Escolar Documentos
Profissional Documentos
Cultura Documentos
ACKNOWLEDGEMENTS
The authors of these lecture notes thank the contribution of Julian Bommer (Section 1);
Andreas Kappos (Section 2.1), Chiara Casarotti, Carlos Blandon, Eleni Smyrou (Section
2.2), Stelios Antoniou, Dario Pietra (Section 2.3), Ricardo Monteiro, Mario Marques
(Section 2.4); Michele Calvi, Guido Magenes, Julian Bommer, Luis Restrepo-Velez
(Chapter 3); Barbara Borzi (Section 4.3); Julian Bommer, Peter Stafford (Section 5.2 and
5.3).
TABLE OF CONTENTS
ACKNOWLEDGEMENTS........................................................................................................................ i
TABLE OF CONTENTS ..........................................................................................................................iii
1. INTRODUCTION ..................................................................................................................................1
2. SEISMIC VULNERABILITY ASSESSMENT OF SINGLE STRUCTURES..............................5
2.1 SEISMIC RESPONSE OF RC BUILDINGS AN OVERVIEW.............................................................5
2.1.1 Structural Configuration of Buildings .................................................................................5
2.1.2 Seismic Response of RC Structures ....................................................................................9
2.1.3 Ductility of RC structures...................................................................................................11
2.1.4 Earthquake-Resistant Properties of Materials..................................................................16
2.1.5 Behaviour of Beams under Seismic Loading ...................................................................32
2.1.6 Behaviour of Columns under Seismic Loading ...............................................................41
2.1.7 Beam-Column Joints ...........................................................................................................46
2.1.8 Seismic Behaviour of Walls under Seismic Loading .......................................................50
2.2 NONLINEAR DYNAMIC ANALYSIS OF STRUCTURES SUBJECTED TO SEISMIC ACTION ..........58
2.2.1 Modelling Approaches for Inelastic Analysis...................................................................58
2.2.2 Fibre Modelling Approach .................................................................................................60
2.2.3 Assembling a FE Model for Dynamic Analyses for Reinforced Concrete Structures62
2.2.4 Case Study 1: Multi-Span Continuous Deck Motorway Bridge ....................................63
2.2.5 Case Study 2: Multi-storey RC Frame ICONS Project ...............................................76
iv
vi
1. INTRODUCTION
In the last few decades, a dramatic increase in the losses caused by natural catastrophes
has been observed worldwide. Reasons for the increased losses are manifold, though
certainly include the increase of world population, the development of new super-cities
(with a population greater than 2 million) - many of which are located in zones of high
seismic hazard - and the high vulnerability of modern societies and technologies, such as
the built environment (e.g. Bommer, 2002; Smolka et al., 2004). Figure 1.1 shows the
actual worldwide losses from great natural disasters from 1950 to 2003 and fitted curves
to show the increased trend of economic and insured losses.
Figure 1.1. Losses from Great Natural Disasters (far exceeding 100 deaths or 100 million US$ in
losses) from 1950-2003. Figures have been adjusted for inflation. Source: Munich Re
2004 (Smolka et al., 2004)
Although Figure 1.1 considers the losses due to all great natural disasters, a large
proportion of the loss can be attributed to the direct and indirect economic impact of
earthquakes; the 1994 Northridge (California, US) earthquake produced the highest ever
insured earthquake loss at approximately US$14 billion, and the US$150 billion cost of
the 1995 Kobe (Japan) earthquake was the highest ever absolute earthquake loss
(corresponding to the spike in Figure 1.1). Although the dollar value of economic losses
in other parts of the world may be far lower than in Japan and the US, the impact on the
national economy may be much greater due to losses being a large proportion of the
gross national product (GNP) in that year. Coburn and Spence (1992) report the
economic losses due to earthquakes from 1972 to 1990; the three largest losses as
proportions of the GNP are in the Central American countries of Nicaragua (1972, 40%
GNP), Guatemala (1976, 18% GNP) and El Salvador (1986, 31% GNP). When the
economic burden falls entirely on the government (such as occurred after the 1999
Kocaeli earthquake in Turkey), the impact on the national economy can be crippling; one
possible solution is to privatise the risk by offering insurance to homeowners and then to
export large parts of the risk to the worlds reinsurance markets (e.g. Bommer et al., 2002).
In order to design such insurance and reinsurance schemes, a reliable earthquake loss
model for the region under consideration needs to be compiled such that the future
losses due to earthquakes can be determined with relative accuracy.
Earthquake loss models should ideally include all of the possible hazards from
earthquakes: amplified ground shaking, landslides, liquefaction, surface fault rupture, and
tsunamis. Nevertheless, strong ground shaking is often the only hazard considered in loss
assessment methods; this is often an acceptable assumption because as the size of the loss
model increases, the relative influence of the secondary hazards such as liquefaction and
landslides decreases (Bird and Bommer, 2004). Constructing an earthquake loss model for
a city, region or country involves compiling databases of earthquake activity, ground
conditions, attenuation equations, building stock and infrastructure exposure, and
vulnerability characteristics of the exposed inventory. The main aim of a loss model is to
calculate the seismic hazard at all the sites of interest and to convolute this hazard with
the vulnerability of the exposed building stock such that the damage distribution of the
building stock can be predicted; damage ratios, which relate the cost of repair and
replacement to the cost of demolition and replacement of the structures, can then be
used to calculate the loss.
Seismic loss can be defined as the possibility or probability of losses due to an
earthquake, whether these losses are human, social, or economic; seismic risk can be
quantified through the convolution of four individual factors:
Seismic Loss = Seismic Hazard * Exposure * Vulnerability * Specific Cost
(1.1)
The seismic hazard represents the effects that earthquakes can produce at the site of a
structure or other engineering project (e.g. strong ground shaking) whilst exposure refers
to the extent of human activity located in the zones of seismic hazard. The vulnerability
represents the susceptibility of the exposed elements to earthquake effects and the
specific cost could represent the economic cost of the repair and restoration of a
structure as a proportion of the cost of demolition and replacement of the structure, or it
could represent the social cost in terms of the percentage of occupants of the building
stock who could be fatally injured.
Earthquake risk or loss modelling based on a single earthquake scenario can be very
useful, particularly for communicating seismic risk to the public and to decision makers.
Nevertheless, for many applications, including decision-making processes within the
insurance and reinsurance industries, and in seismic code drafting committees, it is
necessary to estimate the effects of many, or even all, possible future earthquake
scenarios that could impact upon the urban areas under consideration. In either case, a
significant component of a loss model is a methodology to assess the vulnerability of the
built environment.
This course will look at many different vulnerability methods which are applied in the
estimation of earthquake losses over large geographical areas; however it is noted that
much of the focus of current and future research is on analytical methodologies. In order
to apply an analytical methodology to a group of structures at an urban scale, it is
necessary to first understand the seismic behaviour of single structures. Hence, this
course begins with a detailed look at the seismic response and assessment of single
structures (Chapter 2). When looking at single structures the assessment is most simply
performed in a deterministic manner, as the geometrical and material properties of the
structure are generally known. This part of the course will thus just concentrate on
deterministic methods for seismic assessment of individual structures (such as the
Capacity Spectrum Method, the N2 method, amongst others), whilst probabilistic
methods will subsequently be introduced for the treatment of uncertainties within large
scale loss models where many structures need to be assessed together. Methodologies to
assess the vulnerability of groups of building have evolved over the past 30 years and so
this course will briefly review these methods and then will concentrate on the analytical
methodologies which are currently the focus of most research in this field (see Chapters 3
and 4). The requirement within loss models to assess the influence of ground shaking on
a large number of structures which are spatially distributed over large areas also changes
the way in which the hazard needs to modelled, as compared to conventional site-specific
seismic hazard assessment; this subject is covered in Chapter 5 of these notes. The
specific cost part of a loss model is required to estimate social and economic losses and is
generally based on the collection of a large amount of empirical data from previous
earthquakes; Chapter 6 presents some studies which allow the direct damage (estimated
through the convolution of vulnerability, hazard and exposure) to be transformed into
loss estimates.
CR
CM
Torsional Vibration
Figure 2.1 Mode of torsional vibration under translational excitation (CR is centre of resistance; CM
is centre of mass), and unfavourable plan layouts
The centre of mass is where the earthquake (inertia) force is applied. The structural
resistance is applied at the centre of stiffness of the lateral force resisting elements. If
there is an offset between the centre of force application (centre of mass) and that of
force resistance (centre of stiffness or strength), a couple is applied which increases the
shear force on columns above those considered in design. Therefore, the location of stiff
members, such as walls and stair wells, should be carefully selected to minimise torsional
effects.
2.1.1.2 Elevation
The aspect ratio of the building in elevation affects the over-turning moment exerted on
the foundations. Also, very slender structures suffer from higher mode contributions thus
necessitating the use of more elaborate seismic force calculation methods. This is
illustrated in Figure 2.2 below, where the response spectra analysis for a flexible structure
is shown. It is observed that the elastic contribution of the fundamental mode (modal
contribution of 70%) becomes of the order of 0.018 (i.e. 0.0260.7) whilst the
contribution of the higher modes is 0.04 (i.e. 0.320.13) and 0.022 (i.e. 0.360.06), for the
second and third modes, respectively.
0.50
0.40 0.36
Sa [g]
0.32
0.30
0.20
0.10
0.026
0.00 T3 T2
0.0
0.5
1.0
1.5
Period [s]
T1
2.0
2.5
3.0
Figure 2.2 Spectral response values for periods of vibration of transverse frame
The scenario described above becomes more evident after some stiffness degradation has
taken place, due to accumulation of damage. This causes the periods of the structure to
elongate, thus the amplification of the fundamental mode reduces even further whilst the
higher modes remain relatively unchanged (the smaller the period, the less significant is
the elongation). In the case where the contribution of higher modes to the response of
the structure is significant, the distribution of storey drifts is no regularly distributed
throughout the elevation of the building, as shown in Figure 2.3. Such feature cannot be
identified nor assessed if simplified static methods are used. Therefore, a low aspect ratio
(ratio of height to width) is desirable.
-1.25
104.6
= 0.8%
-0.55
93.6
-32.1
72.7
= 0.6%
-13.1
47.3
-2.77
21.8
-1.58
3.22
(a)
(b)
Figure 2.3 Deformation profiles for transverse frame subjected to: (a) higher modes of vibration; (b)
fundamental modes of vibration
Figure 2.5 Relative behaviour of frames with shear frames and cantilever structures
2.1.2
10
input of seismic energy. The inelastic deformation of the structure (segment de) is
achieved through the development of a plastic hinge at the base of the oscillator where
energy is dissipated through hysteresis. When the ground movement is reversed, only part
of the energy is recovered through kinetics, thus once the base motion terminates, the
structure does not fully recover to its initial state.
Inertia
Force
Horizontal
ground-movement
Recoverable
Energy
Displacement
Time
Elastic Response
(a)
Inertia
Force
Dissipated
Energy
Recoverable
Energy
Displacement
Elasto-plastic Response
(b)
Figure 2.6 Response of a SDOF system under base excitation: (a) elastic and (b) elasto-plastic
behaviour.
11
the top of the structure, the member demands are much higher for the column-sway
mechanism, thus the structure will fail much earlier.
p
Figure 2.7 Collapse mechanisms: (a) beam sway; (b) column sway; (c) mixed mode
12
Moment
Spalling
Failure
First yield
Cracking
Curvature
Figure 2.9
13
The role played by the reinforcement characteristics and axial load is determinant in the
member ductility, though through reasons distinct from those considered at section level.
Higher values of axial load increase second order effects, thus leading to earlier collapse
of the member. On the contrary, higher levels of transverse reinforcement will decrease
the risk of buckling of the reinforcement bars, hence increasing the ductility. Also of
prime importance, is the role played by the material properties of steel in the spread of
inelasticity in the member, which is highly dependent on the strain hardening
characteristics of the reinforcement bars.
2.1.3.3 Structural Ductility
Earthquake-resistant structures are now designed following a capacity design philosophy,
recently implement in several seismic design regulations world-wide. In capacity design,
the structure is viewed as comprising two types of zone; dissipative and non-dissipative.
The dissipative zones are those responsible for the mobilisation of the desired failure
mode, chosen to maximise overall energy absorption capacity and avoid collapse. All
other zones are considered non-dissipative. The dissipative zones are dimensioned first
and carefully detailed to possess maximum ductility. Then their likely overstrength, i.e.
the possibility of these structural elements having, after construction, higher resistance
than calculated and specified in the design, is estimated. The sources of overstrength in
structures include:
higher concrete compressive strength;
confinement;
14
The non-dissipative parts of the structure are then designed to withstand forces
consistent with the strength of dissipative parts, including overstrength. In this way, the
structure is rendered much less sensitive to the characteristics of the input motion, since
it can only respond in the ductile mode that was envisaged in the design phase, resulting
in increased control to seismic response.
Analytical studies recently carried out to evaluate the effectiveness of EC8 rules for the
design of earthquake-resistant structures confirm the efficiency of capacity-design
philosophy (CEB, 1996). In Figure 2.10, the results from the nonlinear dynamic analysis
of an eight-storey frame (Coelho and Carvalho, 1991), confirm that plastic hinge was
restricted to beams, with the exception of ground-floor columns, as desired. Such failure
mode ensures that a ductile dynamic response of structures is achieved.
In addition to the aforementioned capacity design rules, other design features are also
paramount in guaranteeing the attainment of global structural ductility. These are:
continuity and redundancy between members, so as to ensure a clear load path for
horizontal loads and prevent brittle failures;
regularity of mass, stiffness and strength distribution, to avoid adverse torsional
effects and soft-storey mechanisms;
reduced masses and sufficient stiffness, to avoid highly flexible structures which may
lead to heavy non-structural damage and significant P- effects.
A ratio between global yield and ultimate displacements at the top of the structure can
also be used to quantify their ductility. In this case, instead of a moment vs. curvature
(section level) or transverse force vs. drift (member level), the ductility plot consists of
total drift at the top of the structure vs. total base-shear. Yield may be related to the
attainment of a pre-determined inter-storey drift whilst the ultimate criterion is normally
related to the formation of a collapse mechanism.
15
Figure 2.10 Plastic-hinge distribution in an eight-storey frame designed according to EC8 (Coelho
and Carvalho, 1991).
16
Frame Behaviour
Failure Mode
Floor Behaviour
Regularity
(Elevation/Plan)
Connection Behaviour
Deformation
Rotational Ductility
Member Behaviour
Hinge Length
Curvature Ductility
Section Behaviour
Confinement
Figure 2.11 Conceptual relationship between local and global seismic response characteristics
(Elnashai, 1993)
17
18
Confinement by spirals is more effective than that by closed ties (called hoops), since the
latter tend to deflect under the pressure of the expanding concrete core, hence parts of
the core remain practically unconfined, as indicated in Figure 2.13.
Figure 2.14 Uniaxial stress-strain relationship for concrete proposed by Mander et al. (1988)
19
1
Ae = 1
b
cdc
i =1
(wi )2 b
n
(wi )2 1 s 1 s
s
s
d c 2 = bc d c
4
4
6 2bc 2d c
i =1
(2.1)
1
1
bc d c
A
Ke = e =
Acc
(wi )2 1
i =1
s
s
1
6 2bc 2d c
1 cc
(2.2)
where bc and dc represent the core dimensions to centrelines of perimeter hoop in x and y
directions, respectively (bc dc); cc is the ratio of area of longitudinal reinforcement to
area of core of section; wi is the ith clear distance between adjacent longitudinal bars; and
s represents the clear vertical spacing between hoops.
The ratio of total area of transverse reinforcement may be defined as the sum of
transverse confining steel in x and y directions as follows:
s =
Asx bc + Asy d c
bc d c s
Asx Asy
+
= x + y
d c s bc s
(2.3)
where Asx and Asy represent the total area of transverse reinforcement in x and y
directions, respectively. Inversely, the effective section area ratios of transverse
reinforcement to core concrete cut by planes perpendicular to the x and y directions are
given by
20
x =
Asy
Asx
and y =
dcs
dcs
(2.4)
Assuming that the lateral confinement pressure is sufficient to yield the transverse rebars,
then equilibrium is satisfied by guaranteeing the tension force in the bars equals the lateral
stress resultants, as shown in Figure 2.15 below. Therefore, the lateral confining stress on
the concrete in x and y directions becomes
0.5fyhAsx
0.5fyhAsx
5fyhAsy
flxdcs
0.5fyhAsy
flxdcs
flybcs
flybcs
5fyhAsy
0.5fyhAsx
0.5fyhAsy
0.5fyhAsx
Figure 2.15
f lx d c s = 2 0.5 f yh Asx f lx =
Asx
f yh = x f yh and
dcs
f ly bc s = 2 0.5 f yh Asy f ly =
Asy
bc s
f yh = y f yh
(2.5)
(2.6)
21
Figure 2.16 Confined strength from lateral confining stresses for rectangular sections (Mander et
al., 1988)
Hence, the effective lateral confining stresses in the x and y directions are:
f lx = k e x f yh and f ly = k e y f yh
(2.7)
And, finally, the confinement factor is obtained from the general solution of the
multiaxial failure criterion in terms of the two lateral confining stresses presented in
Figure 2.16.
(c) Effect of Strain Rate. The dynamic application of loading gives rise to additional
damping, proportional to strain velocity & = d dt . The CEB (1993) equation for
strength increase is:
(2.8)
Note that typical strain rates in earthquakes are 0.01 to 0.02 sec-1. These effects are
usually negligible and thus are not normally accounted for in earthquake design scenarios.
(d) Cyclic Loading. The main characteristics of concrete under reversed cyclic loading
(Figure 2.17) are:
22
( ,f )
( ,f )
un
re
un
re
( ,f )
ne w
me
plcr
pl
23
(b) Deformation Capacity - su. This property is fundamental to ensure that there is
Stress
Yield stress
Tensile strength
Elastic moduls
Strain hardening
modulus
Yield strain
Strain hardening strain
Strain at maximum load
Strain
Figure 2.19
24
indicated. It is clear that high levels of member ductility will be obtained with increased
deformation capacity of the reinforcement bars. For instance, for this particular example,
reinforcement bars with a deformation capacity corresponding to su = 0.04, provide the
member with only 60% of the ductility level achieved with a value of su = 0.05.
70
60
Force (kN)
50
su = 0.005
40
su = 0.040
30
su = 0.050
20
10
0
0
Displacement ductility
Table 2.1.
25
Properties
su,k (%)
ft / fy
fy,act / fy,nom
DCL
DCM
DCH
5%
6%
9%
1.08*
1.15
1.20
1.35
1.35
1.25
1.20
Esh = 0.10Es
Force (kN)
50
40
Esh = 0.01Es
30
20
10
0
0
20
40
60
80
100
120
Displacement (mm)
Figure 2.21 Influence of strain hardening on the monotonic behaviour of a reinforced concrete
column
Similarly to the requirements for the parameter su, minimum values for the strain hardening ratio
are also defined in EC8, dependent on the ductility class. These are described in
(d) Overstrength ft/fy and fy,act / fy,nom. As mentioned earlier, the likelihood of
materials used in the construction of RC structures possessing higher strength values than
26
those specified in design, should be carefully taken into account. The possible
overstrength of the dissipative members needs to be controlled so that the adjacent nondissipative zones can be designed accordingly. If the capacity of dissipative zones exceeds
the values considered at the design stage, then the pre-defined sequence of plastic hinge
will not occur thus jeopardising the whole design process. Therefore, control on the
overstrength of reinforcement steel is fundamental to advance the design process.
The adverse effects of strength variability where highlighted in a study carried out by
Alexandrou (1991) on the global behaviour of RC frames subjected to seismic loading. In
Table 2.2, the results from a study on a two-storey frame are presented. The steel yield
was varied between practical limits (coefficient of variance was considered as 5%) and the
ensuing population of nominally identical frames was analysed.
Table 2.2 Effect of material variability on response parameters of a RC frame (Alexandrou, 1991).
Displacement
Energy absorption
Behaviour factor
Ductility (d)
(kNm)
Maximum
Minimum
10.11
6.57
73.05
24.89
5.50
3.75
% change
54%
192%
47%
It was observed that although the variations of yield strength correspond to realistic
values (as shown in the following section), their effect on the global behaviour of the
structure are critical, resulting in a change of 47% in the value of the behaviour factor.
As mentioned earlier, overstrength is taken into account in the capacity design procedure prescribed
in EC8, through the use of overstrength factors. In addition, upper bound values are
defined for the tensile-to-yield (ft / fy) and actual-to-nominal (fy,act / fy,nom) strength ratios.
These will not only guarantee economical designs of the non-dissipative zones but
should also prevent brittle failure modes triggered by high shear demands. Such
requirements are shown in
(e)
strength steel are commercially available at present. These are cold-worked, hot-rolled
and Tempcore reinforcement steels. Due to the differences in the manufacturing
processes, their mechanical characteristics are quite distinct, even for steels belonging to
27
Stress (MPa)
the same resistance grade. This is shown in Figure 2.22, where it is observed that hotrolled steels have very high levels of strain hardening and ductility, as opposed to coldworked steel which tends to possess little ductility and strain hardening. Tempcore steel
may be considered as an intermediate case between the two other types.
Tempcore
(grade 500)
Cold-worked
(grade 400)
Strain (%)
Figure 2.22
Several recent studies (Manzocchi and Elnashai, 1994; Plumier and Vangelatou, 1995;
Pipa and Carvalho, 1995) have indicated that the code requirements described previously
are difficult to comply with, particularly in the case of structures designed to the higher
ductility classes. This is illustrated by the results obtained from tests conducted on steel
manufactured by several European and New Zealand producers, shown in Table 2.3.
The majority of the tests were carried out at LNEC (Pipa, 1994), with the exception of
the reinforcement produced in New Zealand (Andriono and Park, 1986). Cold-worked
steel corresponds to a Portuguese manufacturer, with tests being performed on 1327
specimens throughout a ten-year period. Hot-rolled reinforcement from Portuguese and
Spanish manufacturers was also tested at LNEC, with a sample size of 461 and 1200
specimens, respectively. Tempcore steel, which constitutes the majority of commercial
steel presently produced in Europe, was tested using reinforcement bars from 14
different manufacturers, based in six European countries (France, Germany, Italy,
Portugal, Spain and United Kingdom). Only bars with diameters larger or equal to 14 mm
28
were considered and both Grade 400 and Grade 500 were used. A sample size of 350
specimens was considered for each Grade.
Table 2.3 Reinforcement steel mechanical properties (adapted from Pipa, 1994)
Grade 400
Properties
Cold Worked
Grade 500
Hot Rolled
Tempcore
Hot Rolled
Tempcore
Portugal
Portugal
Spain
N.Z.
Europe
Spain
Europe
su,k (%)
3.9
10.2
10.8
12.8
9.0
8.8
7.1
ft / fy
1.14
1.57
1.4
1.52
1.2
1.3
1.16
fy,act /
fy,nom
1.21
1.12
1.16
1.21
1.24
1.09
1.17
29
It is also important to stress that the overstrength ratio (fy,act / fy,nom), as specified in EC8,
should actually be verified through a quality control procedure carried out on the steel
delivered on site, not at the production plant. This would require even tighter control of
the aforementioned steel parameters at the production stage, and makes code
requirements even more difficult to conform to.
(f) Strain-rate Effect (dynamic loading). Equation for strength increase (Soroushian
and Sim, 1986):
(2.9)
hence, for &s = 0.02 sec-1, fy increases by 20% with respect to static loading ( &s 10-5
sec-1) for low strength steel (fym = 240 MPa) and by 12% for higher grade steel (fym = 420
MPa).
(g) Cyclic Loading. Unloading takes place essentially elastically (E 200 GPa), but
reloading in the opposite direction is strongly non-linear (Bauschinger effect), as shown in
Figure 2.23. The resulting behaviour is thus characterised by reduced energy dissipation in
the 2nd and 4th quadrants.
30
(a) Bond Under Monotonic Loading. Bond between concrete and steel bars is initially
due to chemical adhesion. After a bond stress 0 =0.51.0 MPa, adhesion breaks and slip
between the bar and the surrounding concrete starts taking place; further bond is
provided by friction and wedging action between the cement paste and the pitting of
reinforcing bars. More importantly, in deformed bars, mechanical interlock between the
deformations (ribs or indentations) and the surrounding concrete takes place. The
interlock forces eventually lead to internal bond (tensile) cracks next to the deformations;
at about the same time separation of concrete from the bars takes place in the region of
flexural cracks. Subsequent to separation, forces from the deformations to the
surrounding concrete may lead to splitting cracks, typically parallel to the bars. If these
cracks can propagate without restraint, bond splitting failure occurs.
Figure 2.24 Effect of confining pressure on local bond stress () - slip (s) relationship
The presence of pressure transverse to the bars (due to compressive axial loading and/or
confinement) leads to an increase in bond resistance (Figure 2.24), mainly by suppressing
early splitting cracks.
In the presence of transverse pressure, subsequent to the formation of splitting cracks
(which are prevented from propagating to the surface), gradual deterioration of the
31
concrete keys between adjacent ribs occurs and at a certain point these keys fail in shear
(point of max bond stress in Figure 2.24). In properly confined members a descending
branch of the (bond stress)-(bond slip) relationship can be obtained, along which
concrete between adjacent ribs deteriorates completely; the residual strength is due solely
to friction at the cylindrical surface defined by the tips of the deformations.
(b) Bond Under Cyclic Loading. During cyclic loading at stresses in excess of 70 to
80% of max the envelope of the -s curve is no longer the curve corresponding to
monotonic loading (see Figure 2.25). The reduction in bond resistance is more
pronounced as the values of slip between which cycling takes place increase, as well as the
number of cycles increases (Eligehausen et al., 1983).
Of particular relevance to the seismic performance of members is the drastic reduction in
stiffness and energy dissipation capacity in the 2nd and 4th quadrant of the -s relationship.
This reduction occurs subsequent to partial shearing off of the concrete keys between
successive ribs and is due to slip of the bar in the opposite direction until the ribs come
again into contact with intact concrete (e.g. reloading paths (b) and (c) in Figure 2.25).
32
EC2 :
f ck c
Figure 2.26 Favourable and unfavourable collapse mechanisms in buildings: (a) Beam mechanism
(favourable); (b) Column mechanism (unfavourable)
33
The most favourable collapse mechanism involves hinges at all beam ends and also at the
base of the ground storey columns (the latter cannot be avoided), because:
(a) Flexure-dominated Beams. Hysteresis loops from tests by Bertero et al. (1977), on
beams with shear span ratio M/Vd = 4.5 are shown in Figure 2.27. The transverse
reinforcement of the specimens consisted in 6 mm double hoops at 90 mm spacing.
In these beams, the deflection due to shear deformations does not exceed 10% of total,
and the beams are characterized by:
Large ductility ( 5)
34
The T-beam shows a strength increase proportional to the increase in top reinforcement
(due to the presence of bars in the slab) and increased stiffness, mainly at the first load
cycle.
The main factors contributing to stiffness degradation under cyclic loading (for prevailing
flexural behaviour) are:
(b) Shear-dominated Beams. Typical results from tests (Bertero et al., 1977) on
cantilever beams differing only in the shear span ratio M/Vd = l/d are shown in Figure
2.28, where specimens R-5 and R-6 correspond to l/d ratios of 2.75 and 4.46, respectively.
In beam R-5 the fraction of due to shear deformations is up to 37%, and the beam is
characterized by:
Relatively low ductility ( 3.3), despite the presence of closely spaced hoops
Shear type of failure (shear transfer mechanisms destroyed)
Pinching of hysteresis loops, hence reduced energy dissipation
Figure 2.28 Hysteresis loops for beams with different levels of shear stress
35
36
It is clear that a fundamental requirement for ductile behaviour is to keep shear stress as
low as practically feasible. This can be achieved by using compact sections and/or low
longitudinal reinforcement ratios (high flexural strengths lead to correspondingly high
shears).
37
38
If, in members with high shear stress, open cracks in the tension and compression zone
remain open, shear behaviour is governed by a vertical (full depth) crack, which does not
intersect the hoops, however closely spaced they are. This extremely undesirable
behaviour (leads to brittle shear-type of failure) is known as sliding shear, and is
schematically represented in Figure 2.30.
In such cases, cross-inclined reinforcement (bidiagonal bars) has been found to be
effective for preventing considerable drop in shear resistance; these bars have more
favourable anchorage conditions and are subjected to lower flexural stresses than
intermediate longitudinal bars.
39
existing cracks (due to gravity loading), possibly opened further due to the effect
of the vertical component of the earthquake
these cracks do not jeopardize the overall safety of the building
40
41
(a) Effect of Axial Loading. The effects of compressive axial loading (N < -0.1 Acfc) are
both favourable and unfavourable, as follows:
Favourable effects:
i.
Unfavourable effects:
i.
42
ii. For relatively high levels of inelasticity (say > 4) significant 2nd order (P-)
effects. Effective stiffness reduced, hence risk of collapse (as a rule, physical
collapse of a R/C structure is caused by inelastic instability)
Properly designed columns have sufficient ductility ( 4), substantially larger than that
required in capacity designed members (provided the axial load is not too high, see Figure
2.36). High overturning moments (in combination with the vertical component) might
cause tension in columns. Tensile loads, although not harmful from the ductility point of
view, cause significant stiffness degradation and risk of sliding shear!
Figure 2.36 Hysteresis loops for members with various levels of axial loading (Jirsa et al. 1974)
(b) Short Columns. The shear (span) ratio as=M/(Vh) defines the mode of failure
43
Figure 2.37 Failure modes in short columns (Tegos and Penelis, 1988)
(c) Biaxial Seismic Loading. The seismic force (M or V) along the diagonal ( = 45)
of a square column may be up to 42% higher than that corresponding to uniaxial loading
( = 0) and, at the same time, tests have shown that Mu ( = 45) Mu ( = 0).
Therefore, the column may yield and high shear stress may develop.
44
As can be seen in Figure 2.38(b), the resultant of the response in the two directions
practically coincides with the uniaxial loading curve up to the development of the
maximum shear strength, but it is lower than that in the post-peak range. In addition, the
results of Figure 2.38(b) how also that bi-directional load histories (cyclic loading in one
direction followed by cycling in a rectangular or inclined direction) can cause more
damage than diagonal/inclined loading (the one usually used in tests), since lower
capacity is available.
O-U
O-D
O-B
O-BD
Figure 2.38 Diagonal cyclic loading (Jirsa et al., 1980): (a) hysteresis loops in each direction; (b)
envelope of resultant shear vs. displacement.
45
Significant influence of load history (e.g. diagonal, "square", etc.) which is difficult to
determine
Short-column failure
46
47
(a) Seismic Actions and Internal Forces in the Joint Core Area. It is important to
note that when calculating the internal forces in joints, the beam capacity moments must
take into account the actual areas of reinforcement, and the effects of strain-hardening of
steel and higher than specified fy.
48
Figure 2.40(a) shows the actions in a beam-column joint area, whereas the corresponding
internal forces are indicated in Figure 2.40(b), wherein Cci and Cbi are the resultants of
forces in compression zone whilst Tci and Tbi indicate the resultants of forces in tension
zone.
The following relationships can be derived:
(2.10)
(2.11)
Tbi = n f y Asi
(2.12)
V jh = n f y ( As1 + As 2 ) Vcol
(2.13)
Figure 2.40 Seismic actions and internal forces in the joint core area
The relative magnitude of moments acting on the members framing into the joint
changes during the seismic excitation. An approximate average value for column shear
based on capacity design approach is (Paulay et al., 1978):
Vcol
l1 * l 2
M1 +
M 2*
l 1n
l 2n
=
0.5(l c + lc )
(2.14)
49
Regarding the vertical joint shear, and noting that jh = jv and assuming these shear
stresses to be uniformly distributed along each joint face, the following expression can be
derived:
V jh
b j hc
V jv
b j hb
V jv = V jh
hb
hc
(2.15)
(b) Effect of Cyclic Loading. Shear transfer in joint cores takes place mainly through
the development of a concrete strut and truss mechanisms. Contribution of aggregate
interlock and dowel action is practically negligible, since the shear deformation in the
joint core is not large enough to activate these mechanisms.
As the number of inelastic cycles increases, the contribution of the concrete strut
mechanism decreases. The formation of full depth cracks at the beam column interfaces
leads to a concrete compressive stress equal to nil (Vc = 0), and corresponding significant
increase in bond forces at the beam bars. After the first few cycles at high ductility levels
the horizontal joint shear is carried almost exclusively by the truss mechanism. However,
if the column bars do not yield, a significant portion of the vertical joint shear can be
carried by the vertical component of the strut mechanism, leading to reduced
requirements for vertical shear reinforcement (Paulay et al., 1978).
However, further experimental work (e.g. Meinheit and Jirsa, 1981; Paulay, 1986; Park &
Dai, 1988) lead to conclusions which differ from those given above:
increase in joint shear strength caused by increasing the tie reinforcement is relatively
low
increase in the longitudinal reinforcement of the column leads to negligible increase
in the joint shear strength, and also that the value of column axial loading influences
the shear corresponding to first cracking, but not the shear strength of the joint
anchorage deterioration, rather than exceedance of shear capacity, plays an important
role in joint shear failure
for cyclic loading, the contribution of the concrete strut mechanism should not
necessarily be considered as nil
for low displacement ductility ( = 2.0) the influence of the nominal shear stress (j)
is practically independent of the hoop reinforcement ratio
due to extensive diagonal cracking, crushing of the concrete struts between adjacent
shear cracks occurs at early stages of cyclic loading
Hence, for design purposes, it is noteworthy that the combination of low j and a limited
amount of joint reinforcement is more effective than the combination of high j and very
closely spaced hoops; note also that for longitudinal reinforcement ratios in the beam in
excess of about 1.5%, the required hoops in the joint cause congestion of steel.
50
Finally, it is also important to note that adjacent beams offer restraint of joint core
expansion, and develop axial forces which contribute to joint confinement and that slabs
tend equally to increase both stiffness and strength of joints. However, slab
reinforcement increases negative moment capacity of beams (which might cause column
hinging, especially at high deformation levels) and, at exterior joints, torsion induced by
the slab causes torsional cracking of transverse beams, which tend to become ineffective.
2.1.8 Seismic Behaviour of Walls under Seismic Loading
2.1.8.1 Performance Requirements
The main function of RC walls within a frame is to carry horizontal (seismic) loads. The
main advantages associated to its use are:
Significant increase in the stiffness of the structure. Thus, reduced P- effects,
reduced damage to non-structural elements, increased sense of security for occupants
Load bearing capacity maintained, even after significant cracking (especially if
boundary elements are appropriately confined)
Easier to construct than (highly) ductile frames
Seismic behaviour more reliable than that of frames since formation of unwanted
plastic hinges is avoided and the negative influence of asymmetrically arranged infill
panels is significantly reduced
The use of structural walls is not, however, free of some drawbacks:
Concern with regard to reduced ductility due to effect of shear (reason for reduced q
factors in previous codes, e.g. UBC'88 or CEB)
The above is mostly overcome if appropriate design and detailing is used
(approximately the same q factors for frames and dual systems are specified in EC8
and other modern seismic codes)
Reduced favourable effect (on strength and stiffness) of symmetrically arranged infill
panels, with respect to frames
Architectural problems due to the presence of permanent walls (this is probably the
most critical one, in practical terms)
51
Figure 2.41 Hysteresis loops for walls subjected to monotonic and cyclic loading (Bertero, 1980)
The work by Bertero (1980) noted that buckling in the weak direction (after spalling)
constituted one of principal causes for failure. In addition, the contribution of grid
reinforcement to the flexural capacity of the member was found to be small whilst its
influence on the ductility of the web can be considered as relatively important. Finally, as
observed in Figure 2.41 above, the monotonic loading curve is very close to the envelope
of cyclic loading.
The critical parameter for cyclic shear response is the aspect ratio:
hw / lw 2 slender walls - ductile, flexure dominated behaviour
hw / lw < 1 squat walls - (sliding) shear dominated behaviour
2.1.8.3 Behaviour of Slender Walls
The slender wall of Figure 2.42 has the following characteristics:
i. Barbell section
ii. Grid h = 0.63%, v = 0.29%
iii. Boundary elements: f = 3.67% (long. reinforcement) ; w = 1.35% (transverse
reinforcement)
iv. Axial stress 0 = 0.08fc ; Shear stress max = 0.9 f c (MPa)
52
As can be inferred from the loops in Figure 2.42, if the edge regions are adequately
confined (pseudo-columns at the edges, with closely spaced hoops), slender walls
exhibit a highly ductile behaviour, comparable to that of beams. Also, with reference to
Figure 2.41, it is pointed out that the behaviour of the wall specimens subjected to cyclic
loading was satisfactory; despite their relatively low slenderness (hw / lw = 1.3), they
attained displacement ductility factors of the order of 4, under fully reversed cyclic shear.
It is important also to note that in the walls of Figure 2.41 it was observed that
subsequent to the failure of their web, the edge columns remained in a good condition,
maintaining their gravity load carrying capacity.
The contribution of each mode of deformation (flexural, shear, fixed end rotation due to
slippage of bars at the wall base) to the total displacement of the wall of Figure 2.42, can
be seen in Figure 2.43 for two different sections of the specimen, one close to the base
and one at a larger distance. It is quite clear that the fraction of the horizontal
displacement resulting from shear deformations is rather large, especially in the region
close to the base of the wall (shear displacement v amounts to up to 60% tot).
Figure 2.42 Hysteresis loops for slender wall (hw/lw = 2.4) (Oesterle et al., 1980)
53
Figure 2.43 Contribution of various deformation modes to total deflection: (a) Measured
deformations at 1.8 m from the base; (b) Measured deformations at 0.9 m from the
base.
Figure 2.44 summarises the rotational ductility of walls with various cross-sections,
wherein:
Black shapes: N = 0
54
Figure 2.44 Rotational ductility of walls with various cross-sections (Oesterle et al., 1980)
Below, the common failure modes in slender walls under cyclic loading are summarised:
55
Figure 2.45 Hysteresis loops for squat walls (hw/lw = 0.5): (a) Conventionally reinforced wall; (b)
Wall with bidiagonal reinforcement (Vi is the theoretical flexural strength).
In the conventionally reinforced walls the typical failure mode was sliding shear, although
the moment corresponding to max shear was larger than the flexural overstrength
(Mv=1.25Mu). At a ductility 6.0, 65% of the total deflection at the top was due to
sliding at the base crack, and the drop in shear capacity during the 2nd cycle was drastic.
56
The wall with bidiagonal reinforcement was characterized by more ductility and less
pinching than the conventionally reinforced wall; at 6.0 only 40% of the top
deflection was due to sliding, and the strength drop during cyclic was very small.
It is difficult to define the slenderness limit beyond which sliding shear dominates the
response. Recent tests by Salonikios et al. (1996) have shown that EC8-designed walls
with hw/lw= 1.5, although considered as squat in the code, their cyclic load behaviour is
essentially flexure-dominated. Note that the main practical problem associated with
sliding shear modes is the drastic drop in stiffness, which results in significant drop in
strength during cycling at a given displacement amplitude.
2.1.8.5 Typology of Wall Damage
Existing buildings with walls generally perform better than similar ones without walls.
Only 6.5% of the buildings in Thessaloniki after the 1978 earthquake had wall damage
(requiring repair); this corresponds to only 28% of the buildings with damage (Penelis et
al., 1989).
Some typical damage patterns in walls are summarized in the following.
Shear damage
This is a typical damage pattern (30% of the
cases in the Thessaloniki earthquake).
Diagonal cracking may lead to failure if edge
regions are not designed and detailed as
columns (see next section).
57
Flexure damage
This type of damage is rare, although walls in
old multistorey buildings are typically underdesigned in flexure (single storey approach
used in analysis, i.e. all vertical members are
assumed to be fixed both at the top and at the
bottom).
58
59
60
was by means of nonlinear springs at the member ends (Clough and Johnston, 1966,
Giberson, 1967, and Takizawa, 1976). Among the lumped plasticity constitutive models
proposed, some include stiffness degradation in flexure and shear (Clough and Benuska,
1967, Takeda et al., 1970, and Brancaleoni et al., 1983), pinching under load reversal
(Banon et al., 1981, Brancaleoni et al., 1983), and fixed end rotations at the beam-column
joint interface to simulate the effect of bar pull-out (Otani, 1974, Filippou and Issa,
1988). Such concentrated plasticity approach should be used with care, since accuracy of
the analysis may be compromised whenever users are not highly experienced in the
calibration of the available response curves, needed to characterise the lumped plasticity
elements. The limitations of lumped models are discussed in several studies, such as
Charney and Bertero (1982) and Bertero et al. (1984), amongst others.
The distributed inelasticity modelling describes more accurately the continuous
structural characteristics of reinforced concrete members, requiring simply geometrical
and material characteristics as input data. The constitutive behaviour of the cross-section
can be either formulated according to the classical plasticity theory in terms of stress and
strain resultants, or explicitly derived by discretising the cross section into fibres. The
latter approach, known as fibre modelling, represents the spread of material inelasticity
both along the member length and across the section area, thus allowing an accurate
estimation of the structural damage distribution even in the highly inelastic range.
Quoting Spacone (2001) again, the first elements with distributed nonlinearity were
formulated with the classical stiffness method using cubic hermitian polynomials to
approximate the deformations along the element (Hellesland and Scordelis, 1981, Mari
and Scordelis, 1984). Menegotto and Pinto (1973) interpolated both section deformations
and section flexibilities and accounted for the axial force-bending moment interaction.
Shear effects were first included in the model proposed by Bazant and Bhat (1977).
More recently, alternative flexibility-based formulations have been developed by
Mahasuverachai and Powell (1982), Kaba and Mahin (1984), Zeris and Mahin (1988,
1991), however these posed difficulties with regards to their implementation in FE
programs. To overcome such complications, Ciampi and Carlesimo (1986) proposed a
consistent flexibility-based method for formulating frame member models, later applied
by Spacone (1994) to the formulation of a fibre beam-column element. A detailed
discussion on the differences between stiffness-based and flexibility-based approaches
may be found in Papaioannou et al. (2005), for instance.
For the purpose of the current work, a classical stiffness-based formulation, as developed
by Izzuddin (2001), has been adopted.
2.2.2 Fibre Modelling Approach
In fibre modelling, the sectional stress-strain state of the elements is obtained through the
integration of the nonlinear uniaxial stress-strain response of the individual fibres in
61
which the section is subdivided, distinguishing steel, confined and unconfined concrete,
as illustrated in Figure 2.46. The adopted stiffness-based element cubic formulation then
allows both the representation of the spread of inelasticity along the member length as
well as the implicit incorporation of interaction between axial force and transverse
deformation of the element. The use of a sufficient number of elements per structural
member permits the reproduction of plastic hinge (in their full length), typical of
members subjected to high levels of material inelasticity. The spread of inelasticity across
the section and along the member length is thus achieved without requiring expertise
calibration of any lumped plasticity element.
Gauss
Section b
node A
node B
Gauss
Section a
A
L/2 3
L/2
RC Section
Unconfined
Concrete Fibres
Confined
Concrete Fibres
Steel Fibres
Structural members are represented by means of frame elements, with finite length and
assigned cross-sections. Structural and non-structural inertia mass may also be
introduced, in either lumped or distributed fashion, whilst joint/link elements, defined as
spring-type elements joining coincident locations, can be used to model discontinuous
connections. By means of such element types, a number of different element classes
(columns, beams, walls, beam-column joints, etc.), non-structural components (energy
dissipating devices, inertia masses, etc.) and different boundary conditions (flexible
foundations, seismic isolation or structural gapping and pounding) can be represented.
