Bull Earthquake Eng

DOI 10.1007/s10518-015-9812-4
ORIGINAL RESEARCH PAPER

Seismic fragility assessment for reinforced concrete

high-rise buildings in Southern Euro-Mediterranean
zone
Jelena Pejovic1

Srdjan Jankovic1

Received: 29 December 2014 / Accepted: 4 September 2015

Springer Science+Business Media Dordrecht 2015

Abstract This paper presents seismic fragility assessment of RC high rise-buildings for
seismic excitation, typical for Southern Euro-Mediterranean zone. The fragility curves
were derived and log-normal cumulative distribution function parameters were obtained
for the four defined damage states by conducting 3600 nonlinear time-history analyses on
the basis of 60 ground motions with wide range of magnitudes, distance to source and
different site conditions, including in this way uncertainties during ground motion selection. As a prototype buildings, 20-story, 30-story and 40-story RC high-rise buildings with
core wall structural system were chosen. The key points of the process for obtaining the
fragility curves are shown by using algorithm, defined in this paper, and generally applicable to all types of RC high-rise buildings. For the purpose of conducting nonlinear timehistory analyses, non-linear 3D models of the buildings were designed. A detailed probabilistic seismic damage analysis was done and as its result the limit states as well as
corresponding damage states for RC high-rise buildings were defined, where the damage
states were treated as random variables. Inter-storey drifts at threshold of damage state
were defined as random variables with the range of possible values. Since no probabilistic
fragility curves exist for this class of buildings and for this seismic zone, this work partially
fills the void in Southern Euro-Mediterranean seismic risk assessment. The whole approach
presented in this paper may be used for efficient obtaining probabilistic fragility curves for
RC high-rise buildings of different configurations.
Keywords Reinforced concrete high-rise buildings  Seismic fragility assessment 
Fragility curves  Probabilistic seismic damage analysis  Non-linear time-history analysis 
Log-normal distribution function parameters

& Jelena Pejovic

jelenar@t-com.me
Srdjan Jankovic
srdjanj@t-com.me
1

Faculty of Civil Engineering, University of Montenegro, Podgorica, Montenegro

123

Bull Earthquake Eng

1 Introduction
In the last decades, urbanization and massive migrations of people, high land prices and
rapid economic development are just some of the reasons leading to the design of tall
buildings, which have become the norm in the architectural projects worldwide. In
December 2011, The Council on Tall Buildings, (The Council on Tall Buildings and Urban
Habitat 2011) claimed, that only in two decades, from 2000 to 2020, the average height of
building will be doubled. A similar trend has occurred in the Southern Euro-Mediterranean
zone. Since the entire Mediterranean zone is a seismically active zone, it is necessary to
perform detailed seismic risk assessment analyses of these populations of buildings. Within
the process of seismic risk assessment for the certain types of facilities and for certain
locations as well, a seismic vulnerability analysis of buildings is conducted resulted in
obtaining the fragility curves. Today, in literature, there is a lack of information regarding
obtaining fragility curves for RC high-rise buildings. Until today in Southern EuroMediterranean seismic zone, for the population of RC high-rise buildings, such fragility
curves have not been obtained. The topic of this paper is related to the seismic fragility
assessment for RC high-rise buildings and to the derivation of fragility curves for seismic
excitation, specific for Southern Euro-Mediterranean zone. As a prototype buildings,

Fig. 1 Proposed fragility assessment framework

123

Bull Earthquake Eng

20-story, 30-story and 40-story RC high-rise buildings are chosen, with core wall structural
system. The ground motions are chosen from Southern Euro-Mediterranean zone.
Fragility curves are defined as the conditional probability of exceedance of particular
damage (DS) or limit state for a given ground motion intensity measure (IM). Fragility
curves can be written in mathematical form as follows (DM is a demand measure, dmDS is
limit value of demand measure for particular damage state DS) (Eq. 1):


1
PDS=IM  P DM [ dmDS =IM
There are different ways for obtaining fragility curves, from simple to more complex.
Simple methods provide approximated fragility curves while more complicated methods
provide more realistic and accurate curves. Rossetto and Elnashai (2003) classify the
existing fragility curves into four groups: empirical fragility curves, expert fragility curves,
analytical fragility curves and hybrid fragility curves. Empirical fragility curves are
obtained by statistical analyses of damaged buildings in the occurred earthquakes. The
example for empirical fragility curves for the region of Europe and for RC buildings are the
curves done by Rossetto and Elnashai (2003). Expert fragility curves are based partially or
completely on experts opinion and represent the simplest way for obtaining the curves.
The example for expert fragility curves are the curves implemented into HAZUS database
(National Institute of Building Sciences 1999). Analytical fragility curves are derived using
numerical models to simulate the behavior of systems. In the absence of experimental data,
observational data or an opinion of an expert, the only way to explore vulnerability of
buildings is by using analytical methods. Due to the lack of experimental data and data on
field in the occurred earthquake, the fragility curves for RC high-rise buildings could only
be derived by analytical methods (Ji et al. 2007b). Hybrid fragility curves are derived by
combining the above mentioned curves.
For obtaining fragility curves, in this paper analytical method is used. The key points of
the process of obtaining fragility curves are shown by using algorithm, defined in this
paper, and generally applicable to all types of RC high-rise buildings (Fig. 1). The fragility
curves were derived, as well as log-normal cumulative distribution function parameters
were obtained for the four defined damage states by conducting 3600 nonlinear timehistory analyses on the basis of 60 ground motions with wide range of magnitudes, distance
to source and different site condition. In this way uncertainties during ground motion
selection are included. Ground motion uncertainties are usually much higher than other
types of uncertainties in the probabilistic seismic risk analysis (Ji et al. 2009). Therefore,
the ground motions are in this paper treated as the main source of uncertainty in the process
of obtaining the fragility curves. In order to obtain fragility curves, due to their complexity
and large scope of work, the authors of this paper adapted the process by creating the
program in MATLAB (2013).
One of the most important phases in the process of obtaining fragility curves is defining
the limit states and corresponding damage states because they directly affect on derived
fragility curves (Erberik and Elnashai 2004). This paper provides a detailed damage
analysis of the prototype structures in order to define limit states and corresponding
damage states for RC high-rise buildings. A detailed quantitative approach is applied and
certain inter-storey drifts are determined at threshold of each defined damage state. These
analyses resulted in the four damage states for which, the fragility curves are obtained.

