THE STRUCTURAL DESIGN OF TALL AND SPECIAL BUILDINGS

Struct. Design Tall Spec. Build. (2010)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/tal.642

Analytically derived fragility relationships for the modern

high-rise buildings in the UAE

Aman Mwafy*,†
Associate Professor, Department of Civil and Environmental Engineering, United Arab Emirates University, P.O. Box
17555, Al-Ain, UAE

SUMMARY
Investigating and planning for the expected damage that may hit the earthquake-prone areas in the UAE
should be undertaken in order to predict and mitigate earthquake losses. This paper discusses a framework
for developing an essential driving engine in loss estimation systems, namely fragility relationships. Six
reference structures, varying in height from 10 to 60 storeys, are selected due to the concentrated economic
and human assets in this class of buildings. The reference structures are designed according to the building
codes and construction practice adopted in this region. Inelastic fibre-based simulation models are devel-
oped for the buildings using a verified analysis platform, which enables monitoring the spread of yielding
and cracking during the multi-step cyclic analysis. The ground motion uncertainty is accounted for using
20 input ground motions conforming to the latest understanding of the seismo-tectonic characteristics of
the UAE. A large number of inelastic pushover and incremental dynamic collapse analyses are deployed
for the reference structures to derive the fragility relationships. The study illustrates the significance of
assessing the vulnerability of a population of high-rise buildings under the effect of various seismic sce-
narios and the need for expanding this study to cover other classes of structures in this region. Copyright
© 2010 John Wiley & Sons, Ltd.

1. INTRODUCTION

The building and infrastructure inventories in the UAE have expanded at an outstanding timescale to
meet the needs of a rapidly increasing population. The high-rise building stock is the most significant
in the potential consequences from natural hazard events since it represents high levels of financial
investments and population densities. Assessment of the vulnerability of this class of structures is
significant for predicting and mitigating earthquake losses. The UAE seismicity is characterized by
earthquakes originated from different seismic sources, namely long-distance events from Southern
Iran along with earthquakes from local seismic faults (Mwafy et al., 2006). Although no significant
human and monetary losses were reported from recent earthquakes, the repeated reports of seismic
activities in the UAE reflect the possibility of stronger earthquakes striking this area in the future.
While modern technologies have been used in the design and construction of high-rise buildings,
reliable vulnerability studies that account for the latest assessment methodologies alongside local
seismicity and construction practice are urgently need. Vulnerability studies involve the assessment
of the seismic performance of a population of structures representing the inventory. The outcomes
are fragility relationships, which describe the probability of various damage levels due to earthquake
scenarios. Impact of different seismic events on the building inventory can be evaluated from
the vulnerability functions and translated into monetary and human losses through an earthquake
loss estimation system. These studies are not only important for predicting the impact of different
earthquake scenarios, but also an effective mean for mitigating seismic risk in future. The latter objec-
tive can be achieved through the seismic assessment and retrofit of existing buildings as well as the
calibration of the seismic design provisions of new structures (Calvi et al., 2006). The anticipated

* Correspondence to: Aman Mwafy, Department of Civil and Environmental Engineering, United Arab Emirates
University, PO Box 17555, Al-Ain, United Arab Emirates
†
E-mail: [email protected]

Copyright © 2010 John Wiley & Sons, Ltd.

 A.M. MWAFY

losses can be compared with the cost of improving the seismic resistance through retrofit schemes
or increasing seismic design loads, which provide authorities with a tool to compare between the
cost and risk.
The derivation of vulnerability relationships for a wide range of high-rise buildings in the UAE
through Incremental Dynamic Collapse Analyses (IDCAs, e.g., Mwafy and Elnashai, 2001) is dis-
cussed in this paper. The results of designing six reference structures according to the building codes,
material characteristics and construction practice adopted in the study region are presented. The design
information is used to idealize the buildings for multi-degree-of-freedom inelastic simulations under
the effect of 20 input ground motions with increasing severity. The adopted approaches for selecting
realistic limit state criteria and input ground motions for deriving the fragility curves are also dis-
cussed. The measured seismic response is finally related to ground motion intensity through a reliable
statistical model, which enables the derivation of fragility relationships for high-rise buildings in the
UAE. It is noteworthy that a comprehensive regional earthquake assessment and mitigation strategy
requires an interdisciplinary framework that includes the hazard definition, physical damage, and
social and economic consequences. Physical damage should be assessed for the building inventory,
critical facilities, transportation networks and lifeline systems. The focus of the present study is only
on the physical damage of high-rise buildings, which enables the development of an essential com-
ponent in a broad loss estimation system for the UAE and stimulates interdisciplinary research activi-
ties in future.

2. FRAMEWORK OF LOSS ESTIMATION SYSTEMS

The pressing need of comprehensive strategy for regional loss estimation and mitigation as well as
response and recovery plans after earthquake disasters has encouraged researchers to develop meth-
odologies and tools for conducting vulnerability analysis. The driving engines of the earthquake loss
estimation and regional seismic risk systems are (e.g., Kwon and Elnashai, 2006): (a) seismic hazard;
(b) inventory of the exposed systems (e.g., transportation network) or elements (e.g., a building); and
(c) vulnerability relationships. These three components can be integrated using a powerful platform
to enable bridging the gap between researchers, practitioners and policy-makers. A framework for
earthquake loss estimation in the UAE is depicted in Figure 1.
A number of recent studies have addressed the seismic hazard of the UAE (e.g., Abdalla and Al-
Homoud, 2004, Aldama-Bustos et al., 2009, Mwafy et al., 2006). The distribution of earthquakes in
the UAE for the period 734 to 2004 is shown in Figure 1(a) (Mwafy et al., 2006). Natural and syn-
thetically generated input ground motions can be selected based on the abovementioned studies to
represent the UAE seismicity, such as those shown in Figure 1(c). The abovementioned studies indi-
cated that the earthquake hazard in the UAE is dominated by the seismicity of Southern Iran alongside
inland earthquakes originated from local seismic faults. It is worth noting that the seismic events
occurred within or near to the UAE over the period 1924–1999 are 49 (0·65 events per year), only
three of which are located inland within the Arabian Peninsula. Considering the time span from
2000–2006, 18 events were recorded (3·0 events per year), six of which are inland and two just off-
shore (Aldama-Bustos et al., 2009). This apparent increase in the rate of seismic events confirms the
need for the reliable representation of different seismic scenarios to arrive at an accurate estimate of
earthquake losses in the UAE.
The description of the built environment at risk is also a fundamental input in loss assessment
systems. In addition to the building stock, critical and emergency response facilities such as hospitals,
fire stations and schools are significant and should be assessed. Collecting a reliable inventory database
for a large area is one of the major challenges in loss assessment studies due to the lack of detailed
surveys and the rapid changes in the exposed inventory (Cagnan et al., 2008, Elnashai et al., 2008a).
Accordingly, official local surveys, high-resolution satellite images and supplementary field surveys
were utilized in previous studies to collect data regarding the distribution of the building inventory
(e.g. Moharram et al., 2008), as shown in Figure 1(b). The present study only focuses on the high-rise
buildings in the UAE since they are the most significant in the potential consequences from natural
hazard events.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

1
Minor
Moderate
Probability of Exceedance
0.8

Severe Dubai Dubai

0.6 IO
LS UAE
UAE
0.4 CP
Collapse

0.2

Peak Ground Acceleration

0 (a) Seismic Hazard*
0 0.5 1 1.5 2

(d) Vulnerability*

(e) Rehabilitation & (f) Loss Estimation

Re-assessment Platform

Satellite image
for Dubai

(b) Building Inventory

(c) Selection, Design,

Modeling, and Inelastic
Dynamic Simulations*

*: Components considered in the present study

Figure 1. Framework for earthquake loss estimation in the UAE.