The fibre-discretization renders possible a realistic modelling of the different materials,
and their distribution, that make up a given member cross-section. The employable
material models may feature different levels of accuracy/complexity in their definition;
the bilinear, the Menegotto-Pinto (1973) and the Monti-Nuti (1992) models are among
the most used models for steel, whilst concrete may be characterized by tri-linear,
nonlinear with constant or variable confinement constitutive laws (see Scott et al., 1982,
Mander et al., 1987). Many other material constitutive laws are available in the literature.
62
attainment of reliable models: a too fine mesh may cause numerical instabilities,
whilst, on the other hand, if the mesh is exaggeratedly coarse the analysis will not be
sufficiently accurate. The meshing of the structure can be optimally carried out by
refining critical structural locations, such as the zones where high inelasticity is
expected or where abrupt changes in the stiffness of joined elements are present, such
as plastic hinges locations, elements connections, and structural boundaries.
The modelling of each structural element should be based on its expected behaviour:
some examples are the linear behaviour of the deck or the modelling of the plastic
hinge length to account for the flexibility of foundations. In this case, the use of
inelastic fibre elements allows the explicit consideration of the spread of inelasticity.
Model accuracy can be improved by using realistic materials property values and by
63
64
The horizontal displacements of the controlled degrees of freedom are calculated and
then applied to the test structure by servo-controlled hydraulic actuators fixed to the
reaction wall. The PsD testing of the bridge was performed using the sub-structuring
technique, in which the piers were physically tested and the deck was numerically
simulated on-line. Further details and references can be found in Pinto et al. (1996),
Pinho (2000), Sullivan et al. (2004), amongst others.
2.2.4.1 Description of the Model
The tested bridge model labelled as B213C consists of three piers 5.6, 2.8 and 8.4 m high
and a continuous deck with four identical 20 m spans. For what concerns the boundary
conditions, the deck is considered to end at the abutments with shear-keys, with the
extremities free to rotate, as shown in Figure 2.47; the deck-piers connections are
assumed to be hinged (no transmission of moments), transmitting lateral forces due to
the engaging of the sub and superstructure in the transversal direction by means of the
aforementioned shear keys.
The piers have rectangular hollow section with 160 mm wall thickness (Figure 2.47). The
minimum diameter of the longitudinal rebars and of the stirrups is 8 mm and 6 mm,
respectively. The reinforcement layout of the pier models are shown in Figure 2.48. The
mechanical characteristics of materials (B500 Tempcore steel with E = 206 GPa for
longitudinal rebars and C25/30 concrete) and the mechanical characteristics of the pier
cross-sections are shown in Table 2.4 and Table 2.5.
Pier 1
Pier 2
Pier 3
2.8 m
2.8 m
2.8 m
20 m
20 m
20 m
20 m
1.2 m
5.6 m
0.12 m
0.12 m
2.6 m
0.7 m
0.16 m
0.8 m
0.1 m
1.6 m
Deck
Pier
Figure 2.47. Bridge configuration and member cross sections
20 6
65
6 12
14 14
14 10
20 8
Section Type 1
Section Type 4
Diameter
(mm)
(MPa)
(MPa)
(%)
(%)
6
8
10
12
14
363.7
503.4
489.3
558.2
477.2
430.4
563.0
572.3
646.8
577.7
0.177
0.244
0.238
0.271
0.232
15.10
12.30
14.50
12.80
13.00
Hardening
0.0022
0.0024
0.0028
0.0034
0.0038
Table 2.5. Summary of the pier cross section characteristics of the bridge (Guedes, 1997)
Pier
Section
Type
Height
(m)
Longitudinal steel
(%)
Pier 1
Pier 2
Pier 3
4
1
4
14
7
21
1.15
0.50
1.15
37.0 / 3.1
41.2 / 3.1
50.5 / 3.1
(Compressive/tensile)
Table 2.6. Deck cross section geometrical and mechanical characteristics (Guedes 1997)
EA
EI2
EI3
GJ
(kN)
(kNm2)
(kNm2)
(kNm2)
2.7837E+07
1.3544E+07
5.6517E+07
2.8017E+07
The deck is a hollow-core pre-stressed concrete girder 5.6 m wide, as depicted in Figure
2.47. In the PsD test, it was simulated numerically with 32 linear elastic Timoshenko
eccentric beam elements, whose mechanical characteristics are presented in Table 2.6,
where A is the cross-section area, I2 and I3 are the two moments of inertia with respect
to the local principal axes, J is the torsional constant and E is the Young Modulus of 25
66
GPa. The inertia characteristics of the deck are based on a specific weight of 25 kN/m3.
As the deck is assumed to behave elastically, the sub-structured part included a Rayleigh
damping matrix, featuring a damping ratio = 0.016 associated to the two lower
transversal natural frequencies of the complete bridge.
At the top of each pier, an axial force N = 1700 kN was applied by means of actuators,
so as to simulate the vertical load that is transmitted from the deck. The input ground
motion was represented by an adequately scaled accelerogram with duration of 4 seconds
and a nominal peak acceleration of 0.875 g, as shown in Figure 2.49. Two pseudodynamic tests were performed on the structure: one with the input motion corresponding
to the design earthquake and another defined on the basis of the estimated ultimate
capacity of the bridges, and thus equal to 1.2 times the design earthquake.
67
when the transversal size of the deck section is large with respect to the pier height, thus
requiring an additional rotational DOF of the deck, (ii) the top concentrated mass
assumption may be no longer acceptable when piers are massive with respect to the deck,
(iii) the soil-structure interaction can be neglected or modelled with different levels of
complexity, (iv) the influence of the shear deformation needs adequate analytical
characterisation on squat members, for which shear collapse modes and flexure-shear
interaction are relevant, (v) the penetration of plasticization at the base of piers may be
modelled by extending the actual pier length, (vi) the influence of the spatial variability
and/or loss of coherence of the ground motion may be represented by means of
asynchronous input definition, (vii) connections among foundations, piers and deck can
be modelled with different levels of complexity and detail.
In what follows, a description of geometry and discretisation of the model, its element
connections, boundary conditions and loading state is given Adopted nonlinear analysis
procedures and convergence criteria are also explained in some detail.
(a) Modelling the Bridge Piers. As discussed previously, the piers are the elements in
which inelastic deformation will be concentrated, therefore, a good level of accuracy in
the characterization of materials and in the discretization of the mesh should be ensured.
The piers have thus been modelled through a 3D inelastic beam-column element capable
of capturing geometric and material nonlinearities. The number of fibres used in section
equilibrium computations was set to 400; the selection of such number guarantees an
adequate reproduction of the stress-strain distribution across the element cross-section,
considering the shape and material characteristics of the latter, and the degree of material
inelasticity that it is likely to reach.
The pier cross-section has thus been defined through a RC rectangular hollow section of
0.8 m x 1.6 m, with a wall width of 0.16 m, a concrete cover of 8 mm, and a steel layout
reproducing the test specimen of Figure 2.48. The specimen sections contain steel rebars
with different mechanical properties, as illustrated in Table 2.4, however, given the
possibility of specifying only one steel material per section in the computer code used, an
equivalent steel has been defined for each section (see Table 2.7), weighting its properties,
i.e. yielding strength fy and strain hardening parameters, proportionally to the distance
from the sectional centre of gravity and to the area of each rebar.
Table 2.7. Equivalent steel properties per section
Section
Yield
strength
Hardening
section 1
section 4
468 MPa
496 MPa
0.0027
0.0036
68
The stress-strain behaviour of the employed reinforcing steel (see Figure 2.50, left) has
been described by the nonlinear model of Menegotto and Pinto (1973), as modified by
Filippou et al. (1983) to include isotropic strain hardening. This is an accurate and
convenient model, due to its computational efficiency and its very good agreement with
experimental results. It utilises a damage modulus to represent more accurately the
unloading stiffness, and has been modified and improved by Fragiadakis et al. (2006) to
attain better stability and accuracy. The concrete has been represented through a
nonlinear constant confinement concrete model (Figure 2.50, right), as a good
compromise between simplicity and accuracy: it is an uniaxial nonlinear model following
the constitutive relationship proposed by Mander et al. (1988), later modified by
Martinez-Rueda and Elnashai (1997) for reasons of numerical stability under large
deformations. The constant confinement factor is defined as the ratio between the
confined and unconfined compressive stress of the concrete. The model calibrating
parameters, fully describing the mechanical properties of steel and concrete, have been
set as shown in Table 2.8 and Table 2.9, where the concrete cylinder strength has been
estimated as being equal to 85% of the cubic resistances listed in Table 2.9.
-54
800
600
-44
Stress (MPa)
Stress (MPa)
400
200
0
-200
-34
-24
-14
-400
-4
-600
-800
-0.01
0.00
0.01
0.02
0.03
0.04
6
0.0005
-0.0005
-0.0015
Strain (mm/mm)
Figure 2.50. Menegotto-Pinto steel model, with Filippou isotropic hardening (left), and nonlinear
constant confinement concrete model (right)
Table 2.8. Parameters for the Menegotto-Pinto steel model, with Filippou isotropic hardening
Parameter
Sec 1
Sec 4
203000
468
0.0027
20
18.5
0.15
0.025
2
203000
496
0.0036
20
18.5
0.15
0.025
2
69
Table 2.9. Parameters for the nonlinear constant confinement concrete model
Parameter
Pier 1
Pier 2
Pier 3
31.5
3.1
35.0
3.1
42.9
3.1
0.002
0.002
0.002
1.2
1.2
1.2
After the PsD testing of the bridges, the tall and the medium piers were tested cyclically
until failure (Pinto et al., 1996, Guedes, 1997), respectively up to 230 and 150 mm of top
displacement, under the imposed displacement history shown in Figure 2.51. Additional
cyclic tests (up to 72 mm at the top of the pier) were carried out on a short pier similar to
the one tested in pseudo-dynamic fashion. These cyclic tests on the piers are numerically
reproduced herein through a static time-history analysis, so as to enable a first check on
the accuracy of the model being assembled. The numerical reproduction of the cyclic test
has been performed imposing on the piers the displacement history resulting from the
PsD test (Figure 2.51). In addition, the steel young modulus of the medium-height pier
was halved, as suggested by Pinto et al. (1996), in order to reproduce the reduction in the
stiffness due to the shear damage that this pier suffered prior to this cyclic test (it is
recalled that these cyclic tests were carried out after the PsD testing); no reduction in the
steel properties was applied to the tall pier, as it was not damaged during the PsD test.
0.25
0.20
Displacement (m)
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
-0.25
Med Pier
Tall Pier
Short Pier
70
Figure 2.52 to Figure 2.54 show a very good match between experimental and numerical
results for all the piers; only the reduction in member strength at the very last cycle, when
failure occurs, is not perfectly captured.
0.6
Experimental
0.4
Numerical
0.2
0
-0.2
-0.4
-0.6
-0.28
-0.21
-0.14
-0.07
0.07
0.14
0.21
0.28
Displacement (m)
(b) Modelling the Bridge Deck. Normally, the deck can be modelled as linear elastic,
since this is typically the behaviour of real bridges under seismic actions: the deck in fact
is generally pre- and/or post-stressed, which means that no damage nor plastic
deformations are allowed to occur. Moreover, in the case of isolated bridges, the deck is
protected by the isolating system, and it is hardly damaged.
The deck has been modelled with a 3D elastic nonlinear beam-column element, still
capable of modelling local geometric nonlinearities. This type of element is fully
described by the sectional properties values, based on geometric and mechanical
characteristics. In the current application, the Young and shear modules have been taken
respectively as 25 and 10 GPa, and the element parameters have been set as listed in
Table 2.6.
71
0.90
Experimental
0.60
Numerical
0.30
0.00
-0.30
-0.60
-0.90
-0.16
-0.12
-0.08
-0.04
0.00
0.04
0.08
0.12
0.16
Displacement (m)
1.2
Numerical
0.6
0
-0.6
-1.2
-1.8
-0.08
-0.06
-0.04
-0.02
0.02
0.04
0.06
0.08
Displacement (m)
The deck has been located at the height of its centre of gravity, 0.602 m above the pier
top, and connected to it by means of a rigid element. The deck can be either modelled as
described, or located right at the top of the pier, provided that the moment of inertia with
respect to the horizontal axis is translated to that location. In the analysed case this brings
to similar results, but the authors opinion is that the first choice is preferable, in order to
model more accurately the deck displacement in case of non-rigid connections to the
pier, e.g. when a relative rotation of the connection contributes to the drift proportionally
to its vertical location.
72
Before carrying out the nonlinear dynamic analyses, eigenvalue analysis has been run, in
order to compare the first transversal modes of the numerical structure (Table 2.10) with
the initial dynamic characteristics of the bridge specimen, computed analytically: the
match between test and numerical results is precise for the first mode (T = 0.183 s), and
fairly good also for the other two transversal modes.
Table 2.10. First three modal shapes in the transversal direction of the deck
Mode
Period (s)
Test
Analysis
1st
0.183
0.183
2nd
0.146
0.148
3rd
0.085
0.076
Modal shapes
(d) Other Modelling Details. Rigid connections have been modelled either through
elastic frame or link elements: to represent an infinitely stiff connection, avoiding
numerical difficulties, the stiffness of those elements is set 100 to 1000 times that of
adjacent elements. The rigid arm connecting the top of the pier and the centre of gravity
of the deck is constituted by an elastic element, whilst the connection between the base
of the rigid arm and the top of the pier is modelled as a hinge. The link element
73
representing the latter connects two initially coincident structural nodes and requires the
definition of an independent force-displacement (or moment-rotation) response curve for
each of its local six DOF: in order to model the engaging of the sub and superstructure
of the pier-deck connection in the transversal direction by mean of shear keys, the link
element is set as a spring with infinite stiffness in the vertical and transversal direction,
and fully flexible in all the other four DOF. Boundary conditions are defined as restraints
in global coordinates: the deck is simply supported at the first end and hinged at the
other, whilst piers are fully fixed at the base. In Figure 2.55, the above described
connection details are depicted.
direction
Distributed
mass
mass
Longitudinal
direction
Lumped
Mass
Rigid Arm
Deck
Pier-Deck Lateral
Hinge
Pier
Mesh
Nodes
Distributed
mass
Modal and dynamic analyses require as a matter of necessity the definition of the masses,
which can be either lumped nodal masses or distributed. The piers were characterized
with a distributed mass element of 1.664 ton/m, whilst the deck mass was concentrated
at the top of each pier for an amount corresponding to the tributary deck length (56 tons
per pier), and distributed at the two half span deck extremities with an amount of 2.784
ton/m.
Equivalent viscous damping, with values typically ranging around 1 2%, is customarily
introduced in order to somehow represent minor energy dissipation mechanism other
from the hysteretic ones (e.g. friction across cracks, radiation through foundations, and so
on). This usually involves the employment of Rayleigh damping matrices, whereby
damping is defined as proportional to the mass and stiffness of the structural members
(see Clough and Penzien, 1994). However, given the uncertainties associated to the
quantification of such equivalent viscous damping, and considering also the recent doubts
raised with regards to the employment of Rayleigh damping in nonlinear dynamic
74
analyses (Hall, 2005, Priestley and Grant, 2005), this minor source of dissipation has been
conservatively neglected in the numerical simulations.
A static load of 1700 kN has been applied at the top of each pier, representing the deck
weight according to the test setup, whilst the time history described in Figure 2.49,
including the 10 s interval with no acceleration (needed to damp out the structure motion
after the first earthquake run), has been imposed at the pier bases and at the abutments.
In this manner, the cumulative damage effects caused by the testing of the same structure
under to successive earthquake input motions are adequately modelled.
The time step for the dynamic analysis has been selected as 0.004 s, coincident with the
input record sampling time step (hence the input motion is accurately considered), and
sufficiently small with respect to the dominant vibration period of the structure (0.4 secs),
so as to guarantee numerical stability. For what concerns the nonlinear solution
algorithm, the Hilber-Hughes-Taylor (1977) integration scheme was employed, associated
to a displacement and force based convergence criterion.
2.2.4.3 Comparisons between Numerical and Experimental Results
The current section presents the comparisons between numerical and experimental
results, in terms of displacements and forces at the top of the short, medium and tall piers
observed when the bridge was subjected to the second and stronger earthquake input
motion. Figure 2.56 to Figure 2.58 show results of the top displacement (left) and the top
shear (right): there is a good agreement, in terms of both the amplitude and the frequency
content of the response. Table 2.11 illustrates the ratios of the maximum absolute
response obtained from numerical calculation to that from tests. It is noted that the force
response of the squat pier is not reproduced with full accuracy, whereas displacements are
instead very well predicted. The numerical overestimation of the action at the top of the
short pier can be explained by the fact that the fibre-based element formulation employed
did not feature the possibility of modelling shear flexibility (or section torsion/warping):
the fact that the stiffness of this pier is not reduced, as it would be due to the shear
damage.
60
Experimental
45
Experimental
Numerical
500
Numerical
30
Top Shear (KN)
75
15
0
-15
250
0
-250
-30
-500
-45
-60
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
-750
4.0
0.0
0.5
1.0
1.5
Time (s)
2.0
Time (s)
2.5
3.0
3.5
4.0
Figure 2.56. Tall pier top displacements (left) and top shear (right)
Table 2.11. Ratios of the absolute maximum response obtained from numerical calculation to that
from tests
Displacement
Top Shear
Tall
Pier
Med
Pier
Short
Pier
88%
90%
94%
95%
102%
188%
60
750
Experimental
Experimental
Numerical
500
Numerical
30
Top Shear (KN)
45
15
0
-15
250
0
-250
-30
-500
-45
-60
-750
0.0
0.5
1.0
1.5
2.0
Time (s)
2.5
3.0
3.5
4.0
0.0
0.5
1.0
1.5
2.0
Time (s)
2.5
Figure 2.57. Medium pier top displacements (left) and top shear (right)
3.0
3.5
4.0
76
1750
60
Experimental
Experimental
Numerical
1250
Numerical
30
750
Top Shear (KN)
45
15
0
-15
250
-250
-30
-750
-45
-1250
-60
-1750
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0.0
0.5
1.0
1.5
Time (s)
2.0
Time (s)
2.5
3.0
3.5
4.0
Figure 2.58. Short pier top displacements (left) and top shear (right)
77
Figure 2.59. Elevation and plan views of the frame, after Carvalho et al. (1999)
The columns were non-ductile, smooth reinforcing bars were used, capacity design
principles were ignored and lap splicing occurred in critical regions (Figure 2.60). Detailed
information on the frame and the set up of the experimental test can be found in Pinto et
al. (2002) and Pinho and Elnashai (2000).
78
Figure 2.60. Reinforcement detail of the columns, after Carvalho et al. (1999)
Figure 2.61. Scheme of vertical loads for nonlinear analysis, after Carvalho et al. (1999).
79
The materials considered at the design phase were a low strength concrete class C16/20
(CEN, 1991) and smooth reinforcement steel class Fe B22k (Italian standards). The latter
refers to smooth bars with a yield stress of 235 MPa and ultimate strength of 365 MPa.
The vertical loads considered in the design consisted of the self-weight of the slab and
transverse beams, finishes, infill walls and the quasi-permanent static load. Figure 2.61
shows the scheme of vertical loads applied to the structure.
2.2.5.2 Modelling of the Structure in the FE Program
As in the previous case-study, the finite element analysis program SeismoStruct
(SeismoSoft, 2005) is employed to run all analyses. Structural members have been
discretised through the use of beam-column elements, which are recalled to take account
of geometrical nonlinearity and material inelasticity. Also as discussed before, since a
constant generalized axial strain shape function is assumed in the adopted cubic
formulation of the element, it results that its application is only fully valid to model the
nonlinear response of relatively short members and hence a number of elements (usually
three to four per structural member) is required to accurate model the structural frame
members.
In total, 112 inelastic frame elements, capable of representing progressive cracking and
spread of inelasticity, are used to model the RC frame. All member sections are
represented explicitly and, for the purpose of strain/stress evaluation, are subdivided into
a number of fibres (200 300) varying according to the section size. The length of the
elements varies according to their location, with smaller elements being used in the
vicinity of beam-column connections where large levels of inelastic deformation are
expected.
All connections are assumed rigid and fully-fixed boundary conditions are adopted at the
base of the building. Effective slab widths of 1.0 m and 0.65 m were adopted for the long
and short spans, respectively, following the formulae presented in Eurocode 2 (CEN,
1991).
Vertical loads and masses are applied at each beam node and at the beam-column joints
mirroring the load and mass distribution presented previously in Figure 2.61.
Concrete is represented by a uniaxial constant confinement model (Mander et al., 1988)
and is calibrated using the concrete characteristics values obtained during testing. As for
the reinforcement bars, the Menegotto-Pinto (1973) steel model, with an isotropic
hardening constitutive relationship (Filippou et al., 1983), was adopted.
It is worth mentioning that shear strains across the element cross section are not
modelled; in addition, warping strains and warping effects are not considered in the
current formulation, either. Additionally, the elastic torsional rigidity is used in the
80
formulation of the nonlinear frame elements; this clearly involves some degree of
approximation for the case of reinforced concrete sections.
No viscous damping was considered in any dynamic analysis, since energy dissipation
through hysteresis is already implicitly included within the nonlinear fibre model
formulation of the inelastic frame elements, and non-hysteretic type damping was
assumed to be negligible within the scope of the present endeavour (see also previous
discussion on this issue).
2.2.5.3 Nonlinear Dynamic Analysis
(a) Earthquake Input. A collection of artificial records was available for use in the
pseudo-dynamic experimental programme. The records were derived following a
probabilistic seismic hazard analysis carried out by Campos-Costa and Pinto (1999). For
the purpose of the experimental programme, the severity class Moderate-High, typical
of Southern European countries, was chosen. A set of hazard-consistent time histories
was artificially generated to fit the uniform risk spectra (URS) for return periods of 100,
475, 975 and 2000 years.
1.20
300
1.00
250
Displacement [mm]
Acceleration [g]
In Figure 2.62, the acceleration and displacement elastic response spectra for all
accelerograms, computed for an equivalent viscous damping of 5 %, are shown. These
indicate peak acceleration response for periods of vibration of up to 0.5 seconds, with
values ranging from 0.3 g to 1.1g. Figure 2.63 shows the artificial Acc-475 (475 years
return period) and Acc-975 (975 years return period) accelerograms; only these two
records will be considered in the nonlinear dynamic and pushover analyses carried out
herein. As is common with artificial records, a wide range of frequencies is present in the
accelerograms.
0.80
2000 years
975 years
475 years
100 years
0.60
0.40
0.20
0.00
2000 years
975 years
475 years
100 years
200
150
100
50
Period [s]
(a)
Period [s]
(b)
Figure 2.62. Response spectra of input motion: (a) acceleration; (b) displacement
81
0.4
0.3
ACC-475
ACC-975
0.3
0.2
0.2
0.1
0.1
0.0
0.0
-0.1
-0.1
ti
ti
-0.2
-0.2
( )
( )
-0.3
-0.3
0
10
12
14
Time (s)
(a)
10
12
14
Time (s)
(b)
Figure 2.63. Artificial acceleration time histories for (a) 475 year (Acc-475) and (b) 975 year (Acc-975)
return period
82
DISPLACEMENTS (mm)
100
50
-50
-100
0
10
12
14
10
12
14
TIME (S)
EXPERIMENTAL
ANALYTICAL
(a)
TOP LEVEL DISPLACEMENTS
150
DISPLACEMENTS (mm)
100
50
-50
-100
-150
0
8
TIME (S)
EXPERIMENTAL
ANALYTICAL
(b)
Figure 2.64. Analytical and experimental top frame displacement: (a) Acc-475; (b) Acc-975
83
all other analytical analyses will be compared. Note that no post-test calibration has been
carried out. The structure was modelled as it is. The differences between analysis versus
experiment are likely to be due to the fact the third storey developed a shear failure
mechanism, not yet incorporated in the program used.
Storey
Analytical
2
Experimental
0
0
0.2
0.4
0.6
0.8
Drift (%)
(a)
4
Storey
Analytical
2
Experimental
0
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
2.2
2.4
2.6
Drift (%)
(b)
Figure 2.65. Analytical and experimental drift profiles: (a) Acc-475; (b) Acc-975
84
Figure 2.66. Picture of the full-scale, three storey prototype building (Fardis & Negro, 2006)
Plan
C5
Elevation
C1
B1
B11
85
C2
B2
B9
B7
C9
C3
C4
B3
B4
B10
B12
B8
C6
C7
C8
B6
B5
Figure 2.67. Plan and elevation view of the full-scale, three-storey prototype building
The model consists of seven RC columns (0.25 x 0.25 m), one RC column (0.25 x 0.75)
and RC beams (0.25 x 0.50 m).
Concrete material has the below given properties:
(confined): fc = 25000 kPa; ft = 0.001 kPa; c = 0.002 m/m; kc = 1.0
(unconfined): fc = 25000 kPa; ft = 0.001 kPa; c = 0.002 m/m; kc = 1.001
Two various kinds of steel are considered in the model (both use Menegotto-Pinto steel
model):
86
Plan
Lumped Mass
Dynamic
time-history load
Elastic frame element
Dynamic
time-history load
Constraint
(Rigid diaph.)
Mesh Nodes
Dynamic
time-history load
Inelastic
frame
element
87
Figure 2.69. 3D SeismoStruct model of the full-scale, three storey prototype building
In order to run the dynamic time-history analysis, a time-history curve, one for each
direction (X and Y), is loaded. The two curves are defined as follows:
12,000
10,000
8,000
6,000
4,000
Acceleration
2,000
0
-2,000
-4,000
-6,000
-8,000
-10,000
-12,000
-14,000
0
10
Time
15
88
12,000
10,000
8,000
Acceleration
6,000
4,000
2,000
0
-2,000
-4,000
-6,000
-8,000
0
10
15
Time
The time step for the dynamic time-history analysis is set to 0.01 s. The dynamic timehistory loads are applied at the base nodes, in terms of accelerations in X and Y
directions. Incremental dynamic analysis has been applied.
89
4
Experimental
Roof displacement
Analytical
2
1
0
-1
-2
-3
0
10
12
14
16
Time
Figure 2.72 Comparison between experimental and analytical results for the SPEAR building
90
loading using the E-W component of the Friuli earthquake. The model is defined in X-Z
plane. Detailed informations about the material properties and the test arrangements can
be found in Biondi et al. (2000).
91
Strut curve parameters (inf strut response curve): Initial Young modulus Em =
4088000, compressive strength fm = 1500, tensile strength ft = 0.001, strain at
maximum stress em = 0.002, ultimate strain eu = 0.005, closing strain ecl =
0.0000001, strut area reduction strain e1 = 0.001, residual strut area strain e2 =
0.002, starting unloading stiffness factor gu = 2.0, strain reloading factor ar = 1.5,
strain inflection factor ach = 0.6, complete unloading strain factor ba = 2.0,
stress inflection factor bch = 0.6, zero stress stiffness factor gpu = 1.0, reloading
stiffness factor gpr = 1.1, plastic unloading stiffness factor ex1 = 1.5, repeated
cycle strin factor ex2 = 1.0
Shear curve parameters (inf shear response curve): Shear bond strength = 150,
friction coefficient = 0.4, maximum shear resistance = 1000, reduction shear
factor = 1.46
Panel thickness (t): 0.05 m
Out-of-plane failure drift: 40% of ver. panel side
Strut Area 1 (A1): 4 m2
Strut Area 2 (A2): 4% of A1
Equivalent contact length (hz): 50% of vert. panel side
Horizontal and vertical offsets (xo and yo): 0.015% of horiz. panel side and
0.0396% of vert. panel side
Proportion of stiffness assigned to shear: 70%
Specific weight: 0 kN/m3
92
Permanent load
Permanent load
250 kN
250 kN
Displacement load
Inelastic infill
panel element
Inelastic frame
element
Mesh Nodes
Figure 2.74. FE model of the one storey, single bay infilled frame (Colangelo, 1999)
Load Factor
In order to run the nonlinear static time-history analysis, a time-history curve is loaded, as
shown below in Figure 2.75. The time step for the static analysis is selected as
0.00700508s.
0.016
0.014
0.012
0.01
0.008
0.006
0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
-0.012
-0.014
-0.016
-0.018
-0.02
-0.022
0
Time
93
Two permanent loads of 250 kN each are applied at in Z negative direction in terms of
forces in order to simulate the gravity loads, whereas a static time-history load, in terms
of displacements, is imposed in X direction. Nonlinear static time-history analysis has
been used.
2.2.9 Comparison of experimental and analytical results
Experimental and analytical results are compared in the following plot in Figure 2.76. A
fairly good agreement between analytical and experimental results is observed.
94
Plan
Elevation
95
The infill panels are modelled through a four-node masonry panel element (inelastic infill
panel element). This element is defined again in Seismostruct as structs.
The connections between slabs and beam-column joints are modelled with link elements,
in which for F1, F2, and F3 degrees of freedom a linear symmetric response curve is
chosen with a value of stiffness k0 equal to 1.0000E+008, whereas for M1, M2 and M3
degrees of freedom the stiffness k0 is set to 0.01.
Regarding the masses, they are applied to the structure in lumped fashion (for the three
translational inertia values of each lumped mass refer to the SeismoStruct model).
In the Restraints dialog box, all foundation nodes are considered as fully restrained
against rotations and translations.
The FE model of the building is presented below:
Dynamic time-history analysis model
Y
Dynamic
time-history load
Plan
Lumped
Masses
Elastic frame
element
Inelastic
frame
element
Dynamic
time-history load
Mesh Nodes
Infill panel
Dynamic
time-history load
Dynamic
time-history load
In order to run the dynamic time-history analysis, a time-history curve is loaded in the
software.The previous values of acceleration are in units of g, so they will be multiplied
by 0.015 m/s2 in the model. The time step for the dynamic time-history analysis is set to
0.00250187 s. Only dynamic time-history loads are applied at the base nodes, in terms of
accelerations in X direction. Pseudo-dynamic analysis and dynamic time-history analysis
have been applied.
96
97
98
99
Within the framework of earthquake engineering, pushover analysis is employed with the
objective of deriving, with relative ease, an envelope of the response parameters that
would otherwise be obtained through a much more complex and time-consuming
Incremental Dynamic Analysis (IDA) procedure, as can be construed by Figure 2.81.
IDA is a parametric analysis method by which a structure is subjected to a series of
nonlinear time-history analyses of increasing intensity (Vamvatsikos and Cornell, 2002),
with the objective of attaining an accurate indication of the true dynamic response of a
structure subjected to earthquake action.
120
100
80
60
40
20
0
0
0 .5
1.5
2.5
drift (%)
Figure 2.81. Maximum base-shear and top displacement values obtained with incremental dynamic
analysis
100
Recent years have also witnessed the development and introduction of an alternative type
of nonlinear static analysis, which involves running multiple pushover analyses separately,
each of which corresponding to a given modal distribution, and then estimating the
structural response by combining the action effects derived from each of the modal
responses (i.e. each displacement-force pair derived from such procedures does not
actually correspond to an equilibrated structural stress state). Paret et al. (1996) first
suggested the Multi-Modal Pushover procedure, which was then refined by Moghadam
and Tso (2002). Chopra and Goel (2002), on the other hand, have developed and
proposed a Modal Pushover Analysis (MPA) technique, which Hernndez-Montes et al.
(2004) have then adapted into an Energy-based Pushover formulation. A further
refinement of such multiple-pushover procedures, with the aim to account for the
alteration of local resistance and modal characteristics of the structure induced by the
accumulation of damage, consists in the employment of adaptive updating of the loading
pattern (Gupta and Kunnath, 2000; Aydinoglu, 2003), effectively meaning that the
methods may now be categorised as piecewise linear response spectrum analysis. As
highlighted by their respective authors, the main advantage of this category of static
analysis procedures is that they may be applied using standard readily-available
commercial software packages. The associated drawback, however, is that the methods
are inevitably more complex than running a single pushover analysis, as noted by Maison
(2005). Furthermore, some of the proposed multiple-run procedures, either they have
an adaptive or non-adaptive nature, lead to difficulties when applied within capacityspectrum type of procedure due to the need to handle capacity curves associated with
higher mode force patterns that display a reversal of the roof displacement as inelasticity
develops in the structure (Hernndez-Montes et al. 2004; ATC 2005; Goel and Chopra
2005b). For all of the above, these multiple-pushover based approaches do not constitute
the scope of the current presentation, where focus is placed instead on single-run
pushover analysis procedures.
In tandem with the present drive for performance-based seismic engineering, there is also
currently a thrust for the development and code implementation of displacement or,
more generally, deformation-based design and assessment methods. Therefore, it would
seem that applying displacement loading, rather than force actions, in pushover
procedures would be an appropriate option for nonlinear static analysis of structures
subjected to earthquake action. However, due to the unvarying nature of the applied
displacement loading vector, conventional (non-adaptive) displacement-based pushover
analysis can conceal important structural characteristics, such as strength irregularities and
soft storeys, should the displacement pattern adopted at the start of the analysis not
correspond to the structures post-yield failure mechanism. Consequently, when only
non-adaptive static nonlinear analysis tools are available, as has been the case throughout
the past, force-based pushover does constitute a preferable choice over its displacementbased counterpart.
101
On the other hand, however, if one is able to apply displacements, rather than forces, in
an adaptive fashion, that is, with the possibility of updating the displacement loading
pattern according to the structural properties of the model at each step of the analysis,
then a conceptually appealing deformation-based nonlinear static analysis tool is obtained.
2.3.2 Recent Developments in Single-Run Pushover Analysis
According to recently introduced code provisions, such as FEMA-356 (BSSC, 2000) and
Eurocode 8 (CEN, 2002), pushover analysis should consist of subjecting the structure to
an increasing vector of horizontal forces with invariant pattern. Both the force
distribution and target displacement are based on the assumptions that the response is
controlled by the fundamental mode and the mode shape remains unchanged until
collapse occurs. Two lateral load patterns, namely the first mode proportional and the
uniform, are recommended to approximately bound the likely distribution of the inertia
forces in the elastic and inelastic range, respectively.
However, a number of recent studies, summarised in the FEMA-440 report (ATC, 2005),
raise doubts on the effectiveness of these conventional force-based pushover methods in
estimating the seismic demand throughout the full deformation range: (i) inaccurate
prediction of deformations when higher modes are important and/or the structure is
highly pushed into its nonlinear post-yield range, (ii) inaccurate prediction of local
damage concentrations, responsible for changing the modal response, (iii) inability of
reproducing peculiar dynamic effects, neglecting sources of energy dissipation such as
kinetic energy, viscous damping, and duration effects, (iv) difficulty in incorporating
three-dimensional and cyclic earthquake loading effects.
In Figure 2.82 and Figure 2.83, examples of inadequate prediction of both the capacity
curve as well as the deformation response characteristics of a 12-storey reinforced
concrete frame subjected to a natural earthquake recording (case-study RM15-NR2 in
Antoniou and Pinho (2004a)) and of a 4-storey irregular frame subjected to an artificial
accelerogram (ICONS full-scale test specimen, described in Pinho and Elnashai (2000))
are given. It is noted that although the 12-storey building is regular in height, its response
is heavily influenced by higher mode effects, effectively rendering its seismic behaviour
highly irregular in height, as conspicuously shown by Figure 2.82a. The standard
pushover results have been carried out using both triangular and uniform loading
distributions, and are compared with the envelope of results obtained with incremental
dynamic analysis.
102
7000
6000
5000
4000
dynamic
DAP
uniform
triangular
3000
2000
1000
0
0%
1%
2%
total drift
3%
Figure 2.82. Capacity curves of a 12-storey building, obtained with standard pushover
12
DYNAMIC
10
DYNAMIC
TRIANGULAR
TRIANGULAR
UNIFORM
UNIFORM
3
STOREY
STOREY
8
6
4
1
2
0
0.00%
2.00%
4.00%
INTERSTOREY DRIFT (% )
6.00%
0
0.00
0.50
1.00
1.50
2.00
INTERSTOREY DRIFT (% )
(a)
(b)
Figure 2.83. Interstorey drift profiles of (a) 12-storey building and (b) 4-storey irregular frame,
obtained with standard pushover.
The main reason behind the underperformance of these conventional pushover methods
is the fact that they do not account for the effect that damage accumulation, induced by
the increasing deformation levels imposed on the structure, has on the response of the
latter. Cumulative material straining introduces a reduction in stiffness, which, in turn,
causes an elongation of the periods of vibration (Figure 2.84), which then, depending on
the shape of the response spectrum being considered (or on the frequency content of an
input record), may trigger significant changes in the response characteristics of the
buildings (Figure 2.85). Krawinkler and Seneviratna (1998) summarised the above with a
single statement: fixed load patterns in pushover analysis are limiting, be they first modal
103
6
first mode
second mode
third mode
0
0.0%
0.5%
1.0%
1.5%
2.0%
total drift
Figure 2.84. Periods of vibration of 4-storey building under increasing levels of deformation
12
10
8
6
2
0
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
Figure 2.85. Interstorey drift profiles of a 12-storey building subjected to increasing levels of
deformation
104
5000
4000
3000
2000
1000
0
0%
1%
2%
total drift
3%
Figure 2.86. Adaptive pushover: shape of loading vector is updated at each analysis step
F t+ F t
Ft
F k+ F k
Fk
Kt
O 't
Kk
O 'k
dk d k+ d k
dt
d t+ d t
Figure 2.87. The use of tangent stiffness in updating (i.e. incrementing) the loading vector
105
DYNAMIC
DYNAMIC
FAP-SRSS
10
FAP-SRSS
3
STOREY
STOREY
8
6
4
1
2
0
0.00%
2.00%
4.00%
INTERSTOREY DRIFT (% )
(a)
6.00%
0
0.00
0.50
1.00
1.50
2.00
INTERSTOREY DRIFT (% )
(b)
Figure 2.88. Interstorey drift profiles of (a) 12-storey building and (b) 4-storey irregular frame,
obtained with Force-based Adaptive Pushover using SRSS modal combination
With the above in mind, Kunnath (2004) and Lpez-Menjivar (2004) have proposed an
alternative modal combination scheme, consisting of a weighted Direct Vectorial
Addition (DVA) of the different modal shapes that can be mathematically expressed as:
106
Fi = j j j ,i M j Sa j
(2.16)
j =1
where i is the storey number, j is the mode number, n is the highest mode of interest, j
is the modal participation factor for the jth mode, i,j is the mass normalised mode shape
value for the ith storey and the jth mode, Mi is the mass of the ith storey and Saj
represents the acceleration response spectrum ordinate corresponding to the period of
vibration of the jth mode. Finally, j is a weighting factor that aims at accounting for the
varying relative importance that each mode j has on the maximum response of the
structure.
The employment of such alternative modal combination procedure, may indeed lead to
the attainment of improved results, as demonstrated by the interstorey drift profiles given
in Figure 2.89, obtained through consideration of the first three modes of vibration of the
buildings, and using 1 = 1.0, 2 = -1.0 and 3 = 1.0 in Eq. (2.16). However, the arbitrary
nature of these weighting factors j renders the method unfeasible for practical
application, as explicitly acknowledged in Kunnath (2004) and demonstrated in LpezMenjivar (2004). Indeed, in the latter work it is demonstrated how values of j that lead
to optimum results for some building configurations, lead then to poor predictions in
buildings with diverse characteristics. Therefore, and until a general procedure to
correctly determine the values of the weighting factors is found, the DVA adaptive
pushover modality cannot really be deemed as a valid solution for practical application.
12
DYNAMIC
DYNAMIC
FAP-DVA
10
FAP-DVA
3
STOREY
STOREY
8
6
4
1
2
0
0.00%
2.00%
4.00%
INTERSTOREY DRIFT (% )
(a)
6.00%
0
0.00
0.50
1.00
1.50
2.00
INTERSTOREY DRIFT (% )
(b)
Figure 2.89. Interstorey drift profiles of (a) 12-storey building and (b) 4-storey irregular frame,
obtained with Force-based Adaptive Pushover using DVA modal combination
107
U = U0
(2.17)
The normalized modal scaling vector, D , used to determine the shape of the load vector
(or load increment vector) at each step, is computed at the start of each load increment.
In order for such scaling vector to reflect the actual stiffness state of the structure, as
obtained at the end of the previous load increment, an eigenvalue analysis is carried out.
To this end, the Lanczos algorithm (Hughes, 1987) is employed to determine the modal
shape and participation factors of any given predefined number of modes. Modal loads
can be combined by using either the Square Root of the Sum of the Squares (SRSS) or
the Complete Quadratic Combination (CQC) methods.
108
Since application to the analysis of buildings is the scope of the present work, use is made
of the interstorey drift-based scaling algorithm, whereby maximum interstorey drift values
obtained directly from modal analysis, rather than from the difference between notnecessarily simultaneous maximum floor displacement values, are used to compute the
scaling displacement vector. This comes as a reflection of the fact that the maximum
displacement of a particular floor level, being essentially the relative displacement
between that floor and the ground, provides insufficient insight into the actual level of
damage incurred by buildings subject to earthquake loading. On the contrary, interstorey
drifts, obtained as the difference between floor displacements at two consecutive levels,
feature a much clearer and direct relationship to horizontal deformation demand on
buildings. Readers are referred to Antoniou (2002) for further details on this formulation.
In such an interstorey drift-based scaling technique, the eigenvalue vectors are thus
employed to determine the interstorey drifts for each mode ij, as shown in Eq. 2.18,
while the displacement pattern Di at the ith storey is obtained through the summation of
the modal-combined inter-storey drifts of the storeys below that level, i.e. drifts 1 to i:
i
Di = k with i =
k =1
j =1
2
ij
[ (
n
j =1
i, j
i 1, j )
(2.18)
Since only the relative values of storey displacements (Di) are of interest in the
determination of the normalised modal scaling vector D , which defines the shape, not
the magnitude, of the load or load increment vector, the displacements obtained by Eq.
2.18 are normalised so that the maximum displacement remains proportional to the load
factor, as required within a load control framework:
Di =
Di
max Di
(2.19)
Once the normalised scaling vector and load factor have been determined, and knowing
also the value of the initial nominal load vector, the loading vector Ut at a given analysis
step t is obtained by adding to the load vector of the previous step, Ut-1 (existing
balanced loads), a newly derived load vector increment, computed as the product
between the current load factor increment t, the current modal scaling vector D t and
the nominal vector U0, as mathematically translated into Eq. 2.20 and graphically
depicted in Figure 2.90.
Ut = Ut-1+t Dt U0
(2.20)
Ptt = t
U
new increment
of forces
of displacements
nominal
load vector
normalised
shape
normalised
at step
t
shape
at step
t
109
balanced
displacements
existing
forces
UPt=U
+Pt=
t=Pt-1
t-1+U
new increment
of displacements
of forces
new
new displacements
forces applied applied
at step tat step t
The DAP algorithm has been implemented in the computer code SeismoStruct
(SeismoSoft 2005), a fibre element-based program for seismic analysis of framed
structures, freely downloadable from the internet. The program incorporates both local
(beam-column effects) and global (large displacements/rotations effects) sources of
geometric nonlinearity as well as the interaction between axial force and transverse
deformation of the element. The spread of material inelasticity along the member length
is explicitly represented through the employment of a fibre modelling approach, implicit
in the formulation of the inelastic beam-column frame elements adopted in the analyses.
