Structure

Limit state
DL

SD

NC

BF
PI
FI

1.00
1.38
1.32

1.00
2.33
4.77

1.00
2.14
4.53

Table 3
Values of Cf
Hazard

Structure

Limit state
DL

SD

NC

MH

BF
PI
FI

1.29
2.12
3.90

2.38
1.64
1.15

2.77
1.98
1.18

HH

BF
PI
FI

1.33
2.00
2.78

1.97
1.42
1.10

2.11
1.58
1.11

Using Eqs. (1) and (2), and the above parameters, the probability
of exceedance of a given limit state can be calculated. For a 50%
confidence level, Cx equals 1, since Kx = 0 for x = 0.5. Eq.
(1) can
 thusbeinterpreted as the product of the mean hazard
aCg = H aCg CH at the peak ground acceleration causing the
H

Fig. 5. IN2 curves for the partially infilled frame (PI), the fully infilled frame (FI),
and the bare frame. Different limit states are indicated. The demand corresponding
to high hazard (HH) and moderate hazard (MH) locations is shown for the ground
motions with different return periods.

states and for different structures, the values of the parameter k

also vary, although the same hazard curve was used for all three
structures. The k values are shown in Table 1. The parameter k
increases with the severity of the limit state. It can be also observed
that the parameter k is the lowest for the bare frame, and the
highest for the fully infilled frame. Note that the probability of
exceedance increases with increasing k (Eqs. (1) and (2)).
The b values are presented in Table 2. For the bare frame, the IN2
curve is a straight line (see Fig. 5), resulting in b = 1. In the case of
the infilled frames, b is larger than 1. Due to the shape of the IN2
curves (Fig. 5) b is larger in the case of the NC and SD limit states
than in the case of the DL limit state. Note that the probability of
exceedance increases with decreasing b (Eqs. (1) and (2)).

selected limit state, and the correction factor Cf , which takes

into account randomness. Note that the notation for the intensity
measure has been changed in comparison with Eq. (1), since in
these calculations the peak ground acceleration was used as the
seismic intensity measure. CH (Eq. (2)) is equal to 1.046 for all cases
in this study, since H = 0.3. The values of Cf (Eq. (2)) vary between
1.10 and 3.90 (Table 3). In the case of the infilled frames, Cf is
larger for the DL limit state than for the other two limit states. The
opposite applies
  to the bare frame. The median value of the hazard
aCg at the peak ground acceleration corresponding to
function H
any selected limit state (see Fig. 5) can be determined directly from
the hazard curve, as shown in Fig. 6. The values are summarized
in Table 4. Finally, PLS,50 is determined from Eq. (1). The results
are presented in Table 5, both in terms of PLS,50 and in terms of
the probability of exceedence in 50 years, which is determined as
1 (1 PLS,50 )50 .
For a confidence level different from 50%, e.g. 90%, Cx (Eq. (2),
with Kx = 1.28) is different from 1. This factor represents the
ratio between the probability of exceedance of a given limit state
calculated with a 90% confidence level, PLS,90 , and PLS,50 . Constant
values have been adopted for dispersion measures CU and DU , so
for a given confidence level Cx depends only on the k/b ratio. The
values for Cx vary from 1.3 to almost 4, and are presented in Table 6.
The PPL,90 values presented in Table 7 are obtained by multiplying
Cx with PLS,50 . The results are also expressed in percents in 50 years.

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

3191

 
aCg , which correspond to selected limit states.
Fig. 6. Median hazard curves for the moderate (MH) and high seismic hazard (HH) locations with highlighted values H

Table 4
 
aCg for different limit states and both hazard locations
Median hazard H

Table 7
The probability of exceedance of a selected limit state, PLS,90 , for the bare frame and
for both of the infilled frames, for moderate and high hazard