Vulnerability relationships relate the probability of exceeding multiple damage states to a ground
motion intensity, as shown in Figure 1(d). These curves account for the uncertainty and variability
associated with capacity and demand. Fragility curves are, therefore, significant for the estimation of
monetary loss and seismic retrofit decisions. Four possible approaches of damage information can be
used to derive fragility curves: observational, experimental, analytical and hybrid techniques (Calvi
et al., 2006). Although the observational approach is the most realistic since the entire inventory is
taken into consideration with site characteristics, it is practically difficult to collect an observation-
based data for some regions such as the UAE. The experimental option is also costly and time-
consuming since the testing of a wide class of structures is required. Generating damage data through
extensive analytical simulations is the most realistic and cost-effective option, and therefore is
adopted in the current study. Several techniques for deriving vulnerability relationships based on
analytically simulated structural damage statistics have been adopted by researchers, with a variety
in analysis methods, structural idealizations, and seismic hazard and damage models. Most of these
techniques are computationally demanding since a large number of analyses are required to fully
represent the ground motion and structural uncertainties (e.g., Ji et al., 2007, Kwon and Elnashai,
2006). This is particularly true when adopting inelastic multi-degree-of-freedom dynamic simulation
for deriving the vulnerability relationships, which is the most reliable technique, and therefore
is adopted in the present study. An excellent review covering various methods for vulnerability
assessment along with the key advantages and disadvantages of these procedures is presented by
Calvi et al. (2006).

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

Finally, different technologies are available for visualizing the seismic risk. The Federal Emergency
Management Agency’s HAZUS software can be used to integrate the loss assessment components
and provide the required visualization (e.g., Kircher et al., 2006). The Mid-America Earthquake
Center’s MAEviz system is also an open source environment that provides capabilities to develop
risk reduction strategies and implement mitigation plans to minimize the impact of earthquakes
(Elnashai et al., 2008b). Several research projects have been undertaken in different parts of the world
to satisfy the needs of loss assessment process (e.g., Cagnan et al., 2008, Elnashai et al., 2008a).
However, very few studies have been directed to the built environment in the Middle East (e.g.,
Moharram et al., 2008). This brief review reflects the pressing need for developing realistic fragility
relationships for the UAE, not the importation of irrelevant curves from other regions. Vulnerability
functions can be directly integrated with seismic hazard maps and inventory data using earthquake
loss estimation platforms to provide a tool for formulating risk reduction policies in this region.
Six components needed for deriving fragility relationships are discussed in the following section.
These are: (a) selection and design of reference structures; (b) analytical modelling and analysis
procedure; (c) uncertainty modelling; (d) input ground motions; (e) fragility relationship and scaling
approach; and (f) performance criteria.

3. INGREDIENTS OF THE VULNERABILITY STUDY

3.1. Selection and design of reference structures

RC high-rise buildings with shear walls and flat slabs are widespread in the UAE. These modern
high-rise buildings are designed to resist seismic forces according to the Uniform Building (ICBO,
1997) or the International Building (ICC, 2009) codes with a design peak ground acceleration (PGA)
of 0·15 g for 10% probability of exceedance in 50 years. Six RC buildings, ranging from 10 to 60
storeys, are selected to represent a wide range of high-rise buildings in the UAE. The reference struc-
tures have the same layout shown in Figure 2(b). Each building consists of two basements, a ground
storey, and a number of typical storeys. The height of all storeys is 3·2 m except for the ground storey,
which is 4·5 m. The total heights of the six buildings are, therefore, 33·3, 65·3, 97·3, 129·3, 161·3
and 193·3 m, respectively. The buildings are designed and detailed for the purpose of the present
study. The permanent loads used in design include the self-weight of structural members in addition
to a superimposed dead load of 4·0 kN/m2, while the live load is 2.0 kN/m2 except for stairs and exit
ways, which is 4·8 kN/m2.
Wind loads are estimated using ASCE/SEI 7–05 (ASCE, 2006a) based on a basic wind speed of
45 m/s and an exposure category C. Seismic design loads are also calculated using the ASCE/SEI
7-05 (ASCE, 2006a) code. Equivalent mapped spectral values are estimated for the UAE by matching
the ASCE/SEI 7-05 (ASCE, 2006a) elastic design spectrum with the UBC (ICBO, 1997) one. The
seismic zone 2A was recommended in a number of hazard assessment studies for Dubai, and hence
is used with a site class SD (stiff soil profile) to arrive at the UBC spectrum (Abdalla and Al-Homoud,
2004, Mwafy et al., 2006). Both the equivalent lateral force procedure and modal response spectrum
analysis are used in design, as required by ASCE/SEI 7-05. Detailed three-dimensional (3D) analytical
models are developed for the reference structures using the structural analysis and design program
ETABS (CSI, 2008). The six buildings are proportioned and detailed according to various load com-
binations and design provisions recommended by the ACI building code (ACI, 2005). Yield strength
of reinforcing steel is 460 MPa. The concrete strength used for floor slabs and beams is 45 MPa,
while it varies along the height of shear walls from 45 to 60 MPa. The cross-sections of walls and
the corresponding reinforcement also vary along the building height. The floor slab systems comprise
of a 0·28 m cast-in-place flat slabs with a beam at the perimeter. Figure 1(c) shows the 3D models
used in the design of the buildings, while Table 1 summarizes the concrete strength (fc′) and sizes of
vertical structural members used in design.
An important issue in the design of multi-storey buildings with flat slabs is the adequate design of
the floor system to resist the seismic actions and deformations. The slab thickness and reinforcement
are selected in the present study to prevent undesirable modes of failure, particularly punching shear

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

RC flexural wall
cross section

Gauss sections

Concrete and
Concrete and
RC hollow core steel fibers
layers
cross section

0.21 L 0.58 L 0.21 L

L
RC beam
section (a) ZEUS-NL fiber-
based models
Rigid arm

(b) Layout of the

reference structures

Figure 2. Modelling approach of reference structures for inelastic analysis: (a) ZEUS-NL fibre-
based models of the six reference structures in the longitudinal and transverse directions; and
(b) Layout of the buildings showing different lateral force-resisting systems in the longitudinal
and transverse directions.

failure. The bottom reinforcement of the slab is continuous with an adequate amount passing through
the walls to prevent any progressive collapse in the event of a local punching failure. The effect of
considering the seismic loads in the design of floor slabs is also investigated during the design process.
This indicates that the differences in slab reinforcement exceed 100%, particularly at the connection
with shear walls, compared with the case without considering seismic loads. The differences at the
mid-spans are marginal. It is important to note that the slabs of the reference structures are supported
on shear walls with large cross sections, as indicated in Table 1, which significantly improves the
slab shear resistance compared with other systems supported on columns. For the sake of brevity,
additional design information of the reference structures, including detailed structural drawings, are
presented elsewhere (Mwafy, 2009).

3.2. Analytical modelling and analysis procedure

The Inelastic Pushover Analyses (IPAs) and IDCAs of the reference structures are performed using
ZEUS-NL (Elnashai et al., 2009), which is a modern inelastic analysis platform utilizing the fibre

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 Table 1. Periods of vibration in the transverse direction (s), concrete strength (fc′) and sizes of vertical structural members of the reference structures.

Storey
Periods
Building (1st, 2nd, 3rd) Member Characteristics 2nd Bas.-3 4–8 9–18 19–28 29–38 39–48 49–58
10SB 0·63, 0·12, 0·08 SW Cross section 300 × 1600 300 × 1600
Concrete strength 36 36
CORE Thickness 250 200
Concrete strength 36 36

Copyright © 2010 John Wiley & Sons, Ltd.