123

Bull Earthquake Eng

2 Selection and description of prototype RC high-rise buildings

As a prototype buildings, in this paper, 20-story, 30-story and 40-story RC high-rise
buildings with core wall structural system are selected. The specific plan view of the story
characteristic for all prototype buildings, ETABS model (CSI 2013) and PERFORM-3D
model (CSI 2006) of the 30-story prototype RC high-rise building are shown on Fig. 2. RC
core wall structural system is a system which is applicable for RC high-rise buildings up to
the 50 storey (Taranath 2010).
The structural system with RC core wall is very suitable for architectonic reasons and is
very often used in high-rise buildings. RC ductile walls are placed in the central part of the
building around communication core, (lifts, staircase), forming in that way the spatial
system capable to resist to lateral loads in both directions. The space between central RC
core to the building exterior usually stays free or rarely filled with gravity RC columns
connected with beams or flat slab, providing more space in that way. At the building
perimeter, RC frames, used only as gravity load system, are formed. In these systems of
RC high-rise buildings, RC core wall accept all of the lateral load (Taranath 2010). The
main features of the prototype RC high-rise buildings are shown in the Table 1.
Seismic design of the prototype RC buildings was done in accordance with Eurocode 2
(CEN EC2 2004) and Eurocode 8-1 (CEN EC8 2004). The dimensions of the structural
elements of prototype buildings are obtained according to the requirements defined in
Eurocodes provisions. Seismic linear analysis of buildings was done using a multi-modal
response spectrum analysis, which is quite appropriate due to higher-mode effects in highrise RC buildings. The modal periods of buildings and mass participation factors of first
four modes are shown in the Table 2. The elastic flexural and shear stiffness properties of
structural elements are taken to be equal to one-half of the corresponding stiffness of the
uncracked elements, according to Eurocode 8-1 (CEN EC8 2004). For linear analysis and
seismic design of buildings, ETABS spatial models of buildings (CSI 2013) were used.
Seismic forces are dominantly accepted by RC core walls. For this reason, RC core walls
were subject of further detailed seismic design in accordance with relevant provisions of
Eurocode and thereafter nonlinear time-history analyses.

Fig. 2 30-story prototype building ETABS2013 model (left), plan view of the story (middle) and 30-story
prototype building PERFORM3D model (right)

123

Bull Earthquake Eng

Table 1 Main features of the prototype RC high-rise buildings
Features

20-story prototype
building

30-story prototype
building

40-story prototype
building

Total height (m)

60

90

120

Storey height (m)

Floor RC slab thickness

20 cm

20 cm

20 cm

RC beams

40 9 65 cm

40 9 65 cm

40 9 65 cm

RC columns

80 9 80 cm

80 9 80 cm

90 9 90 cm

Core walls thickness

15 storey: 30 cm

15 storey: 40 cm

110 storey: 55 cm

620 storey: 20 cm

630 storey: 30 cm

1140 storey: 45 cm

Coupling beams in X
direction

20 9 80 cm and
30 9 80 cm

30 9 80 cm and
40 9 80 cm

45 9 80 cm and
55 9 80 cm

Concrete fck (fcm) (MPa)

35 (43)

45 (53)

55 (63)

Reinforcement fyk (fym)

(MPa)

500 (575)

500 (575)

500 (575)

Modulus of elasticity Ecm

(MPa)

34,000

36,000

38,000

3 Nonlinear modeling of prototype RC high-rise buildings

In the study, the PERFORM-3D program (CSI 2006) was used for the nonlinear timehistory analyses. The nonlinear models are designed as a spatial models and they consist of
RC core walls. The mathematical models used for elastic analysis are extended to include
the strength of structural elements and their post-elastic behaviour. Element properties are
based on mean values of the properties of the materials. Stressstrain relationship for
unconfined concrete, confined concrete and reinforcement steel are adopted in accordance
with the recommendations of Eurocode 8-2 (CEN 2005). Stressstrain relationship for
unconfined concrete with concrete mean strength of 53 MPa and confined concrete for
typical confining reinforcement in boundary wall elements are shown in Fig. 3. The steel
material is modeled with a bi-linear stressstrain relationship according to Eurocode 8-2
(CEN 2005) with expected yield mean strength of 575 MPa and ultimate strength of
660 MPa, both in compression and tension. The floor slabs are modeled as rigid
diaphragms.
The core walls are modeled using non-linear vertical fiber elements representing the
expected behavior of the concrete and reinforcing steel (CSI 2007). The area and location
of reinforcement within the cross-section and the properties of the concrete are defined
using individual fibers. The shear behavior is modeled as elastic. The behavior for out-ofplane bending and behavior in horizontal transverse plane are assumed to be elastic. The
hinge lengths at the base of the wall are adopted according to Eurocode 8-1 (CEN EC8
2004). The coupling beams are defined as elastic beam elements with a nonlinear displacement shear hinge at the mid-span of the beam. These are connected to the shear walls
using embedded elements as suggested by Powell (CSI 2007). The shear hinge behavior is
based on test results by Wallace (2012).

123

Bull Earthquake Eng

Table 2 Modal periods and mass participation factors for the prototype RC high-rise buildings
Prototype buildings

20-story prototype
building

30-story prototype
building

40-story prototype
building

Period in Y direction (s)

Mode
1

1.652

2.880

4.097

0.389

0.623

0.858

0.181

0.270

0.355

0.117

0.164

0.207

1.641

2.597

3.511

0.480

0.702

0.880

0.250

0.347

0.423

0.164

0.228

0.275

Period in X direction (s)

Mode

Mass participation factors in Y direction (%)

Mode
1

64.26

63.53

63.24

20.32

19.43

18.94

7.04

7.05

7.05

3.23

3.57

3.65

94.85

93.58

92.88

Sum of mass part factors in Y

direction (%)

Mass participation factors in X direction (%)

Mode
1

69.36

67.70

66.08

15.96

17.40

18.78

5.49

5.23

5.68

2.83

2.78

2.64

93.64

93.11

93.18

Sum of mass part factors in X

direction (%)

4 Ground motion records selection

In the study, Southern Euro-Mediterranean seismic zone is selected. Therefore, the ground
motions selection was done within this zone. The data of Seismological Institute of
Montenegro and European strong motion database were used as database of ground
motions (Ambraseys et al. 2002). Ground motions were selected depending on the value of
magnitude M, distance to source R and site conditions. From the large number of available
records, there were chosen 60 ground motions out of which, 25 selected ground motions
were recorded on the rock which correspond to soil type A and 35 selected ground motions
recorded on stiff soil which correspond to soil type B, according to Eurocode 8-1 (CEN
EC8 2004). The values of magnitude of selected ground motions are in the range between
5.1 and 7.0 while the distances to source are in the range from 5 to 70 km. By using larger
number of ground motions with wider range of magnitudes, distance to source and

