Sensitivity of Analytical Fragility Functions to

Capacity-related Parameters
DRAFT FOR COMMENTS

Dina D’Ayala1, Abdelghani Meslem1

1University College London
Sensitivity of Analytical Fragility Functions to
Capacity-related Parameters

Dina D’Ayala, Abdelghani Meslem

Version: 1.0
April, 2013

© GEM Foundation 2010. All rights reserved

The views and interpretations in this document are those of the individual author(s) and should not be attributed to GEM
Foundation. This report may be freely reproduced, provided that it is presented in its original form and that
acknowledgement of the source is included. GEM Foundation does not guarantee that the information in this report is
completely accurate, hence you should not solely rely on this information for any decision-making.

Citation: D’Ayala D., Meslem A. Sensitivity of analytical fragility functions to capacity-related parameters, GEM Technical
Report 2013-X, GEM Foundation, Pavia, Italy.
www.globalquakemodel.org
ii

ABSTRACT
Recent extensive literature review, which has been conducted within the framework of the development of the “GEM Guide
for Selecting of Existing Analytical Fragility Curves and Compilation of the Database”, has shown that often fragility
functions are generated using simplified assumptions to reduce the calculation efforts. Some of the most widely
implemented simplifications are: the use of default values to model structural characteristics-related parameters; the use of
2-D models, and ignoring, for the case of infilled RC buildings, the contribution of infill panels in the seismic response by
modelling them as bare frame structures. However, these assumptions may highly decrease the reliability and accuracy of
the obtained results introducing important epistemic uncertainty in the fragility function construction process. The present
document is devoted to provide, for GEM guidelines users, details on the effect that the choice of building capacity-related
parameters and their expected uncertainty, might have on the results of vulnerability and fragility functions derivation and,
hence, steer users towards a better quantification of such uncertainties. The document presents the result of investigation
on the sensitivity of structure response to variation in structural characteristics-related parameters’ values (i.e. in terms of
mechanical properties, geometric configuration and dimension, structural details) and in mathematical modelling (i.e.
completeness of models). The classes of structures considered are low-ductility RC buildings designed according to earlier
seismic codes, and which are in general characterized by poor quality of materials, workmanship and detailing. This
building class constitutes one of the largest portions of existing residential building stock in earthquake prone countries.

Keywords: sensitivity analysis, uncertainty capacity, analytical fragility functions

 iii

TABLE OF CONTENTS
Page
ABSTRACT..............................................................................................................................................................................ii
TABLE OF CONTENTS..........................................................................................................................................................iii
LIST OF FIGURES ................................................................................................................................................................. v
LIST OF TABLES....................................................................................................................................................................vi
1 Introduction ....................................................................................................................................................................... 7
1.1 Purpose.................................................................................................................................................................... 7
1.2 Scope..................................................................................................................................................................... 10
1.3 Organization and Contents .................................................................................................................................... 12
2 Adopted procedure of analysis........................................................................................................................................ 12
2.1 Selected structural characteristics-related parameters and range of expected values .......................................... 12
2.2 Selected analysis type ........................................................................................................................................... 13
2.3 Choice of mathematical modelling ......................................................................................................................... 14
2.3.1 Modelling RC members .............................................................................................................................. 14
Modelling element-related parameters................................................................................................................... 14
Performance criteria............................................................................................................................................... 16
2.3.2 Modelling unreinforced masonry infill panels.............................................................................................. 16
Modelling element-related parameters................................................................................................................... 16
Performance criteria............................................................................................................................................... 17
2.3.3 Definition of global threshold damage states.............................................................................................. 18
2.4 Index building......................................................................................................................................................... 21
3 Effect of structural characteristics-related parameters .................................................................................................... 23
4 Effect of mathematical modelling .................................................................................................................................... 28
4.1 The use of bare frames model to represent masonry infilled RC building.............................................................. 28
4.2 The use of two-dimensional (planar) model ........................................................................................................... 30
5 Derivation of fragility curves ............................................................................................................................................ 31
5.1 Median values S d ,dsi and dispersion  dsi .......................................................................................................... 31