20SB 1·55, 0·35, 0·14 SW Cross section 350 × 3000 350 × 3000 300 × 3000
Concrete strength 36 36 36
CORE Thickness 300 250 200
Concrete strength 36 36 36
30SB 2·56, 0·65, 0·28 SW Cross section 350 × 4000 350 × 4000 300 × 4000 250 × 4000
Concrete strength 40 40 36 36
CORE Thickness 300 300 250 200
Concrete strength 40 40 40 36
A.M. MWAFY

40SB 3·69, 0·98, 0·44 SW Cross section 400 × 5000 400 × 5000 350 × 5000 300 × 5000 250 × 5000
Concrete strength 48 48 40 36 36
CORE Thickness 350 350 300 250 200
Concrete strength 48 48 40 40 36
50SB 5·10, 1·38, 0·63 SW Cross section 450 × 5000 450 × 5000 400 × 5000 350 × 5000 300 × 5000 250 × 5000
Concrete strength 48 48 40 40 36 36
CORE Thickness 400 400 350 300 250 200
Concrete strength 48 48 40 40 36 36
60SB 6·06, 1·68, 0·77 SW Cross section 500 × 5000 500 × 5000 450 × 5000 400 × 5000 350 × 5000 300 × 5000 250 × 5000
Concrete strength 56 56 48 40 40 36 36
CORE Thickness 450 450 400 350 300 250 200
Concrete strength 56 56 48 40 40 36 36
Concrete strength is in MPa and dimensions are in mm.

DOI: 10.1002/tal
Struct. Design Tall Spec. Build. (2010)
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

modelling approach. The program was developed at Imperial College London and at University of
Illinois at Urbana-Champaign, and has been extensively verified on the member and structure levels
against tests carried out in Europe and the USA (e.g., Jeong and Elnashai, 2005; Kwon and Elnashai,
2006). The idealization adopted in the current study effectively models rebars, unconfined and con-
fined concrete. This approach enables monitoring the stress-strain response of each frame element
over two Gauss sections through the integration of the nonlinear stress-strain response of different
fibres in which the section is subdivided, as shown in Figure 2(a). As explained in the subsequent
section, this modelling approach reduces the modelling uncertainty since assumptions such as the
moment–curvature relationships required by other analysis platforms are avoided. Three cubic elasto-
plastic frame elements capable of representing the spread of yielding and cracking are used to model
each structural member. These enable modelling three different arrangements of reinforcing steel
along the length of each structural member as specified in design (at the two edges and at the mid-
span). Two rigid arms are also utilized to connect the slab/beam ends with shear walls (the length
between the centreline and the face of the vertical element idealizing the wall, as shown in Figure
2(a). Several cross-sections are used from the ZEUS-NL library to model slabs, connecting beams,
shear walls, cores and rigid arms.
The concrete response is represented using a uniaxial constant confinement concrete model,
while a bilinear elasto-plastic model is selected to model reinforcing steel. Actual (mean) rather
than nominal (characteristic) material strengths are employed for assessment of the reference struc-
tures using the ZEUS-NL models. A normal distribution is used in this study for both the unconfined
concrete compressive strength and for steel yield strength, with the characteristic strength being the
lower 5 percentile value of the distribution. The adopted coefficient of variation is 12% for concrete
(assuming average curing conditions and workmanship), and 6% for steel (e.g., Rossetto and Elnashai,
2005).
Although ZEUS-NL is capable of performing 3D inelastic response history analysis of high-rise
structures, such modelling and analysis are computationally demanding particularly with the wide
range of buildings and input ground motions employed in the present study. A two-dimensional ide-
alization is therefore adopted to develop the fragility relationships of the reference structures using
IDCAs. It is assumed that four lateral-force-resisting systems are in the transverse directions of each
building. Each of these framing systems, which is loaded with 25% of the total mass of the building,
consists of two external shear walls (SW1) and an internal core, as shown in Figure 2. The two framing
systems at the left and right margins are assumed to resist gravity loads only. It is also clear from
Figure 2 that a single framing system, which consists of all internal cores and the shear walls SW2,
resists the lateral forces in the longitudinal direction. Other walls are conservatively assumed not
participating in resisting the lateral forces in the longitudinal direction. Therefore, in total, 12 framing
systems are modelled using ZEUS-NL to represent the lateral-force-resisting systems of the six refer-
ence structures (one model for each building in the longitudinal and transverse directions). Figure
2(a) depicts the ZEUS-NL fibre-based models used to idealize the buildings in the longitudinal and
transverse directions of the investigated buildings for inelastic analysis. Table 1 also shows the periods
of vibration of the reference structures in the transverse direction.
The damping issue was investigated before executing IDCAs. Hysteretic damping, which is respon-
sible for the dissipation of the majority of energy introduced by the earthquake action, is accounted
for in the fibre formulation of the inelastic frame elements. A relatively small quantity of non-hys-
teretic type of damping is also added to the inelastic model through stiffness-proportional damping
(e.g., Priestley and Grant, 2005). In addition to the hysteretic damping due to inelastic energy absorp-
tion, a 0·75% stiffness-proportional damping is used based on the information collected from the
literature and comparisons between the response of the six buildings at different damping levels (e.g.,
Willford et al., 2008). Gravity loads are applied as point loads at horizontal frame element nodes,
while the mass is represented by lumped mass elements and distributed in the same pattern adopted
for gravity loads.
The accurate prediction of seismic demand requires an efficient analysis procedure that accounts
for structural ductility and input ground motion variability. The accuracy of many analysis procedures
significantly decrease with increasing the peculiarity of input ground motions. This is particularly
important for long period and irregular structures. Higher mode effects have significant impact on the

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

response of high-rise buildings, and hence static analysis procedures may not adequately predict the
seismic response of such structures. The use of inelastic response history analysis, which accounts
for both the demand and capacity, has therefore several advantages over other alternatives. This
analysis method, which is the most suitable for seismic vulnerability assessment of high-rise buildings
is, therefore, used in the present study.

3.3. Uncertainty modelling

Vulnerability analysis is probabilistic since its capacity and demand ingredients are uncertain. The
most important uncertainties in vulnerability studies are: (a) structural system; (b) structural charac-
teristics (e.g., height and period); (c) analytical modelling; (d) analysis method; (e) material properties;
(f) performance criteria; and (g) input ground motions. Some of the uncertainties are inherently
random (aleatory) while others are due to the lack of knowledge (epistemic), as discussed in Wen
et al. (2004). In order to account for the abovementioned uncertainties, the typical approach is to
conduct Monte Carlo simulations. These require large number of inelastic response history analyses,
which is very demanding and expensive particularly when deriving fragility relationships for large
structures such as high-rise buildings. It is, therefore, more practical to focus on the dominant factors
that control the probabilistic response, while estimating the impact of other uncertainties based on the
conclusions of previous studies.
Several studies concluded that material properties have little impact on the structural response,
particularly at high ground motion intensities, compared with the variability of ground motions (Kwon
and Elnashai, 2006). Researchers also investigated the sensitivity of structures to major variables and
concluded that uncertainties in ground motions is more significant than those in structural properties
(e.g. Kinali and Ellingwood, 2007, Porter et al., 2002). In the present study, only uncertainties in
seismic hazard and structural characteristics (building height or period) are accounted for. Material
properties are considered deterministic, and hence are set equal to their mean values. It is noteworthy
that designing and modelling a wide range of high-rise buildings with different characteristics using
the fibre-modelling approach are challenging. Adopting a rigorous approach for deriving fragility
curves using IDCAs and a broad range of ground motions representing different seismic scenarios
increases the computational effort. Consequently, uncertainties related to analytical modelling and
analysis method are also assumed deterministic. It is important to note that the modelling approach
(fibre-based) and analysis method (inelastic dynamic collapse analysis) adopted in the present study
are the most suitable for deriving vulnerability curves, and significantly contribute in reducing uncer-
tainty compared with other alternatives.