123

Bull Earthquake Eng

Fig. 3 Stressstrain relationship for unconfined and confined concrete with concrete mean strength of
53 MPa

different site conditions, uncertainties during ground motions selection are being included.
High-rise buildings are specific, due to fact that their response frequency range is much
wider than for low-rise or mid-rise buildings. Accordingly, it is necessary to include a
larger number of ground motions, various magnitudes and distances to source. Uncertainties during ground motions selection are usually much higher than other types of
uncertainties in the probabilistic analysis of seismic risk.
The main criterion used in this paper for the selection of ground motions, requires the
mean spectrum of the selected ground motions to be compatible with relevant target
spectrum over the wide spectral period range of interest. As a target spectrum, elastic EC8
spectrum for reference return period of 475 years with design ground acceleration of
0.37 g was chosen. Due to the lack of ground motions in the Southern Euro-Mediterranean
zone, which may be selected without being previously scaled and with mean spectrum to
be in accordance with Eurocode spectrum, it was necessary to scale the ground motions.
The mean squared error method (MSE) was chosen as a mode of scaling of ground motions
(PEER 2010). By this method ground motions are scaled in a way where the mean squared
error is minimized over the whole range of periods. The mean square error represents the
difference between the spectral acceleration of ground motion records and target spectrum
and it is calculated by Eq. (2).
2
Pn  target
Ti  f  Srecord
Ti
a
i1 Sa
2
MSE
n
Parameter f in Eq. (2) is a linear scale factor. The geometric mean spectrum of the selected
ground motions is adopted to be the mean spectrum (PEER 2010). The MSE method is
especially effective in the selection of ground motions since it allows to choose, from the
large number of available records, ground motions whose response spectra the least deviate
from the target spectrum. In this way, the selected original ground motions scaled to
different intensity levels defined by peak ground acceleration (PGA) resulted in the small
dispersion of seismic response parameters (with coefficient of variation less than 0.3).
Except earthquakes which have reference return period of 475 years (or analogously the
seismic action associated with reference probability of exceedance, 10 %, in 50 years), the

123

Bull Earthquake Eng

prototype buildings were tested on the seismic action which have reference return period of
2475 years (or analogously the seismic action associated with reference probability of
exceedance, 2 %, in 50 years). Since in Eurocode 8-1 (CEN EC8 2004) such high level of
seismic intensity is not defined, it was necessary to use the latest literature for defining
appropriate seismic action with 2 %/50. Data for this seismic intensity is defined by the
project called Seismic hazard harmonization in Europe-SHARE (Giardini et al. and the
SHARE consortium 2013). Seismic intensity which corresponds to return period of
2475 years for the territory of Mediterranean is two times higher than seismic intensity
which corresponds to return period of 475 years. In accordance with above mentioned, the
mean spectrum of ground motions with intensity of 2 %/50 is two times higher than the
mean spectrum of ground motions of 10 %50.
In the Figs. 4 and 5, response spectra of selected ground motions scaled by MSE
method for the intensity level of 10 %/50, the mean spectrum and relevant target spectra
(Eurocodes 8-1 elastic spectra) for the intensity level of 10 %/50 and the mean spectrum
for the intensity level of 2 %/50, for certain soil types are shown.
In this way scaled ground motions are directly used in probabilistic seismic damage
analysis of prototype RC high-rise buildings. For obtaining the fragility curves, selected
original ground motions are scaled to different intensity levels defined by PGA.

5 Probabilistic seismic damage analysis of prototype RC high-rise

buildings
One of the most important phases of probabilistic performance-based analysis is defining
the limit states and corresponding damage states. Realistic and comprehensive limit states
determination and thus identification of the performance levels is one of the important

Fig. 4 Response spectra of the selected ground motions for soil type A, mean spectra of the selected ground
motions for intensity levels 10 %/50 and 2 %/50 and elastic EC8 spectrum for soil type A for intensity level
10 %/50

123

Bull Earthquake Eng

Fig. 5 Response spectra of the selected ground motions for soil type B, mean spectra of the selected ground
motions for intensity levels 10 %/50 and 2 %/50 and elastic EC8 spectrum for soil type B for intensity level
10 %/50

steps in the process of obtaining the fragility curves because of their direct influence on
derived fragility curves (Erberik and Elnashai 2004). Limit states are described with
possible and acceptable losses. The losses may be presented in different ways: through
possible levels of damage state (structural damage and nonstructural damage), number of
killed people in the earthquakes, economic losses, the time when the facility is out of
service, repairment costs and other. Limit states present discrete points on continuous scale
of damage states of structures. There are two basic approach of defining the limit states:
qualitative and quantitative approach. Qualitative approach in a description of structures
limit state is the most used in building regulations. FEMA 356 (2000) defines three basic
limit states for structural elements or performance levels: Immediate Occupancy IO, Life
Safety LS and Collapse Prevention CP. HAZUS (National Institute of Building Sciences
1999) defines four levels of structural and non-structural damage as follows: slight damage,
moderate damage, extensive damage and complete damage. Rossetto and Elnashai (2003)
defines seven limit states based on observational data of damages on buildings after the
earthquakes. They defined homogeneous scale of damage for RC structures (HRC scale)
for presenting the level of damage on RC structures with levels from 0 to 100 for four
characteristic RC structural systems.
Due to the lack of research in the case of RC high-rise buildings, particularly for
structural systems with core wall, current qualitative approaches can only serve as references (Ji et al. 2007a). In order to obtain analytical fragility curves for RC high-rise
buildings, it is necessary to apply more detailed quantitative approach for defining limit
states and certain damage states. For quantification of limit states, it is necessary to express
the damage states in terms of deformation. Inter-storey drift is the most used parameter in
the literature for defining the limit states due to its simplicity. Nonlinear time-history
analyses and empirical observations suggest that there is a strong connection between
inter-storey drift and the level of structural damage. In the existing literature there are few

123

Bull Earthquake Eng

examples of defining the limit states by quantitative approach for structural system with
ductile walls using inter-storey drift such as Ghobarah (2004) and Rossetto and Elnashai
(2003). Specifically for RC high-rise buildings, Ji et al. (2007a) proposed the definition for
limit states and they determined relevant inter-storey drifts for certain limit states on the
basis of 54-story building with core wall and RC frames. In this paper limit states and
certain damage states for the prototype buildings with respect to inter-storey drifts at
threshold of damage state are defined. For that purpose, a detailed probabilistic seismic
damage analysis was conducted through performing a large number of nonlinear timehistory analyses.