Procedure 1: Distribution parameters computed from sensitivity analysis ............................................................ 32

Procedure 2: First-Order Second-Moment (FOSM) ............................................................................................... 33
5.2 Effect of structural characteristics parameters and mathematical modeling-based dispersion.............................. 35
6 Basic elements and parameters for modeling and analysis requirement........................................................................ 37
7 Final Comment................................................................................................................................................................ 38
REFERENCES ..................................................................................................................................................................... 39
ANNEX A Quality classification for index building analysis ............................................................................................ 44
iv

Parameters for quality classification/index building................................................................................................ 44

Number of subclass for index building ................................................................................................................... 44
Quality classification for index building .................................................................................................................. 44
ANNEX B Modelling of masonry infill panels .................................................................................................................. 48
 v

LIST OF FIGURES
Page
Figure 1.1 Calculations effort and uncertainty at different steps in analytical vulnerability assessment ................................8
Figure 2.1 Idealisation into fibers of reinforced concrete members......................................................................................14
Figure 2.2 Comparison of load-displacement curves obtained from fiber-based and plastic hinges structural models with
result from experimental test (Colangelo 2005) for RC Bare Frame intended to represent the ground floor of a four-
storey building and is representative of older structures designed using Italian RC non-seismic code provisions.....15
Figure 2.3 Diagonal strut model for masonry infill panel modelling. (a) Equivalent diagonal strut representation of an infill
panel; (b) Variation of the equivalent strut width as function of the axial strain; (c) Envelope curve in compression. 16
Figure 2.4 Parametric analysis for the reduction strut width (ared) parameter. .....................................................................17
Figure 2.5 Effect of envelop curve on the simulation of the capacity functions. (a) Effect of strain at maximum stress, εm;
(b) Effect of ultimate strain, εult....................................................................................................................................18
Figure 2.6 Force-displacement curve for infilled RC building and definition damage conditions at global level. (a) Force-
displacement relationships at global level; (b) Force-displacement relationships for the infill panels.........................21
Figure 2.7 Typical four-storey masonry infilled RC building of 1970’s located in a high-seismically region of Turkey.........22
Figure 3.1 Influence of concrete compressive strength on the deformation capacity of the building. ..................................23
Figure 3.2 Sensitivity of the structure response to the variation in compressive strength of concrete.................................25
Figure 3.3 Sensitivity of the structure response to the variation in tensile strength of steel.................................................25
Figure 3.4 Sensitivity of the structure response to the variation in transverse reinforcement spacing.................................25
Figure 3.5 Sensitivity of the structure response to the variation in storey height. ................................................................26
Figure 3.6 Sensitivity of the structure response to the variation in thickness of infill panels................................................26
Figure 3.7 Sensitivity of the structure response to the variation in compressive strength of infill panel...............................26
Figure 4.1 Comparison of the resulted adaptive pushover curves from infilled frame and bare frame models, considering
concrete compressive strength as variable parameter ...............................................................................................28
Figure 4.2 Comparison between the use of infilled frame and bare frame models for different structural characteristics
configuration ...............................................................................................................................................................29
Figure 4.3 Comparison of capacity curves obtained by 3-D model with those obtained by superposition of 2-D models. ..30
Figure 5.1 Comparisons of fragility curves of the structures with and without considering the contribution of masonry infill
walls............................................................................................................................................................................33
Figure 5.2 Comparisons of fragility curves of the structures using distribution parameters and FOSM method. (a) Resulted
fragility curves from set of infilled frame models; (b) Resulted fragility curves from set of bare frame models...........35
Figure 5.3 Comparisons of structural capacity (structural characteristics and mathematical modelling) uncertainty-based
fragility curves with total uncertainty-based fragility curves. (a) Considering set of infilled frame models; (b)
Considering set of bare frame models........................................................................................................................36

Figure B- 1 Comparison of capacity functions obtained using different formulae for the effective width of equivalent strut.
The RC frame is one-storey, single bay, intended to represent the ground floor of a four--storey masonry infilled
vi

concrete building and are representative of older structures designed using Italian reinforced concrete non-seismic
code provisions. Experimental data was collected from Colangelo (2005).................................................................48
Figure B- 2 Variation of reduction factor for different equations for the effective width of equivalent strut. .........................50