3.4. Input ground motions

Two sets of natural and synthetically generated earthquake records are selected for deriving the fragil-
ity curves of the reference structures, as shown in Table 2. The selected records represent two seismic
scenarios: (a) severe distant earthquakes of magnitude 7·4 with 100 km epicentral distance; and (b)
moderate events of magnitude 6·0 and a distance to source of 10 km. These two seismic scenarios
were recommended in a previous seismic hazard assessment study for Dubai (Mwafy et al., 2006).
This study also recommended Uniform Hazard Spectra (UHS) and two sets of artificial time series
generated using available geophysical information and engineering models to represent the above-
mentioned two seismic scenarios. Ten synthetic accelerograms (five representing severe distant
earthquakes, R6-R10 and five for moderate close events, R16-R20) are selected from the abovemen-
tioned study. Due to the lack of available real records for the UAE, 10 natural accelerograms are
selected from the Pacific Earthquake Engineering Research Center (PEER, 2009) and the European
strong-motion (Ambraseys et al., 2004) databases. Five of the selected natural records (R1–R5) rep-
resent severe distant events, while moderate close earthquakes are represented by R11–R15. The
natural records are selected based on their distance to source, magnitude and spectral amplification
to fit the 10% probability of exceedance UHS.
The elastic response spectra of the 20 selected natural and artificial records are shown in Figure 3
at a PGA of 0·16 g. This intensity level represents the design PGA recommended for Dubai for a

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 Table 2. Characteristics of the selected short- and long-distance earthquakes.

Ref Earthquake Station Comp. Date Mag. (Mw) Site class Ep. Dist. (km) PGA (m/s/s) a/v g/ms−1 a/v class.

Copyright © 2010 John Wiley & Sons, Ltd.

Set 1 R1 Loma Prieta Emeryville 260 18-10-1989 6·93 C 96·5 2·450 0·573 low
R2 Manjil Tonekabun N132 20-06-1990 7·42 C 131 1·2233 0·76
R3 Bucharest Building res. Institute EW 04-03-1977 7·53 D 161 1·7271 0·60
R4 Chi-Chi CWB 99999 ILA013 EW 20-09-1999 7·62 C 135 1·3606 0·52
R5 Izmit Ambarli-Termik EW 17-08-1999 7·64 D 113 1·8008 0·60
R6-10 BEQ3–7 Artificially generated
Set 2 R11 Hollister City Hall 271 28-11-1974 5·14 C 9·8 1·651 1·48 high
R12 Umbria Ma. Castelnuovo-Assisi NE 26-09-1997 6·04 C 22 1·6003 1·25
R13 Lazio Abr. Cassino-Sant’ Elia EW 07-05-1984 5·93 C 16 1·1238 1·59
R14 Preveza OTE building-NS NS 10-03-1981 5·45 B 28 1·4069 1·60
R15 Basso Tirr. Naso NS 15-04-1978 6·10 B 18 1·4737 1·87
R16-20 SEQ1 and SEQ7–10 Artificially generated
FRAGILITY OF MODERN HIGH-RISE BUILDINGS

a/v: PGA/PGV, a/v classification (< 0·8 low & >1·2 high).

DOI: 10.1002/tal
Struct. Design Tall Spec. Build. (2010)
 A.M. MWAFY

Spec. Acc. (g)

Set 1 Set 2

Spectral Acc. (g)

0.8
Severe distant records Moderate close records
0.7
Median Median
0.6 +1 std. dev. spectrum
+1 std. dev. spectrum

0.5 10% prob. of exceedance 10% prob. of exceedance

uniform hazard spectrum uniform hazard spectrum
0.4

0.3

0.2

0.1
Period (s) Period (s)
0
0 1 2 3 4 0 1 2 3 4

Figure 3. Response spectra of 20 natural and synthetically-generated records representing severe

distant (Set 1) and moderate proximate (Set 2) events along with the median spectrum, spectrum
representing one standard deviation above the median, and uniform hazard spectrum for Dubai for
10% probability of exceedance in 50 years.

10% probability of exceedance in 50 years (Mwafy et al., 2006). The UHS for Dubai for 10% probabil-
ity of exceedance in 50 years is also shown in Figure 3. It is clear that the median spectra adequately
represent the two seismic scenarios recommended for Dubai. It is worth noting that the a/v ratio can
be used as a measure for the frequency content, magnitude and epicentral distance of the ground
motion (Tso et al., 1992). The low a/v ratios (<0·8 g/m s−1) signify earthquakes with low predominant
frequencies, broad response spectra, medium-to-high magnitude, long epicentral distance, long
duration and site period. High a/v (>1·2 g/m s−1) ratios represent high predominant frequencies,
narrow band spectra, small-to-medium magnitude, short epicentral distance, short duration and site
period. It is clear from Table 2 that the selected input ground motions follow the abovementioned
classification.
The ground motion duration has also an impact on the maximum seismic response (e.g., Bommer
and Martínez-Pereira, 1999). The significant duration is defined as the interval between the times
at which 5% and 95% of the total integral of the square of acceleration, velocity and displacement
is attained. This range may not be practical for structural analysis since the ground motion could
start at very large acceleration, which may apply unrealistic pulse to the structure. Kwon and
Elnashai (2006) therefore suggested to employ the interval between 0·5% and 95% of the integrals,
which is the approach adopted in the present study. This method significantly reduces the computa-
tional effort, particularly under the effect of Set 1 ground motions, which are characterized with long
duration.

3.5. Fragility relationship and scaling approach

Vulnerability relationships relate the ground motion parameters to the exceedance probabilities. The
input ground motions are scaled in the current study using their PGA, which is selected as the measure
of ground motion intensity, as explained below. The following expression is used for deriving the
fragility relationships (Wen et al., 2004):

(
P ( LS GMI ) = 1 − Φ λ CL − λ D GMI ( β2D GMI + βCL
2
+ β2M )) (1)

where P(LS|GMI) is the probability of exceeding a limit state given the ground motion intensity
(GMI); Φ is the standard normal cumulative distribution function; λCL = ln (median of drift capacity

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

for a particular limit state); λD|GMI = ln (calculated median demand drift given the ground motion
intensity from the fitted power law equation); βD|GMI demand uncertainty = ln (1 + s2 ) , where s2 is
the standard error of the demand drift data; βCL is drift capacity uncertainty; βM is modelling uncer-
tainty. Both βCL and βM are taken as 0·3 in the present study (Wen et al., 2004).
The choice of ground motion parameter is important for the accurate representation of the statistics
along the horizontal axis of the fragility chart. Several intensity measures were used in previous studies
such as PGA, Spectral Acceleration (SA) and Spectral Displacement (SD). Refined scaling approaches
were also proposed in previous studies (e.g., Mwafy and Elnashai, 2001). Nevertheless, it is difficult
to select a single scaling approach for structures with various dynamic characteristics. Moreover, since
the natural and artificially generated ground motions selected in the present study represent two spe-
cific seismic scenarios applicable to the study region, as explained below, normalizing these ground
motion based on their spectral intensities changes their characteristics. Scaling ground motions using
their PGAs relates the seismic forces directly to the input accelerations. This simple scaling method
agrees with the approach adopted by design codes, and hence it was used in several previous studies
and in the preset work (e.g., Ji et al., 2007, Kwon and Elnashai, 2006).
For the derivation of vulnerability relationships using IDCAs, the six analytical models of the refer-
ence structures are combined with the selected two input ground motion sets (refer to Table 2 and
Figure 3). Each input ground motion is scaled to different intensity (PGA) levels. The analysis is
carried out for all reference structures up to the satisfaction of the performance levels discussed
hereafter. It is clear from Figure 3 and Table 2 that each of the two ground motion sets selected in
the current study has different spectral amplification at the most prominent periods of vibration of the
reference structure (refer to Table 1). To effectively utilize the computational power, different scaling
factors are used in IDCA. Therefore, a scaling PGA of 0·08 g, which corresponds to half the design
ground motion suggested in the study of Mwafy et al. (2006), is selected for the Set 1 ground motions.
To improve the resolution of the vulnerability curves, 14 analyses are conducted for each building-
input ground motion, starting from a PGA of 0·08 g and ending with a PGA of 1·12 g. This scaling
approach is too small for the Set 2 records since high PGAs are needed to reach the collapse limit
state. A PGA of 0·32 g, which corresponds to twice the design ground motion, is thus selected
for scaling the Set 2 ground motions for the 10- to 50-storey buildings. Fourteen analyses are also
conducted for each building-input ground motion starting from 0·32 g and ending with a PGA of
4·48 g. The 60-storey building was less affected by the Set 2 records due to its long periods of vibra-
tion. A PGA of 0·64 g is, therefore, selected for this building to scale the Set 2 ground motions.
Fourteen analyses are conducted starting from 0·96 g and ending with a PGA of 9·28 g. This high
level of input ground motion is required to attain all limit states and derive the vulnerability curves.
The abovementioned regions of the probability distribution of each random variable, which is divided
into 14 intervals, insures that all the intervals have equal probability. Several additional analyses
are also carried out to help in deciding the suitable scaling approach as well as to reach the collapse
limit state.