Various verification studies have been carried out with the aforementioned program on a
four-storey reinforced concrete frame (Figure 2.91), a reinforced concrete bridge (Figure
2.92a) and a two-storey steel frame (Figure 2.92b) all of which show the ability of the
analytical models to replicate the seismic response of full-scale structures.
80
Displacement [mm]
60
40
20
0
-20
-40
Analytical
Experimental
-60
-80
10
12
14
T ime [s]
110
60
2000
Experimental
Experimental
Analytical
Numerical
1500
30
1000
15
500
kN
0
-0.4
-15
-0.3
-0.2
-0.1
0.1
0.2
0.3
-500
-30
-1000
-45
-1500
-60
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
-2000
4.0
Time (s)
(a)
(b)
Figure 2.92. Verification of fibre element analytical models of (a) a RC bridge (Casarotti et al., 2005)
and (b) a 2-storey steel frame (Pietra, 2006)
0.4
displacement (cm)
acceleration (g)
45
0.3
0.2
2
1
0.1
0.0
(a)
period (sec)
period (sec)
(b)
Figure 2.93. (a) Acceleration and (b) displacement response spectra of accelerogram employed
in the analysis of 12-storey building.
0.4
111
The second structure is a 4-storey three-bay building refers to a full-scale test specimen,
built to represent typical design and construction practice in most South-European
countries in the 1950's, and tested under pseudo-dynamic conditions (Pinho and
Elnashai, 2000) at the JRC in Ispra (Italy). The frame was designed for gravity loads only,
without any consideration of ductility provisions or capacity design principles.
Consequently, it exhibited a soft-storey type of deformation mechanism at the third
storey level (e.g. Figure 2.83) caused mainly by the drastic stiffness/strength variation
present at such location, as well as by inadequate lap-splicing and defective column shear
capacity. The input motion consisted of artificial accelerograms aiming at being
representative of European seismicity (Campos-Costa and Pinto, 1999).
In Figure 2.94, the interstorey drift profiles of these two case-studies, as obtained with the
employment DAP analyses, are given. It is observed that the predictions now match
much closer the dynamic response of these two structures, which effectively means that
the response irregularities caused by the flexibility of the 12-storey structure, and
subsequent amplification of higher modes, as well as the strength irregularity of the 4storey prototype, have been fully and correctly captured by the proposed static analysis
algorithm.
12
DYNAMIC
DYNAMIC
DAP-SRSS
DAP-SRSS
10
3
STOREY
STOREY
8
6
1
2
0
0.00%
2.00%
4.00%
INTERSTOREY DRIFT (% )
(a)
6.00%
0
0.00
0.50
1.00
1.50
2.00
INTERSTOREY DRIFT (% )
(b)
Figure 2.94. Interstorey drift profiles of (a) 12-storey building and (b) 4-storey irregular frame,
obtained with Displacement-based Adaptive Pushover using SRSS combination.
In Figure 2.95, on the other hand, the capacity curves of the 12-storey building, as
derived by both DAP and standard pushover curves are compared with the Incremental
Dynamic Analysis envelope. The advantages of using an adaptive displacement-based
pushover can be inferred also from this type of results.
112
7000
6000
5000
4000
3000
2000
dynamic
adaptive
uniform
triangular
1000
0
0%
1%
2%
3%
total drift
Figure 2.95. Capacity curves of a 12-storey building, obtained with DAP and standard
pushovers, and compared against IDA envelopes.
The reason behind the most-improved predictions obtained with the displacement-based
adaptive pushover procedure is the fact that storey forces or shears are no longer applied
directly to the structure but rather come as a result of structural equilibrium to the applied
displacement pattern, thus allowing for the reproduction of reversal of storey shear
distributions, observed in dynamic analysis, even if a quadratic rule is employed to
combine the contribution of the different modes. In effect, DAP drift profiles, despite
carrying a permanently positive sign, do, in any case, feature changes of their respective
gradient (i.e. the trend with which drift values change from one storey to the next),
introduced by the contribution of higher modes. When such gradient variations imply a
reduction of the drift of a given storey with respect to its adjacent floor levels, then the
corresponding applied storey horizontal force must also be reduced, in some cases to the
extent of sign inversion, as observed in Figure 2.96.
113
12
11
adaptive
uniform
triangular
10
9
8
7
6
5
4
3
2
1
-500
500
1000
1500
2000
Figure 2.96. Storey shear distributions of a 12-storey building obtained with Displacement-based
Adaptive Pushover as well as with standard non-adaptive pushovers
In other words, given that in DAP, shear distributions are automatically derived to attain
structural equilibrium with the imposed storey drifts, rather than being a result of the
loads directly applied to the structure, the previously described limitations evidenced by
force-based adaptive schemes that use quadratic modal combination rules can be
overcome and, consequently, results as whole (i.e. deformation profiles and capacity
curves) become more accurate.
2.3.3.3 DAP Ease-of-use, Computational Effort and Numerical Stability
When compared with nonlinear time-history analysis, pushover methods are advantaged
by their (i) higher user-friendliness, (ii) reduced running time and (iii) increased numerical
stability. Therefore, it is important that the proposed displacement-based algorithm,
capable of producing improved structural response predictions in comparison with
existing non-adaptive pushover techniques, does also feature these three advantages over
dynamic analysis.
From a usability point-of-view, the proposed displacement-based adaptive pushover
algorithm effectively presents no additional effort and/or requirements with respect to its
conventional non-adaptive counterparts. In effect, the only element of novelty, in terms
of analysis input, is the introduction of the buildings inertia mass, which, however, can
readily be obtained directly from the vertical gravity loads, already included in any type of
pushover analysis.
With regards to computational effort, in general, the amount of time necessary to
complete an adaptive pushover analysis is typically double the time necessary for a
114
115
REGULAR SYSTEM
structural
wall
8 3.0 = 24.0 m
12 3.0 = 36.0 m
coupling beam
IRREGULAR SYSTEM
DUAL SYSTEM
Figure 2.97. Geometric characteristics of the regular, irregular and dual systems.
Table 2.12. Considered building systems
Structural
System
Storeys
(Height)
Regular
Frame
12 (36 m)
Irregular
Frame
8 (25.5 m)
Regular
WallFrame
8 (24 m)
Structure
Reference
RH30
RM30
RM15
RL15
IH30
IM30
IM15
IL15
WH30
WM30
WM15
RH30
Ductility
Level
High
Medium
Medium
Low
High
Medium
Medium
Low
High
Medium
Medium
High
Design
PGA (g)
0.30
0.15
0.30
0.15
0.30
0.15
Behavior
Factor (q)
5.00
3.75
3.75
2.50
4.00
3.00
3.00
2.00
3.50
2.625
2.625
5.00
Tuncracked
(s)
0.697
0.719
0.745
0.740
0.565
0.536
0.613
0.614
0.569
0.557
0.601
0.697
Table 2.13. Bounding characteristics of the employed set of records for buildings
Min
Max
Peak Ground
Acceleration
Peak Response
Acceleration
0.12 g
0.93 g
0.50 g
4.25 g
5% Arias
Intensity
threshold
1.02 s
11.23 s
Significant
Duration
teff
7.24 s
10.43 s
Total
Duration
ttot
10.0 s
40.0 s
teff / ttot
22.3%
72.4%
1.0
1.0
0.8
0.8
acceleration (g)
acceleration (g)
116
0.6
0.4
0.2
0.0
0.6
0.4
0.2
0.0
period (sec)
period (sec)
0.5
3.0
0.4
2.4
acceleration (g)
acceleration (g)
(a) Record AR
0.3
0.2
0.1
0.0
1.8
1.2
0.6
0.0
period (sec)
period (sec)
Figure 2.98. Elastic response spectra of the four records (5% equivalent viscous damping).
(b) Analyses and Results Post-Processing. The two non-adaptive pushover schemes,
proposed in the NEHRP Guidelines (ATC, 1997), were applied to each set of buildings:
the uniform distribution, whereby lateral forces are proportional to the total mass at each
floor level, and the triangular distribution, in which seismic forces are proportional to the
product of floor mass and storey height. The adaptive pushover algorithm was used in
both its force and displacement-based variants, with spectrum scaling, employing SRSS or
CQC modal combination rules.
The inter-storey drift profiles obtained from each pushover analysis are compared to the
drift profiles from the nonlinear dynamic analysis and the standard error of the pushover
results, with respect to the dynamic, is calculated as:
Error(%) = 100
1 n iD iP
n i =1 iD
(2.21)
The interstorey drift profiles are monitored at four different deformation levels: the preyield state (0.5 % total drift), the point of global yielding (1.0 % and 1.5 %), where the
stiffness changes significantly and the local distributions are rapidly updated, and the
deeply inelastic range (2.5 %).
The Standard Error of the non-adaptive and adaptive pushover schemes was computed
for all the structures and earthquake records considered. In order to identify the presence
117
(c) Obtained Results. The Mean Standard Error of the DAP, FAP, Triangular and
Uniform pushovers, considering all structures and ground motions, are 19.11 %, 30.90 %,
21.11 % and 38.76 %, respectively. These overall results seem to indicate only a marginal
advantage of DAP with respect to non-adaptive triangular distribution. However, a closer
inspection of interstorey drift profiles (Figure 2.99) for some particularly difficult cases,
renders much more conspicuous the gains provided by the employment of displacementbased adaptive pushover in the prediction of the seismic demand/capacity of framed
buildings subjected to seismic action.
12
DAP-SRSS
DYNAMIC
TRIANGULAR
UNIFORM
10
STOREY
0
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
INTERSTOREY DRIFT
Figure 2.99. Representative results obtained with model RM15 subjected to one of the natural
accelerograms employed in this study (NR2).
118
been used in such work have been adopted here. Hence, considered ground motions
include Near-Field type records (NF records) as well as ordinary records (Ordinary
Ground Motions, OGMs). Their main properties are summarized in
Table 2.14, whilst displacement response spectra are represented in Figure 2.100. OGMs
(11 records) consist of site class C accelerograms selected from strong-motion records
that do not present near-fault or near-field characteristics (strong velocity pulses, short
duration, high frequency content, etc.). NF accelerograms, on the other hand, consist of
motions recorded close to the epicentre and which contain very strong velocity pulses,
originally included in the FEMA-440 work with the objective of verifying, at least in
preliminary fashion, the validity of employing pushover methods in areas where pulse-like
near-fault ground motions are likely to occur. For further details the reader is referred to
ATC (2005) and Somerville et al. (1997).
EQ
Ms
A1
Superstitn
11-24-87
Northridge
1-17-94
Loma Prieta
10-18-89
Chi-Chi,
Taiwan
8-20-99
Loma Prieta
10-18-89
Northridge
1-17-94
Chi-Chi,
Taiwan
8-20-99
Superstitn
11-24-87
Northridge
1-17-94
Imperial
Valley
10-15-79
Imperial
Valley
10-15-79
6.6
Erzican
3-13-92
Landers
6-28-92
Northridge
1-17-94
Northridge
1-17-94
6.9
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
ERZ
LUC
RRS
SCH
Station
Component
119
PGA
[g]
PGV
[cm/s]
Effective
Length[s]
Source
0.358
46.4
23
CDMG
000
0.41
43
11
USC
7.1
090
0.322
39.1
14
CDMG
7.6
(TCU122)
0.261
34
35
CWB
7.1
090
0.367
44.7
17
CDMG
6.7
196
0.42
60.8
14
USC
7.6
(CHY101)
0.353
70.6
32
CWB
6.6
090
0.258
40.9
27
CDMG
6.7
106
0.356
32.1
16
USC
6.9
140
0.315
31.5
17
USGS
6.9
230
0.38
42.1
18
USGS
0.442
126
7
14
EERL
Caltech
EERL
Caltech
EERL
Caltech
EERL
Caltech
6.7
7.3
280
0.732
147
6.7
213
0.891
186
6.5
6.7
190
0.865
138
5.5
0.8
Sd [m]
0.6
0.4
0.2
0
0
ERZMV1
LUCMV1
RRSMV1
SCHMV1
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
0.5
1.5
2.5
0.5
1.5
T [s]
T [s]
(a)
(b)
2.5
Figure 2.100. Ground Motions: unscaled displacement response spectrum for (a) Ordinary- and (b)
Near-Field-type records (ATC, 2005)
120
The prototype buildings analyzed in the current endeavour consist therefore of two steel
moment-resisting frames (nine and three storeys, Figure 2.101 and Figure 2.102) designed
as a part of the FEMA-funded SAC joint venture project (Gupta and Krawinkler, 1999).
These frames were considered in both regular and irregular (weak-storey at the ground
floor) configurations, thus leading to a total of four frames.
(b) Analyses and Results Post-Processing. The pushover schemes considered include
invariant static load patterns, such as (i) uniform (Uniform), (ii) inverted triangular
(Triangular), (iii) first mode shape (1st Mode), (iv) a code-specified period-dependent
distribution (where lateral forces changes from a linear distribution for low period
systems to a parabolic shape for more flexible models) (Code), and the Displacementbased Adaptive Pushover (DAP). In addition, an alternative adaptive approach has been
tested in the case of OGMs, consisting in the employment of the
121
average response spectrum of all records to compute the modal spectral amplification
that is considered in the calculation of the incremental adaptive loading vector
(Displacement-based Adaptive Pushover with Average Spectrum scaling (DAP-AS)).
Whilst such procedure might be adopted in the case of OGMs, where records have been
scaled to meet the same target displacement, it cannot be used for NF records, which
were employed without scaling and thus involve different drift responses for the same
prototype building.
Verification of the pushover algorithm at the global level was carried out through
comparisons between the base shear vs. top floor IDA envelopes (assumed as
representative of the true behaviour of the frame) and the pushover curves. With the
objective of assessing also the accuracy of DAP in estimating local response parameters,
the dynamic response at different ductility demand levels has been considered, where
each Ductility Level (DL) is identified by means of the total drift value (i.e. top floor
displacement/building height). Three different drift levels are assumed (0.5, 2 and 4%)
and each ordinary record has been scaled in order to get the predefined drift for each
prototype building. Response parameters of interest (displacement, drift, shear and
moment) recorded in time-history analyses are then compared, at each level location, with
those predicted by the pushover procedures at the same roof displacement magnitude.
For the 9-Storey weak frame system an ultimate drift level of 2.7% has been selected
(corresponding to a top floor displacement of 1m), instead of the 4% defined in a
preliminary stage, since higher values were leading to the development of a global failure
mechanism of the structure (corresponding to an ultimate steel strain in frame elements
122
conservatively fixed in 15% (Ballio and Mazzolani 1987)) under several records, and
would thus prevent a complete statistical evaluation of the results.
The effectiveness of the different static procedures in predicting the local dynamic
response is quantified and compared by means of error measure E1, which provides a
direct insight on how inaccurate is the static method (evidently, the mean of a pushover
response estimate is computed only in the case of record-dependent DAP analysis, for all
other pushover schemes the single response value is used):
E1 =
MeanPUSHOVER MeanDYNAMIC
MeanDYNAMIC
(2.22)
(c) Obtained Results. Figure 2.103 summarises the results obtained through this
parametric study, putting in evidence the fact that, when compared with other pushover
procedures, DAP leads to higher accuracy in the prediction of global and local response
parameters of steel buildings, particularly in those cases where the influence of higher
modes of vibration is important (e.g. high-rise buildings). It is also shown that the
employment of an average response spectral shape leads to satisfactory results, thus
rendering the procedure very much applicable within a design application framework,
where standard code spectral shapes are prescribed.
Mean
St.Dev.
75
50
25
0
100
75
50
25
Code
Uniform
Triangular
1st Mode
DAP
DAP-AS
Code
Uniform
Triangular
1st Mode
DAP
DAP-AS
Code
Uniform
Triangular
1st Mode
DAP
DAP-AS
Code
Uniform
Triangular
1st Mode
DAP
0
DAP-AS
E1 - 9Sw [%]
E1 - 3S [%]
100
Figure 2.103. Mean values and standard deviation of error measure E1, averaged along the height of
buildings models 3S and 9Sw at the intermediate drift level (top floor drift level of 2%)
(a) Characteristics of Input Motion and Structural Models. The parametric study has
considered two bridge lengths (50 m spans), with regular, irregular and semi-regular
layout of the pier heights and with two types of abutments; (i) continuous deck-abutment
123
connections supported on piles, exhibiting a bilinear behaviour, and (ii) deck extremities
supported on pot bearings featuring a linear elastic response. The total number of bridges
is therefore twelve, as shown in Figure 2.104, where the label numbers 1, 2, 3 characterise
the pier heights of 7 m, 14 m and 21 m, respectively.
A sufficiently large number of records has been employed so as to bound all possible
structural responses. The employed set of seismic excitation is defined by an ensemble of
14 large magnitude (6 7.3) small distance (13 30 km) records selected from a suite of
historical earthquakes scaled to match the 10 % probability of exceedance in 50 years
uniform hazard spectrum for Los Angeles (SAC Joint Venture, 1997). The bounding
characteristics of the records are summarized in Table 2.15. Further details on modelling
and input can be found in Casarotti et al. (2005).
LONG BRIDGES
SHORT BRIDGES
REGULAR
SEMI
REGULAR
IRREGULAR
Label 222
Label 2222222
Label 123
Label 3332111
Label 213
Label 2331312
Min
Max
Peak Ground
Acceleration
Peak Response
Acceleration
0.30 g
1.02 g
0.84 g
3.73 g
5% Arias
Intensity
threshold
1.25 s
12.5 s
Significant
Duration
teff
5.3 s
19.66 s
Total
Duration
ttot
14.95 s
80.00 s
teff / ttot
9%
52%
(b) Analyses and Results Post-Processing. The response of the bridge models is
estimated through the employment of Incremental Dynamic Analysis (IDA), Force-based
Conventional Pushover with uniform load distribution (FCPu), Force-based
Conventional Pushover with first mode proportional load pattern (FCPm), Force-based
Adaptive Pushover with Spectrum Scaling (FAP) and Displacement-based Adaptive
Pushover with Spectrum Scaling (DAP). Results are presented in terms of the bridge
capacity curve, i.e. a plot of the reference point displacement versus total base shear, and
of the deck drift profile.
124
Each level of inelasticity is represented by the deck centre drift, selected as independent
damage parameter, and per each level of inelasticity the total base shear Vbase and the
displacements i at the other deck locations are monitored. Results of pushover analyses
are compared to the IDA median value out of the responses to the 14 records, of each
response quantity R, be it total base shear or deck drift:
(2.23)
Pushover analyses with spectrum scaling (i.e. adaptive pushovers) are statistically treated
in an analogous way: medians of each response quantity represent that particular
pushover analysis (i.e. FAP or DAP) with spectrum scaling. Finally, the results of each
type of pushover are normalized with respect to the corresponding exact quantity
obtained from the IDA medians, as schematically illustrated in Figure 2.105, and
translated in Eq. 2.24 Representing results in terms of ratios between the approximate
and the exact procedures, immediately indicates the bias in the approximate procedure,
as the ideal target value of the different pushovers is always one.
R i , PUSHOVERtype =
R i , PUSHOVERtype
L ideally
1
R
(2.24)
i , IDA
1 =
1, IDA
2 =
2
2 , IDA
3 =
3
3, IDA
4 =
4 , IDA
Given the fact that a realistic capacity curve does not imply reliable estimations of the
inelastic displacement pattern at increasing levels of inelasticity, the control of the
deformed pattern is of the same relevance of the capacity curve prediction.
Having the same unitary target value, all normalized deck displacements become
comparable, and a bridge index BI can measure the precision of the obtained deformed
shape. Per each level of inelasticity, such bridge index is defined as the median of results
over the m deck locations Eq. 2.25 with the standard deviation measuring the dispersion
with respect to the median Eq. 2.26. The latter indicates the stability of the estimate of
125
displacements along the deck: a small scatter means that predicted normalised
displacements along the deck are averagely close to their median value BI.
BI PUSHOVERtype = median i =1:m ( i , PUSHOVERtype )
m
PUSHOVERtype =
i ,PUSHOVERtype BI PUSHOVERtype
i =1
(2.25)
2
(2.26)
m 1
126
x 10
x 10
3.5
3
1.5
DAP
FAP
FCPu
FCPm
IDA
0.5
0
0
0.02
0.04
0.06
0.08
0.1
2.5
2
1.5
DAP
FAP
FCPu
FCPm
IDA
1
0.5
0.12
0
0
0.045
0.09
0.135
0.18
0.225
(a)
0.27
(b)
Figure 2.106. Capacity curve results
In Figure 2.107, the Bridge Index, as computed at each level of deck centre drift, is
plotted as black filled marks so as to cater for an easier comparison with the IDAnormalised deck displacements, represented as empty marks in the background. In this
manner, it results immediately apparent the level with which each pushover analysis is
able to capture the deformed pattern of the whole bridge, at increasing deformation
levels. For the sake of succinctness, only two analysis types are considered, FCPm and
DAP, which are those leading to the worst and best predictions, respectively.
Table 2.16 provides global averages of means, maximum and minimum values of BI and
respective dispersion as well as of the normalised total base shear, over the whole bridge
ensemble. It is noted that: (i) FCPm heavily underestimates predictions, featuring also a
very high BI dispersion value, (ii) FCPu performs very well for regular bridges and
underestimating otherwise, (iii) DAP features the best overall behaviour, despite the slight
underprediction of deformed shape values, with the lowest values of scatter.
2.3.5 Conclusions Related to Nonlinear Static Analysis of Structures
Given that current performance-based design trends require simple, yet accurate methods
for estimating seismic demand on structures considering their full inelastic behaviour, in
the current work the effectiveness of pushover analysis applied to buildings and bridges
has been investigated. In particular, the effectiveness of applying a displacement-based
adaptive pushover to estimate the seismic response of buildings and bridges subjected to
earthquake action was investigated.
It is observed that the employment of such an innovative adaptive pushover technique
lead to the attainment improved response predictions, throughout the entire deformation
127
Bridge A213
6
5
4
3
2
1
0
1.5
(a)
FCPm
1
0.5
0
0.02
0.04
0.06
0.08
0.1
0.12
0.04
0.06
0.08
0.1
0.12
1.5
1.5
(b)
DAP
0.02
1
0.5
0.5
0
0.02
0.04
0.06
0.08
0.1
0.12
0.02
0.04
0.06
0.08
0.1
0.12
Means
FCPm
FCPu
FAP
DAP
Bridge Index
mean min
0.74 0.57
0.87 0.75
0.88 0.78
0.87 0.78
max
0.92
1.03
1.01
0.99
Dispersion
mean min
0.79
0.58
0.24
0.17
0.22
0.13
0.19
0.14
max
1.00
0.34
0.34
0.27
In other words, within the scope of buildings and bridge applications, whereas the
application of a fixed displacement pattern is a commonly agreed conceptual fallacy, the
present work witnesses not only the feasibility of applying an adaptive displacement
profile, but also its practical advantages, with respect to other pushover methods.
It is important to observe that a static procedure will never be able to completely replace
a dynamic analysis; nevertheless, a methodology has been searched to obtain response
information reasonably close to that predicted by the nonlinear dynamic analyses. The
128
Nonlinear static analysis represents a valuable and appealing tool when seismic
assessment of structural behaviour, focusing on global and local levels, is considered. In
the last decade or so considerable effort has been placed in the development of credible,
yet relatively simple, pushover-based methodologies, an effort that lead to the
introduction of the so-called Nonlinear Static Procedures (NSP) in a number of guideline
documents, such as ATC-40 (ATC, 1996) and FEMA-273 (FEMA, 1973), or design
codes, such as Eurocode 8 (CEN, 2004). Despite unavoidable limitations, when
compared to dynamic nonlinear analyses (Krawinkler and Seneviratna, 1998) due to
conceptual formulation and simplicity, these methods still constitute a meaningful
instrument, particularly in current seismic design applications.
The main application studies that have been conducted on seismic assessment using the
above endeavours have though focused on the structural case of buildings, rather than
bridges. As an effort to overcome such drawback, attempts have been very recently made
to explicitly verify the application of NSPs to bridge structures (e.g. Isakovic and
Fischinger, 2006; Paraskeva et al., 2006; Casarotti and Pinho, 2007; Lupoi et al., 2007).
However, the aforementioned studies have been basically directed at the verification of a
single nonlinear static procedure, and hence there is still a need for assessing how each of
the available NSPs fair with respect to each other, within the scope of either building or
bridge applications.
2.4.2 Nonlinear Static Procedures
Early versions of nonlinear static procedures date back to 1970s (Freeman et al., 1975),
but only in the recent past their potential been recognised by their inclusion in many
guidelines and reports, such as EC8 (CEN, 2005), ATC-40 (ATC, 1996), FEMA-273
129
(FEMA, 1997) and a recent Italian Building code (OPCM, 2003). Among the first
proposals that came out, the Capacity Spectrum Method (CSM) (Freeman et al., 1975;
Freeman et al., 1988) and N2 Method (Fajfar and Fischinger, 1988; Fajfar, 1999) are
typically seen as pioneering, due to their simplicity and introductory character in
nonlinear static analysis. They usually consider a first mode and/or uniform load
distributions in computation of the pushover/capacity curve.
In the past few years, considerable improvements have been made with respect to NSPs,
upgrading the traditional methods, such as the case of the refined version of CSM
presented in FEMA 440 (ATC, 2005), or even with the proposal of new procedures, such
as Modal Pushover Analysis (MPA) (Chopra and Goel, 2002), Adaptive Capacity
Spectrum Method (ACSM) (Casarotti and Pinho, 2007) or Adaptive Modal Combination
Procedure (AMCP) (Kalkan and Kunnath, 2006). These so-called innovative methods
claim to include aspects which had been left out before, such as the consideration of
higher mode effects, progressive damage and alternative definitions of reference node.
Main differences among the existing NSPs are on the capacity calculations. The pushover
approach where the method is based on may also differ (Figure 2.108).
2.4.2.1
The Capacity Spectrum Method was initially introduced by Freeman et al. (1975) as a
methodology to rapidly evaluate seismic vulnerability of buildings. Essentially, it consists
in the graphical, and therefore intuitive, comparison of the capacity, given by the
conversion of the displacements and base shears of the nonlinear pushover curve, with
the demand, represented by an over-damped response spectrum, to account for nonlinear
behaviour given by energy dissipation through hysteresis (Figure 2.109). On this matter,
130
ATC-40 considers three different possibilities based on the duration of the shaking (short
or long) and type of structure (essentially new, average existing or poorly existing
construction). In the recent FEMA-440 report (ATC, 2005), however, new expressions
for the calculation of effective damping and period and for the scaling of the demand
spectrum have been introduced, that are dependent on hysteretic model (bilinear or
stiffness degrading), post-yielding stiffness (parameter ) and ductility achieved by the
system.
The pushover curve is obtained applying a lateral load distribution proportional to the
fundamental vibration mode shape of the structure, and is then transformed into the
capacity curve of a SDOF system in Acceleration-Displacement Response Spectra
(ADRS) format. Classical CSM depends on a reference node for such curve, usually
considered to be located at roof level, for buildings, or at the decks centre of mass, in the
case of bridges. For the latter case, however, such choice may be the subject of debate,
since the adequacy/accuracy of using the decks central node to represent the structure
depends heavily on the piers configuration and characteristics.
The intersection of the two curves is considered to be the performance point of the
structure, to which every structural response parameter is referred to.
Iterative procedure
(until convergence in equivalent
viscous damping is reached)
Figure 2.109 Flow chart illustrating the steps of the Capacity Spectrum Method
2.4.2.2
131
N2 Method
The N2 method was proposed by Fajfar and Fischinger (1988), and features the use of
pushover analysis of a MDOF model combined with the inelastic response spectrum
analysis of its equivalent SDOF system. Similarly to CSM, the N2 method may also be
formulated in the acceleration-displacement format, thus providing a visual interpretation
of the procedure. The main difference with respect to the former, however, is that N2
makes use of ductility-dependent inelastic spectra, rather than over-damped elastic
spectra. It has been included in Eurocode 8 (CEN, 2004) as the recommended nonlinear
static procedure and its implementation, according to such code, foresees the
computation of at least two pushover curves; one derived with a loading vector shape
factor that is uniform along the structure and another where the first mode proportional
shape is used instead.
The capacity curve is represented by plotting the base shear force as a function of the
reference node displacement, the latter being recommended to be taken at the building
roof level or centre of mass of the deformed deck. The performance point is determined
through the computation of the target displacement of the equivalent SDOF system,
obtained through the intersection between the initial period of vibration of the latter and
an inelastic spectrum derived as a function of the ductility achieved by the system (Figure
2.110). A non-compulsory iterative procedure to match the target displacement obtained
for the equivalent SDOF system with that employed in the determination of the idealized
elasto-perfectly plastic force-displacement relationship has also been proposed.
132
2.4.2.3
The Modal Pushover Analysis, one of the more recently improved procedures, was
introduced by Chopra and Goel (2002) and consists in the repeated application of a given
nonlinear static analysis procedure for each of the significant vibration modes of the
structure, followed then by an adequate combination of the results. The procedure starts
with the computation of a number of MDOF pushover curves, each of which is obtained
by employing a load vector across the buildings height or bridges deck that is
proportional to the individual modes of vibration of interest. The equivalent SDOF
capacity curves are then determined by transforming and bilinearizing the modal
pushover-derived base shear-displacement relations, making use of the nth modal
quantities and the reference node displacement (Figure 2.111). The modal performance
points can then be derived by means of (i) nonlinear time-history analysis of each modal
bilinear SDOF system, (ii) inelastic response spectrum and (iii) empirical inelastic-elastic
response ratios. Finally, a quadratic combination rule (e.g. through SRSS or CQC) is
employed to combine the responses obtained for each modal analysis.
Multimodal Pushover
Analysis
Run pushover analysis and get
MDOF pushover curve
133
12
12
12
10
10
10
400000
300000
200000
100000
1st mode
2nd mode
3rd mode
mode 1
mode 2
mode 3
500000
0
0
50
100
150
200
250
300
350
400
450
500
...
2.4.2.4
The Adaptive Capacity Spectrum Method recently proposed by Casarotti and Pinho
(2007), combines elements from Direct Displacement-Based Design (Priestley et al., 2007)
with the original Capacity Spectrum Method formulation. Essentially it consists in using a
displacement-based adaptive pushover analysis (which accounts for higher modes and
stiffness degradation) to then derive an equally adaptive SDOF capacity curve which is in
turn intersected with the acceleration-displacement response spectrum of the design
earthquake, as applied in the traditional CSM. The equivalent SDOF capacity curve is
calculated step-by-step using the actual deformed pattern of the MDOF at any given
instant and hence is not based on the modification of the MDOF capacity curve with
reference to the displacement profile of a specific physical location. In this manner, the
need to define a somewhat arbitrary reference node is avoided.
The computation of seismic forces makes use of the Modified Modal Superposition
(MMS) initially proposed by Eibl and Keintzel (1988) and then further advocated by
Priestley et al. (2007), whereby higher mode elastic forces are combined, through SRSS,
134
with the pushover analysis derived forces at the performance point (intersection between
demand spectrum and capacity curve).
2.4.2.5
The original adaptive method has been proposed by Gupta and Kunnath (2000) and it is
a load-controlled procedure in which load increments are scaled at each pushover step
using elastic spectral accelerations based on the instantaneous dynamic properties of the
system. In the proposed new procedure, AMCP, a displacement-controlled method is
used in which the demand due to individual terms in the modal expansion of the effective
earthquake forces is determined by individual adaptive pushover analyses using the inertia
distribution of each mode, which is continuously updated during the process of loading.
Unlike the adaptive scheme of Gupta and Kunnath (2000) where the contributions of
each mode are combined at the end of each step using square-root-of-sum-of-squares
(SRSS), in the proposed scheme the total seismic demand of the system is obtained at the
end of the analysis by combining the individual responses using SRSS. Please see for a
detailed graphical explanation of the method.
2.4.2.6
135
The DCM is mentioned in the FEMA 356 document. It is based on the statistical analysis
of the results obtained by the time history analysis of SDOF oscillators with various
types. The methos tries to represent the SDOF displacement as fraction of the MDOF
displacement by applying some coefficients on the latter. There are four basic coefficients
which are for
C0: relating the top displacement to the spectral displacement,
C1: takes into account the magnification of the maximum displacement due to
inelastic behaviour,
C2: takes into account the quality of the hysteretic response,
C3: takes into account the displacement increase due to second-order effects,
These coeffecients have been calibrated and improved in the most recent FEMA 440
document.
2.4.3 Validation of Nonlinear Static Procedures
This Section aims at addressing two essential issues, associated to the mentioned
knowledge gaps: the systematic validation of NSPs applicability to buildings and bridges,
still not thoroughly done for the latter, evaluating their ability to match time-history
analysis results, and the parametric comparison of different nonlinear static approaches,
seeking for advantages and pitfalls, allowing one to come up with the choice for a
particular method, over the rest, or at least to establish useful recommendations when
using uch type of procedures. Sixteen real frame buildings and fourteen bridge
configurations, with distinct typology, structural configuration, material and design type,
subjected to an equally wide-ranging set of accelerograms with increasingly varying
intensity have been assessed through at least four distinct nonlinear static procedures.
The study consists in initially scrutinizing in detail each of the selected procedures with a
view to establish their optimum configuration, and then proceeding with the
parametric comparison of the approaches.
2.4.3.1
The building frames, which are real structures, illustrated in Figure 2.113, are from
different countries in Europe, with regular and irregular geometry, seismically designed
and non-designed and concrete to steel-made. A first set consists of five regular
structures with different number of storeys and column alignments, labelled as mod4,
136
mod6, mod12, mod14 and mod16 (Casarotti et al., 2007). The second group is made up
of three different concrete frame buildings with irregular (I), regular (R) and wall (W)
structural systems, designed for 0.3g peak ground acceleration and termed IM30, IH30,
RM30, RH30, WM30 and WH30 (Antoniou and Pinho, 2004). The last group is defined
by three steel (S) frame buildings with different number of storeys, two of them analysed
for a weak (W) labelled as 3S, 3SW, 9S, 9SW and 20S (Gupta and Krawinkler, 1999).
The selected bridges have two possible lengths (viaducts with four and eight 50 m spans),
with regular, irregular and semi-regular layout of the piers height and with two types of
abutments; (i) continuous deck-abutment connections supported on piles, with a bilinear
behavior (type A bridges), and (ii) deck extremities supported on linear pot bearings (type
137
B bridges). The total number of bridges is therefore fourteen, as implied by Figure 2.114,
where the label numbers 1, 2 and 3 stand for pier heights of 7, 14 and 21 meters,
respectively. The fundamental period of vibration (see Table 2.17) ranges approximately
from 0.3 to 0.5 seconds in short configurations and from 0.6 to 0.8 seconds in long ones.
Label 222
Label 2222222
REGULAR
Label 232
SEMI
REGULAR /
IRREGULAR
Label 123
Label 3332111
Label 213
Label 2331312
Configuration
123
213
222
232
2222222
2331312
3332111
Type A
abutments
0.43
0.34
0.41
0.50
0.59
0.65
0.70
Type B
abutments
0.40
0.32
0.42
0.53
0.60
0.67
0.77
The employed set of seismic excitations is defined by an ensemble of ten records selected
from a suite of historical earthquakes scaled to match the 10% probability of exceedance
in 50 years (475 years return period) uniform hazard spectrum for Los Angeles (SAC,
1997) which corresponds, in the current endeavour, to the intensity level 1.0. Additional
intensity levels, linearly proportional to the latter were also considered thus allowing an
overview on how results evolve with increasing seismic intensity. For the analysis of the
building frames, scaling factors of 0.25, 0.5, 0.75, 1.25 and 1.75 have been applied to the
unitary intensity for the entire set of buildings, except for the non-seismically designed,
where the maximum intensity has been limited to 1.25 due to heavy structural damage.
Intensity in bridge analyses was scaled by a factor of 0.5, 0.75, 1.5, 3.0 and 3.5. The
ground motions were obtained from California earthquakes with a magnitude range of 67.3 recorded on firm ground at distances of 13-30 km; their significant duration (Bommer
and Martinez-Pereira, 1999) ranges from 5 to 25 seconds, whilst the PGA (for intensity 1)
varies from 0.23 to 0.99g, which effectively implies a minimum of 0.11g (when intensity
level is 0.5) and a maximum of 3.5g (when intensity level is 3.5). The demand spectrum
was defined as the median response spectrum of the ten records.
138
(2.27)
The aforementioned results normalization consists thus in computing, for each of the
parameters and for each of the considered locations, the ratio between the result coming
from each NSP and the median result coming from THA, as illustrated in Figure 2.115
for a bridge case example, and numerically translated into Eq. (2.27); ideally the ratio
should be unitary.
1 =
2 =
1,IDA
2 , IDA
3 =
3, IDA
139
4 =
4 ,IDA
i =
i , NSP
i , IDA
(2.28)
ideally
This normalization renders also somewhat comparable all the parameters, at different
locations along the structure, since all normalized quantities have the same unitary target
value, thus allowing in turn the definition of a so-called building or bridge index (Pinho et
al., 2007). The building/bridge index (BI) is computed as the median of normalized
results for the considered parameter over the m locations, as shown in Eq. (2.29). The
standard deviation STD measures, on the other hand, the dispersion with respect to the
median Eq. (2.30).
STD , NSP
2.4.3.2
m i , NSP BI
NSP
= i =1
m 1
(2.29)
)
2
0.5
(2.30)
In this Section, the results obtained from the aforementioned parametric study are
scrutinized and interpreted, with a view to evaluate the accuracy of the different NSPs
considered. Prior to such endeavour, however, a preliminary calibration was carried out
140
over all the different response parameters so as to identify which of the variants of each
employed NSP would lead to the attainment of best results. Furthermore, such individual
parametric study intends to establish possible guidelines to the application of each NSP in
future analyses.
Capacity Spectrum Method has two damping dependent modalities for both building
frames and bridges, ATC-40 or FEMA-440 based; two additional reference node
dependent versions are defined for bridges, within a total of four;
In N2 method, the three different variants for building frames are strictly associated
to the load pattern definition, which can be first mode proportional, uniform or the
envelope of the two; within bridge analysis three additional versions come from the
reference dependency;
Modal Pushover Analysis is considered in a single version for buildings, whereas the
two possible bridge application modalities refer, once again, to the reference node
definition;
(b) Capacity Spectrum Method. Figure 2.116 and Figure 2.117 show values of Building
and Bridge Indices for each intensity level, which is the median across the entire set of
frame or bridge configurations (see Intensity level results Section for further details).
It can be clearly observed that FEMA-440 proposed modifications lead to better
performance, standing, therefore, for the best alternative along all the response
parameters. Improvements are highly significant, when predicting displacements,
interstorey drifts and, for low intensity levels, shear forces and moments.
Notwithstanding the increase in dispersion levels when FEMA-440 proposed procedure
is chosen, which is nonetheless always under 0.3, further applications of the Capacity
Spectrum Method will consider this variant, due to the notable increase in the accuracy of
the predictions.
The improvements introduced by the FEMA-440 report are again clearly observed,
especially in the estimations of displacement and deck moments, where results obtained
using the spectrum scaling procedures suggested in FEMA-440 are much closer to unity
141
(which means NSP estimates equal to THA predictions) than those obtained using the
ATC-40 equations (which seem to overestimate the equivalent viscous damping, and
hence the corresponding spectral reduction). The use of the location of central deck
displacement as reference also (marginally) improves the results. On subsequent
applications the CSM will be employed considering its FEMA-440 version (in spite of the
slight increase in dispersion; STD values change from 0.2-0.3 to 0.3-0.4), together with
the centre of mass of the deck as reference node.
Displacements
Storey shear
2
ATC 40
FEMA 440
ATC 40
FEMA 440
1.5
Median BI
Median BI
1.5
0.5
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
Interstorey drift
0.25
0.5
0.75
1.0
Intensity Level
1.25
Storey moments
2
ATC 40
FEMA 440
ATC 40
FEMA 440
Median BI
1.5
0.5
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
0.25
0.5
Base shear
ATC 40
FEMA 440
1.5
Median BI
Median BI
1.5
1.75
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
0.75
1.0
Intensity Level
1.25
1.75
142
Displacements
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
Deck moments
0.5
0.75
1
1.5
Intensity Level
3.5
3.5
1.5
Median BI
1.5
Median BI
0.5
0.5
1.5
Median BI
Median BI
1.5
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
0.5
0.75
1
1.5
Intensity Level
(c) N2 Method. Figure 2.118 and Figure 2.119 show values of Building and Bridge
Indices, obtained for each intensity level considering the entire set of building frames or
bridges.
Displacements
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
Interstorey drift
0.25
0.5
1.25
0.5
1.75
1st mode
Uniform
Envelope
1.5
Median BI
0.75
1.0
Intensity Level
Storey moments
2
1st mode
Uniform
Envelope
1.5
Median BI
0.5
1st mode
Uniform
Envelope
1.5
Median BI
Median BI
1.5
Storey shear
2
1st mode
Uniform
Envelope
143
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
0.25
0.75
1.0
Intensity Level
1.25
1.75
Base shear
1st mode
Uniform
Envelope
1.5
Median BI
0.5
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
From the building results observation, the use of an envelope technique applied to the
uniform and first mode proportional load pattern results seems to lead to better
predictions throughout all response parameters, with particular emphasis in
Displacements and Base Shear estimates. Similar improvements are attained in scatter
levels, which are lower for the envelope modality, ranging from 0.05 to 0.3. Regarding the
other two possibilities, for some parameters, such as storey moments, shear and
interstorey drifts, the uniform load shape performs better, whereas, for other ones, the
opposite occurs. In displacements, for instance, none of the possibilities is always better.
This obviously leads to the conclusion that the envelope of results is the most promising
version, which was indeed the used for the NSP comparison.
144
Displacements
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
Deck moments
0.5
0.75
1
1.5
Intensity Level
3.5
3.5
Abutment forces
1.5
Median BI
1.5
Median BI
0.5
0.5
1.5
Median BI
Median BI
1.5
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
0.5
0.75
1
1.5
Intensity Level
It is observed, for the bridges case, that, whereas in the estimation of deck displacements
and pier shears the differences between the employment of uniform or first mode
proportional load distribution does not influence the results much, when deck moments
and abutment forces are instead considered the influence of pushover load shape is
noticeable (and with opposite trends, which somehow explains why EC8 does not
recommend the use of one loading shape over the other). The envelope shape, on the
other hand, seems to somehow contain the positive aspects of the two EC8recommended distributions, leading to good BI results for all response parameters,
with relatively reduced scatter (STD varying from 0.10 to 0.30). The use of the decks
central node as reference point leads to better predictions, hence this will be adopted on
subsequent applications, together with the envelope pushover loading shape.
(d) Modal Pushover Analysis. Figure 2.120 shows values of Bridge Index for each
intensity level, considering the entire set of bridges. At higher intensity levels, MPA seems
to perform somewhat better when considering the maximum displacement node as
reference, whilst, on the other hand, for low intensity levels, the use of the decks centre
of mass as a reference node seems to yield more consistent results. This scenario is
mainly verified for displacements at pier locations and deck bending moments, given that
for piers/abutments shear predictions the differences stemming from the choice of
reference node are not relevant. Dispersion values proved not to be sensitive to this
145
reference node issue, with STD values typically ranging from 0.10 and 0.30. On
subsequent applications of the method, and considering the lack of a superior variant,
the maximum displacement reference node modality was arbitrarily adopted.