LIST OF TABLES
Table 1.1 Parameters characterizing building capacity and seismic response ......................................................................9
Table 1.2 Uncertainties in capacity parameters considered in the literature for the derivation of fragility curves of RC
buildings......................................................................................................................................................................11
Table 2.1 Range of expected values for the structural characteristics-related parameters associated to the building class
represented by the index building ...............................................................................................................................13
Table 2.2 Definition of limit states at structure level as per several guidelines.....................................................................20
Table 2.3 Definition of different damage conditions at global level proposed in GEM AVM(see also Figure 2.6) ...............21
Table 2.4 Classification of 4-storey RC building according to the GEM Basic Building Taxonomy......................................22
Table 3.1 Implemented models for sensitivity analysis to structural characteristics-related parameters..............................24
Table 3.2 Effect of the variation in the structural characteristics-related parameters’ values on the structure response .....27
Table 4.1 Sensitivity of structural response to the contribution of masonry infill panels.......................................................29
Table 5.1 Median value and dispersion calculated for each threshold of damage state, and used in Figure 5.1................33
Table 5.2 Comparison of median value and dispersion calculated using distribution parameters and FOSM procedure....34
Table 5.3 Comparison of structural characteristics and mathematical modelling dispersion with record-to-record
dispersion, and average value of total dispersion (as per literature)...........................................................................36
Table 6.1 Basic attributes (elements) for modelling and analysis requirement for reinforced concrete buildings ................37
Table 6.2 Basic attributes (parameters) for modelling and analysis requirement for reinforced concrete buildings.............37

Table A- 1 Expected value range and average value associated to structural characteristics-related parameters quality
classification and index buildings analysis..................................................................................................................46

Table B- 1 Formulae of equivalent masonry strut's effective width as per literature ............................................................49
Table B- 2 Range of values of equivalent strut width computed from different relationships for the infill panels. ................49

.
 7

1 Introduction

1.1 Purpose

The process of analytical fragility assessment is essentially based on two components, namely (see Figure 1.1): the
ground motion intensity-to-structural response functions, P(IM|SR), and the structural response-to-damage state functions,
P(SR|DS). These functions are the products of two independent procedures; namely, the Structural Analysis and the
Damage Analysis. At each of these steps, a certain level of uncertainty should be expected and has to be taken into
account by users in the estimation of seismic risk. The level of the uncertainty will depend upon the simplification and
assumption which researchers and engineers do implement in aim to reduce data gathering and calculation efforts.

With regard to the structural analysis, there are a number of uncertainties involved in the estimation of the performance of
the building for given levels of intensity. These uncertainties concerned both the capacity modeling of the examined
building and the demand modelling. The uncertainty in capacity modeling is directly affected by the choice of structural
characteristics-related parameters and mathematical model to compute and estimate the response of a structure. The
uncertainty in the demand is introduced by the record-to-record variability, which captured the variability in the complexity
of the mechanism of the seismic source, path attenuation and site effects of the seismic event (FEMA 2003; NIBE-FEMA
2003; ATC 2011).

With regard to the damage analysis, damage thresholds modelling is in general affected by the conservative character in
defining the different global limit states, the choice of the implemented damage model and its consistency with the type of
analysis, the used damage indicator to represent a structure’s damage states and the correlation with the chosen intensity
measure (FEMA 1999; FEMA 1999; NIBE-FEMA 2003).

Regarding the fragility analysis, this is in general related to the choice of fitting procedure and sampling methods for the
construction of fragility curves, taking into account the identified and quantified uncertainties from structural analysis and
damage analysis (FEMA 1999; Wen et al. 2004; Pagnini et al. 2008; ATC 2011).