5.1 Defining of limit states and damage states

In this paper the limit states are defined in the following way: Immediate Occupancy IO,
Life Safety LS and Collapse Prevention CP (Fig. 6). Structure at point of limit state
Immediate Occupancy IO has slight or no damage. Structure at point called Life Safety LS
can sustain moderate damage but still stays in the zone of high safety against collapse.
Structure at the point of limit state called Collapse Prevention CP has significant and large
damage and it is near to collapse. In compliance with limit states certain damage states are
defined. The four defined damage states are: slight damage DS1, moderate damage DS2,
extensive damage DS3 and complete damage DS4.
Such defined damage states are in compliance with damage scale of selected damage
index Park and Ang DIPA (Park and Ang 1985). Park and Ang index is adopted to be a
damage index in this paper. It is one of the most used damage indexes for its simplicity,
stability and confirmation through experimental research. Park and Ang damage index is
obtained by a linear combination of damage, caused by maximum inelastic deformation
and by the cumulative damage, resulting from repeated cyclic response (Eq. 3) (Park and
Ang 1985).
DIPA

u
Eh
b
uu
Fy  uu

where u is maximum inelastic displacement during a ground motion, uu is an ultimate

displacement capacity of the system under a monotonically increasing lateral deformation,
Eh is hysteretic energy, Fy is yield strength of the structure, b dimensionless constant that

Fig. 6 Defined limit states and corresponding damage states

123

Bull Earthquake Eng

depends on the structural properties and for the usual RC structures has value from 0.10 to
0.25 with mean value of 0.15.
According to the data based on damage of RC buildings, which were slightly or severely
damaged during few earthquakes occurred in America and Japan, Park and Ang defined the
relation between damage level and damage index (Table 3) (Park and Ang 1985).
The value of damage index DIPA = 0 indicates non-damaged structure or structure
remains in elastic behavior during earthquake. For damage index values DIPA \ 0.2 the
structure is exposed to minor damage, accompanied by visible cracks in structural elements. There is no delay in the functioning of facility for this level of damage and
rehabilitation of the structure is relatively simple. The values of damage index between 0.2
and 0.5 correspond to moderate level of damage, accompanied by the appearance of
reinforcement yielding in critical zones of some structural elements, including spalling of
the concrete cover, as well as formation of larger cracks in the plastic hinge zones. The
value of damage index DIPA = 0.5 is considered as a boundary between moderate and
severe damages or boundary between damage that can be repaired and damage that is
irreparable or the costs of damage are not economically justified. In a case of severe
damage it occurs concrete crushing and local buckling of longitudinal bars in the plastic
hinge zones. The collapse of the structure occurs in case of damage index value DIPA [ 1.
At this stage a loss of shear and/or axial load bearing capacity occurs, which results in
partial or complete collapse of structure.
As a deformation measure for defining limit states, inter-storey drift is selected, which is
at the same time the most used seismic response parameter. Specifically, maximum interstorey drift (relative storey displacement divided by the height of the storey) for whole
structure IDRmax is selected. In order to define range of certain damage state it is necessary
to determine inter-storey drift at threshold of damage state. At threshold of each damage
state there is corresponding damage index DIPA according to the Park and Ang scale. For
that reason, it is necessary to maintain correlation between the maximum inter-storey drift
and damage index, assuming that maximum inter-storey drift is required, depending on a
damage index. In this paper the values of inter-storey drifts at threshold of damage state are
described as random variables with range of possible values not as a single value or
deterministic parameter.

5.2 Results of damage analysis and relationship between damage index DIPA
and maximum inter-storey drift IDRmax for prototype buildings
For determination of inter-storey drifts at threshold of certain damage state for the prototype RC high-rise buildings, nonlinear time-history analyses were used. The prototype
buildings are exposed to the selected 60 ground motions with two levels of intensity (10 %/
50 and 2 %/50) in both directions of structures, which, in total include: 720 nonlinear timehistory analyses. These analyses required approximately 180 h of runtime on computer

Table 3 Interpretation of
damage index

Damage states

Damage index

State of structure

Slight damage

0.00.2

Serviceable

Moderate damage

0.20.5

Repairable

Extensive damage

0.51.0

Irreparable

Collapse

[1.0

Loss of storey or buildings

123

Bull Earthquake Eng

Intel CoreTM i5-3470 CPU 3.20 GHz and 8 GB of memory. After performing nonlinear
time-history analyses for selected ground motions result diagram is obtained which presents 720 points of pairs (DIPA,i, IDRmax,i) (Fig. 7). For each ground motion Park and Ang
damage index DIPA,i and maximum inter-storey drift IDRmax,i were determined. Regression
analysis was conducted. According to the constructed diagram it can be noticed that
between these two parameters is possible to establish linear regression model with very
high coefficient of determination R2 = 0.7981 or the coefficient of correlation r = 0.8934
which show very strong mathematical connection of these two parameters. Small values of
dispersion rIDRmax/DI were obtained. Since the distribution of seismic response parameter
IDRmax corresponds to the log-normal distribution, dispersion rIDRmax/DI is defined as the
standard deviation of the natural logarithms of the residuals IDRmax data from the
regression line. Obtained regression line represent median of random variable IDRmax/
DIPA i.e. value for which the probability that the random variable will have less or equal
value is 50 %. In the Fig. 7 are shown the lines which correspond to plusminus one
standard deviation away from the median and in a case of log-normal distribution represent
16 % percentile and 84 % percentile.
With obtained linear relationship between damage index DIPA and maximum interstorey drift IDRmax, the values of inter-storey drifts at threshold of each damage state in
accordance with Park and Ang damage scale are determined. The values of derived interstorey drifts, their mean values (50th percentiles), 16 % percentiles and 84th percentiles for
certain damage states are shown in the Table 4.
Besides that, in the Table 4 confidence intervals for each damage state are shown.
Confidence interval provides information regarding the closeness of the calculated mean
value to the population mean value and is expressed by the probability. Plusminus onesigma confidence interval (L1, L2) which corresponds to the probability of 84 % (calculated by Eq. 5) is shown in Table 4. In analytical practice the acceptable relative width of
confidence interval is 10 %. For derived inter-storey drifts at threshold of damage states
were obtained relative width of confidence interval (calculated by Eq. 6) in the amount

Fig. 7 Relationship between damage index DIPA and maximum inter-storey drift IDRmax for prototype
buildings

123

Bull Earthquake Eng

about 2 % what is a very small relative width of confidence interval and indicates the high
level of accuracy of calculated random variables. On the other hand, the obtained values of
inter-storey drifts are in compliance with drifts defined by Ghobarah (2004) as well as with
inter-storey drifts defined by Rossetto and Elnashai (2003) over homogeneous damage
scale for RC buildings (HRC scale) with small differences because they were not formed
for high-rise RC buildings. If we compare inter-storey drifts at threshold of extensive
damage state and complete damage state defined in HAZUS database (National Institute of
Building Sciences 1999) (0.2 %-slight, 0.5 %-moderate, 1.5 %-extensive, 4.0 %-complete) for structural system with ductile walls and number of storey higher than 8 (HAZUS
definition of high-rise buildings) with herein derived values for damage states DS3 and
DS4, we notice that later values are lower. This is because, the prototype buildings have
20, 30 and 40 stories and belongs to the category of real high-rise buildings which are more
flexible and more fragile than high-rise buildings from HAZUS database, which are
defined as buildings with number of stories [8.