Hazard

Hazard

Structure

Limit state
DL

SD

NC

BF
PI
FI

4.21E03
7.24E04
1.52E04

2.16E04
9.86E05
5.37E05

1.23E04
5.50E05
3.91E05

HH

BF
PI
FI

3.23E02
6.14E03
1.71E03

2.26E03
1.21E03
7.71E04

1.45E03
7.84E04
6.11E04

MH

HH

% in 50 years

Hazard

HH

Limit state
SD

NC

PLS,50
MH

HH

BF
PI
FI
BF
PI
FI

5.65E03
1.61E03
6.22E04
4.49E02
1.29E02
4.99E03

5.37E04
1.70E04
6.49E05
4.64E03
1.81E03
8.88E04

3.56E04
1.14E04
4.88E05
3.19E03
1.30E03
7.17E04

% in 50 years
MH

HH

BF
PI
FI
BF
PI
FI

24.7
7.7
3.1
89.9
47.8
22.1

2.6
0.8
0.3
20.7
8.7
4.3

1.8
0.6
0.2
14.8
6.3
3.5

Table 6
The values of Cx
Hazard

Structure

Limit state
DL

SD

NC

MH

BF
PI
FI

1.98
2.11
2.73

3.54
1.83
1.38

3.94
2.04
1.42

HH

BF
PI
FI

2.07
2.05
2.39

3.06
1.67
1.30

3.23
1.79
1.33

3.6. Discussion of the results

The highest annual probabilities of the exceedance of the
investigated limit states apply to the bare frame, and the smallest

SD

NC

PLS,90

Table 5
The probability of exceedance of a selected limit state, in terms of PLS,50 (above)
and in percents in 50 years (below), for the bare frame and for both of the infilled
frames, for moderate and high hazard

DL

Limit state
DL

MH

Structure

Structure

MH

BF
PI
FI
BF
PI
FI

BF
PI
FI
BF
PI
FI

1.12E02
3.40E03
1.70E03
9.30E02
2.63E02
1.19E02

43.1
15.7
8.2
99.2
73.6
45.0

1.90E03
3.11E04
8.94E05
1.42E02
3.02E03
1.15E03

9.1
1.5
0.4
51.1
14.0
5.6

1.40E03
2.33E04
6.94E05
1.03E02
2.33E03
9.51E04

6.8
1.2
0.3
40.4
11.0
4.6

values apply to the fullyinfilled

frame. These results can be partly

aCg values (Fig. 6), and partly to higher
attributed to higher H
values of Cf and Cx (with the exception of the DL limit state) as
a consequence of higher k/b ratios. The absolute values of the
annual probabilities for the NC limit state are within the range of
the target probabilities defined in Section 3.2 in the case of the
moderate hazard region and a 50% confidence level. In the case of
the moderate hazard and a 90% confidence level, only the annual
probabilities for the infilled structures are within the range of the
target probabilities. In all other cases the computed probabilities
of the NC limit state are larger than the upper bound of the range
of the target probabilities.
The difference between
the values in Table 4, which represent

aCg and neglect all sources of randomness and
median hazard H
uncertainty other than that of the hazard, to the PLS,50 values in
Table 5, that reflect all sources of randomness, and to the PLS,90
values in Table 7, that reflect both randomness and uncertainty, is
illustrated in Fig. 7. Plotted are the ratios between the probability of
exceedance of the peak ground acceleration,
 and
 the probability
  of
aCg and PLS,90 /H aCg for
exceedance of selected limit state PLS,90 /H
all three structures, all three limit states, and both hazard regions.
The first ratio is equal to the factor CH Cf (Eqs. (1) and (2)). The
second ratio represents the product CH Cf Cx (Eqs. (1) and (2)).
Different trends can be observed for different limit states. In the
case of the DL limit state, the ratios are the smallest for the bare
frame and the largest for the fully infilled frame. The opposite
trend applies to the SD and NC limit states. This is mainly the

M. Dolek, P. Fajfar / Engineering Structures 30 (2008) 31863192

3192

(a) Moderate hazard (MH).

(b) High hazard (HH).

Fig. 7. The ratio between the probability of exceedance of a limit state and the probability of exceedance of the peak ground acceleration. aCg .

consequence of the k/b ratio, which increases with more damaging

limit state in the case of bare frame, and decreases in the case of
infilled frames. This observation suggests that the seismic response
of regular infilled frames is, despite the higher inherent uncertainty
in their capacity and response, not only better in median or
deterministic terms (as shown in the companion paper [9]), but
also less sensitive to randomness and uncertainty (as evidenced
by the relatively smaller increase of their annual probability of
SD and NC limit states exceedance due to the various sources of
randomness and uncertainty).

The proposed method has, like other simplified methods,

limitations, which are basically the same as those which apply
to the basic N2 method [4] and to the SAC-FEMA method [1].
Additional studies are needed to determine reliable default
dispersion measures for randomness and uncertainty. Further
studies are also needed to provide guidelines for the idealization
of the hazard and IN2 curves, i.e. for the determination of the
parameters k and b, which strongly influence the results.