Displacements
Central node
Max disp node
1.5
Median BI
Median BI
1.5
0.5
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
Deck moments
0.75
1
1.5
Intensity Level
3.5
3.5
Abutment forces
1.5
Median BI
Median BI
0.5
1.5
0.5
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
0.5
0.75
1
1.5
Intensity Level
(e) Adaptive Capacity Spectrum Method. Eleven different alternatives for applying
ACSM, related to the computation of the spectral reduction factor were considered. Such
studies lead to the results presented in Figures 9 and 10, in terms of global
Building/Bridge Indices, which are the median of the BI for every configuration, in every
intensity level (see Global Results Section for further details). The first five correspond
to damping-based spectral reduction models, with the first four being tried with distinct
reduction factors. The two remaining ones are ductility-based.
Rather than the employed reduction factor, ACSM building frame predictions, mainly
displacements and inter-storey drifts, are strongly influenced by the chosen damping
model. Regarding ductility-based approaches, these prove to behave quite poorly when
compared to their damping-based counterparts. The scatter levels proved not to be
dependent on the chosen model (damping or ductility based), with standard deviation
values ranging from 0.05 to 0.2.
146
Displacements
0.5
ATC40
TakKow
TakGS DwaKowNauPriestley
Mir2000 VidFajFish
Interstorey drift
ATC40
TakKow
Shear moments
RF-EC8
RF-LinChang
Prt-Mrd-Vdc
RF-EC8
RF-LinChang
Prt-Mrd-Vdc
1.5
Median BI
1.5
Median BI
0.5
0.5
RF-EC8
RF-LinChang
Prt-Mrd-Vdc
1.5
Median BI
Median BI
1.5
Storey shear
RF-EC8
RF-LinChang
Prt-Mrd-Vdc
0.5
ATC40
TakKow
TakGS DwaKowNauPriestley
ATC40
TakKow
Base shear
2
RF-EC8
RF-LinChang
Prt-Mrd-Vdc
1.5
Median BI
Mir2000 VidFajFish
0.5
ATC40
TakKow
TakGS DwaKowNauPriestley
Mir2000 VidFajFish
From a bridge structures results perspective, the first conclusion is that, analogously to
building frames, it is the employed damping model, rather than the chosen dampingdependent spectral reduction equation, that conditions the results.
Generally, it seems that the approaches TakKow, DwaKowNau and Priestley lead to
the best global Indices, especially for what concerns displacements, with a global median
BI fairly close to one and median STD that is not excessive. On the other hand, the two
ductility-based approaches (Mir2000 and VidFajFish) and the damping-based ATC40
method show the worst results when compared to nonlinear dynamic analyses. To
proceed with the comparison between NPSs, Priestleys approach, which seemed to lead
to the best predictions, was chosen for the application of ACSM on bridges.
EC8
EC8
LinChang
LinChang
1.5
Median BI
Median BI
1.5
0.5
0.5
ATC40
TakKow
TakGS DwaKowNauPriestleyMiranda2000
VidFajFish
Deck moments
ATC40
TakGS DwaKowNauPriestleyMiranda2000
VidFajFish
LinChang
LinChang
1.5
Median BI
1.5
Median BI
TakKow
EC8
0.5
147
0.5
ATC40
TakKow
TakGS DwaKowNauPriestleyMiranda2000
VidFajFish
ATC40
TakKow
TakGS DwaKowNauPriestleyMiranda2000
VidFajFish
2.4.3.3
Having identified the optimum configuration of all NSPs considered here, it is now
possible to proceed with the parametric comparison of the approaches, with the purpose
of emphasizing relative advantages and disadvantages and, eventually, coming up with
suggestions for possible preferred choices, if any. The study is again carried out on the
basis of Building/Bridge Indices and Standard Deviation comparison, starting from a
somewhat global perspective, where the entire set of results (for all configurations and for
all intensity levels) are first considered together, and then sub-structured in terms of
seismic input and structural model.
(a) Global Results. The global results overview (Figure 2.123 and Figure 2.124) consists
in the computation of the BI per NSP over all the configurations and intensity levels. In
other words, the median BI over all configurations and intensity levels represents the
median of the single BI of every considered structural building or bridge configuration, at
every intensity level. This representation of results caters for (i) comparison with timehistory analyses (it is recalled that BI represents the ratio between NSP and THA results),
(ii) relative comparison of the accuracy of the different NSPs, and (iii) appreciation of the
results dispersion (plots include overall median BI for every method, in filled markers,
and mean BI mean STD error bars).
148
Displacements
1.5
0.5
CSM
MPA
0.5
CSM
N2
MPA
0.5
CSM
N2
MPA
ACSM
Storey moments
1.5
0.5
ACSM
CSM
N2
MPA
ACSM
Base shear
1.5
1.5
ACSM
Interstorey drift
N2
Storey shear
1.5
0.5
CSM
N2
MPA
ACSM
It is noticeable that, for what concerns frame buildings, quite good predictions are
obtained for displacements (BIs very close to unit) regardless the chosen NSP.
Additionally, encouraging results are obtained throughout the rest of the response
parameters, even though with the need to use different NSPs. Such inconsistency along
the considered response parameters renders difficult the choice for a NSP as the best of
the bunch. In fact, ACSM, for instance, manages, in global terms, to perfectly match IDA
displacement estimates whereas, for storey shears and moments predictions, MPA is
clearly ahead. On the other hand, CSM shows the best median results for the interstorey
drift evaluation, but underestimates the remaining quantities. N2 performs mainly in a
mid to low underestimating fashion maintaining, though, a sort of steady behaviour.
With regard to scatter levels, similar conclusions can be drawn. Throughout the analysed
response parameters ACSM and MPA assume themselves as the lowest standard
149
1.5
0.5
CSM
MPA
0.5
CSM
N2
MPA
0.5
CSM
ACSM
N2
MPA
ACSM
1.5
1.5
ACSM
Deck moments
2
Median BI, Mean BI Mean STD
N2
2
Median BI, Mean BI Mean STD
1.5
0.5
CSM
N2
MPA
ACSM
For the bridges case, and observing Figure 2.122, it is conspicuous that all nonlinear static
procedures are able to predict displacement response with relatively good accuracy,
evidencing also reasonable dispersion levels. On the contrary, shear forces, either at piers
or abutments, are underestimated in relatively heavy fashion, by all methods except
ACSM, the latter managing to provide accurate predictions in both cases, most likely due
to the employment of MMS. As for deck bending moments, MPA performs quite well,
slightly underestimating THA results, immediately followed by ACSM, whilst, again, CSM
and N2 seem to heavily underpredict such response quantities.
Regarding dispersion levels, the observed tendency is to have low scatter in shear force
predictions (maximum values of 0.20) except for ACSM, which has approximately the
double of that value. Higher STD values are obtained when predicting displacement and
bending moment (between 0.30 and 0.40). N2 is, in most of the situations, the method
with the lowest dispersion levels.
(b) Intensity Level Results. At each intensity level, the median BI over the total
configurations is computed for each of the NSPs (see Figure 2.125 and Figure 2.126).
The results not only tend to confirm the observations made in the global analysis, but
150
also add some insight on how these may be influenced by the intensity of the input
motion.
Displacements
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
1.75
Interstorey drift
0.25
0.5
Median BI
0.5
1.25
1.75
1.25
1.75
CSM
N2
MPA
ACSM
1.5
0.75
1.0
Intensity Level
Storey moments
CSM
N2
MPA
ACSM
1.5
Median BI
0.5
CSM
N2
MPA
ACSM
1.5
Median BI
Median BI
1.5
Storey shear
CSM
N2
MPA
ACSM
0.5
0.25
0.5
0.75
1.0
Intensity Level
1.25
0.25
0.5
1.25
1.75
0.75
1.0
Intensity Level
Base shear
2
CSM
N2
MPA
ACSM
1.5
Median BI
1.75
0.5
0.25
0.5
0.75
1.0
Intensity Level
The results indeed substantiate the global observations, not only for the generally good
predictions throughout the entire set of NSPs, but also in the behaviour of each modality
according to the response parameter. It is noticeable that, as the intensity of the input
motion increases, the BI maintains stable trends, independently from the considered NSP
and response parameter, with major differences between NSPs being found in shear force
and moment predictions. Once more it is difficult to single out a NSP as the best of the
role, though it is conspicuous that MPA method provides the best results for the storey
moments and storey shear evaluations, while ACSM is the best NSP technique for the
displacements analysis. Regarding dispersion levels, generally low STD values with a
151
maximum boundary of 0.3 are observed. In addition to being the most accurate
methodologies to evaluate median BI, ACSM and MPA revealed the lowest scatter
(ranging from 0.05 to 0.2) whereas CSM stands as that with the highest dispersion.
Displacements
1.5
Median BI
Median BI
1.5
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
Deck moments
0.5
0.75
1
1.5
Intensity Level
3.5
3.5
1.5
Median BI
1.5
Median BI
0.5
0.5
CSM
N2
MPA
ACSM
0.5
0.5
0.75
1
1.5
Intensity Level
3.5
0.5
0.75
1
1.5
Intensity Level
It is observed, for bridge structures, that as the intensity level increases the accuracy in
the predictions of shear forces at the bridge piers decreases significantly. This is because
pier shear forces are highly dependent on higher mode effects, which in turn become
more important as the intensity of the seismic action increases (because the fundamental
period elongates, hence its spectral amplification diminishes, increasing the relative
importance of higher modes). N2 and CSM produce inevitably similar results, because
they are both based on first mode response only, whilst MPA and ACSM, on the other
hand, do consider higher modes, albeit in a different manner with respect to each other,
which leads to underestimation and overestimation of forces, respectively. The latter is
also visible in the variation of abutment forces, underestimated by MPA and
overestimated by ACSM, which thus results conservative.
Important variations are also observed in displacement response estimates, with slight
underprediction at lower intensity levels, for all NSPs, to modest overprediction at high
intensity, for CSM and MPA, while N2 and ACSM seem to keep a steadier closeness to
THA results. These differences between the four methods may be justified with the fact
that major conceptual differences exist between them, such as the reference node choice,
the use of an envelope of different displacement shapes for the case of N2, etc. For what
152
concerns the dispersion of the results, this did not prove to be dependent on intensity
level.
Median BI
1.5
1
0.5
0
mod4
Median BI
mod14
mod16
IH30
IM30
RH30
RM30
Building model
WH30
WM30
3S
3SW
9S
9SW
20S
WH30
WM30
3S
3SW
9S
9SW
20S
WH30
WM30
3S
3SW
9S
9SW
20S
Storey shear
mod4
mod6
mod12
mod14
mod16
IH30
IM30
CSM
N2
MPA
ACSM
2
1.5
Median BI
mod12
0.5
0
RH30
RM30
Building model
Interstorey drift
1
0.5
0
mod4
mod6
mod12
mod14
mod16
IH30
IM30
1.5
RH30
RM30
Building model
Storey moments
CSM
N2
MPA
ACSM
Median BI
mod6
CSM
N2
MPA
ACSM
1.5
1
0.5
0
mod4
mod6
mod12
mod14
mod16
IH30
IM30
RH30
RM30
Building model
WH30
WM30
3S
3SW
9S
9SW
20S
WH30
WM30
3S
3SW
9S
9SW
20S
Base shear
CSM
N2
MPA
ACSM
1.5
Median BI
Displacements
1
0.5
0
mod4
mod6
mod12
mod14
mod16
IH30
IM30
RH30
RM30
Building model
153
It can be noticed for building frames that, with the exception of ACSM, all NSPs perform
worse for steel buildings, with BIs slightly farther from unit, together with higher STD
(maximum of 0.30). Only in storey moments prediction MPA manages to get BIs near to
one for this kind of buildings. It is also conspicuous that seismically designed RC frames
are those with better NSP predictions, both in terms of Building Index (essentially equal
to unity) and Standard Deviation (very close to zero). Globally, but particularly in the
non-seismically designed RC frames, ACSM and MPA come up as the most accurate
methods, not only for median BI (very close to unit) but also in what concerns the
dispersion levels (ranging between 0.05 and 0.20). Finally, according to storey moments
estimation, MPA manages to keep consistently close to a unitary value with lower STD
values, not higher than 0.30.
CSM
A123
A213
A222
A232
Median BI
Median BI
ACSM
B123
B213
B222
B232
B213
B222
B232
B213
B222
B232
B213
B222
B232
A123
A213
A222
A232
B123
Deck moments
2
Median BI
MPA
A123
A213
A222
A232
B123
2
Median BI
N2
Displacements
1
0
A123
A213
A222
A232
On the bridges side, as expected, the response predictions do appear to be very bridgedependent, even if the observations/conclusions previously drawn still hold for the
majority of configurations. The best response estimates (in terms of BIs being close to
unity, and STDs being close to zero) are obtained for the regular bridge configurations
(e.g. 222, 232 and 2222222), as one would expect, though good displacement estimates
are also obtained for semi-regular and irregular bridges; this certainly constitutes good
news for NSPs and their application to assessment of bridge response when the focus is
on deformations. An overestimating trend for short bridges and an underestimating one
for longer ones, can also be observed in the results, whilst the largest scatter, in general, is
associated to long bridges. Finally, no relevant differences between bridges with
154
The ability of four commonly used Nonlinear Static Procedures in predicting the
structural response of building frames and bridges subjected to earthquake action has
been appraised and compared; two pioneering traditional methods (CSM and N2) were
considered along with two of their more innovative contemporary counterparts (MPA
and ACSM). The evaluation was systematically carried out over a relatively large number
of structural configurations, considering different response parameters and using several
accelerograms scaled to a number of intensity levels. A preliminary study was also carried
out with a view to establish the optimum configuration for each of the employed
procedures. The following main observations can be made:
The Capacity Spectrum Method has clearly benefited from the improvements
introduced in the FEMA-440 report, which allowed the attainment of superior
predictions, with respect to those obtained using the antecedent ATC-40 formulae,
something that became quite evident when assessing either building frames or
bridges.
All NSPs proved to be able to predict displacement response with relatively good
accuracy for all sorts of building frames and bridge configurations (regular, irregular,
short, tall or long, etc), something that certainly does lend some reassurance with
regards to the employment of such methodologies for assessing response
displacements and deformations.
Looking at building frames only, an intensity detail level of the results showed that a
sort of stable behaviour was found for every NSP, in every response parameter,
regardless the increase in the ground motion demand. At a structural configuration
detail level, it was equally verified that non-seismically design reinforced concrete
buildings are those where higher scatter in the results should be expected. In fact, a
relatively reduced dispersion throughout the entire analyses was detected, ranging
between 0.05 and 0.30, particularly lower for the case of ACSM and MPA.
If bridges are taken into account, the solid observation is that Adaptive Capacity
Spectrum Method behaved in a slightly more consistent manner, given that,
155
158
Whitman et al. (1973) first proposed the use of damage probability matrices for the
probabilistic prediction of damage to buildings from earthquakes. The concept of a DPM
is that a given structural typology will have the same probability of being in a given
damage state for a given earthquake intensity. The format of the DPM suggested by
Whitman et al. (1973) is presented in Table 3.1, where example proportions of buildings
with a given level of structural and non-structural damage is provided as a function of
intensity (note that the damage ratio represents the ratio of cost of repair to cost of
replacement). Whitman et al. (1973) compiled DPMs for various structural typologies
according to the damaged sustained in over 1600 buildings after the 1971 San Fernando
earthquake.
One of the first European versions of a damage probability matrix was produced by
Braga et al. (1982) which was based on the damage data of Italian buildings after the 1980
Irpinia earthquake and introduced the binomial distribution to describe the damage
distributions of any class for different seismic intensities. The binomial distribution has
the advantage of needing one parameter only which ranges between 0 and 1. On the
other hand it has the disadvantage of having both mean and standard deviation
depending on this unique parameter. The buildings were separated into three vulnerability
classes (A, B and C) and a DPM based on the MSK scale was evaluated for each class.
This type of method has also been termed direct (Corsanego and Petrini, 1990) because
159
there is a direct relationship between the building typology and observed damage. The
use of DPMs is still popular in Italy and proposals have recently been made to update the
original DPMs of Braga et al. (1982). Di Pasquale et al. (2005) have changed the DPMs
from the MSK scale to the MCS scale (Mercalli-Cancani-Sieberg) because the Italian
seismic catalogue is mainly based on this intensity and the number of buildings has been
replaced by the number of dwellings so that the matrices could be used in conjunction
with the 1991 ISTAT data. Dolce et al. (2003) have also adapted the original matrices as
part of the ENSeRVES project (European Network on Seismic Risk, Vulnerability and
Earthquake Scenarios) for the town of Potenza, Italy. An additional vulnerability class D
has been included, using the EMS98 scale (Grntal, 1998), to account for the buildings
that have been constructed since 1980, which should have a lower vulnerability as they
have either been retrofitted or designed to recent seismic codes.
Table 3.1 Format of the Damage Probability Matrix proposed by Whitman et al. (1973)
Damage
State
Structural
Damage
0
1
2
None
None
None
Not
noticeable
4
5
6
7
8
NonStructural
Damage
Damage
Ratio (%)
None
Minor
Localised
Intensity of Earthquake
V
VI
VII
VIII
IX
0 0.05
0.05 0.3
0.3 1.25
10.4
16.4
40.0
0.5
22.5
Widespread
1.25 3.5
20.0
30.0
2.7
Minor
Substantial
Substantial
Extensive
Major
Nearly total
Building condemned
Collapse
3.5 4.5
7.5 20
20 65
100
100
13.2
-
47.1
0.2
-
92.3
5.0
-
58.8
41.2
-
14.7
83.0
2.3
-
Damage Probability Matrices based on expert judgement and opinion were first
introduced in ATC-13 (ATC, 1985). More than 50 senior earthquake engineering experts
were asked to provide low, best and high estimates of the damage factor (the ratio of loss
to replacement cost, expressed as a percentage) for Modified Mercalli Intensities (MMI)
from VI to XII for 36 different building classes. The low and high damage factor
estimates provided by the experts were defined as the 90% probability bounds of a
lognormal distribution, whilst the best estimate was taken as the median damage factor
(Figure 3.1). Weighted means of the experts estimates, based on the experience and
confidence levels of the experts for each building class, were included in the averaging
process, as described in Appendix G of ATC-13 (ATC, 1985). A lognormal distribution
was then used to calculate the probability of a central damage factor by finding the area
160
below the curve within a given damage factor range. Thus, a DPM can be produced for
each intensity level for each building class. Examples of the use of DPMs based on the
ATC-13 approach for the assessment of risk and loss include the city of Basel (Fah et al.,
2001), Bogot (Cardona and Yamin, 1997) and New Madrid (Veneziano et al., 2002).
5%
5%
ML
4.00
MB
10.60
MH
18.20
Figure 3.1 Example of a lognormal distribution of the estimated damage factor for a given intensity
showing the mean low (ML), mean best (MB) and mean high (MH) estimates of the damage factor
(adapted from McCormack and Rad, 1997)
A macroseismic method has recently been proposed (Giovinazzi and Lagomarsino, 2001;
2004) that leads to the definition of damage probability functions based on the EMS-98
macroseismic scale (Grnthal, 1998). The EMS-98 scale defines qualitative descriptions
of Few, Many and Most for five damage grades for levels of intensity ranging from
V to XII for six different classes of decreasing vulnerability (from A to F). Damage
matrices containing a qualitative description of the proportion of buildings that belong to
each damage grade for various levels of intensity are presented in Table 3.2 for
vulnerability class C.
The problems related to the incompleteness of the matrices (i.e. the lack of information for
all damage grades for a given level of intensity) and the vagueness of the matrices (i.e. they
are described qualitatively) has been tackled by Giovinazzi and Lagomarsino (2004) by
assuming a beta damage distribution and by applying Fuzzy Set Theory, respectively. The
damage probability matrices produced for each vulnerability class have been related to the
building stock through the use of an empirical vulnerability index which depends on the
building typology, the characteristics of the building stock (e.g. number of floors,
irregularity etc.) and the regional construction practices. This macroseismic method has
already been applied in the risk assessment of the City of Faro (Oliveira et al., 2004),
Lisbon (Oliveira et al., 2005) and Barcelona (Lantada et al., 2004).
161
Table 3.2 Example of a damage model for vulnerability class C, as presented in EMS-98 (Grnthal,
1998)
Class C
Damage Level
Intensity
V
VI
Few
VII
Few
VIII
Many
IX
X
Few
Many
Few
Many
Few
XI
Many
XII
Most
Following the introduction of DPMs based on intensity, the assessment of seismic risk on
a large scale was made possible in both an efficient and cost-effective manner because in
the past seismic hazard maps were also defined in terms of macroseismic intensity. The
use of observed damage data to predict the future effects of earthquakes also has the
advantage that when the damage probability matrices are applied to regions with similar
characteristics, a realistic indication of the expected damage should result and many
uncertainties are inherently accounted for. However, there are various disadvantages
associated with the continued use of empirical methods such as DPMs:
The derivation of empirical vulnerability functions requires the collection of postearthquake building damage statistics at sites with similar ground conditions for a
wide range of ground motions: this will often mean that the statistics from multiple
earthquake events need to be combined. In addition, large magnitude earthquakes
occur relatively infrequently near densely populated areas and so the data available
tends to be clustered around the low damage/ground motion end of the matrix thus
limiting the statistical validity of the high damage/ground motion end of the matrix.
Seismic hazard maps are now defined in terms of PGA (or spectral ordinates) and
thus PGA needs to be related to intensity in order to use the previously proposed
162
The vulnerability index method (Benedetti and Petrini, 1984; GNDT, 1993) has been used
extensively in Italy in the past few decades and is based on a large amount of damage
survey data; this method is indirect because a relationship between the seismic action
and the response is established through a vulnerability index. The method uses a field
survey form to collect information on the important parameters of the building which
could influence its vulnerability: for example, plan and elevation configuration, type of
foundations, structural and non-structural elements, state of conservation and type and
quality of materials. There are eleven parameters in total, which are each identified as
having one of four qualification coefficients, Ki, in accordance with the quality conditions
- from A (optimal) to D (unfavourable) - and are weighted to account for their relative
importance. The global vulnerability index of each building is then evaluated using the
following formula:
Iv =
11
KW
i
(3.1)
i =1
The vulnerability index ranges from 0 to 382.5, but is generally normalised from 0 to 100,
where 0 represents the least vulnerable buildings and 100 the most vulnerable. The data
from past earthquakes is used to calibrate vulnerability functions to relate the
vulnerability index (Iv) to a global damage factor (d) of buildings with the same typology,
for the same macroseismic intensity or peak ground acceleration (PGA). The damage
factor ranges between 0 and 100% and defines the ratio of repair cost to replacement
cost. The damage factor is assumed negligible for PGA values less than a given threshold
and it increases linearly up until a collapse PGA, from where it takes a value of 100%
(Figure 3).
163
Damage factor, d
Iv = 80
60
40
20
10
0.8
0.6
0.4
0.2
0
0
0.2
0.4
0.6
0.8
PGA
Figure 3.2 Vulnerability functions to relate damage factor (d) and peak ground acceleration (PGA)
for different values of vulnerability index Iv (adapted from Guagenti and Petrini, 1989)
RISK_UE (An advanced approach to earthquake risk scenarios with application to different European
towns) was a major research project financed by the European Commission (see www.riskue.net). Seven European cities (Barcelona, Bitola, Bucharest, Catania, Nice, Sofia and
Thessaloniki) were involved in the project, whose main objective was to develop a general
methodology for the seismic risk assessment of European towns. The vulnerability index
method was adopted as one of the vulnerability assessment procedures which was
developed and successfully applied to all of the aforementioned cities.
The Catania Project, discussed in Faccioli et al. (1999) and GNDT (2000), used an adapted
vulnerability index method for the risk assessment of both masonry and reinforced
concrete buildings. Some modifications to the original vulnerability index procedure were
applied in that a rapid screening approach was used to define the vulnerability scores of
the buildings, following the guidelines of ATC-21 (ATC, 1988). As in the original
procedure, the vulnerability score was obtained from the weighted sum of eleven
parameters; however, only some were directly obtained from a field assessment, while the
rest were based on a range of values according to historic or recent construction practice
within the region, thus leading to a lower and upper bound to the Iv for each building.
Vulnerability functions for old masonry buildings were calibrated to the damage observed
after the 1976 Friuli and the 1984 Abruzzo earthquakes, and damage and peak ground
acceleration were correlated through the relationship proposed by Guagenti and Petrini
(1989), shown in Figure 3.3.
The main advantage of indirect vulnerability index methods is that they allow the
vulnerability characteristics of the building stock under consideration to be determined,
rather than base the vulnerability definition on the typology alone. Nevertheless, the
164
methodology still requires expert judgement to be applied in assessing the buildings, and
the coefficients and weights applied in the calculation of the index have a degree of
uncertainty that is not generally accounted for. Furthermore, in order for the vulnerability
assessment of buildings on a large (e.g. national) scale to be carried out using vulnerability
indices, a large number of buildings which are assumed to represent the national building
stock need to be assessed and combined with census data (e.g. Bernardini, 2000); in a
country where such data are not already available, the calculation of the vulnerability
index for a large building stock would be very time consuming. However, of course, in
any risk or loss assessment model a detailed collected of input data is required for
application at a national scale.
3.1.3 Continuous Vulnerability Curves
Continuous vulnerability functions based directly on the damage of buildings from past
earthquakes were introduced slightly later than DPMs; one obstacle to their derivation
being the fact that macroseismic intensity is not a continuous variable. This problem was
overcome by Spence et al. (1992) through the use of their Parameterless Scale of Intensity
(PSI) to derive vulnerability functions based on the observed damage of buildings using
the MSK damage scale (Figure 3.3). Orsini (1999) also used the PSI ground-motion
parameter to derive vulnerability curves for apartment units in Italy. Both studies
subsequently converted the PSI to peak ground acceleration (PGA) using empirical
correlation functions, such that the input and the response were not defined using the
same parameter.
1.0
0.8
0.6
0.4
D1
D2
D3
D4
D5
0.2
0.0
0
10
15
20
PSI
Figure 3.3 Vulnerability curves produced by Spence et al. (1992) for bare moment resisting frames
using the Parameterless Scale of Intensity (PSI). D1 to D5 relate to damage states in the
MSK scale
165
Figure 3.4 Example of the difference in the vulnerability point distribution using observations of
low and mid-rise building damage for different ground motion parameters: (a) PGA and (b) Spectral
displacement at the elastic fundamental period (Rossetto and Elnashai, 2003)
166
The introduction of vulnerability curves based on spectral ordinates, rather than PGA or
macroseismic intensities, has also certainly been facilitated by the emergence of more and
more attenuation equations in terms of spectral ordinates.
3.1.4 Screening Methods
I S = Eo S D T
(3.2)
IS IS0, corresponds to a low vulnerability condition for all three screening levels;
IS < IS0, is an uncertain condition when IS is only slightly smaller than IS0 (IS falls
within an interval between IS0 and a lower limit that is based on the observed damage
of reinforced concrete buildings) and thus requires a more detailed assessment using
the next screening level or nonlinear dynamic analysis.
I S 0 = ES ZGU
167
(3.3)
where ES was originally taken as 0.6 for the second and third level screening and 0.8 for
the first (hence providing an increase in the demand due to a lack of reliability of the
method); Z is a zone index used to modify the intensity of the ground motion assumed at
the site of the building; G is the ground index used to account for ground-building
interaction, amplification in the top layer of ground, or topographical effects; and U is a
usage index which may be seen as a kind of importance factor concerned with the
function of the building. This seismic judgement index suggests a storey shear force
which is independent of the ground motion, but the force obviously depends on the
response of the building and so in the 1998 revised version of the Japanese Building
Standard Law the index IS0 is taken as the spectral acceleration (divided by g) at the
period of response of the structure in question. This index should be distributed up the
height of the structure, and a triangular distribution is suggested.
The rapid assessment of the seismic performance of existing reinforced concrete
buildings has received particular attention in Turkey in recent years. Several preliminary
assessment methods have been proposed that require the dimensions of the lateral load
resisting elements to be defined. Hassan and Sozen (1997) have proposed a procedure to
define a Priority Index for each building which is a function of a wall index (area of walls
and infill panels divided by total floor area) and a column index (area of columns divided
by total floor area). Damage data can be used to calibrate the Priority Index and define the
level of vulnerability of the building. Yakut (2004) has recently proposed a Capacity Index
which considers the orientation, size and material properties of the lateral load-resisting
structural system as well as the quality of workmanship and materials and architectural
features such as short columns and plan irregularities. Buildings are classified as either at
low risk (and thus should not suffer severe damage) or at high risk (and thus would
probably not meet the life-safety performance level of recent seismic codes). These rapid
vulnerability assessment methods are of particular use for prioritising buildings for
further, more detailed assessments that would be required to design a rehabilitation
scheme.
The Seismic Safety Screening Method (SSSM) has recently been proposed by Ozdemir et
al. (2005), which is an adaptation of the Japanese Seismic Index Method (JBDPA, 1990).
This rapid seismic safety evaluation method can be applied to structures with 6 storeys or
less with either reinforced concrete frame, shear wall or dual frame-shear wall structural
systems. As with the Japanese method, the principle of the method is that the seismic
capacity of a building is represented by a seismic index value, Is, which is a function of its
strength and ductility. If the building has a seismic index value that is lower than a seismic
demand index value (ID, dependent on the seismicity of the region of interest) then
detailed structural analysis would be necessary for further assessment of the vulnerability,
168
otherwise the building can be judged as safe. The method has been calibrated to Turkish
buildings using nonlinear static analyses of 12 buildings from Zeytinburnu, Istanbul
(Ozdemir et al., 2005).
The use of rapid screening methods has an important role to play in the definition of
prioritisation of buildings for seismic retrofit, but the use of such methods in large scale
seismic risk models is limited due to the need to consider buildings individually in a
deterministic fashion, and thus it would not be economically feasible.
3.2 ANALYTICAL/MECHANICAL METHODS
Although vulnerability curves and damage probability matrices have traditionally been
derived using observed damage data, recent proposals have made use of computational
analyses to overcome some of the drawbacks of these methods that have been
highlighted in the previous section. Figure 3.5 summarises the basic components which
are required to analytically derive vulnerability curves or damage probability matrices.
Selection of
computational model
of structure
Selection of earthquake
intensity indicator
169
Selection of model
for definition of
damage
Definition of random
characterisation of structural
parameters
Definition of damage
states
Selection of representative
set of earthquakes
Selection of methodology
for nonlinear analysis
Nonlinear analysis
Definition of probabilistic
distribution of damage
Vulnerability Curves
Damage Probability
Figure 3.5 Flowchart to describe the components of the calculation of analytical vulnerability curves
and damage probability matrices (adapted from Dumova-Jovanoska, 2004)
Singhal and Kiremidjian (1996) developed fragility (or vulnerability) curves and damage
probability matrices for three categories of reinforced concrete frame structures using
Monte Carlo simulation. The probabilities of structural damage were determined using
nonlinear dynamic analysis with an ensemble of ground motions. For the DPMs,
Modified Mercalli intensity was used for the ground-motion parameter, whilst spectral
acceleration was used in the generation of fragility functions. The major components of
the methodology consist of: 1) characterisation of the structure when subjected to
dynamic loads; 2) characterisation of the potential ground motions; and 3) quantification
of the structural response accounting for the variability in the ground motion and the
uncertainty in the structural response. Nonlinear dynamic analyses were carried out using
an ensemble of time-histories, corresponding to a given level of ground motion, for many
buildings with random structural characteristics. The output of each nonlinear analysis
was used to calculate a global damage index (related to a particular damage state), based
on the Park and Ang (1985) model (because it is both simple and has been calibrated
using data from various structures damaged during past earthquakes). Statistical analysis
of the damage indices led to the evaluation of the probability of different damage states
and thus fragility functions and DPMs were evaluated (Figure 3.6). The analytical
170
vulnerability curves produced by Singhal and Kiremidjian (1996) for low-rise frames
described were subsequently updated using observational data obtained from a tagging
survey of 84 buildings damaged by the 1994 Northridge earthquake using a weighting
system (Bayesian updating technique) to take into account the reliability of different data
sources (Singhal and Kiremidjian, 1998).
(a)
(b)
Figure 3.6 (a) Fragility curves and (b) damage probability matrix for sample low rise buildings
(Singhal and Kiremidjian, 1996)
171
172
history records were scaled to PGA values estimated by seismic hazard analysis; intensity
and PGA were correlated using empirical relationships. A global damage index was
derived to correlate the structural response from the dynamic analysis (ductility factors,
displacements etc.) to loss, expressed in terms of cost of repair. A total of 120 analyses of
typical Greek buildings designed to the 1959 code were run (6 structures, 10 ground
motions and 2 intensities), and the statistical damage results were combined with the
observed damage from the 1978 earthquake in Thessaloniki.
Barbat et al. (1996) used the Italian vulnerability index methodology for a hybrid vulnerability
assessment of Spanish urban areas. A post-earthquake study was initially performed for
two earthquakes with a maximum intensity of VII on the MSK scale. The structural and
non-structural damage to masonry structures was analysed and correlated to the
vulnerability and damage indices used in the Italian methodology. Statistical analyses were
then performed to obtain the vulnerability function for intensity level MSK VII. A
computer simulation process was subsequently used to obtain the vulnerability functions
at other intensity levels. Sixty hypothetical buildings with characteristics obtained from
the building stock in the area were generated using Monte Carlo simulation (considering a
uniform probability density function for the capacity parameters) in order to simulate the
behaviour of a complete urban zone and a simplified analytical procedure proposed by
Abrams (1992) was chosen to model the capacity/demand relationship of masonry
structures. The vulnerability index was calculated for each building, and this was plotted
against the global damage index (based on the capacity/demand ratios of the structural
members) for intensity level MSK VII and a curve was obtained by regression analysis;
this curve was then re-calibrated so as to match field observations (Figure 3.7). The
difference between the curves in Figure 3.7 was assumed to be due to the use of the
proposed weighting factors for Italian buildings and so the weighting factors were
modified such that the observed and calculated vulnerability functions converged. Once
the calibration with the 60 random buildings had been carried out, the vulnerability
functions for the other intensity levels were then produced using 2000 hypothetical
buildings in conjunction with the calibrated weighting factors.
173
Figure 3.7 Simulated (thick line) and observed (thin line) vulnerability functions for intensity VII
MSK (Barbat et al., 1996)
The main difficulty in the use of hybrid methods is probably related to the calibration of
the analytical results, considering that the two vulnerability curves include different
sources of uncertainty and are thus not directly comparable. In the analytical curves the
sources of uncertainty are clearly defined during the generation of the curves whilst the
specific sources and levels of variability in the empirical data are not quantifiable. The
method used to calibrate the analytical vulnerability curves using empirical data should
depend on whether the aim is to include the additional uncertainty present in the
empirical data which is not accounted for in the analytical data or whether the aim is in
fact to improve the analytical model used to define the capacity of the building stock. If
the principle goal is in fact the latter, then perhaps it would be preferable to calibrate the
vulnerability curves considering the median (50 percentile) values and an adaptation of
the analytical model could be made such that the two median values coincide. In this way,
the observational data is used to calibrate the analytical model, but analytically-derived
vulnerability curves with their known and specified sources of uncertainty are used in the
loss model. This can be of use when only the uncertainty related to the capacity of the
building stock is required and not that related to the ground motion; for example such
curves are needed when a probabilistic seismic hazard assessment is used to define the
ground motion as this will ensure that the latter variability is not double-counted (see for
e.g. Bommer and Crowley, 2006).
3.2.3 Collapse Mechanism-Based Methods
Many recent proposals for analytical vulnerability assessment methods use collapse
multipliers calculated from mechanical concepts to ascertain whether a mechanism will
174
form and thus damage will occur; these procedures have been particularly applied to
masonry buildings.
VULNUS is a method that has been proposed for the vulnerability assessment of
unreinforced masonry buildings (URMB) using fuzzy-set theory and the definition of
collapse multipliers (Bernardini et al., 1990). A collapse multiplier for in-plane behaviour
(I1) with shear failure at the ground floor is defined as the ratio of the in-plane shear
strength of the system of walls to the total weight:
I1 =
min(Vx ,V y )
W
(3.4)
where W is the total weight of the building and Vx and Vy represent the strength at midstorey height of the ground floor in the x and y directions, which can be calculated
considering the tensile strength of the masonry and the total area of the walls. The
collapse multiplier for out-of-plane behaviour (I2) is found from the ratio of out-of-plane
flexural strength of the most critical external wall to the total weight, evaluated by
summing the resistance of vertical ( I 2' ) and horizontal strips ( I 2'' ) :
I 2 = min ( I 2' + I 2'' )i
i
(3.5)
Following the calculation of the two collapse multipliers, a third index I3 is calculated,
which is the weighted sum of the scores of seven partial vulnerability factors:
I 3i =
Wi Si
45 3.15
(3.6)
where the score S ranges from 0 (good) to 45 (poor) and the weights, W, are assigned
depending on the relative importance of the factors on the vulnerability of the building.
The mean absolute acceleration response of the building (A, maximum base shear
divided by total weight) is calculated as well as an uncertainty factor a, which is obtained
from fuzzy-set theory using I3. The probability of exceeding the D4 limit state (based on
the EMS98 damage scale) for a group of buildings is a function of the aforementioned
parameters: Vg = f(I1, I2, A, a).
FaMIVE, Failure Mechanism Identification and Vulnerability Evaluation method
(DAyala and Speranza, 2002) is another procedure based on collapse multipliers, which is
aimed towards the vulnerability assessment of historic buildings and town centres. The
most probable collapse mechanism is found for both in-plane and out-of-plane failure by
calculating the load factor or collapse multiplier through an equivalent static procedure. A
number of possible out-of-plane collapse mechanisms are assumed (see Figure 3.8); the
175
equivalent shear capacity is calculated for each faade wall of the buildings under
consideration for each collapse mechanism, and the most likely mechanism is identified
as that with the lowest capacity.
B1
B2
Figure 3.8 Out-of-plane failure mechanisms (after DAyala and Speranza, 2002)
The ultimate roof displacement u,i is determined as a function of the ultimate rotation, u
of the structural elements:
176
u ,1 = u ( H n H k )
(3.7)
u ,2 = u H k
(3.8)
u ,3 = u ( H k H k 1 )
(3.9)
The global seismic behaviour is then represented by the base shear coefficient Cb,i (base
shear Vb,i divided by seismic weight, W) and the corresponding lateral (driftu)i (roof
displacement divided by building height):
Cb ,i =
Vb ,i
W
; (driftu )i =
u ,i
Hn
(3.10)
Seismic response is evaluated for 3nz mechanisms, where nz represents the number of
storeys; the seismic capacity corresponds to the lowest Cb of all the considered
mechanisms and the associated drift. The seismic capacity of building stock will vary
from the generic building due to different morphologic/geometric/structural
configurations and material properties. Hence a number of building models are generated
based on the probabilistic distribution of the structural parameters. A Monte Carlo
simulation technique is applied to calculate probabilistic capacity curves which express
the probability of having a capacity lower than the assigned value: this can be likened to
the percentage of buildings with a value of capacity lower than a given threshold value.
A clear framework for the treatment of uncertainty in the geometric and material
properties in an urban environment is accounted for in the method by Cosenza et al.
(2005), whilst this is not the case for the VULNUS and FaMIVE procedures. However,
the main disadvantage of the procedure by Cosenza et al. (2005), and also FaMIVE, is
that no clear indication is given on how the capacity is to be convolved with the demand
to calculate the probability of exceeding given limit states. The VULNUS procedure does
allow the estimation of the probability of damage, but only for one limit state (the
collapse limit state). Thus, the use of these collapse-mechanism methods within a loss
assessment model appears to be somewhat limited at present.
3.2.4 Capacity Spectrum-Based
HAZUS (Hazard US) is the outcome of a project conducted for the National Institute of
Building Science (NIBS), under a cooperative agreement with the Federal Emergency
Management Agency (FEMA), to develop a nationally applicable methodology for
estimating the potential losses from earthquakes on a regional basis (Whitman et al., 1997;
FEMA, 1999; 2003). The main modules of the methodology are as follows:
177
The estimation of the Potential Earth Science Hazards (PESH); these include ground
motion, ground failure (liquefaction, landslides and surface fault rupture) and
tsunami/seiche.
Direct physical damage is calculated for each building class to find the probability of
none, slight, moderate, extensive and complete structural and non-structural damage
due to each type of PESH.
Induced physical damage from inundation, fire, hazardous materials and debris is
estimated.
Indirect economic losses due to downtime, that causes a chain reaction through the
regional economy, are estimated.
The vulnerability assessment component of the procedure is contained within the direct
physical damage module and is based on the Capacity Spectrum Method of ATC-40
(ATC, 1996) wherein the performance point of a building type under a particular groundshaking scenario (or PESH) is found from the intersection of an accelerationdisplacement spectrum, representing the ground motion, and a capacity spectrum
(pushover curve), representing the horizontal displacement of the structure under
increasing lateral load (Kircher et al., 1997), as shown in Figure 3.10. The demand
spectrum is reduced for both damping and duration effects. The reduction of the
spectrum is applied to account for the hysteretic damping that occurs during inelastic
behaviour of the structure, where the damping is based on the area enclosed by the
hysteretic loop at peak response displacement and acceleration. A reduction factor is
applied to the hysteretic damping as a function of shaking duration to simulate
degradation (e.g. pinching) of the hysteresis loop during cyclic response.
178
Figure 3.10 Illustration of the estimation of damage from ground shaking in HAZUS (FEMA, 1999;
2003)
The capacity spectrum has been developed for each building class using model buildings
designed to different levels of design practice in the US. The performance point obtained
from this average building provides the displacement input into limit state vulnerability
curves to give the probability of being in a given damage band. The vulnerability curves
are lognormal curves with a total logarithmic standard deviation, Sds, which combines the
uncertainty in the damage state threshold, M(Sds), the variability in the capacity (response)
properties of the model building type, C, and the uncertainty in the response due to the
spatial variability of the ground-motion demand, D. The latter two components of the
variability are inter-dependent (as the damped demand spectrum is dependent on the
building capacity due to the nonlinearity of the capacity response and thus energy
dissipation), and thus need to be combined using a complex process of convolution
(CONV) of the probability distributions, before being combined with M(Sds) using the
square-root-of-the-sum-of-the-squares (SRSS):
Sds =
(CONV [
D S d , Sds
]) + (
2
M ( Sds )
179
(3.11)
where Sd,Sds is the median spectral displacement for the damage state, ds. Median spectral
displacement (or acceleration) values and the total variability are developed for each of
the model building types and damage states of interest through a combination of
performance data (from tests of building elements), earthquake field data, expert opinion
and judgement.
Once the proportion of buildings in each damage band (none, slight, moderate, extensive,
and complete) has been obtained, a composite measure of damage, which aims to provide
an idea of the expected overall rebuilding cost, is calculated by summing the proportions
which are weighted by 0% for no damage, 2% for slight damage, 10% for moderate
damage, 50% for extensive damage and 100% for complete damage.