6 Fragility curve assessment

Schematic representation of methodology used for obtaining the fragility curves is shown
in Fig. 8. Accordingly, for each level of intensity measure IMj, demand measures (maximum inter-storey drifts IDRmax) are obtained by conducting nonlinear time-history
analyses for all selected ground motions. Thus, scatter plot of nonlinear time-history
analysis results in terms of IDRmax for different levels of intensity measure IMj are
obtained. Then, for each level of intensity measure IMj and corresponding IDRmax values
(obtained as vertical scattered data), the probability of exceedance Pi,j of a certain damage
state (defined by IDRDSi
max) is calculated using Monte Carlo method. In this way, obtained
pairs, intensity measure, and corresponding probability of exceedance (IMj, Pij) are drawn
on the graphic and a dotted fragility curve is obtained. The dotted diagram (diagram of
sample probabilities) is fitted with most similar curve which represents a log-normal
cumulative distribution function. In the process of curve fitting, optimal log-normal distribution function parameter l and r for fragility curves are derived. Due to its simple
implementation and satisfying accuracy, in this research Monte Carlo method was applied
for determination of sample probabilities.
Log-normal cumulative distribution function by which the diagram of sample probabilities is fitted has the following form (Eq. 4):
Table 4 Derived values of inter-storey drifts at threshold of damage states for the prototype RC high-rise
buildings
Inter-storey drift at threshold of damage state
IDRmax (%)

Lower and upper

endpoint of the 84 %
confidence interval

Relative width of
confidence interval

Median
(50th percentile)

16th
percentile

84th
percentile

L1

L2

i (%)

DS1

0.250

0.190

0.330

0.247

0.253

2.40

DS2

0.528

0.398

0.702

0.522

0.534

2.27

DS3

0.945

0.710

1.260

0.935

0.955

2.12

DS4

1.640

1.230

2.190

1.622

1.658

2.19

Damage
states

123

Bull Earthquake Eng

Fig. 8 Schematic representation of obtaining fragility curve





lnIM  l
P DSi IDR [ IDRDSi =IM U
r

U is the standard normal cumulative distribution function. The basic parameters of fragility
curve are the mean value l and standard deviation r. For log-normal distribution, the mean
value l is the median value of intensity measure at which the building reaches the
threshold of the damage state, i.e. the value of intensity measure for which probability of
exceedance threshold of damage state is 50 %. r represents standard deviation of the
natural logarithms of intensity measure for certain damage state. During the process of
fitting of fragility curves, optimal values for these two parameters are derived. In order to
obtain fragility curves and their optimal parameters, the authors of this paper adapted the
process by creating the program in MATLAB (2013) in compliance with schematic representation in the Fig. 8. This program includes the whole procedure of obtaining the
fragility curves.
In the literature, there are several intensity measures that have been used for obtaining
the fragility curves. The most used are the following: PGA, spectral acceleration at some
periods (Sa), and spectral displacement at some periods (Sd). In this paper, PGA was
chosen as an intensity measure for obtaining the fragility curves. PGA was chosen because
it is the most used intensity measure for obtaining the fragility curves and, for the reason of
comparison and incorporation of derived fragility curves into existing databases mostly
based on intensity measure of PGA (for example Syner-G European database). Moreover,
it was chosen due to the simplification of problem regarding obtaining the fragility curves

123

Bull Earthquake Eng

for complex buildings, such as the prototype RC high-rise buildings. The next step in
research of this type of buildings is obtaining fragility curves with other intensity measures.

6.1 Derived fragility curves

The selected prototype RC high-rise buildings are exposed to the selected 60 original
ground motions which are scaled to different intensity levels defined by PGA. As the range
of PGA values a fixed range [0.1, 1.0 g] with increments of 0.1 g is adopted. Ten sets of
ground motions were formed, corresponding to each separate intensity level. For the
selected ground motions, 3600 nonlinear time-history analyses for the both directions of
seismic action are performed. This required approximately 900 h of runtime on computer
Intel CoreTM i5-3470 CPU 3.20 GHz and 8 GB of memory. The fragility curves for each
damage state were obtained, by using created MATLAB program. The derived values of
inter-storey drifts at threshold of each damage state obtained by probabilistic damage
analysis in the previous chapter are used (Table 4). Derived fragility curves of the prototype RC high-rise buildings for intensity measure PGA including fitted log-normal
cumulative distribution function are shown in the Fig. 9. Each of four derived fragility
curves correspond to one of defined damage states (DS1, DS2, DS3 i DS4). Parameters of
fitted log-normal cummulative distribution functions for each damage state are shown in
the Table 5.
The Table 5 and Fig. 9 show, for example, that mean value of intensity measure PGA
for damage state DS2 (moderate damage) is 0.3519 g and represents the median value of
intensity measure at which the building reaches the threshold of the damage state DS2.
This means, that by the probability of 50 % for the value PGA of 0.3519 g, the threshold of
damage state DS2 is reached (i.e. the appearance of reinforcement yielding at critical zones
of some structural elements, including spalling of the concrete cover as well as forming
larger cracks in the plastic hinge zones). 84th percentile for damage state DS2 is 0.6864 g

Fig. 9 Derived fragility curves for the prototype RC high-rise buildings

123

Bull Earthquake Eng

while 16th percentile is 0.1804 g. For the mean value of intensity measure PGA for
damage state DS2 plusminus one sigma confidence interval is [L1,
L2] = [0.33970.3645 g]. Plusminus one sigma confidence interval for log-normal distribution random variable, which corresponds to the probability of 84 %, is calculated by
Eq. (5) in the following way:
pr
5
L1;2 l  e N
where L1 and L2 are lower and upper endpoint of the 84 % confidence interval for the
mean of a log-normal distribution, l is the sample mean value of random variable, r/HN is
standard error of the mean, r is standard deviation of the sample mean value, N is sample
size.
The relative width of confidence interval is calculated by Eq. (6):
i 100 

L2  L1
%
l

For all damage states the relative width of confidence interval is lower than 10 %, which is
acceptable and indicates small width of interval containing the mean value with relatively
high probability of 84 %. All this indicate the achieved great accuracy of obtained fragility
curves as well as their possible implementation for the case of RC high-rise buildings with
RC core wall or similar structural system.
The fragility curves shown in the Fig. 9 are derived by obtained mean values (median
50th percentile) of inter-storey drifts at thresholds of certain damage states. In the probabilistic seismic damage analysis except mean values, the values of 16th percentile and
84th percentile are obtained (Table 4). In the Fig. 10 all three types of derived fragility
curves are shown (fragility curves obtained by mean values, by values of 16th and 84th
percentile of inter-storey drifts at threshold of damage state).

6.2 Analysis and comparison of fragility curves

The derived fragility curves were compared to each other, for the purpose of making
difference in relation to various characteristics of earthquakes. The effects of magnitude,
distance to source and site conditions, on fragility of RC high-rise buildings, were
analysed.