3.6. Performance criteria

Four performance levels are described by the NEHRP provisions (FEMA, 2009). These are termed
the Operational (OP), Immediate Occupancy (IO), Life Safety (LS) and Collapse Prevention (CP)
levels. It is unpractical to design structures to meet the OP performance level since all utilities required
for normal operation must be available after the earthquake. This is particularly true for Occupancy
Category II structures (FEMA, 2009). The NEHRP provisions provide a detailed description of the
expected level of structural damage for the other three performance levels. Several design guidelines
and studies have used the Interstory Drift Ratio (IDR) for the damage evaluation of structures since
it is easier for calibration with experimental data and can be related to local and global damage. IDR
is considered in the present study as the primary performance criterion as it is related to performance
levels in concrete wall structures in ASCE/SEI 41-06 (2006b) and in several other sources. ASCE/
SEI 41-06 (2006b) adopts three performance levels for concrete wall structures, which are IO (0·5%),
LS (1·0%) and CP (2·0%). These are related to minor cracking, extensive damage, and extensive

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

concrete crushing and buckling of reinforcement. The SEAOC blue book (1999) recommends IDR
of 0·4%, 0·9%, 1·4% and 2·1% for concrete shear walls at a performance level SP1 to SP4, respec-
tively. These represent the negligible damage, minor to moderate (reparable) damage, moderate to
major near damage, and collapse performance levels. The latter guideline also suggests IDR greater
than 2·5% for the collapse performance level.
The abovementioned values are not intended for use in the post earthquake evaluation of damage,
but rather as an indication of IDR that typical structures may undergo when responding within various
performance levels. The code recommended drift limits tend to be on the conservative side and do
not necessarily imply actual collapse. Less conservative IDRs have, therefore, been suggested and
used in the literature based on analytical and experimental results (e.g., Ghobarah, 2004, Kircher
et al., 1997, Mwafy and Elnashai, 2002). For low-rise concrete shear wall structures designed to the
UBC (ICBO, 1997) requirements of seismic zone 2B, Kircher et al. (1997) proposed IDRs of 0·4%,
0·8%, 2·3% and 6·0% for the Slight, Moderate, Extensive and Complete damage states. Ghobarah
(2004) also proposed limit states related to ductile concrete wall structures. The suggested IDR limits
associated with ‘no damage’, ‘light reparable damage’, ‘irreparable damage or yield point’, ‘severe
damage or life safe’ and ‘collapse’ are <0·2, 0·4, >0·8, 1·5 and >2·5, respectively. On the other hand,
more conservative performance criteria were also adopted in previous studies. For instance, Ji et al.
(2007) proposed three performance limit states obtained from Inelastic Pushover Analysis (IPA) of a
high-rise building for the derivation of fragility curves. The suggested IDRs for the ‘serviceability’,
‘damage control’ and ‘collapse prevention’ limit states are 0·2%, 0·52 and 1·1%, respectively. Clearly,
these performance levels, particularly serviceability and damage control, are different than those
adopted by ASCE/SEI 41-06 (2006b).
Following the ASCE/SEI 41-06 (2006b) approach, three limit states are adopted in the present
study for the derivation of vulnerability curves, namely: (a) IO; (b) LS; and (c) CP. It is clear from
the abovementioned brief discussion that a CP level ranging from 2·0% to 2·5% is frequently used in
previous studies and by design guidelines to define the CP (or near collapse) limit state. At values in
excess of this limit, P-Δ effects are significant and lead to reduced lateral resistance and failure. Based
on the abovementioned brief review, IDR of 2·5% is adopted in the present study based on the study
of Ghobarah (2004). On the other hand, for a building to be occupied immediately after the earthquake
with little or no repair, it should remain in the elastic range so that non-structural components are not
significantly damaged. The behaviour of concrete structures designed to modern codes deviates from
linearity at the first indication of yielding. First yield is typically assumed when the strain in the main
longitudinal tensile reinforcement exceeds the steel yield strain. The LS limit state, which falls
between the IO and CP, represents a ‘significant damage’ sustained by the structure, while it accounts
for a reasonable margin of safety against collapse. This margin is considered as 50% in ASCE/SEI
41-06 (2006b).
Pushover analysis allows tracing the sequence of yielding and failure as well as the progress of the
capacity curve of the structure. This analysis procedure is, therefore, used in the present study to
define the IO and LS performance levels. Predefined lateral load patterns are incrementally applied
in a step-wise manner on the reference structures in the longitudinal and transverse directions. The
progress of the capacity curves are traced up to the collapse limit state. Following the recommenda-
tions of modern design guidelines (e.g., ASCE, 2006b), two lateral load distributions are used. The
first load is a uniform pattern, representing lateral forces that are proportional with mass. The second
lateral load distribution is an inverted triangular load, resembling the first mode shape. Due to the
significant influence of the second mode of vibration on the seismic response of high-rise buildings,
Mwafy et al. (2006) investigated a third lateral load distribution representing the second mode shape
of a 54-storey building. The study concluded that the uniform lateral load can be used to obtain a
conservative estimate of initial stiffness and lateral capacity of high-rise buildings. Mwafy and Elnas-
hai (2001) also concluded that the inverted triangular load underestimates the capacity of structures
influenced by higher modes, while the uniformly distributed load gives a better prediction of the
ultimate strength. Based on these conclusions, it was decided to use the uniform lateral load to define
the IO and LS limit states.
It is noteworthy that the selection of the reference structures’ layout was motivated by the desire
to arrive at comparable lateral capacity in the two orthogonal horizontal directions. This was con-

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

First slab/beam yielding First wall yielding

Global yielding Ultimate capacity
Uniform load Triangular load
60000

50000 Base Shear (kN) IDR = 1.1%

Base Shear (kN)

IDR = 1.01%

Base Shear (kN)

IDR = 0.81%
40000
IDR = 0.70%

30000 IDR = 0.32 IDR = 0.55%

20000 Design Design

Design
10000
10SB 20SB 30SB
Top Disp. (mm) Top Disp. (mm) Top Disp. (mm)
0
0 250 500 750 1000 1250 0 500 1000 1500 2000 0 500 1000 1500 2000

60000
IDR = 1.2%

Base Shear (kN)

Base Shear (kN)

Base Shear (kN)

50000 IDR = 1.35%
IDR = 1.35%

40000 IDR = 0.80% IDR = 0.87% IDR = 0.92%

30000
Design Design
20000 Design

10000
40SB 50SB Top Disp. (mm) 60SB
Top Disp. (mm) Top Disp. (mm)
0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500

Figure 4. Lateral capacity of the reference structures in the transverse direction along with
interstorey drift ratios at first local and global yield, which are used to define the IO and LS
performance levels.