This present document is devoted to examine the parameters that are associated to the building capacity modeling, only
(Figure 1.1). The purpose is to fill the gap regarding the availability of details on the effect and the choice of these building
capacity-related parameters with regard to the expected uncertainties that might have on the results of vulnerability and
fragility functions derivation and, hence, help GEM end-user to decide with more efficient way for a better quantification of
the uncertainties. Table 1.1 shows the different parameters characterizing building capacity and seismic response. This
classification has been made based on the recent extensive literature review that has been conducted within the
framework of the development of the “GEM Guide for Selecting of Existing Analytical Fragility Curves and Compilation of
the Database” (D’Ayala and Meslem 2012).
In the definition of structure capacity, structural characteristics and mathematical modeling constitute fundamental
attributes in assigning a particular building to a specific class, in determining the representativeness of a particular index
8

building or class of buildings for a given exposed building stock, and ultimately in estimating their seismic performance in
terms of fragility and vulnerability functions.

Structural Analysis
P(SR|IM)
• Uncertainty in structural characteristics modelling
Capacity
modelling

Demand
Modelling • Uncertainty in mathematical modelling

Damage Analysis
P(DS|SR)
• Record-to-record variability
Damage
Thresholds
Modelling

• Uncertainty in definition of damage thresholds

Fragility Analysis
P(DS|IM)
Fragility
Curves
Generation • Sampling and fragility curve fitting model

Monte Carlo Full Reduced

Sampling Partitioning Partitioning

Figure 1.1 Calculations effort and uncertainty at different steps in analytical vulnerability assessment

With reference to the structural characteristics attributes, which is introduced in terms of mechanical properties, structural
details, geometric configuration and dimensions, it has been widely observed that default values, provided in existing
guidelines/codes (e.g. ATC 1996; FEMA 1999; ASCE 2000) and implemented in commonly used structural programs (e.g.
CSI 2009), are assigned to represent the associated parameters, e.g. a default value of concrete strength, or steel
strength, or an estimate of transverse reinforcement spacing, or hinges capacities...etc. (e.g. Inel and Ozmen 2006,
Salvador et al. 2008, Khan & Naqvi 2012). Usually, this is due either to lack of information, especially, for the case of older
 9

structures, where design documents are generally not available, or to expedience. Moreover, it has been also observed
that fragility curves of buildings located anywhere in the world have been generated using, for instance, HAZUS capacity
curves (FEMA 1999) derived for buildings in the US (Lourdes et al. 2007; Vacareanu et al. 2007). This is particularly
common when studies are conducted for large portions of the building stock and resources for direct survey and data
acquisition are modest. Typically, differences in construction techniques and detailing between different countries are
significant, even when buildings are nominally designed to the same code clauses.
Regarding the mathematical modeling, it has been found from many studies (Kircil and Polat 2006, Howary and Mehanny
2011) that for the case of masonry infilled reinforced concrete (RC) buildings, fragility functions were generated from
analysis of bare frames models, ignoring the contribution of infill panels in the seismic response; hence, reducing the
ability of these models to capture the real behaviour of the structures. The literature review has also shown that authors
attempt to simulate buildings as two-dimensional modeling instead of three-dimensional modeling; omit shear failure in
columns or beams...etc.

Table 1.1 Parameters characterizing building capacity and seismic response

Structural Characteristics
Mechanical Characteristics Compressive strength of concrete (fc)
Tensile strength of steel reinforcement bar (fy)
Compressive strength of masonry infill (fw)

Dimension Characteristics Total height (Lz) / Storey height (az)

Number of storeys (nz)
Plan dimensions (Lx, Ly) - Bay length

Structural Detailing Tie spacing at the column (Sc)

Reinforcement ratio at the column (ρ)
Hardening ratio of steel (bh)

Geometric Configuration Perimeter Frame Building - Space Frame Building (PFB/SFB)

Rigid Roof / Deformable Roof (RR/DR)
Column orientation (OR)

Mathematical Modelling
Numerical Modelling Bare frame/infilled frame for masonry infilled RC buildings.
3-D / 2-D element-by-element
2-D storey-by-storey
1-D global model