Table 5 Derived log-normal distribution function parameters l and r for fragility curves of the prototype
RC high-rise buildings
Damage
states

Mean l
(arithmetic
space) PGA (g)

Standard
deviation
r

84th
percentile
PGA (g)

16th
percentile
PGA (g)

Lower and upper

endpoint of the 84 %
confidence interval

Relative width
of confidence
interval

L1

L2

i (%)

DS1

0.1683

0.6304

0.3161

0.0896

0.1628

0.1740

6.65

DS2

0.3519

0.6682

0.6864

0.1804

0.3397

0.3645

7.05

DS3

0.6220

0.6554

1.1979

0.3230

0.6009

0.6439

6.91

DS4

0.9520

0.5133

1.5906

0.5698

0.9266

0.9781

5.41

123

Bull Earthquake Eng

6.2.1 Comparison of fragility curves with respect to the magnitude

The selected 60 ground motions with the values of magnitude in the range between 5.1 and
7.0, are divided into two groups depending on the level of the magnitude: group with range
of magnitudes M = [5.16.2] which contains 29 ground motions (out of which 14 ground
motions with range of distance to source R = [120 km] and 15 ground motions with
range R = [2070 km]) and group with range of magnitudes M = [6.37.0] which contains 31 ground motions (out of which 15 ground motions with range of distance to source
R = [120 km] and 16 ground motions with range R = [2070 km]). Fragility curves for
particular groups are obtained in order to analyze the fragility of high-rise buildings in
relation to the magnitude. The derived fragility curves with two different levels of magnitude, M = [5.16.2] and M = [6.37.0], and for the whole range of distance to source
are shown in the Fig. 11. In the Figs. 12 and 13 are shown derived fragility curves for the
above mentioned levels of magnitude and separately for two range of distance to source as
follows: R = [120 km] and R = [2070 km]. From the Fig. 11 it can be easily noticed
that the fragility of high-rise buildings with higher level of magnitude is larger than the
fragility with smaller level of magnitude for the same damage states and the same PGA
values. The differences become more obvious as the damage state increases (from DS1 to
DS4). For example, for the value PGA of 0.3 g, the probability of exceedance the threshold
of damage state DS2 (moderate damage) for the range of magnitudes M = [5.16.2] is
17 % while for the range of magnitudes M = [6.37.0] the probability is 64 %. The huge
differences are also in the derived parameter l of fragility curves (Table 6). The Table 6

Fig. 10 Derived fragility curves for the prototype RC high-rise buildings obtained by 50th percentile, 16th
percentile and 84th percentile of inter-storey drifts at thresholds of certain damage states

123

Bull Earthquake Eng

shows that medians of intensity measure PGA in the case of larger magnitudes are lower
than in the case of smaller magnitudes, what means that by the same probability of
exceedance of 50 % in case of larger magnitudes, the threshold of certain damage state is
reached for lower values of PGA. Hence magnitude has a significant influence on seismic
vulnerability assessment of RC high-rise buildings and it can be connected with damage
states of RC high-rise buildings.
From the Figs. 12 and 13 it can be noticed that the differences between fragility of RC
high-rise buildings for two observed range of magnitudes are greater when the distance to
source is smaller than when is larger. It is interesting to notice, that, when it comes to the
ground motions with smaller magnitudes larger fragilities are obtained in case of larger
distance to source while regarding the ground motions with larger magnitudes, the larger
fragilities are obtained in case of closer earthquakes.

6.2.2 Comparison of fragility curves with respect to distance to source

The Fig. 14 shows the comparison of derived fragility curves for two different range of
distance to source as follows: R = [120 km] which contains 15 ground motions and
R = [2070 km] which contains 16 ground motions for higher level of magnitude
M = [6.37.0]. It is noticeable from the Fig. 14 that the fragility curves are nearly equal in
case of smaller and larger distance to source. For the higher PGA values, i.e. for values
over median for certain damage states, there is a larger probability of exceedance of
damage states for closer compared to distant earthquakes. For lower PGA values than
median, there is equal or larger probability of exceedance for distant earthquakes. There is
almost no difference in derived parameter l of fragility curves (Table 7). According to this
results, it is evident high fragility of RC high-rise buildings to the impact of distant
earthquakes. This happens because for ground motions with larger distance to source, after

Fig. 11 Fragility curves of the prototype RC high-rise buildings for different magnitude levels

123

Bull Earthquake Eng

Fig. 12 Fragility curves of the prototype RC high-rise buildings for different magnitude levels and distance
to source R = [120 km]

Fig. 13 Fragility curves of the prototype RC high-rise buildings for different magnitude levels and distance
to source R = [2070 km]

decreasing of high frequency, comes lower frequencies of records which still excite high
periodical high-rise buildings to considerable displacements. Such fragilities indicate the
fact that in case of RC high-rise buildings, distant earthquakes indicate the high seismic

123

123

0.1192

0.2499

0.4555

0.7296

DS2

DS3

DS4

1.4350

0.8919

0.5081

0.2427

0.3936

0.5039

0.5250

0.5509

0.5470

0.5941

0.5426

0.4676

M = 5.16.2

M = 6.37

M = 6.37

M = 5.16.2

Standard deviation r

Mean l (arithmetic space) PGA (g)

DS1

Damage states

1.0815

0.7538

0.4225

0.2067

M = 6.37

2.4797

1.6157

0.8742

0.3874

M = 5.16.2

84th percentile PGA (g)

0.4922

0.2752

0.1478

0.0687

M = 6.37

0.8305

0.4924

0.2953

0.1521

M = 5.16.2

16th percentile PGA (g)

Table 6 Log-normal distribution function parameters l and r for fragility curves of the prototype RC high-rise buildings for different magnitude levels

Bull Earthquake Eng

Bull Earthquake Eng

vulnerability. Thus, for high-rise buildings it is necessary to provide an adequate seismic

analysis, even in a case when they are far away from the active fault in the zones of lower
seismic risk.
In the Fig. 15 fragility curves for lower level of magnitude M = [5.16.2] and two
different range of distance to source, R = [120 km] which contains 14 ground motions
and R = [2070 km] which contains 15 ground motions are shown. It is noticeable, from
the Fig. 15, the higher fragility of certain damage states, in case of distant earthquakes
compared to closer earthquakes. All this indicates the high seismic vulnerability of RC
high-rise buildings to the impact of distant earthquakes.

6.2.3 Comparison of fragility curves with respect to site soil condition

The Fig. 16 presents the comparison of fragility curves for two soil types, the rock and stiff
soil (soil types A and B according to Eurocode 8-1). The diagram describes a bit higher
fragility for the case of foundation of high-rise buildings on stiff soil. The differences are in
the derived parameter of fragility curves l (Table 8), where the median of PGA are smaller
in the case of foundation on stiff soil than on rock. It happens because ground motions at
stiff soil sites causes larger responses for longer period modes of structure, as in case of RC
high-rise buildings.
It can be concluded that seismic vulnerability is higher in the case of foundation of RC
high-rise buildings on the stiff soil (soil type A) than on the rock (soil type B).