firmed from the free vibration analyses and IPAs in the longitudinal and transverse directions
(Mwafy, 2009). This enables reducing the number of IDCAs by focusing on a one direction of the
buildings. Although twelve models were modelled using ZEUS-NL to represent the lateral force-
resisting-systems in the longitudinal and transverse directions, the six models representing the
structural systems in the transverse direction are employed in subsequent sections to derive fragility
curves. The number of elements of the selected models is 44% of those used to idealize the refer-
ence structures in the longitudinal direction, and hence a considerable saving in time and computer
resources is achieved.
The lateral capacity envelopes of the reference structures in the transverse direction are depicted
in Figure 4. The first indication of yielding in walls and horizontal members, global yield, ultimate
capacity and IDRs at first local and global yield are shown in Figure 4. The global yield is estimated
from an elastic-perfectly plastic idealization of the capacity envelops, where the initial stiffness
is evaluated as the secant stiffness at 75% of the ultimate strength. The starting point of the
post-elastic branch is considered as the global yield threshold (e.g., Elnashai and Mwafy, 2002). It
is shown in Figure 4 that the first yield gradually shifts from walls to horizontal members with
increasing the building height. Unlike the response of other structures, first yield in horizontal
members of the 10 and 20-storey buildings is observed at significantly higher lateral loads compared
with the yield in shear walls. Clearly, shifting the first indication of yield to be at a higher base
shear level and in slabs/beams rather than in shear walls, as shown from the response of the 40- to
60-storey structures, is more favourable. It is also clear from the IPA results shown in Figure 4 that
the building height has a significant impact on the IDRs at the first indication of local yield. The
IDRs at this limit state are 0·32%, 0·55%, 0·70%, 0·80%, 0·87% and 0·92%. It is also shown that
the building height has less impact on the global yield limit state, particularly for the 20- to 60-storey
buildings.
As a result of the high variability in the limit states observed from IPA results for different buil-
dings, results of IDCAs, presented hereafter, are used to confirm the definition of the IO and LS
performance levels. The formation of plastic hinges in different structural members is monitored
during the IDCAs to provide more insight into the definition of limit states. Table 3 summarizes the
IDCA results at the first indication of member yield. The maximum, minimum and median IDRs are

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

Table 3. Selected IDCA results at the first indication of member yield.

10SB 20SB 30SB 40SB 50SB 60SB

PGA IDR PGA IDR PGA IDR PGA IDR PGA IDR PGA IDR
Set 1 Max. 0·16 0·50 0·40 0·96 0·72 0·98 0·48 1·12 0·48 1·15 0·32 0·90
Min. 0·08 0·13 0·12 0·67 0·16 0·74 0·24 0·83 0·16 0·62 0·24 0·65
Median 0·08 0·41 0·20 0·79 0·30 0·88 0·32 0·98 0·32 0·86 0·24 0·75
Set 2 Max. 0·64 0·65 1·92 0·79 1·92 1·08 2·56 1·05 2·56 0·96 2·88 1·01
Min. 0·16 0·26 0·56 0·60 0·96 0·78 0·96 0·71 1·28 0·72 0·96 0·66
Median 0·32 0·42 0·88 0·68 1·44 0·96 1·28 0·84 1·44 0·80 1·92 0·84
Median of 0.16 0·42 0·48 0·76 0·84 0·93 0·72 0·92 0·88 0·83 0·64 0·77
20 records
PGA: Peak ground acceleration (g); IDR: Interstorey drift ratio (%).

presented for each of the two ground motions sets selected in this study, as discussed hereafter. The
median values of the 20 records (from both Set 1 and Set 2) used are also given. It is clear that
the variability in the IO performance limits is not significant compared with the IPA results under the
effect of the two input ground motion scenarios. This is justified by the uncertainty in the demand
prediction obtained from IPA, particularly for long period structures.
It was decided based on the results presented in Figure 4 and Table 3 to adopt different IO and LS
performance levels for the six reference structures. The adopted IDRs for the IO performance limit
are 0·32%, 0·55%, 0·70%, 0·80%, 0·83% and 0·77% for the 10- to 60-storey buildings, respectively.
These are the most conservative values obtained from the IPAs and the median IDCA results of 20
records. Since estimating the LS limit state from IDCAs is difficult without using IPA results, and
due to the lower variability in this performance limits for different buildings, it was decided to use
LS limit states obtained from IPAs. These performance limits are 0·81%, 1·0%, 1·1%, 1·2%, 1·35%
and 1·35% for the 10- to 60-storey buildings, respectively.
The adopted IO, LS and CP limit states are generally consistent with previous studies (e.g., Gho-
barah, 2004). Although the adopted limit states are slightly higher than the ASCE/SEI 41-06 (2006b)
values, it was decide to use the values derived from the extensive IPAs and IDCAs performed in the
present study. It is worth noting that the structural system selected in the present study (flat slabs with
shear walls) has potentially larger lateral deformation compared with other floor slab systems due to
the absence of beams. Based on the results presented in Figure 4 and Table 3 and the discussion
presented above, the selected limit states should be sufficient to restrict P-Δ effects and limit the
structural damage, particularly for buildings designed to the modern seismic provisions (ACI, 2005;
ICC, 2009). Finally, it should be emphasized that although the abovementioned mapping of IDRs to
the damage levels is based on values recommended in the literature and extensive inelastic analyses
carried out in the current study, this topic is still a significant research issue in the structural engineer-
ing (Kinali and Ellingwood, 2007).

4. DERIVATION OF FRAGILITY RELATIONSHIPS USING IDCA

A large number of Inelastic Response History Analyses are conducted to derive the vulnerability
relationships of the six reference structures (about 1800 IRHAs). In addition to monitoring global
response parameters such as IDR, top displacement and base shear, the formation of plastic hinges
in different structural members is screened to provide more insight into the level of structural damage.
Figure 5 shows sample IDR distributions for the 20- and 50-storey buildings from the two ground
motion scenarios at twice the design PGA. Figure 6 also presents sample IDCA results obtained from
ground motion Set 1 and Set 2 along with the power law used for deriving the fragility relationships
using Equation (1). The horizontal lines in Figure 6 represent the limit states adopted in the present

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

20 50

Story
Story
40
15

30

10
Median of Set 1 20 Median of Set 1
Median of Set 2 Median of Set 2

5
Set 2 10 Set 2
Set 1 Set 1
IDR (%) IDR (%)

2 1 0 -1 -2 2 1 0 -1 -2

Figure 5. Maximum Interstory Drift Ratios (IDRs) of the 20- and 50-storey buildings from two
input ground motion scenarios scaled to twice the design PGA along with the median values.

study, namely IO, LS and CP from bottom to top, respectively. A total of 280 points are plotted for
each building, where each point represents a PGA-IDR value obtained from an inelastic response
history analysis.
It is clear that the differences between the IDRs obtained from the two seismic scenarios are quite
significant for all buildings. Figure 6 shows that different limit states are observed at significantly
higher PGAs under the effect of Set 2 ground motions compared with Set 1, particularly for the taller
buildings. This is attributed to the significant spectral amplifications of ground motion Set 1 up to
2·0 s, which match the first or/and second mode periods of the reference structures, as shown from
Figure 3. Despite the lower mass participation of the second mode of vibration, it is prominent when
high-rise structures are excited by the first seismic scenario. The seismic response of the six reference
structures under the Set 2 records is insignificant, particularly for long-period structures at practical
PGA levels. The results emphasize the vulnerability of high-rise buildings in the UAE to severe distant
earthquakes and confirm that this impact applies to a wide class of multi-storey buildings ranging
from 10 to 60 storeys. This observation is further emphasized from the fragility relationships derived
for the reference structures, as explained below.
Figure 6 suggests a direct relationship between the building height and the damage level. Different
limit states are exceeded at higher PGAs for taller buildings. This implies that earthquakes have
more impact on low-rise buildings than high-rise structures. This observation is confirmed under
the effect of both severe distant and moderate proximate events. This is attributed to the lower
contribution of the predominant mode(s) of vibration to seismic response with increasing the building
height.
The statistical distributions shown in Figure 6 are used to calculate the probability of exceeding
each limit state at different intensity levels. The vulnerability curves are derived by plotting the cal-
culated probability data versus PGAs. The derived fragility relationships of the reference structure
using the methodology outlined above are shown in Figures 7–9. Figure 7 compares between the
vulnerability curves obtained from the two seismic scenarios employed in the present study for the
30-storey building. Figures 8 and 9 compare between the fragility curves of the six reference struc-
tures using ground motion Set 1 and Set 2, respectively. The steepness of the developed fragility
curves increases as the limit state shifts from CP to IO. The steep shape of the IO limit state curve,
particularly for the shorter buildings, is due to the high lateral stiffness in the elastic range (before
the first yield), which significantly decreases the variability of IDRs from different input ground
motions.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