Performance Criteria Yield and ultimate capacities

Out-of-plane failure mechanisms in masonry buildings
Shear failure mechanisms

As a consequence, these different choices of assumptions and simplifications may highly decrease the reliability and
accuracy of the obtained results introducing important aleatoric and epistemic uncertainties in the fragility function
construction process (Dolsek 2012). Although these uncertainties have been accounted in some previous seismic
vulnerability studies (see Table 1.2), apart from Liel and Deierlein (2008), very few of these studies have consistently
analyzed or examined the effect of the variability of several structural characteristics or of the simplified modelling
assumptions on the generated fragility curves, with the scope of estimating the level of uncertainties that should be taken
into account. Moreover, the uncertainties were accounted for in different ways based on the nature of considered
parameters and the needs of each study. Table 1.2 summarizes some examples on how these different capacity-related
uncertainties have been accounted for as per several references from literature. In most of these previous studies, the
uncertainty has not been fully accounted for by considering all the attributes (i.e. structural characteristics-related
10

parameters, and mathematical modeling) in accordance to what has been recommended in the existing guidelines, such
as NIBE-FEMA (2003), HAZUS-MH (FEMA 2003), ATC-58 (ATC 2011).

In most of the references from literature, the aleatoric uncertainties associated to the structural characteristics-related
parameters are accounted for by considering the probabilistic variability in their values (Rossetto, and Elnashai 2004;
Bakhshi and Karimi 2006; Iervolino et al. 2007; Ay and Erberik 2008; Polese 2008; Rajeev and Tesfamariam 2011; Uma et
al. 2011; Jiang et al. 2012; Dolsek 2012). In some others vulnerability studies the effect of dispersion in structural
characteristics-related parameters are accounted for by survey of a large number of existing buildings and definition of a
median and standard deviation of the sample of buildings, after calculation of the capacity and damage threshold for each
building in the sample (D’Ayala et al. 1997; Lang and Bachmann 2004; D’Ayala 2005; D’Ayala, and Paganoni 2010). With
regard to the mathematical modeling, the associated epistemic uncertainties are in general underlined with regards to the
parameters of the hysteric models, modeling shear failure mechanisms in concrete elements and the consideration of
masonry infill walls contribution in RC buildings; modeling out-of-plane failure mechanisms in masonry buildings. With
respect to the contribution of masonry infill panels to the response of RC structures, some studies have highlighted the
different performance of the building due to the irregularities and distribution of infill panels (e.g. Dymiotis et al. 1999; Ellul
2006, Dolsek and Fajfar 2008, D’Ayala et al 2009, Mulgund and Kulkarni 2011), however, only few studies have shown
their direct effect on fragility curves (e.g. Sattar and Liel 2010, Haldar et al. 2012).

1.2 Scope

the main scope of this present document is to analyze the effect of the variation in structural characteristics-related
parameters’ values, and of the mathematical modelling choices, with particular reference to inclusion of infill and model
dimension, on the structure response and generated fragility curves. The objective is to provide guidance for the GEM
end-user in estimating the level of uncertainties that should be taken into account, with respect to the different choices of
simplifications and assumptions modelling that can be made in evaluating seismic performance and generating fragility
curves.

Accordingly, the sensitivity study presented in this document has been conducted for the class of structures that are
defined as low-ductility RC buildings designed according to earlier seismic codes, and which are in general characterized
by poor quality of materials, workmanship and detailing. The main reason for the choice of this class of buildings is that it
constitutes one of the largest portions of existing residential building stock in earthquake prone countries.

To best identify the expected mean and range for the various parameter analysed a real frame in Turkey, is the reference
prototype, however the methodology and results obtained are applicable to other typologies and locations, once the basic
data is available. The sensitivity study is based on nonlinear static adaptive pushover analysis selecting forty-two (42) 3D
models. Indeed, when using this type of analysis the variation in the structural stiffness at different deformation levels, and
consequently the system degradation can be better accounted for. The observations of the influence of variability of the
selected parameters are conducted in terms of deformation capacity, considering different damage thresholds. The effect
of mathematical model is investigated by performing a comparative analysis of fragility curves derived with and without
considering the contribution of masonry infill panels, and between 3-dimension and 2-dimension models.