6.3 Comparison of derived fragility curves with existing fragility curves

There is a lack of information in the literature regarding derived fragility curves for RC
high-rise buildings. The largest database of fragility curves in America is HAZUS database

Fig. 14 Fragility curves of the prototype RC high-rise buildings for large magnitude level M = [6.37.0]
and different distances to source

123

123

0.1206

0.2452

0.4340

0.7197

DS2

DS3

DS4

0.7403

0.4800

0.2567

0.1170

0.3672

0.4377

0.4339

0.4641

0.4199

0.5602

0.5982

0.6431

R = 2070

R = 120

R = 120

R = 2070

Standard deviation r

Mean l (arithmetic space) PGA (g)

DS1

Damage states

1.0390

0.6724

0.3784

0.1919

R = 120

1.1266

0.8405

0.4669

0.2225

R = 2070

84th percentile PGA (g)

0.4985

0.2802

0.1589

0.0758

R = 120

0.4865

0.2741

0.1411

0.0615

R = 2070

16th percentile PGA (g)

Table 7 Log-normal distribution function parameters l and r for fragility curves of the prototype RC high-rise buildings for large magnitude level M = [6.37.0] and
different distances to source

Bull Earthquake Eng

Bull Earthquake Eng

Fig. 15 Fragility curves of the prototype RC high-rise buildings for small magnitude level M = [5.16.2]
and different distances to source

Fig. 16 Fragility curves of the prototype RC high-rise buildings for selected ground motions recorded at
soil type A and soil type B

(National Institute of Building Sciences 1999). HAZUS represents the database of expert
fragility curves and includes wide range of various structural systems. This database
includes fragility curves for structural system with ductile walls for the three types of

123

Bull Earthquake Eng

height range (low rise, medium rise, and high rise) where high-rise building type is defined
with number of stories [8 floors with typical value of 12 floors.
In the Fig. 17 HAZUS fragility curves with derived fragility curves for the prototype
RC high-rise buildings are compared. It can be noticed that fragility curves do not vary
significantly in terms of damage states (DS1-slight and DS2-moderate) where HAZUS
fragility curves are above the fragility curves for the prototype buildings, because the interstorey drifts at the thresholds of these damage states for HAZUS fragility curves are less
and thus, the fragility is increased. For the damage state DS3 which corresponds to HAZUS
extensive damage state fragility curves are also close, where HAZUS curve is below
derived curve due to the fact that inter-storey drift for HAZUS curve is bigger and the
fragility is decreased as well. Fragility curve for HAZUS collapse damage state, with interstorey drift in the amount of 4 %, is noticeably lower than fragility curve of DS4 damage
state for the prototype buildings, where the inter-storey drift is 1.64 %. Such results
indicate the high degree of matching of derived fragility curves for the prototype buildings
with HAZUS fragility curves. The comparison of HAZUS parameters with derived
parameters of fragility curves l and r is done in the Table 9.
It is interesting to compare these two types of fragility curves assuming the same interstorey drifts at the threshold of defined damage states which correspond to HAZUS drifts
(Fig. 18). It may be concluded from the Fig. 18, the great level of matching of fragility
curves for each damage state. The matching is also noticeable in the median values of PGA
(Table 10). Considering that derived fragility curves are analytically obtained, and HAZUS
curves are expert fragility curves, such results indicate the high level of matching and
accuracy. In literature there are empirical fragility curves as well, for the region of Europe
and RC structures done by Rossetto and Elnashai (2003). Rossetto and Elnashai (2003)
pointed out that their fragility curves are not applicable for the structural systems with
ductile walls, due to the lack of observational data from the site for this type of structural
systems.

6.4 Comparison of fragility curves with respect to number of stories

The derived fragility curves, separately for each considered prototype RC high-rise
building (20-story, 30-story and 40-story), are shown in the Fig. 19. From the Fig. 19 it can
be easily noticed that the fragility increases with buildings height increase, but with no
significant differences in fragility for certain heights of the building. Since three considered
prototype buildings represent the same type of height range (high-rise) and are designed
according to the same rules defined in Eurocode provisions, similar levels of fragility are
Table 8 Log-normal distribution function parameters l and r for fragility curves of the prototype RC highrise buildings for selected ground motions recorded at soil type A and soil type B
Mean l (arithmetic
space) PGA (g)

Standard deviation r

84th percentile PGA

(g)

16th percentile PGA

(g)

Soil type
A

Soil type
B

Soil type
A

Soil type
B

Soil type
A

Soil type
B

Soil type
A

Soil type
B

DS1

0.1697

0.1679

0.7473

0.5463

0.3583

0.2899

0.0804

0.0972

DS2

0.3688

0.3423

0.7660

0.6013

0.7934

0.6245

0.1715

0.1876

DS3

0.6536

0.6025

0.7050

0.6229

1.3228

1.1233

0.3229

0.3232

DS4

0.9438

0.9578

0.4601

0.5525

1.4952

1.6642

0.5958

0.5512

Damage
states

123

Bull Earthquake Eng

Fig. 17 Comparison of HAZUS fragility curves with derived fragility curves for the prototype RC high-rise
buildings

expected and suggest clearly the possibility of using fragility curves obtained on integrated
sample of three prototype high-rise buildings.

7 Conclusions
In this study, probabilistic fragility curves of RC high-rise buildings with RC core wall
structural system were obtained for seismic excitation typical for Southern EuroMediterranean zone. The fragility curves and parameters of fitted log-normal cumulative
distribution functions for four damage states were derived by conducting 3600 nonlinear
time-history analyses on the basis of 60 ground motions with wide range of magnitudes,
distances to source and different site conditions. In the process of obtaining fragility curves
it was performed detailed probabilistic seismic damage analysis of the prototype RC highrise buildings and as a result of it, limit states and corresponding damage states for RC

Table 9 Comparison of HAZUS and derived log-normal distribution function parameters for fragility
curves
Damage
states

Mean l (arithmetic space)

PGA (g) derived

Mean l (arithmetic space)

PGA (g) HAZUS

Standard
deviation r
derived

Standard
deviation r
HAZUS

DS1

0.1683

0.1200

0.6304

0.6400

DS2

0.3519

0.2900

0.6682

0.6400

DS3

0.6220

0.8200

0.6554

0.6400

DS4

0.9520

1.8700

0.5133

0.6400

123

Bull Earthquake Eng

Fig. 18 Comparison of HAZUS fragility curves with derived fragility curves for the prototype RC high-rise
buildings assuming the same inter-storey drifts at the threshold of damage states which correspond to
HAZUS drifts

Table 10 Comparison of HAZUS and derived log-normal distribution function parameters for fragility
curves assuming the same inter-storey drifts at the threshold of damage states which correspond to HAZUS
drifts
Damage
states