6 6

Max Interstory Drift Ratio (%)

Max Interstory Drift Ratio (%)
5 5

4 4

y= 3.4699x0.919 y = 1.0214x 0.8524

3 R² = 0.9274 3 R² = 0.9104

2 2

1 1

10SB-Set 1 PGA (g) 10SB-Set 2 PGA (g)

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
6

Max Interstory Drift Ratio (%)

6
Max Interstory Drift Ratio (%)

5 5

4 4

y= 2.6438x0.8815 y = 0.6567x 0.8988

3 R² = 0.8335 3
R² = 0.8901

2 2

1 1

30SB-Set 1 PGA (g) 30SB-Set 2 PGA (g)

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
6
Max Interstory Drift Ratio (%)

6
Max Interstory Drift Ratio (%)

5 5

4 4

y = 2.5403x 0.916 y = 0.5651x0.7901

3 R² = 0.9037 3 R² = 0.8128

2 2

1 1

50SB-Set 1 PGA (g) 50SB-Set 2 PGA (g)

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Figure 6. Selected IDCA results obtained from ground motion Set 1 and Set 2 along with the
power law equations and limit states (IO, LS and CP from bottom to top, respectively).

The differences between the vulnerability relationships obtained from the two seismic scenarios
employed in the present study are clear from Figure 7. The slops are steeper and the probability of
exceeding various limit states is much higher under the effect of the Set 1 ground motions compared
with Set 2. The large disparities are due to differences in characteristics between Set 1 and Set 2
records, as explained above. The seismic response of the reference structures under the Set 2 records
when scaled to the design PGA is insignificant, particularly for long-period structures, as shown from
Figure 9. The earthquake magnitude and distance have a significant impact on the seismic risk evalu-
ation of multi-storey buildings. The results confirm the observations discussed above regarding the
vulnerability of the reference buildings to severe distant earthquakes compared with moderate close
events, which suggest expanding this study to investigate the seismic risk of other types of structures
in this region.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

1
30SB-Set 1

P(Limit State|GMI)
0.8

0.6
IO - Set 1
LS - Set 1
0.4 CP - Set 1
IO - Set 2
LS - Set 2
0.2 CP - Set 2
30SB-Set 1
PGA (g)
0
0 0.5 1 1.5 2

Figure 7. Fragility relationships of the 30-storey building obtained from IDCAs using two ground
motion scenarios.
1 1
Minor Minor Moderate
Moderate
P(Limit State|GMI)

P(Limit State|GMI)
0.8 0.8

Severe Severe
0.6 0.6 IO
IO
LS Collapse LS
Collapse
0.4 CP 0.4 CP

10SB-Set 1 20SB-Set 1
0.2 0.2

PGA (g) PGA (g)

0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
1 1
Minor Minor
Moderate Moderate
P(Limit State|GMI)

P(Limit State|GMI)

0.8 0.8

Severe Severe
0.6 IO 0.6 IO
LS LS
0.4 CP 0.4 CP
Collapse Collapse
30SB-Set 1 40SB-Set 1
0.2 0.2

PGA (g) PGA (g)

0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
1 1
Minor
Minor Moderate
Moderate
P(Limit State|GMI)

P(Limit State|GMI)

0.8 0.8

Severe Severe
0.6 IO 0.6 IO
LS LS
0.4 CP 0.4 CP
Collapse Collapse
50SB-Set 1 60SB-Set 1
0.2 0.2

PGA (g) PGA (g)

0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2

Figure 8. Fragility relationships of the reference structures obtained from IDCAs using Set 1
ground motions.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

1 1

Minor

P(Limit State|GMI)
P(Limit State|GMI)
0.8 0.8
Minor
Moderate

0.6 IO 0.6
Moderate IO
LS LS
Severe
0.4 CP 0.4 CP

10SB-Set 2
20SB-Set 2
0.2 0.2
Severe
Collapse PGA (g)
PGA (g) Collapse
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
1 1

P(Limit State|GMI)
P(Limit State|GMI)

0.8 0.8

Minor Minor
0.6 0.6 IO
IO
LS
Moderate LS
CP
0.4 CP 0.4
Moderate

30SB-Set 2 40SB-Set 2
0.2 0.2
Severe Severe
PGA (g) PGA (g) Collapse
0 Collapse 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
1 1
P(Limit State|GMI)

P(Limit State|GMI)

0.8 0.8

Minor Minor
0.6 IO 0.6
IO
LS LS
0.4 CP 0.4 CP

Moderate 50SB-Set 2 60SB-Set 2

Moderate
0.2 0.2
Severe
PGA (g) PGA (g) Severe
0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2

Figure 9. Fragility relationships of the reference structures obtained from IDCAs using Set 2
ground motions.

5. CONCLUSIONS

High-rise buildings are the most significant in the potential consequences from natural hazard events
since they represent high levels of financial investments and population densities. The increase in the
rate of seismic activities along with expanding the high-rise building inventory in the earthquake-
prone areas of the UAE necessitate the detailed assessment of seismic risk in order to effectively
predict and mitigate earthquake losses. This paper focused on the development of fragility relation-
ships, which constitute an essential driving engine in loss assessment systems. The study included
the selection, full structural design and developing verified fibre-based simulation models of six refer-
ence structures representing a wide spectrum of the high-rise buildings in the UAE. The selection and
scaling of twenty input ground motions representing long (Set 1) and short (Set 2) source-to-site
distance earthquake scenarios anticipated in the study area were discussed. Limit state criteria for

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 FRAGILITY OF MODERN HIGH-RISE BUILDINGS

deriving fragility curves were selected based on the mapping of local and global response from
Inelastic Pushover Analyses (IPAs) and Incremental Dynamic Collapse Analyses (IDCAs). The mea-
sured seismic response from a large number of IDCAs was related to ground motion intensity through
a reliable statistical model to derive the fragility relationships of the reference structures.
The disparities between the vulnerability relationships obtained from the two seismic scenarios
employed in the present study were quite significant for all buildings. The probability of exceeding
various limit states was much higher and the slops were steeper under the effect of the Set 1 earth-
quake scenario compared with Set 2. These are attributed to the high spectral amplifications and
effective durations of the Set 1 ground motions, which amplify the most significant modes of vibration
of high-rise buildings. At the design PGA, the response of the reference structures under the Set 2
seismic scenario was insignificant, particularly for higher buildings. Under the effect of both Set 1
and 2 events, limit states were exceeded at higher ground motion intensities for taller buildings, which
imply that earthquakes have higher impact on low-rise buildings.
The presented study confirmed the vulnerability of a wide range of high-rise buildings, ranging
from 10 to 60 storeys, to the severe distant seismic scenario anticipated in the earthquake-prone areas
of the UAE. The significance of assessing the seismic risk of a population of buildings under the
effect of anticipated seismic scenarios and the need for expanding this study to cover other classes
of structures in this region were emphasized. While the focus of the study was on the physical damage
of representative high-rise buildings, it stimulates interdisciplinary research activities in future, aiming
at formulating earthquake loss assessment and mitigation strategy for the UAE.