Table 1.2 Uncertainties in capacity parameters considered in the literature for the derivation of fragility curves of RC buildings
Compressive strength of concrete (fc)
Reference
Jiang et al. (2012)
Rajeev & Tesfamariam (2011)
Howary and Mehanny (2011)
Verbicaro et al. (2009)
Ozer and Erberik (2008)
Kappos et al. (2006)
Rossetto and Elnashai (2005)
Inel and Ozmen (2006)
Polese et al. (2008)
Iervolino et al. (2007)
Liel and Deierlein (2008)
NIBE-FEMA (2003)
ATC-58 (ATC 2011)
11
Uncertainty in Capacity (βC)
Uncertainty in Structural Characteristics Uncertainty in Mathematical Modelling
Dispersion in
Dispersion in Dispersion in
Mechanical Dispersion in Dimension Dispersion in Numerical Dispersion in Performance
Structural Detailing Geometric
Characteristics Characteristics (βDM) Modelling (βNM) Criteria (βPC)
(βST) Configuration (βGE)
(βME)
Compressive strength of masonry infill (fw)
Perimeter Frame Building - Space Frame
Rigid Roof / Deformable Roof (RR/DR)
Reinforcement ratio at the column (ρ)
Tensile strength of reinforcement (fy)
Infilled frame/Bare frame system
Tie spacing at the column (Sc)
Hardening ratio of steel (bh)
Yielding/Ultimate capacities
3D/2D element-by-element
Shear failure mechanisms
Plan dimensions (Lx, Ly)
Column orientation (OR)
Number of storeys (nz)
2-D storey-by-storey
Building (PFB/SFB)
Storey height (az)
1-D global model
Total height (Lz)
√ √
√ √ √ √
√
√ √
√ √
uncertainty in capacity (√)
√ √ √
√
√ √ √ √
√ √ √ √
√ √
Uncertainty associated with the capacity curve (√)
Uncertainty in modelling
Uncertainty of the in situ constructed building with the construction documents, material properties (√) degradation, mechanisms and
system interactions (√)
12

1.3 Organization and Contents

The present documented is divided in five chapters besides the present. The second one explains the procedure that has
been adopted to conduct the sensitivity analysis, regarding the selected structural characteristics-related parameters to be
examined, choice of analysis type, choice of mathematical modelling, and choice of index building and quality. Chapter 3
present and discuss results of different sensitivity of structure responses to the structural characteristics-related
parameters. These results are presented in terms of deformation capacities for different damage conditions. Chapter 4
present and discuss results of sensitivity of structure response to the numerical modelling completeness related to the
consideration or non-consideration of masonry infill panels contribution, and the use of planer model (i.e. 2-D models).
Chapter 5 discuss the influence of capacity-related parameters uncertainty on the derived analytical fragility curves and
compares the different options in considering the dispersion as per several references from literature. Chapter 6 provides
the basic attributes (in terms of elements parameters) for modelling and analysis requirement, And, Chapter 7 provides
general comments regarding the outcomes of this present document.

In addition, the document provides ANNEXES. ANNEX A with regards to the quality classification for index building
analysis, and which provide: (a) the procedure that has been followed to identify the basic parameters for the quality
classification of building, and hence, which will be considered for the sensitivity analysis; (b) the choice of the most
expected values range and their means. ANNEX B with regards to the modelling of masonry infill panel, and which
provide: (a) result of comparative analysis of several existing formula from literature for the modelling of infill panels, and to
justify the choice of the selected one to be implemented in the rest of study within framework of this document; (b) the
basic parameters in modelling of infill panels, and which have identified based on their result of sensitivity on the structure
response.

2 Adopted procedure of analysis

In the followings, details are provided regarding the different steps, and assumed assumption that have been adopted in
the implementation of sensitivity analysis; in terms of the parameters that were considered to examine their effects on the
structure response performance and choice of range of expected values; the choice of analysis type that has been
selected; the choice of mathematical modeling adopted in the different analyses (the adopted modeling techniques and
their differences, and definition of global damage states); and the selected index building.