4. Conclusions

The results presented in this paper are based on the work

supported by the Slovenian Research Agency and by the European
Commission within the 6th Framework project LESSLOSS (505448GOCE).

The simplified method for a probabilistic seismic assessment

of buildings, summarized in this paper, is basically the SAC-FEMA
approach in which the most demanding part, i.e. the Incremental
Dynamic Analysis (IDA), is replaced by the incremental N2 method,
which is based on pushover analysis and the response spectrum
approach. Default values for dispersion measures are needed for
practical applications. The method was applied to a bare frame and
two infilled frames.
The results of the analyses indicate that the infills can
completely change the distribution of damage throughout the
structure. In the case of a regular distribution of infills throughout
the building, as in the case of the example infilled frames, the effect
of the infills is beneficial. The estimated probability of exceedance
of the NC limit state is the lowest for the fully infilled frame
and the highest for the bare frame. The beneficial influence of
infill, observed in the deterministic analysis which is presented
in the companion paper, is even more evident in the case of
the probabilistic analysis, due to a more favourable (smaller) k/b
ratio, which is characteristic for infilled frames. Despite the higher
inherent uncertainty in their capacity and response, regular infilled
frames are less sensitive to randomness and uncertainty.
The calculated probabilities of failure are influenced by the
conservative assumptions used in the test examples, especially
those related to the definition of failure. On the other hand it should
be noted that, in addition to damage due to flexure, damage in
the joints and due to shear also contribute to the total damage,
and may cause failure of reinforced concrete frames, especially
those with masonry infill. In the test examples, these sources of
potential damage were not considered since the tests and analyses
performed on the test structures indicated that the structural
behavior was dominated by flexure in the columns.

Acknowledgements

References
[1] Cornell CA, Jalayar F, Hamburger RO, Foutch DA. Probabilistic basis for 2000
SAC federal emergency management agency steel moment frame guidelines.
J Struct Eng ASCE 2002;128(4):52633.
[2] Deierlein G. Overview of a comprehensive framework for earthquake
performance assessment. In: Fajfar P, Krawinkler H (Eds.) Proceedings of the
international workshop on performance-based seismic designconcepts and
implementation. PEER report 2004/05, Berkeley; 2004. p. 1526.
[3] Vamvatsikos D, Cornell CA. Incremental dynamic analysis. Earthq Eng Struct
Dyn 2002;31:491514.
[4] Fajfar P. A nonlinear analysis method for performance-based seismic design.
Earthq Spectra 2000;16(3):57392.
[5] Dolek M, Fajfar P. IN2A simple alternative for IDA. In: Proceedings of the
13th world conference on earthquake engineering. 2004. p. 3353.
[6] Dolek M, Fajfar P. Simplified probabilistic seismic performance assessment of
plan-asymmetric buildings. Earthq Eng Struct Dyn 2007;36:202141.
[7] Dolek M, Fajfar P. Simplified non-linear seismic analysis of infilled reinforced
concrete frames. Earthq Eng Struct Dyn 2005;34(1):4966.
[8] Dolek M, Fajfar P. Inelastic spectra for infilled reinforced concrete frames.
Earthq Eng Struct Dyn 2004;33(15):1395416.
[9] Dolek M, Fajfar P. Effects of masonry infills on the seismic response of a four
storey reinforced concrete framedeterministic assessment. Eng Struct 2008;
30(7):19912001.
[10] Dymitios C, Kappos AJ, Chryssanthopoulos MK. Seismic reliability of masonryinfilled RC frames. J Struct Eng 2001;127:296305.
[11] FEMA 350. Recommended seismic design criteria for new steel moment
frame buildings. SAC Joint Venture. Federal Emergency Management Agency,
Washington (DC); 2000.
[12] Carvalho EC, Coelho E (Eds.). Seismic assessment, strengthening and repair of
structures. ECOEST2-ICONS report No.2, European CommissionTraining and
Mobility of Researchers Programme; 2001.
[13] Yun S-Y, Hamburger RO, Cornell CA, Foutch DA. Seismic performance
evaluation for steel moment frames. J Struct Eng ASCE 2002;128(4):53445.
[14] Joint committee on structural safety. Probabilistic model code, Part 1Basis of
design (12th draft); March 2001. http://www.jcss.ethz.ch/.