ACKNOWLEDGEMENT

This study was supported by the Research Affairs at the UAE University under a contract no. 07-01-
07-11/09. This financial support is appreciated. The author would also like to thank Eng. A.A. Ashry
for his contribution to related research. The opinions expressed in this paper are those of the author
and do not reflect the views of the sponsors.

REFERENCES
Abdalla JA, Al-Homoud AS. 2004. Seismic hazard assessment of United Arab Emirates and its surroundings. Journal of
Earthquake Engineering 8(6): 817–837.
ACI 2005. Building Code Requirements for Structural Concrete and Commentary (318-05). American Concrete Institute:
Detroit, MI.
Aldama-Bustos G, Bommer JJ, Fenton CH, Stafford PJ. 2009. Probabilistic seismic hazard analysis for rock sites in the cities
of Abu Dhabi, Dubai and Ra’s Al Khaymah, United Arab Emirates. Georisk: Assessment and Management of Risk for
Engineered Systems and Geohazards 3(1): 1–29.
Ambraseys NN, Douglas J, Sigbjornsson R, Berge-Thierry C, Suhadolc P, Costa G, Smit PM 2004. Dissemination of European
strong-motion data, vol. 2, using strong-motion datascape navigator, CD ROM collection. Engineering and Physical Sci-
ences Research Council: Swindon, UK, Feb. 2004.
ASCE 2006a. Minimum design loads for buildings and other structures. ASCE Standard ASCE/SEI 7-05 Including Supplement
No. 1. American Society of Civil Engineers: Reston, VA.
ASCE 2006b. Seismic rehabilitation of existing buildings, ASCE Standard ASCE/SEI 41-06. American Society of Civil Engi-
neers: Reston VA (formerly FEMA 356).
Bommer JJ, Martínez-Pereira A. 1999. The effective duration of earthquake strong motion. Journal of Earthquake Engineer-
ing 3(2): 127–172.
Cagnan Z, Sesetyan K, Zulfikar CAN, Demircioglu MB, Kariptas C, Durukal E, Erdik M. 2008. Development of earthquake
lossmap for Europe. Journal of Earthquake Engineering 12(2): 37–47.
Calvi GM, Pinho R, Magenes G, Bommer JJ, Restrepo-Velez LF, Crowley H. 2006. Development of seismic vulnerability
assessment methodologies over the past 30 years. ISET Journal of Earthquake Technology 43(3): 75–104.
CSI. 2008. ETABS—Integrated Building Design Software. Computers and Structures, Inc.: Berkeley, CA.
Elnashai AS, Cleveland LJ, Jefferson T, Harrald J 2008a. Impact of earthquakes on the Central USA. Mid-America Earthquake
Center Report 08-02, University of Illinois at Urbana-Champaign, Urbana, IL.
Elnashai AS, Hampton S, Lee JS, McLaren T, Myers JD, Navarro C, Spencer B, Tolbert N. 2008b. Architectural overview of
MAEviz–HAZTURK. Journal of Earthquake Engineering 12: 92–99.
Elnashai AS, Mwafy AM. 2002. Overstrength and force reduction factors of multistorey reinforced-concrete buildings. The
Structural Design of Tall and Special Buildings 11(5): 329–351.
Elnashai AS, Papanikolaou V, Lee D 2009. Zeus-NL—A System for Inelastic Analysis of Structures—User Manual. Mid-
America Earthquake Center, University of Illinois at Urbana-Champaign: Urbana, IL.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal
 A.M. MWAFY

FEMA. 2009. NEHRP Recommended Seismic Provisions for New Buildings And Other Structures, FEMA P-750 2009 Edition,
Part 1: Provisions and Part 2: Commentary to ASCE/SEI 7-05. Federal Emergency Management Agency: Washington,
DC
Ghobarah A. 2004. On drift limits associated with different damage levels. In Proceedings of the Bled ‘04 International
Workshop on Performance-Based Seismic Design, Bled, Slovenia, 28 June 28–1 July.
ICBO 1997. Uniform Building Code, 1997 Edition. International Conference of Building Officials: Whittier, CA.
ICC. 2009. International Building Code. International Code Council: Country Club Hills, IL.
Jeong SH, Elnashai AS. 2005. Analytical assessment of an irregular rc frame for full-scale 3D pseudo-dynamic testing. Part
I: analytical model verification. Journal of Earthquake Engineering 9(1): 95–128.
Ji J, Elnashai AS, Kuchma DA. 2007. An analytical framework for seismic fragility analysis of RC high-rise buildings. Engi-
neering Structures 29(12): 3197–3209.
Kinali K, Ellingwood BR. 2007. Seismic fragility assessment of steel frames for consequence-based engineering: a case study
for Memphis, TN. Engineering Structures 29(6): 1115–1127.
Kircher CA, Nassar AA, Kustu O, Holmes WT. 1997. Development of building damage functions for earthquake loss estima-
tion. Earthquake Spectra 13(4): 663–682.
Kircher CA, Whitman RV, Holmes WT. 2006. HAZUS earthquake loss estimation methods. Natural Hazards Review 7(2):
45–59.
Kwon OS, Elnashai AS. 2006. The effect of material and ground motion uncertainty on the seismic vulnerability curves of RC
structure. Engineering Structures 28(2): 289–303.
Moharram AM, Elghazouli AY, Bommer JJ. 2008. A framework for a seismic risk model for Greater Cairo. Soil Dynamics
and Earthquake Engineering 28(10–11): 795–811.
Mwafy AM 2009. Seismic Design Response Factors and Vulnerability Functions of Typical UAE Buildings. Phase I: Response
Factors. UAE University: Al Ain, UAE.
Mwafy AM, Elnashai AM, Sigbjornsson R, Salama A. 2006. Significance of severe distant and moderate close earthquakes
on design and behavior of tall buildings. The Structural Design of Tall and Special Buildings 15(4): 391–416.
Mwafy AM, Elnashai AS. 2001. Static pushover versus dynamic collapse analysis of RC buildings. Engineering Structures
23(5): 407–424.
Mwafy AM, Elnashai AS. 2002. Calibration of force reduction factors of RC buildings. Journal of Earthquake Engineering
6(2): 239–273.
PEER. 2009. PEER NGA database. Pacific Earthquake Engineering Research Center, University of California, Berkeley, CA,
http://peer.berkeley.edu/nga [30 January 2009].
Porter KA, Beck JL, Shaikhutdinov RV. 2002. Sensitivity of building loss estimates to major uncertain variables. Earthquake
Spectra 18(4): 719–743.
Priestley MJN, Grant DN. 2005. Viscous damping in seismic design and analysis. Journal of Earthquake Engineering 9(2):
229–255.
Rossetto T, Elnashai AS. 2005. A new analytical procedure for the derivation of displacement-based vulnerability curves for
populations of RC structures. Engineering Structures 27(3): 397–409.
SEAOC. 1999. Recommended Lateral Force Requirements AND Commentary, 7th Edition. Seismology Committee Structural
Engineers Association of California: Sacramento, CA.
Tso WK, Zhu TJ, Heidebrecht AC. 1992. Engineering implications of ground motion A/V ratio. Soil Dynamics and Earthquake
Engineering 11(3): 133–144.
Wen YK, Ellingwood BR, Bracci J 2004. Vulnerability function framework for consequence-based engineering. Mid-America
Earthquake Center Project DS-4 Report. University of Illinois at Urbana-Champaign: Urbana, IL.
Willford M, Whittaker A, Klemencic R 2008. Recommendations for the Seismic Design of High-Rise Buildings. Council on
Tall Buildings and Urban Habitat: Chicago, IL.

Copyright © 2010 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2010)
DOI: 10.1002/tal