2.1 Selected structural characteristics-related parameters and range of expected values

In the present document, the investigated structural characteristics-related parameters are those associated to mechanical
properties, geometric configuration, and structural details; and which are in general affected by the quality of workmanship;
i.e. compressive strength of concrete, yield strength of reinforcement, strength and stiffness of infill walls (in terms of
compressive strength and thickness), story height, and transverse reinforcement spacing. The choice of range of expected
values for each parameter (see Table 2.1) is based on the results of structural characteristics assessment available from
different literature sources such as direct studies(Ay 2006, Bal et al. 2008), post-earthquakes surveys (EERI 2000, EEFIT
2003, Ellul 2006), the requirement from different versions of earlier seismic codes, e.g. TS500 (TSE 1985), and values
adopted in previous similar studies on seismic vulnerability (Gulkan et al. 2002, Erol et al. 2004, Kappos 2006).
 13

The values shown in Table 2.1 have been selected as a result of a process that was followed in defining the different
ranges, and which has been reported with details in ANNEX A. Such value ranges represent the most feasible range of
expected values characterizing the low-ductility RC buildings class, typically designed according to earlier seismic codes
and, in general, characterized by poor quality of materials, workmanship and detailing.

Table 2.1 Range of expected values for the structural characteristics-related parameters associated to the building class represented
by the index building
Range of most expected values for Poor Quality Class of Buildings
Parameters Central value
Lower Bound Upper Bound
Compressive strength of concrete (fc) 14 MPa 20 MPa 17 MPa
Tensile strength of steel (fy) 200 MPa 320 MPa 260 MPa
Transverse reinforcement spacing (S) 150 mm 250 mm 200 mm
floor-to-floor Story height (h) 2.5 m 3.2 m 2.8 m
Thickness of infill walls (tw) 13 cm 19 cm 16 cm
Compressive strength of infill walls (fw) 1.0 MPa 1.5 MPa 1.25 MPa

2.2 Selected analysis type

The accuracy of any selected procedure for structural response analysis might depend on the type of the selected analysis
approach, and the adopted mathematical model that must be consistent with the type of analysis implemented. In present
work, the sensitivity analyses have been based on the implementation of Static Adaptive Pushover Analysis (SAPA), which
is an extension and advanced from conventional pushover analysis. In fact, when using SAPA method the lateral load
distribution is not kept constant but rather continuously updated during the analysis, according to the modal shapes and
participation factors derived by eigenvalue analysis carried out at each analysis step. Due to its ability to update the lateral
load patterns according to the constantly changing modal properties of the system, it overcomes the intrinsic weaknesses
of fixed-pattern displacement pushover and provides a more accurate performance-oriented tool for structural assessment,
providing better response estimates than existing conventional methods, especially in cases where strength or stiffness
irregularities exist in the structure (Papanikolaou and Elnashai 2005; Bento et al. 2008).

Nevertheless, adaptive pushover analysis may not be exempt from limitations which can be summarized as follows:
excessive force concentration at the locations of the structure where the damage first develops; the combination of the
modal contributions; the updating procedure of the lateral load vector. This being directly related to the frequency and
modal shapes computed at each step, the procedure defeats its purpose in presence of extensive nonlinear deformation
or brittle failure as the modal shapes include imaginary components and hence the vector is not updated. (Papanikolaou et
al. 2006).

It is worth to mention that in literature one can find other more sophisticated methods which in general are based on
nonlinear dynamic analysis, such as Incremental Dynamic Analysis (IDA) which has been recommended in recent
guidelines, e.g. ATC-58 (ATC 2011). These types of analysis are more complex and time consuming, requiring more input
data. However, for the purpose of the present study, i.e. sensitivity analysis, the use of adaptive pushover analysis is quite
sufficient to ensure the accuracy and reliability of the outcomes of this document. It is useful to recall that adaptive
pushover analysis has been widely used recently in many nonlinear studies, for masonry buildings (e.g. Lourenco et al.
2011), steel buildings (e.g. Shakeri et al. 2012), and RC buildings (e.g. Chaulagain et al. 2013). Recent work by Abbasnia
et al. 2013, has shown that the adaptive pushover analysis method can capture the results of IDA analysis with a
14

reasonable accuracy (e.g. high ability to reproduce the capacity curve obtained with IDA and reproduce IDA envelops,
accuracy in estimation of interstorey drift).