The HAZUS methodology was originally derived as a tool for estimating the impact of
individual earthquake scenarios; it has been subsequently adapted (FEMA, 2001) to a full
loss assessment methodology wherein the hazard from all possible earthquakes, derived
from probabilistic seismic hazard assessment (PSHA), is considered. In this case, the
variability in the demand is already modelled in the PSHA and so needs to be removed
from the vulnerability curves to avoid double counting; the curves thus become
lognormal functions with a logarithmic standard deviation given by:
Sds =
( C )2 + ( M ( Sds ) )2
(3.12)
The use of PSHA in regional loss models, although computationally efficient, is less
theoretically robust than the representation of the seismic hazard by triggering a large
number of earthquake scenarios that are compatible in magnitude, location and
associated frequency of occurrence with the regional seismicity (see e.g. Crowley and
Bommer, 2006; Bazzurro and Luco, 2005 and Chapter 5 of these notes).
A potential weakness of the method is that the capacity curves and vulnerability functions
published in the HAZUS manual have been derived for buildings in the US having a
limited range of storey heights, thus the application of this method to other parts of the
world requires additional research to be carried out. Hence, capacity curves and
vulnerability functions would need to be derived for the building stock under
consideration which would require a significant analytical effort.
HAZUS has been adopted all over the world for the loss assessment of urban areas. The
methodology in itself has not been adapted in any way, but the capacity curves and
fragility functions have been calibrated to the building stock under consideration.
180
Examples include the loss assessment of Turkey carried out by Bommer et al. (2002), the
seismic risk assessment of Oslo documented by Molina and Lindholm (2005), the loss
estimation of Taiwan (using a modified version of HAZUS called Haz-Taiwan) discussed
in Yeh et al. (2000) and the RISK_UE project for the risk assessment of 7 European
towns that has been discussed previously.
The LNECloss tool is an automatic seismic scenario loss estimation methodology that
makes use of the capacity spectrum method and has been implemented in a GIS
environment (Sousa et al., 2004). However, an equivalent approach to the capacity
spectrum method to evaluate the performance point has been implemented in the
LNECloss code. This alternative equivalent scheme to compute the performance point
starts with the definition of an input motion in terms of a power spectral density function
(PSDF) and an equivalent stationary duration, T. The seismic input needs to be defined in
terms of response spectra, associated to some damping coefficient, and thus a linear
stochastic approach has been implemented in LNECloss to accomplish this
transformation. This methodology illustrated in Figure 3.11 was developed and
implemented within the LESSLOSS Integrated Project (Calvi and Pinho, 2004) under the
earthquake disaster scenario predictions and loss modelling for urban areas sub-project,
and was applied to the city of Lisbon (see www.lessloss.org/main).
Giovinazzi (2005) presents a mechanical procedure for the risk assessment of both
masonry and reinforced concrete frames. The procedure uses simplified bilinear capacity
spectra (capacity or pushover curves converted to plots of spectral acceleration versus
spectral displacement) which are derived using the equations and parameters available in
seismic design codes. The base shear coefficient (which can be related to the yield
spectral acceleration) is generally a function of seismic zone, soil conditions, building
dynamic response, structural type and building importance, all of which can be obtained
from the code. The yield spectral displacement is a function of the yield spectral
acceleration and the yield period of vibration; the latter being calculated using the simple
formulae available in seismic design codes which usually give the period as a function of
the building height. The ultimate spectral displacement is a function of the yield
displacement and the building ductility capacity which has been assumed to be 2.5 for
reinforced concrete buildings that have not been seismically designed, or it can be derived
as a function of the behaviour factor specified in the code. The use of seismic design
codes to calculate simplified pushover curves of the building stock is a cost-effective
method for calibrating the vulnerability of buildings designed to different design codes.
However, code-based definitions of the yield period of vibration and base shear capacity
of RC structures could differ greatly from the actual properties of the building stock (see
for e.g. Crowley, 2003) and in many countries the existing building stock has been
constructed without conforming to a seismic design code.
181
Figure 3.11 The modules involved in the LNECloss tool (Sousa et al., 2004)
In the Giovinazzi (2005) approach, once capacity spectra for the building classes have
been derived the next step in the proposed procedure is to use the Capacity Spectrum
Method which has been developed and employed in HAZUS (1999; 2003). As has been
described in the review of HAZUS, the performance point of the building class is
obtained by comparing the capacity spectrum with the overdamped AccelerationDisplacement Response Spectrum (ADRS) which is produced using codified spectral
shapes anchored to the peak ground acceleration (PGA) obtained from the hazard
analysis. Once the performance point is obtained it is inserted into vulnerability curves in
order to obtain the probability of exceeding pre-defined damage states. The method
proposed by Giovinazzi for this stage of the risk assessment differs from that proposed
by HAZUS, as discussed in what follows.
182
The mean values for the displacement thresholds (Sd,i) employed by Giovinazzi (2005) are
functions of yield and ultimate displacement:
S d ,1 = 0.7 d y
(3.13)
S d , 2 = 1 .5 d y
(3.14)
S d ,3 = 0.5( d y + d u )
(3.15)
S d ,1 = d u
(3.16)
These thresholds have been based on expert judgement and have been verified on the
basis of the results of pushover analyses performed on prototype buildings, though
details of these analyses have not been reported. In order to model the variability in the
estimation of the damage suffered by the building (which is due to the uncertainty in the
capacity curve, the damage state definition and the seismic demand spectra), rather than
combining the logarithmic standard deviations assigned to each of these uncertainties (as
carried out in HAZUS), the proposed procedure is to evaluate the overall uncertainty in the
damage estimation in such a way that it can represent the same dispersion of observed damage data
that.are well fitted by a binomial distribution. Giovinazzi (2005) discusses that the binomial
distribution has been shown to fit damage data by Braga et al. (1982) and Spence et al.
(2003); however, Giovinazzi admits that there are problems with the use of this
distribution as it does not allow a variation of the scatter around the mean damage band
(i.e. it is fixed) and Spence et al. (2003) have shown that frame structures may follow a
very different, more complex distribution (see Figure 3.12 where observed damage data
from the Kocaeli earthquake of 1999 is presented and compared with the results from
various risk assessment procedures for reinforced concrete frames).
Indeed, the authors believe that constricting the damage distribution to follow a binomial
distribution could lead to erroneous damage predictions in many cases where non-ductile
reinforced concrete frames are considered in a loss model. The observed damage
distribution presented in Figure 3.12 can be easily understood by considering the
behaviour of such frames: once the first/yield limit state has been exceeded, the
inadequate confinement in these buildings means that they will very rapidly also fail the
higher limit states due to the low level of attainable limit state strains (and consequently
low levels of ductility). For this reason, the vulnerability curves for the different damage
states for these frames should be steep and close together. It is therefore expected that
the damage distribution will show larger proportions at the extreme damage bands (i.e.
slight and complete damage) with lower proportions being predicted for the intermediate
damage states for moderate ground-motion input. For well-designed buildings, however,
183
there should be a greater separation between the curves, representing the greater postyield capacity due to the higher limit state strains of adequately confined members, and
thus a binomial/normal/lognormal type of damage distribution could be expected.
Figure 3.12 Kocaeli earthquake (1999), Glck, Coastal Zone: results of various analyses and
comparison with damage data for mid-rise reinforced concrete frames (Spence et al., 2003)
184
mfi
Fi
meff
j
F
Fy
eff
hT
hi
Keff
LS
(a)
(b)
(c)
(d)
Figure 3.15 Deformed shapes for different limit states and in-plane failure modes
Calvi (1999) considered the inherent variability in the structural properties within an
urban environment by assigning maxima and minima to the variables and assuming a
185
uniform probability distribution function. The period of vibration was calculated using
the empirical formula in EC8 (CEN, 2003) which directly relates the height of a building
to its period; again maxima and minima were applied to the parameters in this equation.
The range of the limit state displacement capacity for the building class can be plotted
against the range of the period of vibration, as presented in Figure 3.16.
Figure 3.16 Example of intersection of capacity areas and demand spectrum (Calvi, 1999)
This capacity area can then be directly compared with the displacement response
spectrum; the area below the spectrum represents the proportion of buildings failing, or
exceeding, the limit state. The displacement response spectra were adjusted to include
nonlinear response, wherein a reduction of the spectral ordinates was applied to account
for the energy dissipation capacity of the structure as a function of the target
displacement and the structural response.
The methodology proposed by Calvi (1999) has subsequently been developed for
reinforced concrete buildings by Pinho et al. (2002) and Crowley et al. (2004, 2006),
leading to the Displacement-Based Earthquake Loss Assessment (DBELA) procedure.
For the masonry component of the methodology, a parallel extensive development was
carried out by Restrepo-Velez and Magenes (2004; 2005a; 2005b), leading to the
procedure MeBaSe. Borzi et al. (2008a) took aspects of the MeBaSe method and the
186
DBELA method and produced the SP-BELA method, which is a simplified pushoverbased method. Chapter 4 looks at these recently proposed mechanics-based
methodologies in more detail.
3.2.6 General Evaluation of Analytical Methods
Perhaps one of the main reasons for which analytical/mechanical methodologies have
only recently been proposed for use in risk and loss assessment is due to developments
which have occurred in the field of seismic hazard assessment, such as the derivation of
seismic hazard maps in terms of spectral ordinates as opposed to macroseismic intensity
or PGA. However, the benefits which the use of more advanced seismic hazard and
vulnerability models offer need to be weighed against the increased amount of
information, and thus time and money, which is required to construct an earthquake loss
model based on such analyses. For example, apart from the compilation of databases of
building stock inventory (which are required for any loss model), information on the
structural characteristics of the building stock are also required in order to calibrate the
analytical model and databases of the earthquake activity, attenuation equations and
ground conditions are needed to carry out a seismic hazard assessment in terms of
spectral ordinates. Nevertheless, the use of such detailed and transparent loss models
allows detailed sensitivity studies to be undertaken which provide valuable insight into
how much the loss results depend on the models, data, uncertainties and assumptions
employed.
However, there are still some outstanding issues with some analytical/mechanical
methods which need to be addressed, such as the capability of numerical models to
accurately predict the response of real structures; the accuracy in transforming numerical
indices of damage into actual damage of real structures; the capability of accounting for
human errors in the design and construction of buildings, which are often the main
causes of catastrophic collapses; and the need to extend the results obtained for few
reference models to a large class of structures, when accurate analyses are carried out.
3.3 CONCLUSIONS
187
Following the presentation of the major advantages and drawbacks of each methodology,
it may be summarised that an optimal or ideal vulnerability assessment methodology
should feature the following main characteristics:
The most recent developments in the field of seismic hazard assessment should be
incorporated.
The model should be easily adaptable to the different construction practices around
the world, as well as allow for the inclusion of new construction types and the
influence of retrofitting on the response of existing structures.
A balance should be struck between the computational intensity and the amount of
detailed data which is required and the consequent degree of confidence in the
results.
However, it is unlikely that a single methodology can be produced which fulfils all of
these requirements. For example, many analytical/mechanical models, though
theoretically superior, require a large amount of detailed data, but the benefit of collecting
such data is often not proven through validation of the methodology with empirical
methods based on observed damage data. On the other hand, the derivation of
vulnerability curves from observed data does not always consider the frequency
characteristics of the building stock, and the influence of incorrectly modelling the
seismic demand experienced by the buildings in the damaged region is normally
unaccounted for. Furthermore, many methodologies do not explicitly model the various
sources of uncertainty; this can be a problem when, for example, the uncertainty in the
seismic demand needs to be removed from the vulnerability assessment calculations.
Thus, it appears that the ideal approach for the future needs to be a combination of the
positive aspects of different vulnerability assessment methodologies. In other words,
reliable vulnerability assessment of a given region is likely to call for the employment of at
least two different approaches, which should complement and/or verify each other.
This chapter looks in more detail at analytical or mechanics-based methods for the
vulnerability assessment of urban populations of buildings. Particular focus is given to the
methodologies being developed by the Seismic Risk Section of the EUCENTRE.
4.2 DISPLACEMENT-BASED EARTHQUAKE LOSS ASSESSMENT (DBELA)
,
y
y
t
i
t
i
ci
l
t
,ac
p
d
u
oad
i
rc
td
e
pn
n
ea
em
,
n s
e
i
f
s
ce
e
an
Dl
pf
f
st
i
i
d
s
y
a
e
.
i
mw
S
si
n
m
aa
he
cB
er
mo
ehn
t
m
eu
n
l
io
f
C
e
D.
3
S
L
2
S
L
1
S
L
e
vt
i
t
h
c
g
ei
f
e
f
H
E
tt
a
h
ti
g
n
ee
mH
ee
v
ci
a
lt
c
p
e
sf
f
i
D
E
3
S
L
2
S
L
1
S
L
The substitute structure approach (Shibata and Szen, 1976) is thus applied which
requires the calculation of the effective period of vibration of the building at different
limit states to damage (Figure 4.1); the damping of the structure is accounted for not in
190
the capacity of the system but in a reduction of the demand by reducing the displacement
response spectrum with a damping reduction factor. This reduction factor is a function of
the predicted equivalent viscous damping of the system, which in turn is a function of the
system ductility.
In the DBELA methodology, different building classes are defined as a function of the
assumed response mechanism; for reinforced concrete buildings which were designed to
modern capacity design principles a global response mechanism (beam-sway) may be
assumed, whilst those which were not seismically designed may predominantly
beassumed to have a soft ground storey response mechanism (column-sway) (Figure 4.2a
and b, respectively). However, a sway potential index - which relates the
strength/stiffness of the beams to the strength/stiffness of the columns - can be used to
identify the mechanism of a given building within the building class. The mechanism may
still be column-sway even if the building has been seismically designed and vice-versa, as
will be discussed in more detail in Section 4.2.2. Masonry buildings are currently assumed
to have a storey-sway response mechanism at the ground floor (Figure 4.2c).
The generic formula for the displacement capacity at the centre of seismic force of the
beam-sway mechanism is presented in Equation (4.1) whilst that for column-sway/storey
sway mechanisms is shown in Equation (4.2):
sway
bm
= by 1 H + bp 1 H
LS
(4.1)
(4.2)
where by and cy are the yield chord rotation capacities of the beams and columns (or
walls), respectively; 1 is the effective height coefficient (to obtain the equivalent height
of the deformed SDOF system); H is the height of the building; bp and cp are the plastic
rotation capacities at a given post-yield limit state of the beams and columns, respectively;
hs is the ground floor storey height for RC frames and the pier height for masonry
buildings; 2 is the effective height coefficient of the masonry piers (for RC buildings this
is taken as 1). This concept is further expanded below, whilst further information may be
found in Glaister and Pinho (2003) and Crowley et al. (2004).
(a)
(b)
191
(c)
Figure 4.2 (a) Beam-sway mechanism and (b) column-sway mechanism for RC frames and (c) storey
sway mechanism for masonry buildings
For reinforced concrete buildings, the calculation of the yield and plastic rotation
capacities of the beams and columns for the two mechanisms is based on simple
principles of mechanics related to the behaviour of reinforced concrete (RC) members.
As an introduction to the deformational behaviour of RC members, the yield
displacement and plastic displacement capacity of a cantilever in simple bending will be
presented using a RC cantilever column, as shown in Figure 4.3. It should be noted here
that many of the assumptions and choices made in the derivation of the displacement
capacity equations described in the following passages (such as the effective height
coefficient, the equation to calculate the plastic hinge coefficient, the empirical factors to
account for shear and joint deformations, amongst others) can be easily adapted
depending on the assumptions chosen by the user due to the flexible, transparent nature
of the methodology.
A typical cross section of a column is shown in Figure 4.3a and the strain profile at that
section is shown in Figure 4.3b. At any cross section, the curvature can be computed by
the sum of steel and concrete strains (s and c, respectively) at the two extremes of the
section, divided by the effective depth, d. The column member is presented in Figure
4.3c, with a height hs. When the section at the base of the column reaches the yield
curvature, y, the first limit state is reached and the curvature distribution with height can
be conservatively approximated as triangular (Figure 4.3d). By integrating the curvature
distribution along the length of the deformed member, the tangent yield rotation shown
in Figure 4.3f, ty, is obtained and the yield displacement at the top of the column can be
computed by the moment-area method, as described in what follows.
192
Priestley (1998) first showed how the yield curvature, y, of reinforced concrete sections
is independent of the strength, but rather dependent on the yield strain of the
reinforcement steel and the geometry of the section, as shown in the following equations:
y = 1.7
y = 2.14
y
hb
y
hc
(4.3)
(4.4)
where hb is the height of the beam section, hc is the depth of the column section and y is
the yield strain of the reinforcement steel.
ty
y
hc
p
p
V
b
LS = y + p
s
hc
d'
Extent of
plasticity
hs
lp
s
= (c+s)/d'
(a)
(b)
(c)
y
(d)
y
(e)
(f)
Figure 4.3 (a) typical poorly-confined reinforced concrete column section, (b) strain profile and
definition of curvature, (c) prismatic reinforced concrete cantilever, (d) curvature distribution with
height when the section at base reaches yield curvature (limit state 1), (e) curvature distribution with
height at post-yield limit states, (f) components of total lateral tip deflection. (Not to scale, adapted
from Paulay and Priestley 1992).
4.2.1.2
The tangent yield rotation at the top of the column, ty, shown in Figure 4.3f, is calculated
by integrating the curvature distribution at yield for a column (Equation 4.5) and then this
is increased using empirical factors proposed by Priestley (2003) to account for shear and
joint deformations (Equation 4.6).
ty = y
y hs
hs
h
= 1.07 y s
= 2.14
hc 2
hc
2
y =1.35 ty =1.44 y
hs
hc
193
(4.5)
(4.6)
The moment-area method is then used to find the yield displacement capacity, y, at the
top of the column:
2
y = y
h
2
hs = 0.96 y s
hc
3
(4.7)
The curvature distribution of the column at the post-yield limit states is presented in
Figure 4.3e. The plastic curvature, p, can be found from the difference between the limit
state curvature, LS, and the yield curvature at the base of the section, as shown in the
following equation.
2
y = y
h
2
hs = 0.96 y s
hc
3
(4.8)
The limit state curvature has been approximated by the sum of the limit state steel and
concrete strains at the two extremes of the section (S(LSi) and C(LSi), respectively), divided
by the total depth of the column section. The dotted lines in Figure 4.3e show the
approximate curvature distribution that has been assumed in order to simplify the
integration of the actual curvature profile. In this way, the plastic curvature may be
multiplied by a plastic hinge length, lp, assumed here to be half of the section depth (see
Paulay and Priestley 1992), to give the plastic rotation capacity:
(4.9)
As can be noted from Figure 4.3e, the plastic hinge length does not represent the total
extent of plasticity, but may be considered to be a representative equivalent/effective
length used for obtaining in a simplified manner the correct rotation value. The plastic
displacement at the tip of the cantilever shown in Figure 4.3f is then found by multiplying
the plastic rotation by the height of the column:
P = p hs = ( C ( LSi ) + S ( LSi ) 2.14 y )0.5hs
(4.10)
The total limit state displacement capacity given in Equation (4.11) is finally obtained by
adding the yield displacement, from Equation (4.7), to the plastic displacement.
194
LSi = y + p = 0.96 y
hs2
+ ( C ( LSi ) + S ( LSi ) 2.14 y )0.5hs
hc
(4.11)
In summary, Equation (4.7) is used to define the capacity at the top of the column for
Limit State 1 (yield) and Equation (4.11) for Limit States 2 and 3 (post-yield), where the
two post-yield states are distinguished by the specified allowable concrete and steel limit
state strains.
4.2.1.3
When beam-sway and column-sway RC frames need to be considered, the single-degreeof-freedom system has an assumed curvature distribution for both mechanisms that
varies linearly with a zero value at the mid-span of the member (beam or column,
respectively). Thus the beam and column yield curvatures presented in Equations (4.3)
and (4.4) can be used in conjunction with the assumed curvature distributions, which are
integrated to predict the yield chord rotation of either the beam-sway (by) or the columnsway (cy) frame mechanisms which are used in Equations (4.1) and (4.2), respectively, to
calculate the yield displacement capacity at the centre of seismic force. The resulting yield
displacement capacity (LS1) formulae for beam- and column-sway frames are presented
in Equations (4.12) and (4.13) respectively; these are used to define the first structural
limit state.
sway
bm
= 0.5ef hH T y
Sy
lb
hb
(4.12)
hs
hc
(4.13)
lb
+ 0.5( C ( LSi ) + S ( LSi ) 1.7 y )ef h H T
hb
(4.14)
sway
clSLSi
= 0.43ef h H T y
hs
+ 0.5( C ( LSi ) + S ( LSi ) 2.14 y )hs
hc
195
(4.15)
In most of the recent codes, the strong column-weak beam provision is satisfied by
comparing the beam end moment capacities with column end moment capacities,
including an uncertainty of 15-20% due to slab contribution and possible material
strength variations.
The soft storey condition commonly occurs in buildings with open fronts at the ground
floor or with particularly tall first storeys. Soft storeys are usually revealed by an abrupt
change in interstorey drift. Basically, the columns which have the same section
dimensions in successive floors must also have the same stiffness (essentially this means
the same or similar storey heights) so that a certain floor does not develop significantly
more drift than the others. Although a comparison of the stiffnesses in adjacent storeys is
the direct approach, a simple first step might be to compare the interstorey drifts.
According to FEMA (2000) and the New Zealand draft code (NZSEE, 2002), the
stiffness of lateral force resisting system in any storey shall not be less than 70% of the
stiffness in an adjacent storey above or below, or less than 80% of the average stiffness of
196
the three storeys above or below. Eurocode 8 (CEN, 2003) specifies that there should
not be significant difference in the lateral stiffness of individual storeys and at any storey;
the maximum displacement in the direction of the seismic forces should not exceed the
average storey displacement by more than 20%.
A weak storey irregularity occurs mostly in the absence of infill walls in a certain floor,
which is often the first floor. The storey strength is the total strength of all the lateral
force-resisting elements in a given storey for the direction under consideration. Weak
storeys are usually found where vertical discontinuities exist, or where member size or
reinforcement has been reduced. The result of a weak storey is a concentration of
inelastic activity that may result in the partial or total collapse of the storey. According to
FEMA (2000) and the New Zealand draft code (NZSEE, 2002), the strength of lateral
force resisting system in any storey shall not be less than 80% of the strength in an
adjacent storey, above or below.
The sway potential of a building is best defined by a pushover analysis, or better still by
an adaptive pushover analysis; nevertheless, some approximate prediction methods are
available. The first prediction method, a strength-based index, has been proposed by
Priestley et al. (2007) as described below in detail. Sway potential indices depending on the
element dimensions, or consequently on element stiffness or deformation properties,
have been proposed in different forms by several researchers. Bal (2005) proposed an
index where the possibility of soft storey behaviour is predicted by using element
moment of inertia values in the critical storey. Abo El Ezz (2007) has recently proposed a
deformation-based index in which the sway potential of a certain floor is calculated by
comparing the yield rotation capacities of beam and column members around a joint of a
floor, as described in Section 4.2.2.2.
4.2.2.1
The strength-based sway index proposed by Priestley (1997), relating the relative
strengths of beams and columns at the centroids of all beam/column joints at a given
floor for a given frame, is as follows:
Si =
(M
(M
bl
ca
+ M br )
+ M cb )
(4.15)
where Mbl, Mbr are the beam expected flexural strengths at the left and right of the joint,
respectively, extrapolated to the joint centroid, including the contribution of slab
reinforcement to the strength, and Mca, Mcb are the expected column flexural strengths
above and below the joint, also extrapolated to the joint centroid. If the storey heights
197
above and below the considered floor are the same and the strength-based sway index is
smaller than 0.85, then a column-sway mechanism is unlikely to form, whilst sway index
values higher than 1.00 indicate a column-sway behaviour for the building (Priestley et al.,
2007).
It can be seen from Equation (4.16) that the strength-based index is a section based
formulation which does not take into account the member response. This index is valid
under the assumption that all the floors have similar height, and this consequently means
that the member stiffness is neglected. Nevertheless, as it will be shown in the following
sections, the strength-based sway potential index may fail even for buildings which
have the same storey height for all floors.
4.2.2.2
There are two different ways to arrive at the same formula of the deformation-based sway
index. The first method relies on the observation that the sway index can be related to
some general properties of the building. The probability of having a column-sway
mechanism increases with:
hb / Lb
hc / Lc
(4.16)
where hb and hc are the beam and column section depths, respectively, whilst Lc and Lb
are the column and beam lengths, respectively. The value of the index for ith joint for a
certain floor is:
Ri =
hb , L / Lb , L + hb , R / Lb , R
2(hc , B / Lc , B )
(4.17)
198
where sub-indices L and R refer to Left and Right and B refers Below,
respectively. The index per floor could then be obtained by averaging the result of
Equation (4.17) for each floor:
n
S def , j =
i, j
i =1
(4.18)
Ri =
2 cy , B
Beams
2 cy , B
by , L + by , R
(4.19)
where cy,B is the yield rotation of the column below the joint, by,L is the yield rotation of
the beam to the left of the joint, and by,R is the yield rotation of the beam to the right of
the joint. As given by Priestley et al. (2007), member rotation is a function of a constant,
steel yield strain, member section depth and member length. It can be readily noted that
the steel yield strains and constants would be cancelled, assuming that columns and
beams are constructed with the same steel material. In this case, Equation (4.19) leads to
Equation (4.18).
Regardless of the difference in storey heights, the deformation-based sway index is
assumed to indicate a beam-sway mechanism below 1.00 and a column-sway mechanism
above 1.20. Considering the fact that beams carry almost zero axial force, the columnsway limit is shifted to a little higher than 1.00 in order to represent the increase in
column strength and stiffness due to axial forces on columns.
4.2.2.3
The two aforementioned indices have been applied to 14 different frames chosen from
real buildings or from research work on existing buildings. The details of case study
frames can be found in Bal (2008) where Case Study Set-III and V have been explained.
The case study RC frames vary between 2 and 8 storeys. All frames are analyzed by using
the Displacement-based Adaptive Pushover (DAP) method in SeismoStruct.
199
In some cases the index varies significantly from floor to floor. This renders the
determination of a single index value for the whole building rather difficult. There are
possible solutions of using the minima/maxima of all floors or an average index;
however, another approach has been adopted within this report. The index value of the
floor which has the highest drift has been used as a representative index of the whole
structure.
It can be seen from Figure 4.4 that there are 3 frames that are predicted to form a beamsway mechanism (i.e. sway-indices are found to be less than 0.85) when the strengthbased sway index is used; however, adaptive pushover results indicate that these frames
will develop a column-sway mechanism. By using the strength-based index, one of the
case study frames is predicted to form a column sway mechanism (ie. the sway-index is
larger than 1.00) though the pushover results show that the frame has the potential to
develop a beam-sway mechanism. However, the mechanism of the very same frame is
also wrongly predicted by the deformation-based index. Briefly, the mechanism of 4 out
of 14 frames has been wrongly predicted by the strength-based sway index. One of the
mixed-mechanism frames falls into the transitive zone, whilst the others are spread above
and below the limit values.
3.00
Column-Sw ay
2.00
Mixed
Beam-Sw ay
Upper Limit
1.00
Low er Limit
0.00
Figure 4.4 Results of the strength-based index for 14 case study buildings
Figure 4.5 shows the results for the deformation-based sway index where it can be seen
that this index fails to predict correctly the mechanism of 2 frames, which exhibited a
beam sway mechanism according to the pushover analyses but were predicted to form a
column-sway mechanism according to the index. Overall, the mechanism of 2 out of 14
frames was wrongly predicted by the deformation-based index. The mixed-sway
mechanisms were also not successfully predicted by the method.
200
3.00
Column-Sw ay
Mixed
2.00
Beam-Sw ay
Upper Limit
Low er Limit
1.00
0.00
Figure 4.5 Results of the stiffness-based index for 14 case study buildings
It is seen from Figure 4.6 that 3 of the 14 frames are predicted as having a column-sway
mechanism and another 3 of the 14 are predicted as beam-sway by both methods. The
predictions of these 6 frames are found to be in good agreement with the pushover
results. The two methods predict different mechanisms for 4 out of 14 frames.
column-sway
3.00
Column-Sway
2.00
Mixed
Beam-Sway
1.00
beam-sway
0.00
0.00
1.00
2.00
3.00
4.2.2.4
(a) Strength-Based Index. The advantage of the strength-based sway potential index is
that the sectional behaviour is correctly defined. The main disadvantage of the method is
the negligence of the member response which could simply change the prediction of the
overall response of a certain floor. Another shortcoming of the strength-based index is
that it is difficult to apply. All existing end sections of all the elements within the building
must be modelled to run moment-curvature analyses. Once all of the sections have been
201
modelled, only a small amount of additional effort would be required to carry out a
pushover analysis instead and thus obtain a more realistic prediction of the sway
mechanism, depending on the software used. However, the effort to calculate the
strength-based index could be reduced by using simple calculations to estimate the yield
moment based on just the yield strain of steel and section dimensions.
A column-sway mechanism may even occur in a regular building (with regular storey
heights and regular bay lengths) where the column sections are stronger than the beam
sections. The fact that the column sections are stronger than the beam sections does
not necessarily mean that the column members will fail (or yield) after the beam
members. The sway potential is directly related to the member response which is also a
function of the sectional response. In other words, the sectional response represents the
capacity of the section whilst the member properties, mostly the stiffness, define the
demand (i.e. distribution of moments among members around a joint and within a
storey) which will be created on that section.
Another characteristic example where the strength-based index fails would be the case
with a higher ground floor, perhaps used for commercial purposes. A significant
proportion of the existing building stock (56% of the Turkish RC building stock, has
been estimated by Bal et al., 2008) consists of buildings with a higher ground floor. Due
to the provisions in recent codes, higher ground floors are designed with more strength
and ductility as compared to upper floors in new constructions. However, the existing
building stock is a mixture of non-engineered buildings, gravity load designed buildings
and buildings designed with little earthquake actions. This simply entails no or little
change in the section strength when the ground floor height is increased by the designer
(or constructor). Buildings with a higher ground floor and with the same column sections
along the height of the building are rather common.
(b) Stiffness-Based Index. The main advantage of the stiffness-based index is its ease
in application, whilst also being sufficiently reliable. Considering that sway-potential
indices are approximations which do not require significant computational effort, it
appears that the use of a stiffness-based index, such as the one presented herein, is a
meaningful solution for defining the mechanism in a simple manner.
There are cases where the stiffness-based index fails to predict the mechanism accurately.
The reason for this failure is mostly the inconsistency of the single assumption upon
which the whole index is based. It is assumed that the rotational capacity and stiffness are
proportional for a given member. This implicitly indicates a relationship between section
depth and stiffness as well as section depth and strength. It is therefore assumed
indirectly that the increasing section depth causes an increase in member stiffness and
this means a constant or increasing member strength. This assumption is not necessarily
202
valid, especially in the cases where a general design trend is not followed. A beam element
with length L and section depth d for example, could also be redesigned with a section
depth of 1.5d. The stiffness-based index would immediately assume in this case that the
new design with 1.5d causes more stiffness of the element. This is not certain from a
theoretical point of view since any type of reinforcement amount and configuration can
be used for this new design of the beam leading thus to a strength value which could be
lower or higher than the previous one. Nevertheless, from a practical design point of
view, the new design of the beam with 1.5d depth is expected to have at least the same
strength with the previous case since the beam has to sustain some certain level of loads.
The newer design with 1.5d depth could even have more strength than the previous case
because of minimum reinforcement provisions or practical difficulties in construction (i.e.
the tradition of not using beam reinforcement smaller than 12mm diameter). The
stiffness-based index thus fails where a set of members (or a single member) is heavilyreinforced and over-designed or the opposite occurs.
There are 2 frames where the stiffness-based index failed to predict the correct
mechanism. One of the frames is from a two-storey building where the strength-based
index has also failed to predict correctly a beam-sway mechanism. The second example
where the stiffness-based index failed is a frame which exhibited a beam-sway mechanism
in the push-over analysis whilst the mechanism was predicted as a column-sway by the
stiffness-based index. The frame has been newly-designed according to the Portuguese
Code. It should be noted that the stiffness-based sway index has the tendency to be more
conservative in the prediction of the mechanism of the frames (i.e. it gives a column-sway
mechanism for beam-sway frames) which are designed according to the modern codes
where higher strength is required for columns.
Consequently, the stiffness-based sway index is found to be a useful and optimum
solution to be used in displacement-based loss assessment calculations of existing
buildings.
4.2.3 Derivation of Period of Vibration Formulae
Simple empirical relationships are available in many design codes to relate the
fundamental period of vibration of a building to its height. However, these relationships
have been realised for force-based design and so produce lower bound estimates of
period such that the base shear force will be conservatively predicted. In assessment, the
displacement demand on a structure needs to be accurately estimated; however, with a
conservative period-height relationship the displacement demand would generally be
under-predicted. The derivation of a robust relationship between the yield period and
height of reinforced concrete buildings has thus been undertaken, as discussed in the next
section.
4.2.3.1
203
Yield Period.
An extensive analytical study has been carried out (Crowley and Pinho, 2004) whereby
the yield period of vibration of 17 reinforced concrete bare frames, representative of
existing European building stock, has been estimated by means of different procedures
involving eigenvalue, pushover and nonlinear dynamic analyses using the fibre-modelling
Finite Elements program SeismoStruct (SeismoSoft, 2009). The RC frames considered
correspond to actual buildings from five different European countries exposed to
earthquake activity (Greece, Italy, Portugal, Romania and ex-Yugoslavia), the vast
majority of which were designed and built between 1930 and 1980, thus following
outdated seismic regulations, where seismic action was either ignored or inadequately
accounted for (see Crowley and Pinho, 2004).
The results obtained with each of the different analytical methodologies employed to
estimate the yield period of the 17 case study frames being considered, have been
represented in period versus height plots, such as that shown in Figure 4.7, which
features the period predictions using gross stiffness eigenvalue analysis. Regression
analysis was then carried out, for each analysis case, so as to derive a best-line curve fit to
the analytically-computed data. The curve has been taken as a power series of the
unconstrained form T = , one of the forms suggested by Goel and Chopra (1997).
2.5
Eigenvalue - kgross
2.0
T = 0.0354H 1.154
Period (s)
(s est = 0.110 s)
1.5
1.0
0.5
0.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
Height (m)
Figure 4.7 Period-height relationship obtained from gross stiffness eigenvalue analysis
A description of the analytical procedures that have been used to calculate the yield
period of vibration versus height relationships is given in what follows, and a comparison
of all results is presented in Figure 4.8.
204
The use of the gross section (i.e. uncracked) stiffness (see Eigenvalue kgross in Figure
4.8) in the computation of elastic vibration periods of reinforced concrete structures is
inadequate. Indeed, as reported by Priestley (2003), amongst others, cracking of critical
elements such as beams generally occurs under gravity loading alone. Even in those cases
where cracking is not found to have occurred before the design seismic level of excitation
(considered unlikely as this level of excitation would with all probability have been
preceded by a number of lower intensity events), it will occur early on in the response to
excitation and thereafter the stiffness will reduce rapidly. Since the tension stiffening
effect of the concrete is lost following the initiation of seismic excitation and will not be
regained without post-earthquake intervention, the stiffness of RC members can only be
reliably estimated if taken as the yield stiffness.
The actual moment of inertia of a given reinforced concrete section is known to be
influenced by the magnitude and sign of the moment applied to the structural member,
the amount of flexural reinforcement, the section geometry and the level of axial load P.
In addition, diagonal cracking of the member due to shear and reversed cyclic loading do
equally affect the moment of inertia and consequently the stiffness of the member
(Paulay and Priestley, 1992). A relatively common procedure that has been widely used to
take account of all of the aforementioned effects on the stiffness of RC structures is that
of reducing the moment of inertia of the gross section Ig to obtain an equivalent (cracked)
moment of inertia, Ie. Within such framework, Paulay and Priestley (1992) have
suggested reduction factors ranging from 0.35 to 0.80 for beams and columns. Similar
approaches have been proposed in a number of design codes and guidance documents
such as the New Zealand concrete design code (NZSA, 1992), Eurocode 8 (CEN, 2003)
and FEMA 356 (ASCE, 2000), to name but a few. The effective moment of inertia of
columns was taken as half of its uncracked counterpart, whilst for beams, reduction
factors between 0.35 and 0.40 were considered instead and then the eigenvalue analyses
were repeated.
(b) Eigenvalue - analytical kyield
The stiffness reduction factors described above do not take into account the flexural
strength of the member. However, as demonstrated by Priestley (1998), such strengthindependent stiffness assumption is in fact erroneous since experimental results indicate
that it is the yield curvature of members, rather than its yield stiffness, that can be
considered as being independent of the strength
Priestley (1998) carried out an analytical parametric study whereby moment-curvature
curves of a reinforced concrete member with varying levels of longitudinal reinforcement
(l) and axial load (N/fcAg) were plotted. The attained results led to the development of
the dimensionless stiffness ratio curves which allow the effective stiffness (EI) of
205
In Japan, an empirical expression for the ratio of secant stiffness at yielding and initial
stiffness has been widely used since it was first presented by Sugano (1970). This equation
has been derived from the results obtained by testing of 93 beams and 112 columns
featuring a tensile reinforcement ratio (t) ranging from 0.4 to 2.8 %, a shear span to
section depth ratio (a/D) varying from 2.0 to 5.0 and normalised axial force levels
(=N/Acfc) starting at 0.0 and going all the way up to 0.55.
This empirical expression is given by Equation (4.20), below, where Es/Ec stands for the
reinforcement to concrete modular ratio and d represents the effective depth of the
member section.
K yield / K gross
d 2
Es
a
t + 0.043 + 0.33 .
= 0.043 + 1.64
Ec
D
(4.20)
Being experimentally-derived, the above equation does account for flexural yielding,
deformation caused by bar slip within the anchorage zone and member shear
deformation. It is noted that the latter effectively means that the member gross stiffness
considered in the estimation of the corresponding yield counterpart, should, in principle,
consider both flexural and shear components (i.e. 1/Kgross=1/Kflex+1/Kshear). However,
taken into account the almost negligible contribution of shear stiffness in the slender
structural members of the structures considered in the present case study, only the
flexural stiffness component was considered.
It is also worth noting that the only conceptual difference between the proposal by
Sugano (1970) and that of Priestley (1998) is that the former has chosen to consider the
shear span to section depth ratio (a/D) as a third parameter, in addition to the
reinforcement ratio and axial load levels, influencing the values of stiffness reduction
factors. This seems to be logical since stockier members (i.e. members with low a/D
values) are more likely to feature a large number of cracks (due to shear deformations),
thus presenting a further reduced yield stiffness. This behaviour is fully captured by
Equation (4.20), which does predict higher reduction factors for members with low a/D
values.
The ratio between the yield and initial stiffnesses proposed by Sugano (1970) has been
computed for each column and beam of the case-study frame models and eigenvalue
analyses have been repeated.
206
Pushover analyses can be used to calculate the fundamental period of vibration of multidegree of freedom (MDOF) structures, provided the latter are converted into a substitute
single degree of freedom (SDOF) structure featuring an effective mass, Meff, and an
effective stiffness, keff. Considering that the yield period is sought, the first step required
in the procedure is to calculate the yield displacement of the MDOF structure which has
been found using two different procedures, as will be described below. Once the yield
displacement, and corresponding structures deformed shape, are known, the effective
mass of the SDOF system can be computed from classical eigenvalue analysis theory and
the effective stiffness can then be obtained through the ratio between the base shear at
point of yield (Vy) and the yield displacement (dy,heff) at the effective height (i.e. the height
to the centre of seismic force, heff).
Finally, the familiar equation for the period of vibration of a simple oscillator can be
employed to estimate the fundamental period of vibration of the frame, at point of yield:
T =2
M eff
keff
= 2
M eff
V y / d y ,heff
(4.21)
Park (1988), amongst others, suggests the yield displacement to be taken as that
corresponding to the attainment, in the pushover curve, of 75% the ultimate base shear
capacity (Vu) of the frame. Using this value as the starting point to the procedure outlined
above, the yield period of vibration of the 17 buildings was computed, and the
corresponding period-height best-fit curve determined.
(e) Pushover semi-empirical dy
Equations (4.12) and (4.13) give the yield displacement capacity of beam-sway and
column-frames and so these equations have been used to define the yield point of the
pushover curves. However, the effective height coefficient (efh) has been removed from
the aforementioned equations for the purpose of this period-height study so that the top
displacement of the structure is found and can be used in conjunction with the pushover
curves. It is also noted that the decision on which type of sway mechanism to adopt (i.e.
which yield displacement equation to employ), was made by observing the results of the
pushover analysis.
(f) Dynamic
207
former, the effects of higher modes of vibration are not accounted for and, in the latter,
all members within the building are assumed to be at yield stiffness. On the other hand,
however, the frequency with which a building frame subjected to a given accelerogram
vibrates is dictated not only by its natural period of vibration, an intrinsic property of a
structure, but can also be marginally influenced by the frequency characteristics of the
input motion itself, for which reason results may become more difficult to interpret.
Two real accelerograms, (normalised to the same peak ground acceleration value, for
easier comparison), have been chosen for this study; Sakaria time-history record from the
1999 Kocaeli/Izmit earthquake (Turkey) and Emeryville time-history record from the
1989 Loma Prieta earthquake (USA) (see Crowley and Pinho (2004) for further details).
Each of the 17 frames has undergone dynamic time-history analysis under the two
accelerograms. For each frame, the record has been scaled accordingly so that the
maximum response displacement at the top of the building corresponds to the yield
displacement, as defined previously by Equations (4.12) and (4.13) (without the effective
height coefficient, as discussed above). The Fourier Amplitude Spectrum of the
displacement response time history at the top node of each building was then computed
(using the program SeismoSignal (SeismoSoft, 2009)), so that the dominant period of
vibration of the structure, at yield, could be identified.
In Figure 4.8, a graphical summary of all period-height equations described herein is
shown alongside the period-height relationship for displacement-based assessment
proposed by Chopra and Goel (2000). The formulation by Chopra and Goel (2000) as
well as the gross stiffness eigenvalue best-fit are markedly outside of the range of results
obtained with other methods. This is expected since neither of these two curves has been
derived with yield period determination in mind; the former represents the mean plus 1
standard deviation best-fit curve of measured periods of vibration, whilst the latter refers
to uncracked elements (yielding of reinforced concrete members occurs well after
cracking has taken place). Moreover, the work of Chopra and Goel (2000) focused on
Californian, as opposed to European, building stock, rendering its applicability to the case
of the latter effectively unfeasible.
The period-height curve obtained from cracked stiffness (i.e. using strength-independent
member stiffness reduction factors) eigenvalue analyses fails also to provide a reliable
estimate of yield period, providing period estimates that are noticeably below those
obtained using one of the remaining five methods. This is certainly related to the fact that
the building stock considered in this study refers to existing non-seismically designed
frames, whilst these strength-independent member stiffness reduction factors, provided
in a number of codes and design regulations have been calibrated for application to new
construction, which is, by inception, stiffer than existing old structures.
208
By excluding the three aforementioned relationships, one is left with a set of five periodheight equations that do offer a relatively high degree of confidence, due to the explicit
consideration of both cracking and member strength-stiffness proportionality in their
derivation. In effect, it is reassuring to observe in Figure 4.8, above, that four out of these
five equations provide very similar period predictions, even if they all arise from diverse
approaches. The single exception to this close match is the equation obtained from
pushover analysis with semi-empirical yield displacement capacities, which provides
slightly larger, but nonetheless conservative, predictions.