Mean l (arithmetic space)

PGA (g) derived

Mean l (arithmetic space)

PGA (g) HAZUS

Standard
deviation r
derived

Standard
deviation r
HAZUS

DS1

0.1290

0.1200

0.6648

0.6400

DS2

0.3290

0.2900

0.6686

0.6400

DS3

0.8960

0.8200

0.5545

0.6400

DS4

1.5080

1.8700

0.2908

0.6400

high-rise buildings were defined, where damage states were treated as random variables.
The inter-storey drifts at threshold of each defined damage state were defined as random
variables with the range of possible values. Calculated confidence intervals for all interstorey drifts of certain damage states, with relative width less than 5 %, indicate the high
level of accuracy of derived inter-storey drifts. In this way, per three types of fragility
curves (the fragility curves obtained by mean values, by 16th percentile and 18th percentile
of inter-storey drifts at threshold of certain damage states) were derived for four damage
states of the prototype RC high-rise buildings. Calculated relative width of confidence
interval (lower than 10 %) for the derived log-normal cumulative distribution function
parameters l of fragility curves indicate high accuracy of derived fragility curves and their
possible implementation for RC high-rise buildings with RC core wall or similar structural
systems.

123

Bull Earthquake Eng

Fig. 19 Comparison of fragility curves of prototype RC high-rise buildings with respect to number of
stories

The derived fragility curves were compared to each other in order to make difference
regarding various characteristics of earthquakes. The effects of magnitude, distance to
source and site conditions, on fragility of RC high-rise buildings, were analysed. The
fragility of RC high-rise buildings with large magnitude is larger than fragility with smaller
magnitude for each damage state and same PGA values. Magnitude has significant influence on seismic vulnerability assessment of RC high-rise buildings and can be connected
with damage states of RC high-rise buildings. The fragilities of RC high-rise buildings for
smaller and larger distance to source are nearly equal. It is evident high fragility of RC
high-rise buildings to the impact of distant earthquakes. Accordingly, high-rise buildings
require an adequate seismic analysis, even in the case when they are located far away from
active faults in the zones of lower seismic risk. The fragility of RC high-rise buildings is
larger in case of foundation of high-rise buildings on stiff soil (soil type B) than on the rock
(soil type A according to Eurocode 8). Derived analytical fragility curves are mostly
matched with expert HAZUS curves.
Since no probabilistic fragility relationships exist for this class of building and for this
seismic zone, the aim is to incorporate derived fragility curves into existing European
database of fragility curves.
Acknowledgments The authors sincerely thank Seismological Institute of Montenegro for sharing data
from its ground motions database.

References
Ambraseys N, Smit P, Sigbjornsson R, Suhadolc P, Margaris B (2002) Internet-site for European strongmotion data. European Commission, Directorate-General XII, Environmental and Climate Programme,
Brussels. http://www.isesd.cv.ic.ac.uk

123

Bull Earthquake Eng

CEN EC8 (2004) Eurocode 8-1design of structures for earthquake resistance. Part 1: general rules,
seismic actions and rules for buildings. European standard EN 1998-1, December 2004, European
Committee for Standardization, Brussels
CEN EC2 (2004) Eurocode 2design of concrete structures. Part 1: general rules and rules for buildings.
European standard EN 1992-1-1, December 2004, European Committee for Standardization, Brussels
CEN EC8 (2005) Eurocode 8-2design of structures for earthquake resistance. Part 2: bridges Annex E
probable material properties and plastic hinge deformation capacities for non-linear analyses. European
standard prEN 1998-2, November 2004, European Committee for Standardization, Brussels
CSI (2006) PERFORM 3D nonlinear analysis and performance assessment for 3D structures. Computers &
Structures Inc., Berkeley
CSI (2007) PERFORM 3D detailed example of a tall shear wall building by Dr. Graham H. Powell.
Nonlinear modeling, analysis and performance assessment for earthquake loads. Computers &
Structures Inc., Berkeley
CSI (2013) ETABS 2013 integrated analysis, design and drafting of buildings systems. Computers &
Structures Inc., Berkeley
Erberik M, Elnashai A (2004) Fragility analysis of flat-slab structures. Eng Struct 26(2004):937948. doi:10.
1016/j.engstruct.2004.02.012
FEMA 356 (2000) Prestandard and commentary for the seismic rehabilitation of buildings. Federal
Emergency Management Agency, Washington
Ghobarah A (2004) On drift limits associated with different damage levels. In: Proceedings of an international workshop on performance-based seismic design concepts and implementation, Bled Slovenia,
pp 321332
Giardini D, Woessner J, Danciu L, Crowley H, Cotton F, Grunthal G, Pinho R, Valensise G and the SHARE
consortium (2013) SHARE European seismic hazard map. European Commission. www.share-eu.org
Ji J et al. (2007) Seismic fragility assessment for reinforced concrete high-rise buildings. Report 07-14, MidAmerica Earthquake Center, University of Illinois at Urbana-Champaign
Ji J, Elnashai AS, Kuchma DA (2007) An analytical framework for seismic fragility analysis of RC high-rise
buildings. Eng Struct 29(2007):31973209. doi:10.1016/j.engstruct.2007.08.026
Ji J, Elnashai AS, Kuchma DA (2009) Seismic fragility relationships for reinforced concrete high-rise
buildings. Struct Des Tall Spec Build 18:259277. doi:10.1002/tal.408
MATLAB (2013) MATLABthe language of technical computing, version R2013b 8.2.0.701. MathWorks
Inc., Natick
National Institute of Building Sciences (1999) HAZUS technical manual, prepared for Federal Emergency
Management Agency, Washington, DC
Park YJ, Ang AHS (1985) Mechanistic seismic damage model for reinforced concrete. J Struct Eng
111(4):722739
PEER (2010) Technical report for the peer ground motion database web application. Pacific Earthquake
Engineering Research Center, University of California, Berkeley
Rossetto T, Elnashai A (2003) Derivation of vulnerability functions for European-type RC structures based
on observational data. Eng Struct 25(2003):12411263. doi:10.1016/S0141-0296(03)00060-9
Taranath BS (2010) Reinforced concrete design of tall buildings. International Code Council, Concrete
Reinforcing Steel Institute, CRC Press Taylor & Francis Group, Boca Raton
The Council on Tall Buildings and Urban Habitat (2011) The tallest 20 in 2020: entering the era of the
Megatall. Illinois Institute of Technology, Chicago. http://www.ctbuh.org/TallBuildings/
HeightStatistics/BuildingsinNumbers/TheTallest20in2020/tabid/2926/language/en-US/Default.aspx
Wallace JW (2012) Behavior, design, and modeling of structural walls and coupling beamslessons from
recent laboratory tests and earthquakes. Int J Concr Struct Mater 6(1):318. doi:10.1007/s40069-0120001-4

123