5.0
4.5
Pushover - semi-empirical dy
4.0
3.5
3.0
Pushover - dy at 0.75Vu
2.5
Eigenvalue - kcracked
Period (s)
Eigenvalue - kgross
2.0
1.5
1.0
0.5
0.0
0
10
15
20
25
30
35
40
Height (m)
Finally, and perhaps most significantly, further observation of Figure 4.8 shows that there
appears to be an almost linear proportionality between period, in seconds, and height, in
metres. Indeed, a linear relationship would lead to period estimates exactly poised in
between the reliable lower- and upper-bound values already identified, as can be observed
in Figure 4.9, below.
Thus a straightforward choice of a linear yield period vs. height (HT in metres) formula
for European RC moment-resisting frames has been chosen, as presented in Equation
(4.22):
T y = 0.1H T
(4.22)
209
5.0
4.5
4.0
3.5
Period (s)
3.0
T=0.1H
2.5
2.0
1.5
1.0
0.5
0.0
0
10
15
20
25
30
35
40
Height (m)
Figure 4.9
A further study was carried out by Crowley and Pinho (2006) to produce a period vs.
height equation to predict the fundamental period of vibration of reinforced concrete
buildings with moment resisting frames and infill panels. The same frames employed in
the period-height study for bare frames described above were used. For bare RC
buildings, these 2D frames can be used to adequately estimate the period of vibration of
the whole building in the principle direction in which these frames act. However, in a real
building with infill panels, some of the frames are bare, some are fully infilled and some
have infill panels with openings for windows and doors. Hence, the influence of infill
panels cannot be simply modelled using a single 2D frame and should ideally be
calculated using 3D models of real structures. A simplified solution to this problem has
been adopted herein such that the 2D frames which have already been modelled can be
employed. The 2D frames are modelled as bare frames, fully infilled frames and infilled
frames with openings and a weighted average of the period of vibration of these three
types of frames is then calculated by taking into account the their frequency of
occurrence within a typical building stock. A detailed study of the characteristics of
Turkish reinforced concrete building stock in the Marmara region has been carried out by
Bal et al. (2008) and it has been found that the proportion of each frame in a building
stock of 30 buildings has been found to be 34% bare frames, 28% fully infilled frames
and 38% infilled frames with openings. In the frames with openings, the ratio of area of
the openings compared to the area of the bays (within a single storey) has been found to
210
have a mean value of 20%. The size of these openings ranges between (1.0-1.7) (1.32.4) m for windows and between (0.9-1.2) (2.1-2.3) m for doors.
All the eigenvalue analyses described in the subsequent sections of this paper were carried
out with SeismoStruct (SeismoSoft, 2009). SeismoStruct models the masonry infill panels
using a macro model that represents the infill panel as a diagonal strut (see Figure 4.10a)
with a given equivalent area/stiffness, resistance and nonlinear behaviour. The diagonal
element consists of two compression struts in each diagonal of the panel, as shown in
Figure 4.10b. The element has 4 external nodes, 4 internal nodes (connected to the frame
corner) and 4 dummy nodes (located at a distance hz from the frame corner); such
formulation is based on the proposals of Crisafulli (1997) and Crisafulli et al. (2000). The
dummy nodes are used to represent the local effect between the infill and the frame and
are placed at a distance equal to a third of the contact length, z (illustrated in Figure
4.10a). The internal nodes represent the frame-infill contact at the exterior part of the
beam and column and the internal forces are then transformed to the 4 exterior nodes
where the element is connected to the frame. The hysteretic material model which
represents the nonlinear behaviour of the strut has been defined using the default values
suggested/calibrated by Blandon-Uribe (2005) and Smyrou et al. (2006). The geometric
and material properties of the diagonal struts have been defined using the values and
formulae reported in Table 4.1.
V
Yoi
U
Internal
Internal
hz Xoi
hw
Dummy
dw
Dummy
dm
bw
Dummy
Ae
Dummy
hz
Internal
Internal
lw
(a)
(b)
Figure 4.10 (a) Diagonal strut representation of an infill panel (Paulay and Priestley, 1992), (b)
Description of implemented infill panel (adapted from Crisafulli, 1997)
211
Table 4.1 Values and formulae used to define the properties of the diagonal masonry strut
Parameters required to
define masonry strut
Mean diagonal
compressive strength,
fm
Adopted value/formula
Reference
1.5MPa
Elastic Youngs
Modulus, Em
1000fm = 1500MPa
Length of strut, dw
=4
Relative stiffness
parameter,
E m t w sin(2 )
4 E c I p hw
Crisafulli (1997)
Calvi et al. (2004)
Calvi et al. (2004)
Bal et al. (2008)
See Figure 1a
Width of compressive
strut, bw
bw =
0.95hw cos
hw
Reduction coefficient
for openings
Ams2 = Ams1(bwcracked/bwuncracked)
z=
Horizontal offset, X0
Vertical offset, Y0
212
0.90
bwcracked /bwuncracked
0.85
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0
10
hw
Figure 4.11 Reduction factor to be applied to the equivalent width of diagonal struts due to cracking
of masonry panel (Decanini and Fantin, 1986)
Two boundary values of infill panel thickness have been adopted (0.1m and 0.25m), as
this parameter has a large influence on the stiffness of the diagonal struts and varies from
building to building. The material properties adopted for the masonry infill aim to reflect
the characteristics of highly perforated clay blocks typically used in European earthquakeprone countries. The elastic modulus of this type of infill panel has a large coefficient of
variation (Calvi et al., 2004); however, a mean value has been adopted herein and further
studies will be carried out in the future to study the influence of this variation on the
period of vibration of the infilled frames.
The study by Crowley and Pinho (2004) considered five different procedures (using
eigenvalue analysis, pushover analysis and dynamic analysis) to estimate the yield stiffness
of RC frames and found that all five approaches led to very similar period-height
predictions, as presented previously in Figure 4.9. The dependency of the yield stiffness
on the strength of the RC members (see for e.g. Priestley, 2003) was the main assumption
that was applied in all methods to calculate the yield period of the RC frames.
Considering the similarity of the five methods used in the bare frame yield period study
only one of these methods needs to used herein, hence the approach suggested by
Priestley (1998) has been adopted to extend the study to calculate the yield period of
vibration of infilled buildings (Eigenvalue analytical kyield described previously).
The same models with reduced member stiffness were used to create the fully infilled
frames and the infilled frames with openings. To take into account the cracking of the
masonry, the residual strut area, Asm2, has been calculated for each strut using the reduced
strut width obtained from Figure 4.11. The yield period-height results for these two types
of frames are presented in Figure 4.12, and again the weighted average period for infilled
213
buildings has been calculated using the ratios presented previously, leading to the curve
presented in Figure 4.13.
2.5
Bare frames
Ty = 0.089H
Period (s)
2.0
1.5
1.0
0.5
0.0
0
10
15
20
25
30
Height (m)
Figure 4.12 Analytical yield period-height relationships for cracked RC buildings obtained with
yield stiffness eigenvalue analyses for bare frames, fully infilled frames and frames with openings
2.5
Period (s)
2.0
1.5
1.0
0.5
0.0
0
10
15
20
25
30
Height (m)
Figure 4.13 Analytical yield period-height relationship for cracked infilled RC buildings obtained by
calculating a weighted average of the results in Figure 4.12.
214
The aforementioned study has led to a simplified period-height equation for use in the
assessment of existing RC buildings, taking due account of the presence of infill panels:
T y = 0.055 H T
4.2.3.2
(4.23)
Post-Yield Period.
For post-yield limit states, the limit state period of the substitute structure can be
obtained from the secant stiffness to the point of maximum deflection on an idealised bilinear force-displacement curve. It has been observed during several analyses on existing
structures that it is more likely that a negative post-yield stiffness ratio is achieved. Postyield stiffnesses would be over-estimated were a elastic-perfectly plastic behaviour to be
assumed. Inclusion of this aspect needs an additional parameter (the post-yield stiffness
ratio) which can only be obtained following an extensive number of nonlinear analyses on
several case-study buildings. Taking advantage of several analyses conducted by Bal
(2008), post-yield periods will be presented in a form where the negative post-yield
stiffness values can also be considered.
The ratio between the yield period, Ty, and Limit State Period, TLS, can be directly related
to the ratio of the yield stiffness, Ky, to the ratio of the limit state stiffness, KLS. This ratio
would purely depend on the square-root of the Limit State ductility, LS in cases where
the post-yield stiffness ratio is neglected (i.e. assumed as zero). Otherwise a relationship can
be derived as given below (see Figure 4.14 for detailed representation of the terms used);
T LS = T y
Ky =
K LS =
K LS / K y =
Ky
K LS
Fy
y
FLS F y + K y ( LS y )
=
LS
LS
F y / LS + K y ( LS y ) / LS
Fy / y
(4.24)
(4.25)
(4.26)
(4.27)
TLS = Ty
LS
1 + LS
215
(4.28)
where TLS is limit state period, Ty is yield period, KLS is limit state stiffness, Ky is yield
stiffness, LS is limit state displacement at the effective height, y is yield period at the
effective height, and is post-yield stiffness ratio. The formula takes the simplified form
below if post-yield stiffness is neglected:
(4.29)
TLS = T y LS
/
F
/y /L
S
F F
S
L
tt
a
h
ti
g
n
ee
mH
ee
v
ci
ac
t
l
p
e
sf
f
i
D
E
S
L
Figure 4.14 Yield and limit state stiffness values for the case of negative post-yield stiffness ratio
The main concepts of the SP-BELA methodology (Borzi et al., 2008a; 2008b) are the
same as those of DBELA, especially those related to the calculation of the displacement
capacity, but the main differences lie within the calculation of the yield period of
vibration and the response mechanisms, which are both strength-based. In order to
estimate the strength of buildings in an urban area, where detailed drawings of the
buildings are not available, a simulated design of a typical structure can be carried out
based on the design codes used within the region or country. An outline of the
methodology for reinforced concrete buildings is described in the following sections.
216
P1
P2
P3
P3
P2
P1
Edge beam
P4
P5
P5
P5
P5
P4
Central beam
P1
P2
P3
P3
P2
P1
Edge beam
ly
ly
lx
lx
lx
lx
lx
Figure 4.15 Plan view of the RC frame building assumed as representative of the building structural
type for seismically designed buildings in Mediterranean countries.
217
In order to define the distribution of structural bending moments and shear forces in the
elements of the frame, the seismic loads have been assumed to correspond to horizontal
forces linearly distributed along the height, noting however that different distributions
may be easily assumed, when relevant (e.g. for taller buildings where the effects of higher
modes become important). The procedure, which inspires itself on the work of Priestley
and Calvi (1991), then calculates for each column of the frame the maximum value of
shear that the column can withstand as the smallest of:
The shear corresponding to the flexural capacity of the beams supported by the
column.
For the beams only the flexural collapse mechanism is taken into account, given that the
beams tend to be less prone to shear failure than the columns since gravity load design
typically features high shear forces in the beams and thus these elements have
traditionally been provided with an adequate amount of shear reinforcement.
The checks conducted during the procedure to define the cause of failure in each column
are illustrated in Figure 4.16, wherein the subscript R is for resistance and the subscripts
C and B represent column and beam, respectively.
Once the shear capacity has been calculated for every storey, the collapse multiplier is
defined by the following relationship:
n
Vi
= C
WT
i
Wz
j j
j =1
n
(4.30)
W z
k k
k =i
where WT is the global building weight, Wi is the weight associated with floor i located at
height zi. The final collapse multiplier used to define the capacity curve will be the
smallest i.
218
External columns
MR,B-4
MR,B+4
MR,B-3
MR,B+3
MR,B-2
MR,B+2
MR,B-1
MR,B+1
Last storey
2 M iR ,C j
hi
2 M nR ,C j
hi
M Ri , B
+ M iR , C j
2
hi
(M
) h1
n
R , B
+ M nR ,C j
Internal columns
MR,B+4
MR,B-4
MR,B+3
MR,B-3
MR,B+2
MR,B-2
MR,B+1
MR,B-1
Last storey
2 M Ri ,C j
hi
2 M nR ,C j
hi
M Ri , B + M Ri , B+
1
+ M iR , C
hi
2
M nR , B + M nR , B+
1
+ M nR , C
,
hn
2
Figure 4.16 Maximum shear force that the columns in a frame can withstand accounting for (i) shear
and flexural failure mechanism in columns and (ii) flexural failure mechanism in beams.
Finally, in order to evaluate the collapse mechanism of the building the procedure uses
the following criteria:
If there is a shear failure mechanism detected in at least one column, the capacity
curve will be interrupted at the lateral force that produces this failure. This choice
is consistent with the fact that the shear failure mechanism is brittle and does not
have associated dissipative capacity. Therefore, the structure cannot enter the
nonlinear range.
If all the columns within a certain storey fail in bending, than a column-sway
collapse mechanism will be activated (Figure 4.17b).
If after the development of plastic hinges in all beams, plastic hinges form in all
columns at a certain level, a beam-sway collapse mechanism will be activated
(Figure 4.17a).
219
(a)
(b)
Figure 4.17 Possible collapse mechanisms for a frame (a) beam-sway collapse mechanism, (b)
column-sway collapse mechanism.
There could be a situation in which at the storey corresponding to the smallest i some of
the columns are stronger than the beams, or vice versa. Therefore, it cannot be clearly
identified whether a beam or a column-sway mechanism will be activated. Various
analyses undertaken to validate the SP-BELA procedure have shown that in such cases
the collapse mechanism is an average between the column and the beam-sway.
4.3.3 Simplified Pushover Curves with DBELA and SP-BELA
Simplified pushover curves, such as the one shown in Figure 4.14 (with or without postyield stiffness) can be generated with both DBELA and SP-BELA. The previous section
discussed the calculation of the lateral strength of reinforced concrete buildings. Once the
strength and the displacement capacity have been calculated, the pushover curve can be
easily produced. In SP-BELA the yield period of vibration can then be calculated using
the following formula:
TLSy = 2
m y
W y
y
m
= 2
= 2
= 2
Ky
V
Vg
g
(4.31)
Simplified pushover curves can also be generated with the DBELA methodology. The
base shear force divided by the seismic weight of the building (F/) can be calculated
directly from the displacement capacity (y) and the period of vibration at yield (Ty) using
the following formula (which is a rearrangement of Equation (4.31):
=
4 2 y
gT 2
4 2 y
g ( 0.1H )2
(4.32)
220
These pushover curves can be used to validate the methodology. Crowley et al. (2008)
present the comparison of pushover curves of a 4 storey building, designed with and
without seismic design, obtained with the fibre-elements structural analysis program
SeismoStruct (SeismoSoft, 2009) and those obtained with DBELA and SP-BELA. Some
example results are shown in Figure 4.18 where it can be seen that the base shear force of
the buildings is well predicted, whilst the yield displacement capacity is slightly
overestimated with the SP-BELA methodology. Such analyses allow the uncertainty in
the displacement capacity equations to be estimated (see Section 4.5).
0.4
0.20
Collapse Multiplier,
Collapse Multiplier,
0.25
0.15
0.10
SP-BELA
DBELA
FE Analysis
Bilinear (EC8)
0.05
0.00
0.00
0.05
0.10
Displacement (m)
0.15
0.20
0.3
0.2
SP-BELA
DBELA
FE Analysis
Bilinear (EC8)
0.1
0.0
0.00
0.05
0.10
0.15
0.20
0.25
Displacement (m)
(a)
(b)
Figure 4.18 Comparison of pushover analyses in x direction of a four storey building (a) nonseismically designed (b) designed with base shear coefficient c = 5% (Crowley et al., 2008)
When calculating the damage to buildings within urban areas in a loss assessment model,
the variability of the properties of the structures within these areas needs to be modelled.
This can be carried out by collecting statistics on a sample of the buildings. Such a study
has been undertaken by Bal et al. (2008) for Turkish buildings; the results of this study led
to probability density functions to describe the variability in the geometrical and material
properties, such as those shown in Figure 4.19.
350
1200
300
1000
800
600
400
200
0
250
200
150
100
50
3.30
3.20
3.10
3.00
2.90
2.80
2.70
2.60
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0
2.50
# of Buildings
221
Figure 4.19 Historgam and fitted probability distributions from a sample of Turkish buildings (a)
beam length and (b) regular storey height for all RC buildings (Bal et al., 2008)
The limit state strains can also be assigned a probability density function, as can the
coefficients in the period-height relationships used in DBELA, based on the results of
experimental and numerical analyses (see Crowley et al., 2004). Once all of the random
variables have been defined, random populations of hundreds or thousands buildings can
be generated with Monte Carlo simulation; a random sample from each distribution is
made for each building.
4.4.1 Direct Damage Distribution
Once the random population of buildings has been generated, the displacement capacity
and period of vibration of each can be calculated using either the DBELA or SP-BELA
methodologies.
The displacement demand for each limit state can be estimated from a displacement
response spectrum, with the appropriate damping at each limit state. In order to estimate
the limit state displacement demand to the equivalent linear system, the spectrum is
modified by spectrum correction factors, i :
LS =
7
2+
( 4.32 )
, the damping value of each limit state, if found as a function of the limit state ductility,
LS, according to the equivalent viscous damping equations provided by Priestley et al.
(2007).
222
( 4.33 )
( 4.34 )
eq = 0.05 + 0.565
eq = 0.05 + 0.444
Equivalent damping values for masonry structures are assumed as 5%, 10% and 15% for
first, second and third limit states, respectively (Priestley et al., 2007).
Figure 4.20 shows how, for a single building in the random population, the displacement
demand can be compared with the displacement capacity at each limit state to see if the
limit state is exceeded or not. In the example shown in Figure 4.20 the first and second
limit states are exceeded, but not the third (collapse). If four damage states of damage,
slight, moderate, extensive and complete are associated with the thee limit states to
damage, the building in the example would be predicted to be extensively damaged. This
calculation is repeated for all random buildings in the population and allows the damage
distribution to this building typology to be directly estimated.
t
n
e
m
le
a
c
ra
t
l
c
p
es
pi
SD
e
g
a
m
a
D
e
t
a
r
e
d
o
M
1
,
d
3
,
d
e
g
a
m
a
D
y
v
a
e
H
2
,
d
3
,
d
S S S
2
,
d
1
,
d
d
o
i
r
e
P
Figure 4.20 Illustration of the estimation of the damage state of a single building through the
comparison of limit state displacement capacity and demand.
Vulnerability curves can also be generated with the DBELA and SP-BELA
methodologies. The steps for producing these curves in terms of peak ground
acceleration are reported below:
223
i) Generate a random population of 1000 buildings using Monte Carlo simulation, based
on probability density functions of the geometric and material properties defined a priori
(e.g. the beam length varies uniformly between 3 and 5 metres, the regular storey height
follows a lognormal distribution with a mean of 3m and a standard deviation of 0.15m)
ii) Calculate simplified bilinear capacity curves for each random building and calculate the
three limit state periods of vibration and displacement capacities.
iii) For a given level of PGA, produce a 5% damped displacement response spectrum.
iv) For each random building, beginning with the first limit state, compare the
displacement capacity with the damped displacement demand at the limit state period of
vibration (see Figure 4.20); if the demand exceeds the capacity the one moves to the next
limit state until the demand no longer exceeds the capacity and the ultimate damage state
of the building is attained.
v) Repeat for all random buildings and calculate the number of the buildings within each
damage state, divide by the total number of random buildings to obtain the proportion
and then calculate the percentage which exceed each limit state.
vi) Repeat for a number of levels of PGA and plot probability of exceeding each limit
state versus PGA.
vii) Apply regression analysis to obtain the parameters of a lognormal distribution for
PGA vs. probability of damage exceedance for each limit state.
4.5 DEFINITION OF UNCERTAINTIES
There are several uncertainties associated with the loss assessment calculations. These
uncertainties may be related to the demand, or to the capacity of the exposure. It is
relatively straightforward to mention the uncertainties qualitatively. On the other hand, it
needs much more technical and analytical effort to frame these uncertainties and come up
with quantified values to be used in the loss estimation calculations. There are numerous
uncertainties; however, an effort to quantify two of major uncertainties in loss estimation
with DBELA has been carried out by Bal (2008). These major uncertainties are related to:
Variation of the results of pushover analyses that are used for the estimation
and calibration of the displacement capacity formulae presented in Section 4.2.
Uncertainties in the calculation of the response of structures to a given demand due
to the random nature of earthquakes.
224
Variation of the analyses results are due to three different reasons: variation in the
modelling choices, variation in the selection of the nonlinear static analysis method, and
use of different mesh sizes in the fibre-based sections. The last variation is cancelled in
the calculations presented in this study due to the fact that the finest meshing has been
introduced in the models used. The first two variations however, have been investigated
in detail by Bal (2008). Figure 4.21 shows how the choices in the modelling can change
the limit state displacement which is obtained from the pushover analysis (Bal, 2008).
500
450
400
350
300
250
200
150
LS2 Points
100
50
0
0.00
0.05
0.10
0.15
0.20
0.25
Figure 4.21 Analyses results for sensitivity analysis based on different element types with various
subdivisions and number of integration points (Bal, 2008)
According to the results which are obtained from analyses conducted on a simple twobay two-storey frame, the standard deviation for calculation of the second limit state
displacement is around 0.034m where the mean displacement is 0.074m (results of the
runs with single displacement-based elements have been excluded due to their high level
of inaccuracy). This simply provides a high level of uncertainty where the standard
deviation of the logarithmic residuals, n=ln(obs)-ln(mean), is calculated as 0.52. The is
assumed to follow a standard normal distribution. The displacement capacity formulae
given in Equations (4.1) and (4.2) will then take the form:
ln ( LS ) = ln( LS ) + n
(4.35)
where is the standard error variable, n is a variable representing the possible errors
caused by numerical modelling issues, LS is the displacement capacity by taking into
225
account the numerical modelling uncertainty and LS is the basic displacement capacity
coming directly from the capacity equation formulae. Another possible error due to the
selection of nonlinear static method is neglected here due to its relatively lower variability,
but it can be added in the same fashion as well.
In the steps of the DBELA methodology presented thus far the displacement demand to
a multi-degree-of-freedom structure in terms of the displacement at the effective height
of the building is assumed to be equal to the displacement demand of an equivalent
SDOF system from an overdamped displacement response spectrum. However, there are
a large number of uncertainties in this procedure which would render the actual
displacement of the MDOF structure different from the prediction, which are related to:
226
0.20
0.20
0.15
y = 0.95x
R2 = 0.43
0.10
0.05
0.00
0.00
0.05
0.10
0.15
0.20
0.15
y = 0.87x
R2 = 0.72
0.10
0.05
0.00
0.00
0.05
0.10
0.15
0.20
Figure 4.22 Relationship between the predicted Intensity Measure for two five storey buildings and
the Engineering Demand Parameter obtained from 50 earthquake records (Bal, 2008)
ln (S d , LS ) = ln (S d , LS ) + p
(4.36)
where is the standard error variable, Sd,LS is the displacement demand by taking into
account the possible uncertainty, and Sd,LS is the median displacement demand that is
obtained directly from the over-damped spectrum.
Seismic hazards encompass all the potentially destructive effects of earthquakes (e.g.
strong ground-motion, liquefaction, landslides) and seismic hazard analysis is concerned
with the prediction of the severity and likelihood of occurrence of these effects at a
particular site.
The seismic hazard component of an earthquake loss model is generally modelled using
scenario earthquakes; either single scenarios, which might be the repetition of an
historical earthquake or the Maximum Credible Earthquake, or the consideration of all
possible earthquake scenarios which might affect the area under consideration. A
probabilistic seismic hazard assessment considers the ground motion at a particular site
due to all possible earthquake scenarios; however, when the losses to a number of
locations need to be aggregated, the calculations need to be carried out considering
separate earthquake scenarios, as discussed further in this Chapter. The concepts of a
traditional deterministic and probabilistic seismic hazard assessment are first summarised
and then focus is given to the aleatory variability in ground-motion prediction equations,
which has a large influence on the modelling of hazard in loss models.
5.1.1 General Requirements for a Seismic Hazard Assessment
The main requirement for any seismic hazard assessment is a seismicity model of the
region of interest wherein the occurrence in time and space of earthquakes of different
sizes are modelled. There are three components that are generally combined in a
seismicity model: earthquake catalogues, seismic source zones and recurrence relations.
Earthquake catalogues report the origin time, location (i.e. epicentral co-ordinates and
focal depths) and magnitude of earthquakes that have occurred in or near to the region of
interest. Instrumentally, earthquakes have been measured and recorded for around 100
years, though reports of historical seismicity spanning several hundreds of years are
available in many countries. Seismic source zones capture the spatial aspect of a
seismicity model by defining regions within which future earthquakes are expected to
occur at any location with an equal likelihood (i.e. the seismicity is uniform). Tectonics,
observed seismicity and a degree of judgement are all called upon in the determination of
source zones. Once the earthquake catalogue is compiled and the source zones have
228
been defined, the temporal element of the seismicity model can be provided by
recurrence relationships. Within each source zone, a recurrence relationship can be
derived through the extraction of the number of earthquakes of different magnitudes
from the earthquake catalogue. Effectively, the recurrence relationship provides
information on the annual rate of exceedance of earthquakes of different magnitudes;
recurrence relationships will be discussed further in Section 5.1.3.
Once the seismicity model has been derived for the study area, a choice of whether to
implement a deterministic or a probabilistic seismic hazard assessment needs to be made.
As will be seen in Sections 5.1.2 and 5.1.3, there are actually many similarities between the
two methods and both have their advantages and disadvantages. An important aspect of
both procedures, which cannot be ignored, is the treatment of uncertainty; the origin of
different types of uncertainty is discussed in Section 5.2.
5.1.2 Deterministic Seismic Hazard Assessment (DSHA)
229
Figure 5.1 The four-step process to a typical DSHA (after Reiter, 1990)
Once the scenario earthquake magnitude and distance for each source zone have been
defined, the ground motions at the site can be estimated using a ground-motion
prediction equation. These equations provide estimates of ground motion (e.g. peak
acceleration, peak velocity, response spectrum ordinates) for an earthquake of a given
magnitude at different distances by means of a curve fitted to observed data. As will be
described in Section 5.2, there is a large amount of scatter in the data (see Figure 5.1) and
whilst many studies use the median values from the predictive equations, the current
trend is to use the 84-percentile motions, corresponding to one logarithmic standard
deviation above the logarithmic mean (Krinitzsky, 2002).
The motions at the site for each of the controlling earthquakes from the various sources
are considered in the design, unless one scenario is seen to dominate the seismic hazard at
all response frequencies.
230
In the previous section it was shown how a few isolated M-D- triplets (i.e. magnitude,
distance, and number of logarithmic standard deviations) are considered in a
deterministic seismic hazard assessment for a single site. In a probabilistic seismic hazard
assessment (PSHA), the fundamental difference is that all possible M-D- triplets are
considered in the calculation of hazard (see e.g. Cornell, 1968; Reiter, 1990; Kramer,
1995). Another difference is that the results of a PSHA have units of time, though it is
arguable that units of time can also be applied to a deterministic assessment through the
assignment of recurrence intervals to the scenario earthquakes.
Figure 5.2 The four-step process to a typical PSHA (after Reiter, 1990)
A PSHA comprises four basic stages, as illustrated in Figure 5.2. The first step involves
the identification and characterisation of the earthquake sources in much the same way as
has been described for a DSHA. The main difference, however, is that a probability
231
distribution of potential rupture locations within each source is defined. This distribution
is usually uniform, leading to an equal likelihood of earthquake occurrence at any point
within the source. In a DSHA, on the other hand, it is assumed that the probability of an
earthquake is 1 at the point on the source closest to the site.
The second step of a PSHA involves the description of the temporal distribution of
earthquake recurrence through the use of a recurrence relationship. These relationships
specify the average rate at which an earthquake of some magnitude will be exceeded and
the first was proposed by Gutenberg and Richter (1944):
log( N m ) = a bM
(5.1)
where Nm is the mean annual rate of exceedance of the magnitude M, 10a is the mean
yearly rate of earthquakes of magnitude greater than or equal to zero, and b describes the
relative likelihood of large and small earthquakes. The a and b values can be obtained
through regression analysis of a historical earthquake catalogue which must be corrected
for completeness in magnitude and the removal of fore- and aftershocks.
The validity of the Gutenberg-Richter law for an individual fault has been questioned by
Schwarz and Coppersmith (1984): it has been observed through paleoseismic studies that
points on fault segments tend to move by approximately the same distance in each
earthquake, thus suggesting that individual fault segments generate earthquakes of a
characteristic magnitude that is close to their maximum magnitude. A modified recurrence
relationship is required for fault segments, as illustrated in Figure 5.3; the characteristic
earthquakes are shown to occur more frequently than would be predicted from
extrapolation of the seismicity data from low to high magnitude.
The next step in the PSHA process is the definition of the ground motions at a site; this
is carried out in a similar fashion to deterministic seismic hazard assessment whereby
ground-motion prediction equations are used. However, with PSHA a range of
earthquakes is considered and so a family of ground-motion prediction curves is required
for each magnitude, as illustrated in Figure 5.2. The aleatory uncertainty that arises when
the data are fit to a particular curve is accounted for in PSHA, as will be described in the
next step.
232
Figure 5.3 Recurrence relationship for faults that fit the characteristic earthquake model (Reiter,
1990).
The fourth step lies in determining the hazard at a particular site through the integration
of the effects of all earthquakes of different sizes, occurring at different locations in
different source zones at different probabilities of occurrence. A hazard curve is
produced which shows the probability of exceeding different levels of ground motion at
the site during a specified period of time. The annual expected number of exceedances
of ground motion level (z) can be calculated from:
N
mu r =
(5.2)
mo r = 0
where i is the mean rate of occurrence of earthquakes between lower and upper bound
magnitudes (mo and mu) being considered in the ith source, fM(m) is the probability density
function (PDF) of magnitude (recurrence relationship) within source i, fR(r) is the PDF of
distance between the various locations within source i and the site, and P(Z>z/m,r) is the
probability that a given earthquake of magnitude m and distance r will exceed ground
motion level z (due to the aleatory variability in the prediction equation).
In practice, the integral shown in Eq. (5.2) is computed numerically and the range of
values of M and R, are divided into bins or segments of width m, and r respectively
and so the integrals in Eq. (5.2) are replaced by summations.
233
An assumption often made in PSHA is that each earthquake occurs independently of any
other earthquake, such that the annual frequency of exceedance (z) can be used in
conjunction with the Poisson model to calculate the probability of exceedance of a given
ground motion level q(z), during the design life L of a project placed at the site:
q( z ) =1 e
( z )
(5.3)
Two sources of uncertainty are generally considered in all seismic hazard assessment:
aleatory and epistemic. The former is a random source of uncertainty, a significant
proportion of which is due to the scatter in ground-motion prediction equations. This
scatter results from randomness in rupture mechanisms and from variability and
heterogeneity of the source, travel path and site conditions and is quantified using the
standard deviation of the predicted ground-motion parameter. The latter uncertainty is
due to the considerable degree of subjective judgement required in the definition of a
seismicity model, and is the result of incomplete knowledge therein.
5.2.1 Epistemic Uncertainty in Hazard Analysis
Uncertainties related to the subjective decisions that are inevitably made as part of the
process of carrying out the assessment include the following:
The determination of the Mmax value for each seismic source zone.
234
logic tree, however, Abrahamson and Bommer (2005) discuss how this is the least
appropriate choice, and they suggest that the fractile should reflect the desired degree of
confidence that the safety level implied by the selected annual frequency of exceedance (or return period) is
being achieved in light of the uncertainty in the estimation of the hazard.
Figure 5.4 Logic-tree formulation for a probabilistic seismic hazard study of the Mississippi
embayment (Toro et al., 1992)
The logarithmic spectral accelerations at a given site may be estimated using groundmotion prediction equations that generally have the following form:
ln S a (T ) = ( M , R, T , ) + total (T ) total (T )
(5.4)
where the function (M,R,T,) represents the expected value of the logarithmic groundmotion as a function of magnitude (M), some measure of distance (R), period (T) and
other parameters () such as local site conditions and style-of-faulting. Regression analysis
is carried out using the measured ground motions at a number of sites from different
earthquakes in order to calculate the parameters of the function in Equation (5.4) that
give the best-fit to the data. If one plots the ground motions observed at a number of
sites against distance from any particular earthquake, it is not uncommon to observe
departures from the median predictions (i.e., residuals) that can exceed 3 standard
deviations (Strasser et al., 2008). Thus, the term total(T) represents the standard deviation
of the logarithmic spectral accelerations and quantifies this observed variability. The
random variable total(T) defines the number of standard deviations that each observation
is away from the median.
235
(5.5)
Figure 5.5 Spectral acceleration at 0.05s from a number of recordings from two different events to
illustrate the inter-event (between-earthquake) and intra-event (site-to-site) residuals
236
A further consideration made herein concerns the macrospatial correlation between the
ground-motion residuals at pairs of sites due to their proximity. This is a phenomenon
whose importance in seismic risk analysis has been recognized for many years (e.g.,
McGuire, 1988), but which has only recently begun to be modeled due to the increased
number of dense arrays of recording stations that have been set up around the world.
Macrospatial correlation between the ground-motion residuals at pairs of sites arises
because the sites are affected by the same earthquake and thus may be affected by similar
azimuthal and directivity effects of the propagating waves, similar geological profiles, etc.
Boore et al. (2003) summarize this phenomenon as follows: The spatial variability in ground
motions reduces to zero as the distance between two sites decreases to zero. On the other hand, for a great
enough separation distance the spatial correlation of the ground motions reduces to zero and the additional
uncertainty reaches that for an individual observation about the overall change of motion with distance (as
given, for example, by fitting the data to a function using regression analysis).
A recent proposal for the correlation between the residuals of peak ground velocity
(PGV) at two sites by Wang and Takada (2005) has the following form:
(h, k ) = e( r ( h ,k ) / r )
0
(5.8)
where r(h,k) is the distance between two sites h and k and r0 is the standard correlation
length (the separation distance at which the correlation coefficient is equal to e-10.368).
Wang and Takada (2005) obtained the ground-motion residuals from hundreds of
recordings from six earthquakes in Japan and Taiwan by considering two different
attenuation relationships. For each earthquake, they grouped the pairs of sites into bins
with separation distance intervals of 4km and calculated the covariance between the
ground-motion residuals in each bin for the two ground-motion relationships (Figure
5.6). The covariance functions for each earthquake and for each ground-motion
relationship were found to lead to similar standard correlation lengths (around 20-40km).
The uncertainty in the correlation length is nevertheless influenced by many factors such
as wave propagation and ground conditions and is thus expected to change for different
regions around the world.
(h,k)
= exp(-r(h,k)/27.3)
(h,k)
= exp(-r(h,k)/27.8)
237
r(h,k)
r(h,k)
Figure 5.6 Normalised auto-covariance functions for the Chi-Chi earthquake for (a) the Annaka
ground-motion relationship and (b) the Midorikawa-Ohtake ground-motion relationship (Wang and
Takada, 2005).
The appendix of Boore et al. (2003) presents a summary of the work carried out by Boore
(1997) in which the spatial variability of peak motion from the 1994 Northridge
mainshock was studied and from which the following formula for the correlation of the
residuals of peak motion at pairs of sites, (h,k), separated by kilometres can be obtained:
(h, k ) = 1 1 e[
0.6 r ( h , k ) ]
(5.11)
A comparison of these three former models (i.e., McGuire, 1988; Wang and Takada,
2005; Boore et al., 2003) for macrospatial correlation with separation distance is presented
in Figure 5.8 where the correlation length, r0, in the former two models has been taken as
5km. Figure 5.8 shows that the Boore et al. (2003) equation has a correlation length of
approximately 4km and is furthermore a fixed distribution. The other two models have
been plotted with a correlation length of 5km, but in the Wang and Takada (2005) study
this length has been found to be up to 40km and thus it appears to be heavily influenced
by regional factors such as wave propagation, site conditions and perhaps the fault
rupture mechanism. McGuire (1988) does not actually propose a value for the correlation
length, most probably for the aforementioned reasons.
238
Figure 5.7 Standard deviation of difference of log of the larger peak horizontal acceleration as a
function of interstation spacing leading to the F(r(h,k)) function (Boore et al., 2003)
1.0
Boore et al. (2003)
Wang & Takada (2005) r0 = 5km
Covariance, (h,k)
0.8
0.6
0.4
0.2
0.0
0
10
15
20
Figure 5.8 Comparison of the covariance of the residuals of peak ground motion with separation
distance of two sites for the models of Wang and Takada (2005), McGuire (1988) and Boore et al.
(2003)
239
The models presented in Figure 5.8 can be used to predict the macrospatial correlation of
the residuals of peak ground motion parameters such as PGA, PGV, etc. However,
recently proposed analytical vulnerability assessment methods make use of response
spectra. Hence, a model is required that would enable one to generate a correlated
ground-motion field for different spectral ordinates. Figure 5.9 shows positive and
negative residuals of spectral ordinates (also known as epsilon, the difference between the
spectral ordinate from a record at a given period of vibration and the median spectral
ordinate from a ground-motion prediction equation). Two ground motion measures are
said to be correlated if there is a systematic relationship between their epsilon values.
Baker and Cornell (2006b) have studied the correlation between the total residuals of
spectral ordinates at two different periods, and have proposed a model to predict this
correlation (see Figure 5.10).
Figure 5.9 Example of a positive epsilon record (left) and a negative epsilon record (right) (Baker
and Cornell, 2005)
Figure 5.10 Correlation contours for horizontal epsilons in the same direction at two periods (T1 and
T2) (Baker and Cornell, 2006b)
240
If one assumes that there is no correlation between the spectral ordinates then the
simulated spectra would take the form shown in Figure 5.11a; if a single value of epsilon
is applied to all of the spectral ordinates then the spectrum would be fully correlated as
shown in Figure 5.11b; if the correlation contours of Figure 5.10 are applied when
generating epsilon values for each period then the spectrum would look like Figure 5.11c;
these simulated spectra are compared with real earthquake records (Figure 5.11d) where it
can be seen that the correlated spectra have the closest resemblance in shape to real
records.
Completely Uncorrelated
Correctly Correlated
Completely Correlated
Figure 5.11 Samples of 20 response spectra from magnitude 6.5 earthquakes with a source-to-site
distance of 8 km. (a) Simulated spectra using correlation coefficients equal to zero between all
periods. (b) Simulated spectra using correlation coefficients equal to one between all periods. (c)
Simulated spectra using correlation coefficients from equation (9). (d) Real spectra from recorded
ground motions with magnitude 6.5 and distance 8 km. (Baker and Cornell, 2006b)
Such a correlation between spectral ordinates can easily be included when generating
ground motions at given sites for a given earthquake scenario (one needs only to include
the correlation when sampling epsilon values at each site for each spectral ordinate);
however, one issue that needs to be resolved is how this correlation should be considered
in combination with the macrospatial correlation of the ground-motion residuals at pairs
of sites. A more robust procedure would be to generate the correlation at each response
period as a function of the separation distance, as proposed by Goda and Hong (2008).
The outcome of their research were correlation contours for different separation
distances between sites, as shown in Figure 5.12
241
Figure 5.12 Spatial intraevent correlation for separation distances equal to 2.5, 7.5, and 12.5 km: (a)
calculated by using Californian records and (b) calculated by using a proposed approximate relation
(Goda and Hong, 2008)
242
The simulation of earthquake catalogues that are compatible with the seismicity of a given
area is necessary to determine multiple scenario earthquakes that can be used in loss
calculations. The historical catalogue cannot be used for the definition of such event sets
as it does not describe the full range of events in time and space that could occur within a
specified region. Various possibilities for the definition of a complete catalogue of
earthquakes, either through modification of the historical catalogue or through stochastic
generation of synthetic catalogues, are presented below.
5.3.2 Modified Historical Catalogue
Bommer et al. (2002) modified the historical catalogue to eliminate spatial incompleteness
for their Turkish loss model based on multiple earthquake scenarios. A total of 1039
earthquake scenarios were triggered with a minimum magnitude of 5.5.
Each scenario in the catalogue had an annual frequency of occurrence assigned to it,
calculated from the recurrence relationship. A projected fault rupture plane was used as
the model of each earthquake event source based on the predominant mechanism and
average dip angle for each source zone. For an event of given magnitude, slip type and
dip angle, surface fault rupture length and width were calculated from relationships
provided by Wells and Coppersmith (1994). Each event was triggered with an epicentral
location chosen randomly: anywhere within a source zone for magnitude < 7 and along a
mapped fault of a certain minimum length for magnitude 7. The estimated length of
the fault rupture for a given magnitude defined the selection of available faults within a
zone; it was assumed that the rupture would not exceed the mapped fault length. For a
given (sub-)zone multiple events were triggered for a given magnitude and the event
annual frequency was calculated by dividing the frequency for that magnitude by the
number of events triggered in the (sub-)zone, so as to maintain the temporal description
of the seismicity (see Figure 5.13).
243
Rupture length
Historical earthquake
Annual rate of occurrence = 1/1000 years
Identified fault
Figure 5.13 Illustrative diagram showing the modification of historical earthquakes to eliminate
spatial incompleteness
A similar method is presented in the next section which uses the Monte Carlo method
such that a simple program can be coded to calculate the magnitude and location of
earthquakes in a synthetic catalogue using a seismicity model as the input.
5.3.3 Stochastic catalogues using Monte Carlo simulations
The Monte Carlo simulation method, also known as stochastic modelling, can be used to
generate large numbers of synthetic earthquake catalogues or stochastic event sets.
Although the use of stochastic catalogues is not widely documented in scientific journals,
their use for earthquake risk assessment appears to be common in the commercial sector
(e.g. Zolfaghari, 1998; Eugster et al., 1999; Liechti et al., 2000; Windeler et al., 2004).
However, Musson (1998, 1999) has provided an insight for earthquake engineers into the
mechanism of the Monte Carlo method for the generation of stochastic earthquake
catalogues and their use in probabilistic seismic hazard analysis.
5.3.4 The generation of stochastic earthquake catalogues
A seismic source zone model prepared for a conventional probabilistic seismic hazard
analysis, as discussed in Section 5.1.3, describes with a certain degree of completeness the
spatial and temporal distribution of earthquakes within a given region. The temporal
distribution is modelled using recurrence relationships; in general a Gutenberg-Richter
relationship is used for area sources and a characteristic magnitude recurrence
244
relationship is used for active faults. The seismicity model of a region can be used to
generate synthetic earthquake catalogues using the Monte Carlo method. The method
described herein assumes a Poisson model, such that the occurrence of earthquakes is
independent of the time elapsed since the last event, although time-dependent models
can also be used.
For active faults modelled using a characteristic magnitude recurrence relationship, the
procedure to define a stochastic catalogue for each fault segment is as follows:
1) For each year in the catalogue, a random number between 0 and 1 is generated from a
uniform distribution.
2) If the random number is less than the annual probability of earthquakes on that fault
then this implies an earthquake in that year, otherwise no earthquakes occur.
3) The spatial distribution of the earthquakes is already determined from the fault
segmentation model and the characteristic magnitude is as defined for that fault segment.
If the whole process is repeated for a very large number of years (e.g. 100 000 years of
data), then the ratio between the number of times that the random number falls below
the annual rate of earthquakes and the total number of simulations should be equal to the
annual probability of earthquakes on the fault.
For distributed seismicity in area sources, a two-step random number process is required
to define the catalogue for each source zone:
1) For each year in the catalogue, a random number, Prandom, between 0 and 1 is generated
from a uniform distribution.
2) If the random number is less than the annual probability of events above a chosen
minimum magnitude, Mmin, (i.e. Pmin, obtained from the annual rate Nmin by assuming a
Poisson model) and greater than the annual probability of events, Pmax, below a chosen
maximum magnitude, Mmax, then there is an earthquake within the source zone in that
year:
N min =10( a bM
min
N max =10 ( a bM
max
(5.12)
(5.13)
245
Pmin =1 e N
min
(5.14)
Pmax =1 e N
max
(5.15)
earthquake
(5.16)
where a and b are calculated for the area source from the historical catalogue, using the
Gutenberg Richter relationship, as described in Section 5.1.3. The concept is illustrated in
Figure 5.14; random numbers between 0 and 1 are generated, and the number of times
that these fall between Pmax and Pmin divided by the total number of generated numbers
will approximate the annual probability of earthquakes between the minimum and
maximum magnitude range for the area under consideration.
0 Pmax
Pmin
Annual probability of
exceedance
Figure 5.14 Illustration to show how Monte Carlo method can be used to generate earthquakes. If a
random number is generated between Pmax and Pmin then an earthquake occurs.
a log( N random )
b
(5.17)
(5.18)
In this way, even though the random numbers are uniformly distributed between 0 and 1,
the magnitudes are exponentially distributed between Mmin and Mmax.
4) Each epicentre within the source zone is located by a random latitude and longitude
co-ordinate (N, E) by generating uniformly distributed numbers within the latitude and
246
longitude bounds of the source. The assumption is thus that any location within the
source zone has an equal probability of being the epicentre of the next earthquake,
though this does not have to be the case and other assumptions can be incorporated into
the model.
The use of these stochastically-generated earthquake scenarios to define the ground
motions, and ultimately the loss, at multiple sites is discussed in the next section.
5.3.5 Stochastic Ground Motion Fields
Once a synthetic earthquake catalogue of sufficient length has been generated then, for
each earthquake generated, the ground shaking at each site can be simulated using a
ground-motion prediction equation and the aleatory variability in the equation. For each
earthquake generated:
i) A random number is picked using the standard normal probability distribution to
obtain a random number of standard deviations (): this represents the number of interevent standard deviations, inter, which is multiplied by the inter-event logarithmic
standard deviation of the ground motion parameter (inter).
ii) A ground motion prediction equation is used to calculate the logarithmic mean value
of the ground shaking parameter (e.g. the spectral ordinate at a given period) at the site
using the magnitude and distance from source to site.
iii) At each site, a second random draw from the standard normal probability distribution
is made to define the number of intra-event standard deviations, intra, which is multiplied
by the intra-event logarithmic standard deviation in the ground motion parameter (intra).
It would be possible to include the spatial and inter-period correlation between the
ground motions at different sites here, though at the expense of greatly increasing the
computational effort.
iv) The two random measures of logarithmic scatter are then added to the logarithmic
mean of the ground motion parameter and the exponential is taken to find the predicted
ground motion at each site.
5.4 PSHA VS. STOCHASTIC EARTHQUAKE CATALOGUES IN LOSS MODELS
One approach for defining the hazard in loss models, which is attractive because of the
great computational efficiency that it offers, is to use PSHA to obtain hazard curves at
many locations and then calculate the exceedance frequency of the resulting losses by
convolution of the ground motions at specified return periods with the exposure
247
248
Figure 5.15 Loss curves obtained for total exposure at three sites in the Sea of Marmara region
obtained using PSHA and Monte Carlo simulations (Crowley and Bommer, 2006)
6. SOCIO-ECONOMIC LOSSES
In the last few decades, a dramatic increase in the losses caused by natural catastrophes
has been observed worldwide. Reasons for the increased losses are manifold, though
certainly include the increase of world population (Figure 6.1), the development of new
mega-cities (with a population greater than 10 million), many of which are located in
zones of high seismic hazard, and the high vulnerability of modern societies and
technologies, such as the built environment [e.g. Smolka et al., 2004]. Figure 1.1 showed
the actual worldwide losses from great natural disasters from 1950 to 2003 and fitted
curves to show the increased trend in economic and insured losses. Human losses from
earthquakes are also on the increase and over 420,000 people lost their lives in the past 8
years from earthquakes or tsunamis.
250
losses, on the other hand, is highest, and dramatically increasing, in the industrialized
world, driven primarily by the growing interdependence of national economies on their
built infrastructures.
6.1 ESTIMATION OF SOCIAL LOSSES
Coburn and Spence (2002) discuss the many earthquake-induced causes of death which
include fires following earthquakes, tsunamis, rockfalls and landslides. However, it is clear
that in most large-scale earthquake disasters the main cause of death is the collapse of
buildings (see Figure 6.2). Over the last century about 75% of fatalities attributed to
earthquakes have been caused by the collapse of buildings (Coburn and Spence, 2002),
and the greatest proportion of victims die in the collapse of masonry buildings.
Unfortunately these buildings are very common in seismic areas and make up a large
proportion of the building stock around the world. It should be noted, however, that in
some areas of the world the buildings constructed in reinforced concrete, which has
become the material of choice for new construction, are actually highly vulnerable and
when they do collapse can be more lethal and kill a higher percentage of their occupants
than masonry buildings. Figure 6.2 shows how there has been an increase in fatalities
caused by the collapse of reinforced concrete buildings in the second half of the last
century.
Figure 6.2 Breakdown of earthquake-induced fatalities by cause (Coburn and Spence, 2002)
Coburn and Spence (2002) provide a plot of the total number of people killed versus the
total number of heavily damaged buildings (Figure 6.3), where it can be readily seen that
deaths can be correlated to the destruction caused by earthquakes. However, it should be
noted that the casualty values are highly dispersed when less than 5000 buildings were
damaged.
251
Figure 6.3 Relationship between the number of fatalities and the number of buildings damaged
heavily in earthquake (Coburn and Spence, 2002)
Coburn and Spence (2002) have extensively studied data from past earthquakes and have
found that the lethality ratio (ratio of the number of people killed to the number of
occupants present in the collapsed building, for a given class of building) depends on the
building type and function, occupancy levels, type of collapse mechanism, ground motion
characteristics, occupant behaviour, amongst others. They propose a model to estimate
the number of fatalities (NVT) using a set of M parameters that define the population
per building (M1), the occupancy at the time of the earthquake (M2), the expected
proportion of occupants who will be trapped (M3), the injury distribution at collapse
(M4) and the mortality proportion post-collapse (M5).
N VT = M1M 2 M 3 [ M 4 + M 5 (1 M 4 )]
(6.1)
Figure 6.4 shows the details of each of these M factors, where it can be seen that they are
dependent on the building class.
252
Figure 6.4 Factors M1 to M5 given in Equation (6.1) used to estimate the number of human
casualties due to an earthquake (Coburn and Spence, 2002)
A more updated model recently proposed by Spence (2007) is summarized in Table 6.1.
These proportions give the proportion of the total number of occupants (not those that
are trapped, as considered in the previous model), with different levels of injury as a
function of construction type and number of storeys. To estimate the number of deaths,
the proportions presented in Table 6.1 (column I5) can be multiplied by the proportion of
completely damaged buildings and the estimated number of people in the buildings.
253
Table 6.1 Injury distributions for specific building types (Spence, 2007). UI=uninjured; I1=slight
injuries; I2=moderate injuries; I3=serious injuries; I4=critical injuries; I5=deaths
UI
23.6%
16.5%
9.4%
32.9%
20.8%
9.7%
I1
50.0%
50.0%
50.0%
30.0%
30.0%
30.0%
I2
12.0%
15.0%
18.0%
19.0%
23.0%
27.0%
I3
8.0%
10.0%
12.0%
3.0%
4.0%
5.0%
I4
0.4%
0.5%
0.6%
0.2%
0.2%
0.3%
I5
6.0%
8.0%
10.0%
15.0%
22.0%
28.0%
Earthquake
Managua
Year
Loss ($bn)
GNP that
year ($bn)
Loss
(% GNP)
1972
2.0
5.0
40.0
El Salvador
San Salvador
1986
1.5
4.8
31.0
Guatemala
Guatemala City
1976
1.1
6.1
18.0
Greece
Athens
1999
14.1
110.0
12.8
Yugoslavia
Montenegro
1979
2.2
22.0
10.0
Iran
Manjil
1990
7.2
100.0
7.2
Italy
Campania
1980
45.0
661.8
6.8
Romania
Bucharest
1977
0.8
26.7
3.0
Mexico
Mexico City
1985
5.0
166.7
3.0
USSR
Armenia
1988
17.0
566.7
3.0
Japan
Kobe
1995
82.4
2900.0
2.8
Philippines
Luzon
1990
1.5
55.1
2.7
Greece
Kalamata
1986
0.8
40.0
2.0
China
Tangshan
1976
6.0
400.0
1.5
Quindio
Colombia
1999
1.5
245.0
0.6
USA
Los Angeles
1994
30.0
7866.0
0.3
USA
Loma Prieta
1989
8.0
4705.8
0.2
Turkey
Kocaeli, Izmit
1999
20.0
184.0
0.1
254
The economic losses in Table 6.2 are a combination of direct and indirect losses; direct
losses are correlated to the damage from earthquakes (e.g. cost of repair, cost of
downtime due to loss of function of damaged building), whilst indirect losses arise due
the interruption of business and the effect that has on dependent activities down the line
(e.g., a damaged factory cannot supply materials to other businesses). From an
engineering point of view, the focus is on estimating direct losses due to damage as the
study of indirect losses requires significant input from economic sectors. The direct loss
due to repair or replacement of the building stock is directly related to the damage
distribution from the earthquake; damage ratios which relate the cost of repair to the cost
of replacement are applied to the proportion of buildings in each damage band and the
results are summed to give a Mean Damage Ratio (MDR) which can be multipled by the
value of the buildings to get an absolute estimate of the economic loss. A study carried
out by Bal et al. (2006; 2007) to estimate the damage ratios for each damage state based
on the Turkish building stock is presented in the next section.
6.2.1 A Turkish Study of Cost of Damage for Economic Loss Models
6.2.1.1
The Mean Damage Ratio (MDR) is an indicator that is frequently used to relate damage
calculations with financial loss. The MDR gives the total average loss divided by the value
of the affected buildings or the building stock. In practice, the absolute loss and the
absolute value of the building stock are not calculated separately but a ratio between the
cost of repairing and the reconstruction cost for a given damage state is applied to the
proportion of buildings in each damage state and the results are integrated. In this study,
damage ratios which represent the cost of recovery (repairing a slightly damaged building,
or retrofitting a moderately damaged building and so on) to the cost of reconstruction are
proposed for the Turkish building stock. Four damage states have been considered
herein: slight, moderate, extensive and complete, as this damage grading in accordance
with the official damage definition of the Ministry of Public Works and Settlement of
Turkey.
6.2.1.2
Costs of Retrofitting
The cost of retrofitting is the starting point of the damage ratio calculations. In Turkey,
all moderately damaged buildings should be retrofitted following an earthquake. The data
for the cost of retrofitting can be easily found from different academic and practical
sources in terms of expected cost; however, experience shows that the expected cost of
retrofitting almost never matches the real final cost. This sometimes leads to serious
conflict between the owner and the constructor and/or designer. Differences in the cost
arise because it is not always possible for the designer or the constructor to calculate all
255
the areas of the building which need to be retrofitted since these parts are often covered
by plaster, furniture, walls or sometimes even by soil.
Considering the difficulty of obtaining the cost of retrofitting from the expected design
values, the authors have followed a more straightforward approach by directly asking the
municipalities, design offices or other institutions, who apply retrofitting operations
within the country, how much completed retrofitting projects have cost. A total of 231
sample buildings have been investigated from 4 different governmental and private
sources. A summary of the different types of buildings for which retrofitting scheme
costs were obtained is presented in Table 6.3.
Table 6.3 The building types for which the cost of repair/retrofitting has been obtained
Building Type
Number of buildings
Residential
192
Commercial
School
16
Dormitory
Fire Brigade
16
Town Hall
Total
231
This cost data can be used to calculate a ratio between the cost of repair/retrofitting to
cost of replacement of a building: an important ratio for loss estimation studies as when
this ratio is multiplied by the proportion of buildings with moderate damage and the
value of the building stock, the loss due to moderate damage can be directly obtained.
In order to calculate the reconstruction costs of the buildings, the area of each of the
buildings reported in Table 6.3 has been multiplied by the Approximate Unit
Construction Costs for New Buildings published by the Ministry of Public Works and
Settlement (see Table 6.4). In some cases, the cost of retrofitting needed to be converted
from US dollars to Turkish Lira using dollar rates, and furthermore the influence of
inflation rate at the time of the retrofitting has been taken into account (for those
buildings where retrofitting was carried out some years ago). The results of the
retrofitting cost ratios are presented in Table 6.5.
256
Using the ratio obtained from each source presented in Table 6.5, a mean ratio of 33%
has been obtained considering all 231 buildings, with a standard deviation of 8.2%.
Table 6.4 Approximate Unit Construction Costs for New Buildings, 2006
Cost (*)
Class
Explanation
Ia
28.4
Ib
49.5
IIa
78.4
IIb
146.4
( / m2 )
Residences (up to 4-storey, w/o elevator and central heating) local sportive
facilities, multi-storey car parks, office buildings etc.
Residences and office buildings(w/ elevator and w/o central heating or
w/o elevator and w/ central heating), hotels (1 and 2-star), integrated
agricultural and industrial buildings, official administration buildings, simple
medical facilities, residences etc.
Private dwellings (villas, governor residences etc.), office buildings (w/
elevator and central heating), big shopping malls etc.
200.5
IVb
250.0
IVc
300.5
Va
372.6
Vb
Vc
451.6
515.3
Vd
615.3
IIIa
IIIb
IVa
(*) 1
175.8
226.3
257
6.2.1.3
Source
Number of
Buildings
187
31
19
26
Metropolitan Municipality of
Istanbul
22
53
Tan-Con Co.
20
Damage ratios must be calculated by considering the conditions of the country, i.e. the
use of country specific damage ratios is strongly suggested. HAZUS (FEMA, 2003) and
Yucemen (2005) provide damage ratios for heavily damaged buildings as around 50 %
and 70%, respectively; however, the current laws in Turkey make this assumption
impossible since all heavily damaged buildings should be demolished following an
earthquake. Theoretically, a ratio of cost of repair/retrofitting to cost of replacement of
100% would thus be obtained, as a heavily damaged building would need to be rebuilt.
In reality, additional costs arising from the need to demolish the existing building (which
is not completely collapsed) and transport the rubble away from the site, are incurred. An
exercise to calculate the order of such additional costs has been carried out and can be
summarised as follows:
1. Buildings need to be divided into two groups: those with 4 storeys or without an
elevator and central heating system, and those with an elevator and central heating
system.
2. For each group, approximate construction costs per m2 of reinforced concrete can be
obtained from the Ministry of Public Works and Settlement, and then an average unit
cost of construction can be found for all buildings.
3. The cost of demolishing per m3 of reinforced concrete can also be obtained from the
Ministry of Public Works and Settlement. In the case of complete damage, this cost needs
to be reduced as it is assumed that a proportion of the building is already totally collapsed
(75% has been assumed herein), whilst for extensive damage the cost is multiplied by 1 as
no collapse has occurred and thus the whole building needs to be demolished.
258
4. For the removal and carriage of the rubble, it is assumed that each m2 of construction
produces 0.35m3 of rubble (due to an assumed floor depth of 0.35m), and this has a
weight of 0.875 tons. It is also assumed that the rubble needs to be transported 30km on
roads with a slope 10.
5. The formula used to calculate the unit cost of rubble transportation, F (in Turkish
Lira/ton) is as follows:
F = K( 0.0007 M + 0.01)
(6.2)
The parameter which is probably the hardest to calculate is the cost ratio for slightly
damaged buildings. Slightly damaged buildings are not recorded in any source since they
are constantly ignored because of their lower importance; nevertheless this type of
damage can cause a significant amount of economic loss. Considering the variety of the
damage and the uncertainties involved, an analytical method to calculate the cost of
repairing a slightly damaged building is practically impossible. The chosen manner to
obtain the cost ratio for slightly damaged buildings has thus been to observe the loans
which were given by the government to the householders. Before the TCIP (Turkish
259
Catastrophe Insurance Pool), which insures the properties of the citizens against any
earthquake damage, the Turkish Governments used to cover all of the loss following an
earthquake. A financial source was found, mainly from the World Bank, and the
moderately damaged buildings were assessed, designed and retrofitted. The retrofitting
projects were originally prepared by universities, and since 1999 it has also been possible
for authorised design offices to carry out the design of these projects. The repair of
slightly damaged buildings is not controlled by the government but instead the owners of
these buildings are funded by a constant level of loan which has been found to be more
or less half of the retrofitting cost of moderately damaged buildings (Ministry of Public
Works and Settlement, personal communication). Thus, it is recommended that the cost ratio
of slightly damaged buildings is assumed to be half of the cost of retrofitting (i.e. a
damage ratio of 16%). This value can serve as an initial recommendation but further
research is required in order to study the standard deviation of this value for slightly
damaged buildings.
6.2.2 Discussion
The damage ratios for slight, moderate, extensive and complete damage proposed in the
previous section are compared with those in HAZUS in Figure 6.5. The mean damage
ratio (MDR) has been found to be very sensitive to the cost ratios assigned to each
damage state (Crowley et al., 2005) and considering the large difference in the damage
ratios proposed herein and those used in HAZUS, the importance of defining countryspecific damage ratios is shown to be of utmost importance.
An important issue which needs to be considered when defining damage ratios (or so
called replacement cost ratios) relates to the description of the damage ranges. In this
report, a damage grading in accordance with the official damage definition of the Ministry
of Public Works and Settlement of Turkey has been selected, as was also considered by
Yucemen (2005). This grading starts from no damage and continues to slight damage,
moderate damage, heavy damage and total collapse. Other studies (e.g. Durukal et al.
(2002)) have used an EMS98 compatible damage scale, but the authors have some
reservations with the use of damage scales which do not correspond to the legal criteria
within a country.
260
120
100
105
104
80
60
33
40
20
16
Slight
Moderate
Severe
Collapse
Figure 6.5 The damage ratio comparison for different damage states
Another important issue that has been considered herein is the definition of damage
ratios (or replacement costs) according to the real practice applied in the country. The
inconsistency between the HAZUS parameters for damage ratios and the real Turkish
practice has already been demonstrated above in Figure 6.5. In addition, Yucemen (2005)
claims that the damage ratio defined for heavily damaged buildings in Turkey would be
on average around 70 %. Despite the fact that their damage definitions are not exactly the
same, Durukal et al. (2002) provide a value of 50% for this damage state. However, if one
considers legal limitations (which require buildings to be demolished if they are heavily or
extensively damaged) then, as demonstrated herein, the damage ratio for extensive
damage should be at least 100%, or even higher since those buildings have to be
demolished following the earthquake.
REFERENCES
Abo Al Ezz, A. (2007) Deformation and strength based assessment of seismic failure
mechanisms for existing RC frame buildings, MSc Thesis, ROSE School, Pavia, Italy.
Aboutaha, R.S., M.D.Engelhardt, J.O. Jirsa & M.E. Kreger (1996) Retrofit of concrete columns
with inadequate lap splices by the use of rectangular steel jackets, Earthquake Spectra, 12(4),
693-714.
Abrahamson, N.A. and Bommer, J.J. (2005) Opinion Paper: Probability and uncertainty in
seismic hazard analysis. Earthquake Spectra; 21(2):1-5.
Abrams, D.P. (1992). Strength and behaviour of unreinforced masonry elements, Proc. 10th
World Conf. on Earthq. Engg., Madrid.
ACI Committee 408 (1991) Abstract of State-of-the-art report: Bond under cyclic loads, ACI
Materials Journal, Vol. 88, No. 6, pp. 669-673.
Alexandrou, E. (1991) Seismic response of RC frames with material variability, MSc Dissertation,
Civil Engineering Department, Imperial College.
Allen, E.W. and Bailey, J.S. (1988) Seismic rehabilitation of the Salt Lake City and County
Building using base isolation, Proceedings of the Ninth World Conference on Earthquake Engineering,
Tokyo-Kyoto, Japan, Vol. 5, pp. 639-644.
Ambraseys, N.N., Elnashai, A.S., Broderick, B.M., Salama, A.I. and Soliman, M.M. (1992) The
Erzincan (Turkey) earthquake of 15 March 1992, Research Report ESEE-11/92, Engineering
Seismology and Earthquake Engineering Section, Imperial College, London, UK.
American Society of Civil Engineers (2000) FEMA 356 Prestandard and Commentary for the Seismic
Rehabilitation of Buildings, Federal Emergency Management Agency, Washington, D.C.
Anagnostopoulos, S.A., J. Petrovski & J.G Bouwkamp (1989) Emergency earthquake damage and
usability assessment of buildings, Earthquake Spectra, 5, 461-476.
262
263
Applied Technology Council (1997) NEHRP Guidelines for the Seismic Rehabilitation of
Buildings, FEMA Report No. 273, Federal Emergency Management Agency, Applied
Technology Council, Washington, D.C.
Aviles, N.B., K. Maruyama & L.E. Rojas (1996) Improving strength and ductility by using steel
plate wrapping, Proceedings of the Eleventh World Conference on Earthquake Engineering, Acapulco,
Paper No. 742.
Aydinoglu, M. N. (2003) An incremental response spectrum analysis procedure based on inelastic
spectral deformation for multi-mode seismic performance evaluation, Bulletin Earthquake
Engineering, 1(1), 3-36.
Ayoama, H. (1981) A method for the evaluation of the seismic capacity of existing reinforced
concrete buildings in Japan, Bulletin of the New Zealand National Society for Earthquake
Engineering 14(3), 105-131.
Baker, J.W. and Cornell, C.A. (2006a) Which spectral acceleration are you using? Earthquake
Spectra 22, 293-312.
Baker, J.W. and Cornell, C.A. (2006b) Correlation of response spectral values for
multicomponent ground motions, Bulletin of the Seismological Society of America 96(1), 215-227.
Bal .E., Crowley, H., Pinho, R. and Glay G., (2008) Detailed assessment of structural
characteristics of Turkish RC building stock for loss assessment models, Soil Dynamics and
Earthquake Engineering, 28(10-11), 914-932.
Bal, .E. (2005) Rapid assessment techniques for collapse vulnerability of reinforced concrete
buildings, MSc Thesis, stanbul Technical University, Civil Engineering Department (in
Turkish).
Bal, .E. (2008) Displacement-Based Earthquake Loss Assessment: Method Development and
Application to Turkish Building Stock, PhD Thesis, ROSE School Pavia, Italy.
Ballio, G., and Mazzolani, F. M., 1987. Strutture in acciaio, Ulrico Hoepli editore S.p.A., Milano,
Italy. (in Italian)
Banon, H., Biggs, J., and Irvine, M., (1981) Seismic Damage in Reinforced Concrete Frames,
Journal of Structural Engineering, ASCE, 107 (ST9), 1713-1729.
Barbat, A.H., Ypez Moya, F. and Canas, J.A. (1996). Damage scenarios simulation for seismic
risk assessment in urban zones, Earthquake Spectra, Vol. 12, No. 3, pp. 371-394.
264
Bazant, S., and Bhat, P., (1977) Prediction of Hysteresis in Reinforced Concrete Members,
Journal of Structural Engineering, ASCE, 103 (ST1), 151-167.
Bazzurro, P. and Luco, N. (2005). Accounting for uncertainty and correlation in earthquake loss
estimation, Proc. 9th Int. Conf. on Structural Safety and Reliability (ICOSSAR05), Rome, Italy.
Benedetti, D. and Petrini, V. (1984). Sulla vulnerabilit di edifici in muratura: Proposta di un
metodo di valutazione, Lindustria delle Costruzioni, Vol. 149, pp. 66-74.
Bernardini, A. (2000). La vulnerabilit degli edifici: valutazione a scala nazionale della vulnerabilit
sismica degli edifici ordinari, CNR-Gruppo Nazionale per la Difesa dai Terremoti Roma,
Italy.
Bernardini, A., Gori, R. and Modena, C. (1990). An application of coupled analytical models and
experimental knowledge for seismic vulnerability analyses of masonry buildings, Engineering
Aspects of Earthquake Phenomena, A. Koridze (ed.); Vol. 3, pp. 161-180.
Bertero, V., Aktan, A., Charney, F., and Sause, R., (1984) Earthquake Simulator Tests and
Associated Experimental, Analytical and Correlation Studies of One-Fifth Scale Model, in
Earthquake Effects on Reinforced Concrete Structures, American Concrete Institute, SP-84-13,
Detroit, 375-424.
Bertero, V.V. (1980) Seismic behaviour of R/C wall structural systems, Proceedings of 7th World
Conference on Earthquake Engineering, Istanbul, Turkey, Vol. 6, pp. 323-330.
Bertero, V.V. (1992) Seismic upgrading of existing structures, Proceedings of the Tenth World
Conference on Earthquake Engineering, Madrid, vol. 9, 5101-5106.
Bertero, V.V. (1997) Performance-based seismic engineering: A critical review of proposed
guidelines, Proceedings of the International Workshop on Seismic Design Methodologies for the Next
Generation of Codes, Bled, Slovenia, pp. 1-31.
Bertero, V.V. and Popov, E.P (1977) Seismic behaviour of ductile moment-resisting reinforced
concrete frames, in ACI SP53: Reinforced Concrete Structures in Seismic Zones, ACI, Detroit,
Michigan, pp. 247-291.
Bertero, V.V., Cowel, A. and Popov, E. (1980) Repair of bond in RC structures by epoxy
injection, Proceedings of the First Seminar on Repair and Retrofit of Structures, US/Japan Cooperative Earthquake Engineering Research Program, Dept. of Civil Engineering, Univ. of
Michigan, Ann Arbor, pp. 234-240.
265
Bertoldi, S.H., Decanini, L.D., Santini, S. and Via, G. (1994) Analytical Models in Infilled
Frames, Proceedings of the 10th European Conference in Earthquake Engineering, Vienna.
Biondi S., Colangelo F., Nuti C. (2000) La Risposta Sismica dei Telai con Tamponature Muraie, Research
Report, GNDT, Rome, Italy. (in Italian)
Bird J.F. and Bommer, J.J. (2004). Earthquake losses due to ground failure, Engineering Geology,
Vol. 75, No. 2, pp. 147-179.
Bird, J.F., Bommer, J.J., Crowley, H. and Pinho, R. (2006). Modelling liquefaction-induced
building damage in earthquake loss estimation, Soil Dynamics and Earthquake Engineering, Vol.
26, No. 1, 15-30.
Bird, J.F., Crowley, H., Pinho, R. and Bommer, J.J. (2005). Assessment of building response to
liquefaction-induced differential ground deformation, Bulletin of New Zealand Society of
Earthquake Engineering, Vol. 38, No. 4.
Blandon-Uribe, C.A. (2005) Implementation of an Infill Masonry Model for Seismic Assessment
of Existing Buildings, Individual Study, ROSE School, Pavia, Italy.
Bommer, J. J., and Acevedo, A. B., (2004) The use of real earthquake accelerograms as input to
dynamic analysis, Journal of Earthquake Engineering, 8 (Special Issue 1), 43-91.
Bommer, J.J. & Elnashai, A.S. (1999) Displacement spectra for seismic design, Journal of
Earthquake Engineering 3(1), 1-32.
Bommer, J.J. & Ruggeri, C. (2002) The specification of acceleration time-histories in seismic
design codes, European Earthquake Engineering 16(1).
Bommer, J.J. and Crowley, H. (2006). The influence of ground motion variability in earthquake
loss modelling, Bulletin of Earthquake Engineering, Vol. 4, No. 3, pp. 231-248.
Bommer, J.J., A.S. Elnashai & A.G. Weir (2000) Compatible acceleration and displacement
spectra for seismic design codes, Proceedings of the Twelfth World Conference on Earthquake
Engineering, Auckland, Paper No. 206.
Bommer, J.J., and Martinez-Pereira, A. (1999) The effective duration of earthquake ground
motion, Journal of Earthquake Engineering, 3 (2), 127-172.
266
Bommer, J.J., Pinho, R. and Crowley, H. (2005). Using displacement-based earthquake loss
assessment in the selection of seismic code design levels, Proceedings of 9th International
Conference on Structural Safety and Reliability (ICOSSAR 05), Rome, Italy.
Bommer, J.J., Spence, R., Erdik, M., Tabuchi, S., Aydinoglu, N., Booth, E., Del Re, D. and
Peterken, O. (2002). Development of an earthquake loss model for Turkish catastrophe
insurance, Journal of Seismology, Vol. 6, No. 3, pp. 431-446.
Bommer, JJ., Spence, R. and Pinho, R. (2006). Earthquake loss estimation models: time to open
the back boxes?, Proceedings of 1st European Conference on Earthquake Engineering and Seismology,
Geneva, paper no. 834.
Boore, D.M. (1997) Estimates of average spectral amplitudes at FOAKE sites, Appendix C in An
evaluation of methodology for seismic qualification of equipment, cable trays, and ducts in
ALWR plants by use of experience data, K. K. Bandyopadhyay, D. D. Kana, R. P. Kennedy,
and A. J. Schiff (Editors), U.S. Nuclear Regulatory Commission NUREG/CR-6464 and
Brookhaven National Lab BNL-NUREG- 52500, C-1-C-69.
Boore, D.M. (2005) Erratum: Seismological Research Letters, Vol. 68, No. 1, pp. 128-153,
January/February 1997, Equations for estimating horizontal response spectra and peak
acceleration from western north American earthquakes: A summary of recent work, D.M.
Boore, W.B. Joyner, and T.E. Fumal Seismological Research Letters 76, 368-369.
Boore, D.M., Gibbs, J.F., Joyner, W.B., Tinsley, J.C. and Ponti, D.J. (2003) Estimated Ground
Motion From the 1994 Northridge, California, Earthquake at the Site of the Interstate 10 and
La Cienega Boulevard Bridge Collapse, West Los Angeles, California, Bulletin of Seismological
Society of America 93(6), 2737-2751.
Booth, E.D., Kappos, A.J., Park, R., and Moehle, J.P. (1998) Seismic design of concrete
structures: a comparison of Eurocode 8 with other international practice, Proceedings of 6th
SECED Conference (Oxford, UK), Balkema.
Borzi, B., Pinho, R. and Crowley, H. (2008a) Simplified pushover-based vulnerability analysis for
large scale assessment of RC buildings, Engineering Structures, Vol. 30, No. 3, pp. 804-820.
Borzi, B., Crowley, H. and Pinho, R. (2008b) Simplified pushover-based earthquake loss
assessment (SP-BELA) method for masonry buildings, International Journal of Architectural
Heritage, Vol. 2, No. 4, pp. 353-376.
Borzi, B., Elnashai, A.S., Faccioli, E., Calvi, G.M. and Bommer, J.J. (1998) Inelastic spectra and
ductility-damping relationships for displacement-based seismic design, Research Report
267
ESEE/98-4, Engineering Seismology and Earthquake Engineering Section (joint report with
Politenico di Milano, Italy), Imperial College, London, UK.
Borzi, B., G.M. Calvi, A.S. Elnashai, E. Faccioli & J.J. Bommer (2001) Inelastic spectra for
displacement-based seismic design, Soil Dynamics & Earthquake Engineering 21, 47-61.
Bracci, J. M., Kunnath, S. K., and Reinhorn, A. M., (1997) Seismic performance and retrofit
evaluation for reinforced concrete structures, ASCE Journal of Structural Engineering, 123(1),
3-10.
Braga, F., Dolce, M. and Liberatore, D. (1982). A statistical study on damaged buildings and
ensuing review of the MSK-76 scale, Proceedings of 7th European Conference on Earthquake
Engineering, Athens, September.
Brancaleoni, F., Ciampi, V. , and Di Antonio, R., (1983) Rate-Type Models for Non Linear
Hysteretic Structural Behavior, EUROMECH Colloquium, Palermo, Italy.
Brundson, D.R. & M.J.NM. Priestley (1984) Assessment of seismic performance characteristics
of reinforced concrete buildings constructed between 1936 and 1975, Bulletin of the New
Zealand National Society for Earthquake Engineering 17(3), 163-184.
BSSC (1992) NEHRP Handbook for the seismic evaluation of existing buildings. FEMA Report 178,
Building Seismic Safety Council for the Federal Emergency Management Agency,
Washington, D.C.
Building Seismic Safety Council, (2000) Prestandard and Commentary for the Seismic Rehabilitation of
Buildings, FEMA-356, Washington D.C., USA.
Calvi, G.M. (1999). A displacement-based approach for vulnerability evaluation of classes of
buildings, Journal of Earthquake Engineering, Vol. 3, No. 3, pp. 411-438.
Calvi, G.M. (1999) Performance-based approaches for seismic assessment of existing structures,
Proceedings of the Eleventh European Conference on Earthquake Engineering: Invited Lectures, Bisch,
Labb & Pecker eds., Balkema, 3-19.
Calvi, G.M. and Pavese, A. (1995) Displacement based design of building structures, Proceedings
of the Fifth SECED Conference on European Seismic Design Practice, Chester, UK, pp. 127-132.
Calvi, G.M. and Pinho, R. (2004). LESSLOSS. A European Integrated Project on Risk Mitigation
for Earthquakes and Landslides, ROSE Research Report 2004/02, IUSS Press, Pavia, Italy.
268
269
CEB (1985) Model Code for Seismic Design of Concrete Structures. CEB Bull. d' Inf., No. 165,
Paris.
CEB (1993) CEB-FIP Model Code 1990. CEB Bull. d' Inf., No. 213/214, Lausanne.
CEN - Comite Europeen de Normalisation (2003). Eurocode 8, Design of Structures for
Earthquake Resistance Part 1: General rules, seismic actions and rules for buildings, PrEN 1998-1. Final Draft. December 2003.
CEN (1991) European Prestandard ENV 1992-1-1: Eurocode 2 Design of concrete structures,
Part 1-1: General rules and rules for buildings. Comite Europeen de Normalisation,
Brussels.
CEN (1996) European Prestandard ENV 1998-1-4: Eurocode 8 Design provisions for
earthquake resistance of structures, Part 1-4: Strengthening and repair of buildings. Comite
Europeen de Normalisation, Brussels.
CEN Techn. Comm. 250 / SC8 (1995) Eurocode 8: Earthquake Resistant Design of StructuresPart 1: General rules (ENV 1998-1-1/2/3), CEN, Brussels.
CEN, (2002) Eurocode 8: Design of Structures for Earthquake Resistance - Part 2: Bridges,
PrEN 1998-2: 2003, 2 April 2002, CEN, Brussels, Belgium.
CEN (2005) Eurocode 8: Design of Structures for Earthquake Resistance - Part 2: Bridges, EN
1998-2, Brussels, Belgium.
Charney, F., and Bertero, VV., (1982) An Evaluation of the Design and Analytical Seismic
Response of a Seven Story Reinforced Concrete Frame-Wall Structure. EERC Report 82/08,
Earthquake Engrg. Research Center, Univ. of California, Berkeley.
Chen, S.-J., Huang, T. and Lu, L.-W. (1988) Diaphragm behaviour of reinforced concrete floor
slabs. Proceedings of 9th World Conf. on Earthq. Engng., Tokyo-Kyoto, Japan, Vol. IV, pp. 565570.
Chen, Y.-H. and Tsai, C.-C.P. (2002) A new method for estimation of the attenuation
relationship with variance components, Bulletin of the Seismological Society of America 92(5),
1984-1991.
Cheung, P.C., Paulay, T. and Park, R. (1992) Some possible revisions to the seismic provisions of
the New Zealand Concrete Design Code for moment resisting frames, Bulletin of the New
Zealand National Society for Earthquake Engineering, Vol. 25, No. 1, pp. 37-43.
270
Chopra A. K., and Goel R. K., (2002) A modal pushover analysis procedure for estimating
seismic demands for buildings, Earthquake Engneering and Structural Dynamics, 31, 561-582.
Chopra, A. K. and Goel, R. K. (2000) Building period formulas for estimating seismic
displacements, Earthquake Spectra 16(2), 533536.
Chopra, A.K., Goel, R.K., and Chintanpakdee, C. (2004) Evaluation of a Modified MPA
Procedure Assuming Higher Modes as Elastic to Estimate Seismic Demands, Earthquake
Spectra, 20 (3), 757-778.
Ciampi, V., and Carlesimo, L., (1986) A Nonlinear Beam element for Seismic Analysis of
Structures, Proceedings of 8th European Conference on Earthquake Engineering, Lisbon.
Clough, R. W., and Penzien , J., (1994) Dynamics of Structures, 2nd Edition, McGraw Hill.
Clough, R., and Benuska, L., (1967) Nonlinear Earthquake Behavior of Tall Buildings, Journal of
Engineering Mechanics, ASCE, 93(EM 3), 129-146.
Clough, R., and Johnston, S., (1966) Effect of Stiffness Degradation on Earthquake Ductility
Requirements, Transactions of Japan Earthquake Engineering Symposium, Tokyo, 195-198.
Coburn, A. & R. Spence (1992) Earthquake Protection. John Wiley & Sons.
Coburn, A. and Spence, R. (2002). Earthquake Protection, Second Edition, Wiley and Sons.
Cook, D., Malkus, D. S., and Plesha, M. E., (1989) Concepts and applications of Finite Element Analysis,
3rd Edition, Wiley & Sons, Inc.
Cornell, A. (1968) Engineering seismic risk analysis. Bulletin of the Seismological Society of America;
58(5): 1583-1606
Corsanego, A. and Petrini, V. (1990). Seismic vulnerability of buildings work in progress, Proc.
Workshop II on Seismic Risk Vulnerability and Risk Assessment, Trieste, Italy, December, pp. 577
598.
Cosenza, E., Manfredi, G., Polese, M. and Verderame, G.M. (2005). A multi-level approach to
the capacity assessment of existing RC buildings, Journal of Earthquake Engineering, Vol. 9,
No. 1, pp. 1-22.
Cowell, A. D.; Popov, E. P.; Bertero, V. V. (1980) Repair of bond in R/C structures by epoxy
injection, Proceedings of the First Seminar on Repair and Retrofit of Structures, US/Japan Co-
271
272
Crowley, H., Pinho, R. and Bommer, J.J. (2004). A probabilistic displacement-based vulnerability
assessment procedure for earthquake loss estimation, Bulleting of Earthquake Engineering, Vol.
2, No.2, pp. 173-219.
Culver, C., Lew, H.S., Hart, G.C. and Pinkham, C.W. (1975) Natural Hazards Evaluation of
Existing Buildings. Building Science Series, National Bureau of Standards, Washington.
Cusson, D. and Paultre, P. (1995) Stress-strain model for confined high-strength concrete,
Journal of Structural Engineering, ASCE, Vol. 121, No. 3, pp. 468-477.
DAyala, D. and Speranza, E. (2002) An integrated procedure for the assessment of seismic
vulnerability of masonry structures, Proceedings of 12th European Conference on Earthquake
Engineering, London, paper no. 561.
Decanini, L.D. and Fantin, G.E. (1987) Modelos Simplificados de la Mamposteria Incluida en
Porticos. Caractreisticas de Rigidez y Resistencia Lateral en Estrado Limite, (in spanish)
Jornadas Argentinas de Ingenieria Estructural, 2, 817-836
Del Valle Caldern, E. (1980) Some lessons from the March 14, 1979 Earthquake in Mexico
City. Proceedings of the Seventh World Conference on Earthquake Engineering, Istanbul, Turkey, Vol.
4, pp. 545-552.
Di Pasquale, G., Fralleone, A., Pizza, A.G., Serra, C. (1999a) Synthesis of the code evolution
from the Royal Decree issued after the Messina and Reggio earthquake up to the first
Ministry decree issued after the Messina and Reggio earthquake up to the first decree issued
after the law n. 64/74, in La Classificazione e la normative sismica italiana dal 1909 al 1984, De
Marco, R. and Martini, M.G. (eds.) Istituto Poligrafico e Zecca dello Stato, Roma.
Di Pasquale, G., Fralleone, A., Pizza, A.G., Serra, C. (1999b) Relevant changes to the Italian
Seismic Code from 1909 to 1975 a synoptic table in La Classificazione e la normative sismica
italiana dal 1909 al 1984, De Marco, R. and Martini, M.G. (eds.) Istituto Poligrafico e Zecca
dello Stato, Roma.
Di Pasquale, G., Orsini, G. and Romeo, R.W. (2005). New developments in seismic risk
assessment in Italy, Bulletin of Earthquake Engineering, Vol. 3, No. 1, pp. 101-128.
Doherty K.T., Griffith M.C., Lam N. and Wilson, J. (2002). Displacement-based Seismic Analysis
for Out-of-plane Bending of Unreinforced Masonry Walls, Earthquake Engineering and
Structural Dynamics, Vol. 31, pp. 833-850.
273
Dolce, M., Masi, A., Marino, M. and Vona, M. (2003). Earthquake damage scenarios of the
building stock or Potenza (Southern Italy) including site effects, Bulletin of Earthquake
Engineering, Vol. 1, No. 1, pp. 115-140.
Donea, J., Magonette, G., Negro, P., Pegon, P., Pinto, A. and Verzeletti, G. (1996)
Pseudodynamic capabilities of the ELSA laboratory for earthquake testing of large
structures. Earthquake Spectra, Vol. 12, No. 1, pp. 163-180.
Dumova-Jovanoska, E. (2004). Fragility curves for RC structures in Skopje region, Proceedings of
13th World Conference on Earthquake Engineering, Vancouver, Canada.
Durukal, E., Erdik, M., Sesetyan, K and Fahjan, Y. (2002), Building loss estimation for eathquake
insurance pricing, Proceedings of the Seventh US National Conference on Earthquake Engineering,
Boston.
Earthquake Engineering Research Institute (EERI), (1997) Recovery and Reconstruction from
Recent Earthquakes, Proc. 5th United States / Japan Workshop on Urban Earthquake Hazard
Reduction, Pasadena, CA, United States, pp. 455
EERI (1996) Spectra Theme Issue: Repair and rehabilitation research for seismic resistance of
structures, J.O. Jirsa (ed.), Earthquake Spectra, Vol. 12, No. 4.
Eibl, J., and Keintzel, F. (1988) Seismic Shear Forces in RC Cantilever Shear Walls, Proceedings of
the Ninth World Conference in Earthquake Engineering, Tokyo-Kyoto, Japan, Paper 9-1-1.
Eligehausen, R., Popov, E.P and Bertero, V.V. (1983) Local Bond Stress-Slip Relationships of
Deformed Bars under Generalized Excitations. Report EERC-83/23, Univ. of California,
Berkeley.
Elnahsai, A.S. & R. Pinho (1998b) Repair and retrofitting of RC walls using selective techniques.
Journal of Earthquake Engineering 2(4), 525-568.
Elnashai, A. S., (2001) Advanced inelastic static (pushover) analysis for earthquake applications,
Structural Engineering Mechanics, 12(1), 51-69.
Elnashai, A.S. & R. Pinho (1998a) Current status and future developments for repair and
strengthening of RC structures in European seismic design practice. Proceedings of the Eleventh
European Conference on Earthquake Engineering, Paris.
274
275
Fajfar, P., and Fischinger, M. (1988) N2 A method for non-linear seismic analysis of regular
buildings, Proceedings of the Ninth World Conference in Earthquake Engineering, Tokyo-Kyoto,
Japan, 5, pp. 111-116.
Fardis, M. (1999) Seismic assessment and retrofit of RC structures. Proceedings of the Eleventh
European Conference on Earthquake Engineering: Invited Lectures, Bisch, Labb & Pecker eds.,
Balkema, 53-64.
Fardis, M. N. (1994) Analysis and design of reinforced concrete buildings according to Eurocode
2 and 8. Configuration 3, 5 and 6, Reports on Prenormative Research in Support of Eurocode 8.
Fardis, M.N. (1998) Seismic assessment and retrofit of RC structures. Proceedings of the Eleventh
European Conference on Earthquake Engineering Invited Lectures, Paris, France, pp. 53-64.
Federal Emergency Management Agency (FEMA) (1997) NEHRP Guidelines for the Seismic
Rehabilitation of Buildings, FEMA 273, Washington, DC, USA.
FEMA (1999). HAZUS 99. Earthquake Loss Estimation Methodology. Technical Manual,
Federal Emergency Management Agency, Washington D.C.
FEMA (2000), FEMA 310 - Handbook for the Seismic Evaluation of Buildings A Prestandard,
Federal Emergency Management Agency, Washington DC, USA.
FEMA (2001). FEMA 366: HAZUS estimated earthquake losses for the United States, Federal
Emergency Management Agency, Washington D.C.
FEMA (2003). HAZUS-MH Technical Manual, Federal Emergency Management Agency,
Washington D.C.
Filippou, F. C., and Issa, A., (1988) Nonlinear Analysis of Reinforced Concrete Frames Under
Cyclic Load Reversals, EERC Report 88/12, Earthquake Engineering Research Center,
Univ. of California, Berkeley.
Filippou, F. C., Popov, E. P., and Bertero, V. V., (1983) Modelling of R/C joints under cyclic
excitations, Journal of Structural Engineering, 109(11), 2666-2684.
Foutch, D.A., Hjelmstad, K.D. and Del Valle Caldern, E. (1988) Seismic retrofit of a RC
building: A case study. Proceedings of the Ninth World Conference on Earthquake Engineering,
Tokyo-Kyoto, Japan, Vol. 7, pp. 451-456.
276
Fragiadakis, M., Pinho, R., and Antoniou, S., (2006) Modelling Inelastic Buckling of Reinforcing
Bars under Earthquake Loading, Journal of Structural Engineering, ASCE. (submitted for
publication)
Freeman, S.A. (1998) Development and use of capacity spectrum method, Proceedings of the Sixth
U.S. National Conference on Earthquake Engineering, Seattle, Oakland, USA.
Freeman, S.A., Nicoletti, J.P. and Tyrell, J.V. (1975) Evaluations of Existing Buildings for Seismic
Risk A Case Study of Puget Sound Naval Shipyard, Bremerton, Washington. Proceedings of
First the U.S. National Conference on Earthquake Engineers, Oakland, California, pp. 113-122.
Frosch, R.J, Li, W., Jirsa, J.O and Kreger, M.E. (1996) Retrofit of non-ductile moment-resisting
frames using precast infill wall panels, Earthquake Spectra, Vol. 12, No. 1, pp. 163-180.
Gergely, P. (1977) Experimental and analytical investigations of reinforced concrete frames
subjected to earthquake loading. Proceedings of Workshop on Earthquake Resistant Reinforced
Concrete Building Construction, Univ. of California, Berkeley, July 1977, Vol. III, pp. 1175-1195.
Ghobarah, A., Biddah, A. and Mahgoub, M. (1997) Rehabilitation of reinforced concrete
columns using corrugated steel jacketing. Journal of Earthquake Engineering, Vol. 1, No. 4, pp.
651-673.
Ghobarah, A., N.M. Aly & M. El-Attar (1997) Performance level criteria and evaluation. Seismic
Design Methodologies for the Next Generation of Codes, Fajfar & Krawinkler eds.,
Balkema, 207-215.
Giberson, M., (1967) The Response of Nonlinear Multi-Story Structures Subjected to Earthquake
Excitations, Earthquake Engineering Research Laboratory, Pasadena.
Giovinazzi, S. (2005). The vulnerability assessment and the damage scenario in seismic risk
analysis, PhD Thesis, Technical University Carolo-Wilhelmina at Braunschweig and
University of Florence.
Giovinazzi, S. and Lagomarsino, S. (2001). Una metodologia per lanalisi di vulnerabilit sismica
del costruito, X Congresso nazionale LIngegneria Sismica in Italia, Potenza-Matera.
Giovinazzi, S. and Lagomarsino, S. (2004). A macroseismic method for the vulnerability
assessment of buildings, Proceedings of 13th World Conference on Earthquake Engineering,
Vancouver, Canada.
277
Glaister, S., and Pinho, R. (2003) Development of a simplified deformation based method for
seismic vulnerability assessment, Journal of Earthquake Engineering, 7(S1), 107-140.
GNDT. (1993). Rischio Sismico di edifici Pubblici-Parte I Aspetti Metodologici, CNR Gruppo Nazionale per la Difesa dai Terremoti, Roma.
GNDT. (2000). The Catania Project: earthquake damage scenarios for a high risk area in the
Mediterranean, CNR - Gruppo Nazionale per la Difesa dai Terremoti, Roma.
Goda, K. and Hong, H. P. (2008) Spatial Correlation of Peak Ground Motions and Response
Spectra, Bulletin of Seismological Society of America, 98(1), 354365.
Goel, R. K., and Chopra, A. K., (2005a) Response to B. Maisons Discussion of Evaluation of
Modal and FEMA Pushover Analysis: SAC Buildings, Earthquake Spectra, 21(1), 277-279.
Goel, R. K., and Chopra, A. K., (2005b) Role of Higher-Mode Pushover Analyses in Seismic
Analysis of Buildings, Earthquake Spectra 21(4), 1027-1041.
Goel, R.K. and Chopra, A.K., (1997) Period formulas for moment-resisting frame buildings,
Journal of Structural Engineering, 123(11), 1454-1461.
Goel, S.C. and Masri, A.C. (1996) Seismic strengthening of an RC slab-column frame with ductile
steel bracing. Proceedings of the Eleventh World Conference on Earthquake Engineering, Acapulco,
Mexico, Paper No. 506.
Griffith, M.C., Magenes, G., Melis, G. and Picchi, L. (2003). Evaluation of out-of-plane stability
of unreinforced masonry walls subjected to seismic excitation, Journal of Earthquake
Engineering, Vol. 7, Special Issue 1, pp. 141-169.
Grnthal, G. (1998). European Macroseismic Scale 1998, Cahier du Centre Europen de
Godynamique et de Sismologie, Vol. 15, Luxembourg.
Guagenti, E. and Petrini, V. (1989). The case of old buildings: towards a damage-intensity
relationship, Proceedings of 4th Italian National Conference on Earthquake Engineering, Milan, pp.
145-153 (in Italian).
Guedes, J., (1997) Seismic behaviour of reinforced concrete bridges. Modelling, numerical
analysis and experimental assessment, PhD Thesis, Univ. of Porto, Porto, Portugal.
Gulkan, P. & M. Sozen (1974) Inelastic response of reinforced concrete structures to earthquake
motions. ACI Journal 71, 604-610.
278
Gutenberg, B. and Richter, C.F. (1944) Frequency of earthquakes in California. Bulletin of the
Seismological Society of America; 34(4): 1985-1988.
Hall, J. F., (2005) Problems encountered from the use (or misuse) of Rayleigh damping,
Earthquake Engineering and Structural Dynamics, 35(5), 525-545.
Hamburger, R.O. & C.A. Cole (2001) Seismic upgrading of existing structures. In: The Seismic
Design Handbook, Naeim ed., second edition, Kluwer Academic Publishers, 624-679.
Hamburger, R.O. (1997) Defining performance objectives. Proceedings of the International Workshop
on Seismic Design Methodologies for the Next Generation of Codes, Bled, Slovenia, pp. 33-42.
Hanson, R.D. (ed.) (1980) Proceedings of the First Seminar on Repair and Retrofit of
Structures, US/Japan Co-operative Earthquake Engineering Research Program, Dept. of Civil
Engineering, Univ. of Michigan, Ann Arbor, Vol. 1.
Hassan, A.F. and Sozen, M.A. (1997). Seismic vulnerability assessment of low-rise buildings in
regions with infrequent earthquakes, ACI Struct. J., Vol. 94, No. 1, pp. 31-39.
Hellesland, J. and, Scordelis, A., (1981) Analysis of RC Bridge Columns Under Imposed
Deformations, IABSE Colloquium, Delft, 545-559.
Hernndez-Montes E., Kwon, O. S., and Aschheim, M. (2004) An energy-based formulation for
first and multiple-mode nonlinear static (pushover) analyses, Journal of Earthquake Engineering,
8(1), 69-88.
Higashi, Y., Endo, T. and Shimizu, Y. (1984) Experimental studies on retrofitting of reinforced
concrete buildings. Proceedings of Eighth World Conference on Earthquake Engineering, San
Francisco, United States, Vol. 4, pp. 477-484.
Hilber H. M., Hughes T. J. R., and Taylor R. L., (1977) Improved numerical dissipation for time
integration algorithms in structural dynamics, Earthquake Engineering and Structural Dynamics,
5(3), 283-292.
Hoshikuma J. et al. (1997) Stress-strain model for confined reinforced concrete in bridge piers,
Journal of Structural Engineering, ASCE, Vol. 123, No. 5, pp. 624-633.
Hughes, T. J. R., (1987) The Finite Element Method, Linear Static and Dynamic Finite Element Analysis,
Prentice-Hall, Inc.
279
280
281
Kowalsky, M.J., Priestley, M.J.N. and MacRae, G.A. (1995) Displacement-based design of RC
bridge columns in seismic regions. Earthquake Engineering and Structural Dynamics, Vol. 24,
No. 12, pp. 1623-1643.
Kramer, S.L. (1996). Geotechnical Earthquake Engineering. Prentice-Hall.
Krawinkler H., and Seneviratna G. D. P. K., (1998) Pros and cons of a pushover analysis of
seismic performance evaluation, Engineering Structures, 20(4-6), 452-64.
Krinitzsky, E.L. (2002) How to obtain earthquake ground motions for engineering design.
Engineering Geology; 65: 1-16.
Kunnath, S. K., (2004) Identification of modal combination for nonlinear static analysis of
building structures, Computer-Aided Civil and Infrastruc. Eng., 19, 246-259.
Kunnath, S. K., and John, A. Jr., (2000) Validity of static procedures in performance-based
seismic design, Proceedings of ASCE Structures Congress, Philadelphia, USA.
Lantada, N,.Pujades, L.G. and Barbat, A.H. (2004). Risk scenarios for Barcelona, Spain,
Proceedings of 13th World Conference on Earthquake Engineering, Vancouver, Canada, paper no.
423.
Ledbetter, S.R. & J.J. Bommer (1987) The San Salvador earthquake of 10th October 1986. A
Field Report by EEFIT. Earthquake Engineering Field Investigation Team.
Liauw, T.C. and Kwan, K.H. (1984) Nonlinear Behaviour of Non-Integral Infilled Frames,
Computers and Structures, 18, n 3, 551-560.
Liechti, D., Rttener, E. and Zbinden, A. (2000) Disaggregation of average annual loss, XXV
European Geophysical Society General Assembly, Nice, France, 25-29 April, 2000.
Lpez-Menjivar, M. A., (2004) Verification of a displacement-based Adaptive Pushover method
for assessment of 2-D Reinforced Concrete Buildings, PhD Thesis, European School for
Advances Studies in Reduction of Seismic Risk (ROSE School), University of Pavia, Italy.
Lupoi, A., Franchin, P., and Pinto, P.E., 2007. Further probing of the suitability of push-over
analysis for the seismic assessment of bridge structures, Proceedings of the First US-Italy
Workshop on Seismic Design of Bridges, Pavia, Italy.
282
283
McGuire, R.K. (1988) Seismic risk to lifeline systems: critical variables and sensitivities,
Proceedings of 9th World Conference on Earthquake Engineering Vol. VII, pp. 129-134.
Medvedev, S. and Sponheuer, W. (1969). MSK scale of seismic intensity, Proceedings of 4th World
Conference on Earthquake Engineering: Chilean Assoc. Seism. and Earthq. Engg., Santiago.
Medvedev, S.V. (1969) Engineering Seismology. US Department of Commerce publication.
Meinheit, D.F. and Jirsa, J.O. (1981) Shear strength of R/C beam-column connections. Journal of
the Structural Division, ASCE, Vol. 107, No. 11, pp. 2227-2244.
Menegotto, M., and Pinto P. E., (1973) Method of analysis for cyclically loaded R.C. plane frames
including changes in geometry and non-elastic behaviour of elements under combined
normal force and bending, Symposium on the Resistance and Ultimate Deformability of Structures
Acted on by Well Defined Repeated Loads, International Association for Bridge and Structural
Engineering, Zurich, Switzerland, 15-22.
Miranda, E. and Bertero, V.V. (1990) Post-tensioning technique for seismic upgrading of existing
low-rise buildings. Proceedings of the Fourth US National Conference on Earthquake Engineering,
Palm Springs, California, Vol. 3, pp. 393-402.
Moehle, J.P. (1992) Displacement-based design of RC structures subjected to earthquakes.
Earthquake Spectra, Vol. 8, No. 3, pp. 403-428.
Moghadam, A. S., and Tso, W. K, (2002) A pushover procedure for tall buildings, Proceedings of
12th European Conference on Earthquake Engineering, London, UK, Paper 395.
Mokka, A.S., Amin, N., Constantinou, M.C. and Zayas, V. (1996) Seismic isolation retrofit of
large historic building. Journal of Structural Engineering, Vol. 122, No. 3, pp. 298-308.
Molina, S. and Lindholm, C. (2005). A logic tree extension of the capacity spectrum method
developed to estimate seismic risk in Oslo, Norway, Journal of Earthquake Engineering, Vol. 9,
No. 6, 877-898.
Monti, G., Nuti, C., (1992) Nonlinear cyclic behaviour of reinforcing bars including buckling,
Journal of Structural Engineering, 118(12), 3268-3284.
Musson, R.M.W (1998) On the use of Monte Carlo simulations for seismic hazard assessment
Proceedings of 6th U.S. National Conference on Earthquake Engineering.
284
Musson, R.M.W. (1999) Determination of design earthquakes in seismic hazard analysis through
Monte Carlo simulation Journal of Earthquake Engineering; 3(4): 463-474.
Nakano, Y. (1995) Damage to buildings due to 1994 Sanriku-harukaoki earthquake. Building
Disaster 211, 6-15, Japan Building Disaster Prevention Association (in Japanese).
Newmark, N. M., (1959) A method of computation for structural dynamics, Journal of the
Engineering Mechanics Division, ASCE, Vol. 85, No. EM3, 67-94.
NZNSEE (2002) New Zealand National Society for Earthquake Engineering, The assessment
and improvement of the structural performance of earthquake risk buildings, Draft prepared
for the NZ Building Industry Authority.
NZNSEE (1996) The assessment and improvement of the structural performance of earthquake
risk buildings. Draft for General Release. National Zealand National Society for Earthquake
Engineering.
NZSA (1992) Code of practice for general structural design and design loadings for buildings,
NZS4203, New Zealand Standards Association, Wellington.
Oesterle, R.G., Fiorato, A.E., Aristizabal-Ochoa, J.D. and Corley, W.G., (1980) Hysteretic
Response of Reinforced Concrete Structural Walls, in: ACISP-63: Reinforced Concrete Structures
subjected to Wind and Earthquake Forces, American Concrete Institute, Detroit, pp. 243-273.
Okada, T., Nakano, Y. and Kumazawa, F. (1992) Experimental study of 1/10 scaled R/C frames
retrofitted with steel framed Y-shaped bracing system. Proceedings of the International Symposium
on Earthquake Disaster Prevention, Mexico City, pp. 200-209.
Oliveira, C.S., Ferreira, M.A. and Mota de S, F. (2004). Seismic vulnerability and impact analysis:
elements for mitigation policies, Proceedings of the XI Congresso Nazionale Lingegneria Sismica in
Italia, Genova, Italy.
Oliveira, C.S., Mota de S, F. and Ferreira, M.A. (2005). Application of two different vulnerability
methodologies to assess seismic scenarios in Lisbon, Proceedings of 250th Anniversary 1755
Lisbon Earthquake, Lisbon, Portugal.
Ordaz, M., Miranda, E., Reinose, E. and Prez-Rocha, L. E. (2000). Seismic loss estimation
model for Mexico City, Proceedings of. 12th World Conference on Earthquake Engineering,
Auckland, New Zealand, paper no. 1902.
285
Orsini, G. (1999). A model for buildings vulnerability assessment using the parameterless scale of
seismic intensity (PSI), Earthquake Spectra, Vol. 15, No. 3, pp. 463-483.
Otani, S., (1974) Inelastic Analysis of R/C Frame Structures, Journal of Structural Engineering,
ASCE, 100 (ST7), 1433-1449.
Ozdemir, P., Boduroglu, M.H. and Ilki, A. (2005). Seismic safety screening method, SPEAR
Workshop An event to honour the memory of Jean Donea, Ispra, 4-5 April, 2005.
Pantazopoulou, S. and Imran, I. (1992) Slab-wall connections under lateral forces. ACI Structural
Journal, Vol. 89, No. 5, pp. 515-527.
Papaioannou, I., Fragiadakis, M., and Papadrakakis, M., (2005) Inelastic analysis of framed
structures using the fiber approach, Proceedings 5th GRACM International Congress on
Computational Mechanics, Limassol, Cyprus.
Papia, M. and Russo, G. (1989) Compressive concrete strain at buckling of longitudinal
reinforcement, Journal of Structural Engineering, ASCE, Vol. 115, No. 2, pp. 382-397.
Paraskeva, T., Kappos, A., and Sextos, A. (2006) Development and evaluation of a modal
pushover analysis procedure for seismic assessment of bridges, Earthquake Engineering and
Structural Dynamics, 35 (10), 1269-1293.
Paret, T. F., Sasaki, K. K., Eilbeck, D. H., and Freeman, S. A., (1996) Approximate inelastic
procedures to identify failure mechanisms from higher mode effects, Proceedings of 11th World
Conference on Earthquake Engineering, Acapulco, Mexico, Paper 966.
Park, R. (1988) State-of-the-art report: ductility evaluation from laboratory and analytical testing,
Proceedings of the Ninth World Conference on Earthquake Engineering, Tokyo-Kyoto, Japan, VIII,
605616.
Park, R. and Dai, R. (1988) A comparison of the behaviour of reinforced concrete beam-column
joints designed for ductility and limited ductility. Bulletin of the New Zealand National Society for
Earthquake Engineering, Vol. 21, No. 4, pp. 255-278.
Park, R. and Paulay, T. (1975) Reinforced Concrete Structures, J. Wiley & Sons, New York.
Park, R., Priestley, M.J.N. and Gill, W.D. (1982) Ductility of square confined concrete columns.
Journal of the Structural Division, ASCE, Vol. 108, No. 4, pp. 929-950.
286
Park, Y.J. and Ang, A.H.S. (1985). Mechanistic seismic damage model for reinforced concrete,
Journal of Structural Engineering, ASCE, Vol. 111, No. 4, pp. 722-739.
Park, Y.-J. and Ang, A.H.-S. (1985) Mechanistic seismic damage model for reinforced concrete.
Journal of Structural Engineering, ASCE, Vol. 111, No. 4, pp. 722-739.
Paulay, T. (1982) Lapped splices in earthquake-resisting columns. Journal of the ACI, Vol. 79, No.
6, pp. 458-469.
Paulay, T. (1986) A critique of the special provisions for seismic design of the Building Code
Requirements for Reinforced Concrete (ACI 318-83) Journal of the ACI, Vol. 83, No. 2, pp.
274-283.
Paulay, T., Priestley, M.J.N. (1992) Seismic Design of Reinforced Concrete and Masonry Buildings, John
Wiley & Sons, Inc., New York.
Paulay, T., Park, R. and Priestley, M.J.N. (1978) Reinforced concrete beam-column joints under
seismic actions. Journal of the ACI, Vol. 75, No. 11, pp. 585-593; also: Discussion by D.F.
Meinheit and Authors' closure, Journal of the ACI, Vol. 76, No. 5, May 1979, pp. 662-667.
Paulay, T., Priestley, M.J.N. and Synge, A.J. (1982) Ductility in Earthquake Resisting Squat Shearwalls, ACI Journal, Vol. 79, No. 4, pp. 257-269.
Penelis G.G. and A.J. Kappos (1997), Earthquake-resistant Concrete Structures, E & FN SPON
(Chapman & Hall), London.
Penelis, G.G., D. Sarigiannis, E. Stavrakakis and K.C. Stylianidis (1989) A statistical evaluation of
damage to buildings in the Thessaloniki, Greece earthquake of June, 20, 1978. Proceedings of
9th World Conference on Earthquake Engineering, (Tokyo-Kyoto, Japan, Aug. 1988),
Tokyo:Maruzen, Vol. VII, pp. 187-192.
Pietra, D. (2006) Evaluation of pushover procedures for the seismic design of buildings, MSc
Dissertation, ROSE School, IUSS, Pavia, Italy.
Pinho, R. (2000) Selective retrofitting of RC structures in seismic areas. PhD Thesis, Imperial
College, University of London.
Pinho, R., (2000) Shaking Table Testing of RC Walls, ISET Journal of Earthquake Technology, 37(4),
119-142.
287
Pinho, R., and Elnashai, A. S. (2000) Dynamic collapse testing of a full-scale four storey RC
frame, ISET Journal of Earthquake Engineering, Special Issue on Experimental Techniques, 37(4),
143-164.
Pinho, R., Bommer, J.J. and Glaister, S. (2002). A simplified approach to displacement-based
earthquake loss estimation analysis, Proceedings of 12th European Conference on Earthquake
Engineering, London, England, paper no. 738.
Pinho, R., Casarotti, C., and Antoniou, S. (2007) A comparison of single-run pushover analysis
techniques for seismic assessment of bridges, Earthquake Engineering and Structural Dynamics,
36 (10), 1347-1362.
Pinho, R., Elnashai, A.S. and Vaz, C.T. (1999) Shaking-table tests on selectively retrofitted RC
walls. Research Report ESEE/99-4, Engineering Seismology and Earthquake Engineering
Section, Imperial College, London, 1999.
Pinto, A. V., Verzeletti, G., Molina, J., Varum, H., Pinho, R., and Coelho, E., (2002) Pseudo-dynamic
tests on non-seismic resisting RC frames (bare and selective retrofit frames), Research Report EUR 20244
EN, Joint Research Centre, Ispra, Italy.
Pinto, A.V. (1998) Earthquake performance of structures behavioural, safety and economical
aspects. PhD Thesis, Instituto Superior Tcnico, Technical University of Lisbon, Portugal.
Pinto, A.V., Verzeletti, G., Molina, J. and Varum, H. (1999) Pseudo-dynamic tests on non-seismic
resisting RC frames (bare and selective retrofitted frames) Special Publication, ISIS, EC,
JRC, Ispra, Italy.
Popov, E. P. et al. (1977) On seismic design of R/C interior joints of frames. Proceedings of the
Sixth World Conference on Earthquake Engineering, New Delhi, India, Vol. II, pp. 1933-1938.
Priestley, M. J. N. (1997) Displacement-based seismic assessment of reinforced concrete
buildings, Journal of Earthquake Engineering, (1)1, pp.157-192.
Priestley, M.J.N. (1998) Brief comments on elastic flexibility of reinforced concrete frames and
significance to seismic design, Bulletin of the New Zealand National Society of Earthquake
Engineering, 31(4), 246-259.
Priestley, M.J.N. (1993) Myths and fallacies in earthquake engineering conflicts between design
and reality. Bulletin of the New Zealand National Society for Earthquake Engineering 26(3), 329-341.
288
Priestley, M.J.N. (1999) Displacement-based approaches to rational limit states design of new
structures. Proceedings of the Eleventh European Conference on Earthquake Engineering: Invited
Lectures, Bisch, Labb & Pecker eds., Balkema, 317-335.
Priestley, M.J.N. (2003). Myths and fallacies in earthquake engineering revisited, The Mallet
Milne Lecture, IUSS Press, Pavia, Italy.
Priestley, M.J.N. and Park, R. (1987) Strength and ductility of concrete bridge columns under
seismic loading. ACI Struct. Journal, Vol. 84, No. 1, pp. 61-76.
Priestley, M.J.N. & G.M. Calvi (1991) Towards a capacity-design assessment procedure for
reinforced concrete frames. Earthquake Spectra 7(3), 413-437.
Priestley, M.J.N. & G.M. Calvi (1997) Assessment of existing buildings. CEB Bulletin 236
Seismic Design of Reinforced Concrete Structures, CEB Task Group 3.2, Comit EuroInternational du Bretn, Laussane, Switzerland, 159-204.
Priestley, M. J . N., and Grant, D. N., (2005) Viscous damping in seismic design and analysis,
Journal of Earthquake Engineering, 9 (Special Issue 2), 229-255.
Priestley, M.J.N., F. Seible & G.M. Calvi (1996) Seismic Design and Retrofit of Bridges. John Wiley &
Sons Inc., New York.
Priestley, M.J.N., Calvi, G.M., Kowalsky, M.J. (2007) Displacement-Based Seismic Design of Structures,
IUSS Press, Pavia, Italy.
Public Building Services (1978) Earthquake Resistance of Buildings. Volume II: Evaluation of
Existing Structures. General Services Administration, Washington DC, March.
PWRI (1986) Manual for Repair Methods of Civil Engineering Structures Damaged by Earthquakes, Public
Works Research Institute, Ministry of Construction, Japan.
Reinhorn, A., (1997) Inelastic analysis techniques in seismic evaluations, in Seismic design
methodologies for the next generation of codes, Krawinkler and Fajfar (editors), Balkema, 277-287.
Reiter, L. (1990) Earthquake Hazard Analysis: Issues and Insights. Columbia University Press, New
York.
Requena, M., and Ayala, G. (2000) Evaluation of a simplified method for the determination of
the nonlinear seismic response of RC frames, Proceedings of 12th World Conference on Earthquake
Engineering, Auckland, New Zealand, Paper No. 2109.
289
Restrepo-Vlez, L.F. and Magenes, G. (2004) Simplified procedure for the seismic risk
assessment of unreinforced masonry buildings, Proceedings of 13th World Conference on
Earthquake Engineering, Vancouver, Canada, paper no. 2561.
Restrepo-Vlez, L.F. and Magenes, G. (2005a) Experimental testing in support of a mechanicsbased procedure for the seismic risk evaluation of unreinforced masonry buildings,
International Seminar SAHC 2004 Structural Analysis of Historical Constructions, Modena,
Loureno and Roca eds, Vol. 2, Taylor and Francis, London, pp. 1079-1089.
Restrepo-Vlez, L.F. and Magenes, G. (2005b) A Mechanics-Based Methodology for the
Evaluation of the Seismic Risk of Unreinforced Masonry Buildings, ROSE Research Report
No. 2005/05, IUSS Press, Pavia, Italy
Rodriguez, M. and Park, R. (1991) Repair and strengthening of reinforced concrete buildings for
seismic resistance. Earthquake Spectra, Vol. 7, No. 3, pp. 439-459.
Rossetto, T. and Elnashai, A. (2003). Derivation of vulnerability functions for European-type RC
structures based on observational data, Engineering Structures, Vol. 25, pp. 1241-1263.
Rossetto, T. and Elnashai, A. (2005). A new analytical procedure for the derivation of
displacement-based vulnerability curves for populations of RC structures, Engineering
Structures, Vol. 7, No. 3, pp. 397-409.
Rota, M., Penna, A. and Strobbia, C. (2006). Typological fragility curves from Italian earthquake
damage data, Proceedings of 1st European Conference on Earthquake Engineering and Seismology,
Geneva, paper no. 386.
Saadatmanesh, H., M.R. Ehsani & L. Jin (1997) Seismic retrofitting of rectangular bridge columns
with composite straps. Earthquake Spectra 13(2), 281-304.
Sabetta, F. & J. Bommer (2002) Modification of the spectral shapes and subsoil conditions in
Eurocode 8. Proceedings of the Twelfth European Conference on Earthquake Engineering, London,
Paper No.518.
Sabetta, F. and Pugliese, A. (1987). Attenuation of peak horizontal acceleration and velocity from
Italian strong motion records, Bulletin of Seismological Society of America, Vol. 77, No. 5, pp.
1491-1513.
Sabetta, F., Goretti, A. and Lucantoni, A. (1998). Empirical fragility curves from damage surveys
and estimated strong ground motion, Proceedings of 11th European Conference on Earthquake
Engineering, Paris.
290
SAC Joint Venture (1997) Develop Suites of Time Histories, Project Task: 5.4.1, Draft Report,
March 21, Sacramento, CA, USA.
Salama, A.I. (1993) Repair of earthquake-damaged RC structures. PhD Thesis, Imperial College,
University of London, London.
Salonikios, T., Tegos, I., Kappos, A. and Penelis, G. (1996) Squat R/C walls under inelastic shear
reversals, Proceedings of 11th World Conference on Earthquake Engineering (Acapulco, Mexico,
June 1996), CD ROM Proceedings, Pergamon, Paper No. 749.
Sasaki, K. K., Freeman S. A., and Paret T. F., (1998) Multi-mode pushover procedure (MMP) a
method to identify the effects of higher modes in a pushover analysis, Proceedings of 6th US
Conference on Earthquake Engineering, Seattle, Washington Earthquake Engineering Research
Institute, Oakland, California.
Satyarno, I., Carr, A. J., and Restrepo, J., (1998) Refined pushover analysis for the assessment of
older reinforced concrete buildings, Proc. NZSEE Technology Conference, Wairakei, New
Zealand, 75-82.
Scawthorn, C., Iemura, H. and Yamada, Y. (1981). Seismic damage estimation for low- and midrise buildings in Japan, Earthquake Engineering and Structural Dynamics, Vol. 9, pp. 93-115.
Scott, B. D., Park, R., and Priestley, M. J. N., (1982) Stress-Strain Behavior of Concrete Confined
by Overlapping Hoops at Low and High Strain Rates, ACI Journal, 79(1), 13-27.
SEAOC (1995) Vision 2000: Performance Based Seismic Engineering of Buildings Part 2:
Conceptual Framework. Vision 2000 Committee, Structural Engineers Association of
California, Sacramento, California.
SeismoSoft (2009) SeismoStruct: A computer program for static and dynamic nonlinear analysis
of framed structures, Available from URL: http://www.seismosoft.com.
Sheikh, S.A. and Yeh, C.- C. (1990) Tied concrete columns under axial load and flexure. Journal
of Structural Engineering, ASCE, Vol. 116, No. 10, pp. 2780-2800.
Shibata, A., and Szen, M. (1976) Substitute structure method for seismic design in reinforced
concrete, ASCE Journal of Structural Engineering, 102(1), pp.1-8.
Shinozuka, M., Chang, S.E., Eguchi, R.T., Abrams, D.P., Hwang, H.H.M. and Rose, A. (1997).
Advances in earthquake loss estimation and application to Memphis, Tennessee,
Earthquake Spectra, Vol. 13, No. 4, pp. 739-758.
291
Singhal, A. and Kiremidjian, A.S. (1996). Method for probabilistic evaluation of seismic structural
damage, Journal of Structural Engineering, ASCE, Vol. 122, No. 12, pp. 1459-1467.
Singhal, A. and Kiremidjian, A.S. (1998). Bayesian updating of fragilities with application to RC
frames, Journal of Structural Engineering, ASCE, Vol. 124, No. 8, pp. 922-929.
Smolka, A., Allmann, A., Hollnack, D. and Thrainsson, H. (2004). The principle of risk
partnership and the role of insurance in risk mitigation, Proceedings of 13th World Conference on
Earthquake Engineering, Vancouver, B.C., Canada.
Smyrou E., Blandon-Uribe C., Antoniou S. and Pinho R. (2006) Implementation and verification
of a masonry panel model for nonlinear dynamic analysis of for infilled RC frames,
Proceedings of the First European Conference on Earthquake Engineering and Seismology, Geneva,
Switzerland, Paper no. 355.
Sobiah, M.E., Soliman, M.M., Elnashai, A.S. (1992) Seismic Vulnerability Evaluation of Existing
Reinforced Concrete Buildings. International Symposium on Earthquakes and Ways of Protection,
Aleppo, Syria.
Soliman, M.M. (1992) Seismic Vulnerability Evaluation for Existing Reinforced Concrete
Buildings, PhD Thesis, Cairo University, 252 pp (joint supervision, Imperial College)
Sommerville P., Smith N., Punyamurthula S. and Sun J., (1997) Development of ground motion time
histories for phase 2 of the FEMA/SAC steel project, SAC Background Document, Report No.
SAC/BD/-7/04, Applied Technology Council, Redwood City, CA.
Soroushian, P. and Sim, J. (1986) Axial behaviour of reinforced concrete columns under dynamic
loads. Journal of the ACI, Vol. 83, No. 6, pp. 1018-1025.
Sousa, M.L., Campos Costa, A., Carvalho, A. and Coelho, E. (2004). An automatic seismic
scenario loss methodology integreated on a geographic information system, Proceedings of.
13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada.
Spacone, E., (1994) Flexibility-Based Finite Element Models for the Nonlinear Static and
Dynamic Analysis of Concrete Frame Structures, Ph.D. Dissertation., Department of Civil
Engrg., Univ. of California, Berkeley.
Spacone, E., (2001) A module for Analysis and Design of Segmental Prestressed Concrete
Bridges (CASI-TR-01-04), Final report of a CASI FY00 Technology Transfer Grant. Fort
Collins, CO: Colorado Advanced Software Institute.
292
Spence, R. Ed (2007) Earthquake disaster scenario predictions and loss modelling for urban
areas, LESSLOSS Report 7, IUSS Press, Pavia, Italy.
Spence, R., Bommer, J., Del Re, D., Bird, J., Aydinoglu, N. and Tabuchi, S. (2003). Comparing
loss estimation with observed damage: A study of the 1999 Kocaeli earthquake in Turkey,
Bulletin of Earthquake Engineering, Vol. 1, pp. 83-113.
Spence, R., Coburn, R.W. and Pomonis, A. (1992). Correlation of ground motion with building
damage: the definition of a new damage-based seismic intensity scale, Proceedings of 10th
World Conference. on Earthquake Engineering, Rotterdam.
Stafford-Smith, B. and Carter, C. (1969) A Method for the Analysis of Infilled Frames,
Proceedings of the Institution of Civil Engineers, London, England, 44, 31-38.
Standards Association of New Zealand (1982) (a) Code of Practice for the Design of Concrete
Structures (NZS 3101 - Part 1:1982); (b) Commentary on Code of Practice for the Design of
Concrete Structures (NZS 3101 - Part 2:1982), Wellington.
Standards New Zealand (1995) Concrete Structures Standard (NZS 3101) - Part 1: The Design of
Concrete Structures; Part 2: Commentary on The Design of Concrete Structures, Wellington.
STAP (1999) Repair and Strengthening Products Catalogue, S.T.A.P., Lisbon, Portugal (in
portuguese)
Strasser, F.O., Bommer, J.J. and Abrahamson, N.A. (2008) Truncation of the distribution of
ground-motion residuals. Journal of Seismology 12(1), 79-105.
Strasser, F.O., Bommer, J.J., Sesetyan, K., Erdik, M., Cagnan, Z., Irizarry, J., Goula, X., Lucantoni,
A., Sabetta, F., Bal, I.E., Crowley, H. and Lindholm, C. (2008) A comparative study of
European Earthquake Loss Estimation Tools for an Earthquake Scenario in Istanbul,
Journal of Earthquake Engineering, Volume 12, Issue S2, pp. 246-256.
Sugano, S. (1970) Experimental study on restoring force characteristics of reinforced concrete
members, PhD Thesis, University of Tokyo, Japan (in Japanese).
Sugano, S. (1996) State-of-the-art techniques for rehabilitation of buildings. Proceedings of the
Eleventh World Conference on Earthquake Engineering, Acapulco, Paper No. 2178.
Sullivan, T., Pinho, R., and Pavese, A., (2004) An Introduction to Structural Testing Techniques in
Earthquake Engineering, Educational Report ROSE 2004/01, IUSS Press, Pavia, Italy.
293
Takeda, T., Sozen, M. A., and Nielsen, N., (1970) Reinforced Concrete Response to Simulated
Earthquakes, Journal of Structural Engineering, ASCE, 96 (ST12), 2557-2573.
Takizawa, H., (1976) Notes on Some Basic Problems in Inelastic Analysis of Planar RC
Structures, Transaction of Architecture Institute of Japan, 240, Part I in February, 51-62, Part II in
March, 65-77.
Tassios, T.P (1989a) Specific rules for Concrete Structures - Justification Note No. 5: Anchorage
of beam's reinforcements across a column, in: Background document for Eurocode 8 - Part 1, Vol.
2 - Design Rules, CEC DG 111/8076/89 EN, 18-22.
Tassios, T.P (1989b) Specific rules for Concrete Structures - Justification Note No. 6: Required
confinement for columns, in: Background document for Eurocode 8 - Part 1, Vol. 2 - Design Rules,
CEC DG 111/8076/89 EN, 23-49.
Tassios, T.P (1991) Seismic design of concrete structures, Ch. 4 in Structures subjected to Dynamic
Loading - Stability and Strength (R. Narayanan and T.M. Roberts, Eds.), Elsevier, London, pp.
115-165.
Tegos, I. and Penelis., G.G. (1988) Seismic Resistance of Short Columns and Coupling Beams
Reinforced with Inclined Bars, ACI Journal, Vol. 85, No. 2, pp. 82-88.
Teran, A. and Ruiz, J. (1992) Reinforced concrete jacketing of existing structures. Proceedings of
the Tenth World Conference on Earthquake Engineering, Madrid, Spain, Vol. 9, pp. 5107-5112.
Tolis, S.V. & E. Faccioli (1999) Displacement design spectra. Journal of Earthquake Engineering
3(1), 107-125.
Tsonos, A.G., Tegos, I.A. and Penelis, G.G. (1992) Seismic Resistance of Type 2 Exterior BeamColumn Joints Reinforced with Inclined Bars, ACI Journal, Vol. 89, No. 1, pp. 3-12.
Uang, C.-M. (1993) An evaluation of two-level seismic design procedures Earthquake Spectra 9(1),
121-135.
Umehara, H. and Jirsa, J.O. (1984) Short rectangular RC columns under bidirectional loading.
Journal of Structural Engineering, ASCE, Vol. 110, No. 3, pp. 605-618.
UNIDO (1983) Building Construction Under Seismic Conditions in the Balkan Region, Vol. 4:
Post-earthquake damage evaluation and strength assessment of buildings under seismic
conditions, United Nations Industrial Developing Organisation, UNDP/UNIDO project
RER/79/015, Vienna, Austria.
294
Vamvatsikos D., and Cornell C. A., (2002) Incremental dynamic analysis, Earthquake Engineering
and Structural Dynamics, 31(3), 491-514.
Varum, H. and Pinto, A. (1999) Dynamic nonlinear analyses for the 4-storey RC frame. Bare,
infilled and retrofitted frames. ICONS Report, Topic 2 Assessment, Strengthening and
Repair, ISIS, EC, JRC, Ispra, Italy.
Veneziano, D., Sussman, J.M., Gupta, U. and Kunnumkal, S.M. (2002). Earthquake loss under
limited transportation capacity: assessment, sensitivity and remediation, Proceedings of 7th US
National Conference on Earthquake Engineering, Boston.
Wang, M. and Takada, T. (2005) Macrospatial correlation model of seismic ground motions,
Earthquake Spectra 21(4), 1137-1156.
Warner, J. (1996) Shotcrete in seismic repair and retrofit. Seismic Rehabilitation of Concrete
Structures, American Concrete Institute, G.M Sabnis, A.C. Shroff and L.F. Kahn (eds),
Farmington Hills, Michigan, pp. 299-313.
Weichert, D.H. (1980) Estimation of the earthquake recurrence parameters for unequal
observation periods for different magnitudes, Bulletin of the Seismological Society of America;
70(4): 1337-1346.
Wells, D.J. and Coppersmith, K.J. (1994) New empirical relationships among magnitude, rupture
length, rupture width, rupture area, and surface displacement, Bulletin of the Seismological
Society of America; 84: 974-1002.
Whitman, R.V., Anagos, T., Kircher, C.A., Lagorio, H.J., Lawson, R.S. and Schneider, P. (1997).
Development of a national earthquake loss estimation methodology, Earthquake Spectra,
Vol. 13, No. 4, pp. 643-661.
Whitman, R.V., Reed, J.W. and Hong, S.T. (1973). Earthquake Damage Probability Matrices,
Proceedings of 5th World Conference on Earthquake Engineering.
Wiggins, J.H. (1964) Effect of Site Conditions on Earthquake Intensity. Journal of the American
Society of Civil Engineers, ST Division.
Windeler, D., Morrow, G., Williams, C.R., Rahnama, M., Molas, G., Pea, A. and Bryngelson, J.
(2004) Earthquake risk estimates for residential construction in the western United States,
Proceedings of 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada, August
1-6, 2004.
295
Wong, P.K.C., Priestley, M.J.N. and Park, R. (1990) Seismic resistance of frames with vertically
distributed longitudinal reinforcement in beams. ACI Struct. Journal, Vol. 87, No. 4, pp. 488498.
Wood, H.O. and Neumann, F. (1931). Modified Mercalli Intensity Scale of 1931, Bulletin of
Seismological Society of America, Vol. 21, No. 4, pp. 277-283.
Xian,D. (1992) Inelastic response and effective design asymmetric buildings under strong
earthquake loading. PhD Thesis, Civil Engineering Department, University College London.
Yakut, A. (2004). Preliminary seismic performance assessment procedure for existing RC
buildings, Engineering Structures, Vol. 26, No. 10, pp. 1447-1461.
Yeh, C.H., Jean, W.Y. and Loh, C.H. (2000). Damage building assessment for earthquake loss
estimation in Taiwan, Proceedings of 12th World Conference on Earthquake Engineering, Auckland,
New Zealand
Yucemen, M. S. (2005) Probabilistic assessment of earthquake insurance rates for Turkey,
Natural Hazards, 35:291-313.
Zahn, F.A. Park, R. and Priestley, M.J.N. (1989) Strength and ductility of square reinforced
concrete column sections subjected to biaxial bending,ACI Structural Journal, Vol. 86, No. 2,
pp. 123-131.
Zeris, C. A., and Mahin, S. A., (1988) Analysis of Reinforced Concrete Beam-Columns Under
Uniaxial Excitation, Journal of Structural Engineering, ASCE, 114 (ST4), 804-820.
Zeris, C. A., and Mahin, S. A., (1991) Behavior of Reinforced Concrete Structures Subjected to
Biaxial Excitation, Journal of Structural Engineering, ASCE, 117 (ST9), 2657-2673.
Zezhen, N. (1986) Seismic Safety Evaluation Criteria of Existing Buildings in China. Proceedings of
the Eight Regional Symposium on Earthquake Engineering, Vol. I, pp. 301-307.
Zolfaghari, M.R. (2000) Earthquake loss estimation model for southern Europe. Proceedings of 6th
International Conference on Seismic Zonation.