2.3 Choice of mathematical modelling

The reliability of mathematical models relies on the rigorous and logical representation of all part of the structures with
comparable level of complexity so that their influence on the behavior and seismic vulnerability is adequately accounted
for. An account of the choices made in each step of the mathematical modelling is provided in the following.

2.3.1 Modelling RC members

Modelling element-related parameters

A reinforced concrete member is composed of three types of materials: unconfined concrete (corresponding to the cover),
confined concrete (corresponding to the core concrete) and reinforcing steel. All reinforced concrete components are
detailed with transverse steel which provide both shear resistance and confining action for the core concrete. The
response of RC components and consequently the frame system is a function of the behavior of the confined core and the
longitudinal steel. In the case of non-ductile sections, the response of the core will be only marginally different from the
response of the cover concrete.

In the framework of this present study, fiber-based structural modeling was adopted to model reinforced concrete
members (see Figure 2.1). Indeed, this numerical technique of modeling allows characterizing in higher detail, the
nonlinearity distribution in RC elements by modelling separately the different behaviour of the materials constituting the RC
cross-section (.i.e. cover and core concrete and steel) and, hence, to capture more accurately response effects on such
elements. Fibre-based modelling models a structural element by dividing it into a number of two-end frame elements, and
by linking each boundary to a discrete cross-section with a grid of fibres. The material stress-strain response in each fibre
is integrated to get stress-resultant forces and rigidity terms, and from these, forces and rigidities over the length are
obtained through finite element interpolation functions which must satisfy equilibrium and compatibility conditions. Figure
2.1 shows an example of idealization into fibers of reinforced concrete members using SeismoStruct (SeismoSoft 2013).
The cover concrete is modelled using unconfined properties while the core concrete is modelled with properties depending
on confinement models.

Figure 2.1 Idealisation into fibers of reinforced concrete members

 15

In fact, there are several advantages which justify the use of fiber-based structural modelling:
- Fiber-based modelling may easily take into account the case of complex cross-sections. Indeed, a fiber cross-
section can have any general geometric configuration formed by sub-regions of simpler shapes; geometric
properties are calculated through the numerical integration.
- Since each fiber is associated to a given uniaxial stress-strain material response, higher accuracy and more
realistic behavior effects can be captured in a fiber-based model.
- This technique has also the ability to take into account the case of complex stress-strain behaviour. Since each
fiber can have any stress-strain response, this technique allows modelling nonlinear behavior in steel members,
reinforced concrete members (unconfined and confined concrete), and composite members as shown in Figure
2.1.
- Since the length of the fiber is not constrained, the cross-section defined at each of the two ends can be different,
and therefore, the response can be roughly estimated. Precision can be increased with more integration points.

Figure 2.2 Comparison of load-displacement curves obtained from fiber-based and plastic hinges structural models with result from
experimental test (Colangelo 2005) for RC Bare Frame intended to represent the ground floor of a four-storey building and is
representative of older structures designed using Italian RC non-seismic code provisions.

It is worth to mention that users might employ, as an alternative strategy to model reinforced concrete members, the well-
known lumped plastic-hinge structural modeling of which many application to derivation of fragility curves can be found in
literature. The characterization of these lumped plastic hinges requires a moment-curvature diagram to be defined, which
can be obtained from the monotonic loading of the cross-section, and an assumed plastic hinge length. The simplifications
and hence limitations of applicability of this method is recognised by sevral authors (e.g. Charney and Bertero 1982,
Bertero et al. 1984, Monteiro et al. 2008)

By way of example, considering as reference a simple experimental portal frame with well-defined hinge formation and
failure mechanism tested by Colangelo (2005) as shown in Figure 2.2, the rresults obtained with the fiber-based model
shows a good agreement with experimental data, regarding the ability to predict the gradual transition from initiation of the
plasticity (at 0.3% drift) to the complete formationof the hinge (estimated at 1% drift) whereas for the lumped-plastic hinge
model the transition from linear to plastic occurs over a much narrower range of drift and for higher lateral capacity
(estimated at 0.5% drift). Hence, use of such model would introduce epistemic uncertainty in the computation of fragility
curves for states of damage prior to collapse. The accuracy would be further reduced in the case of elements with complex
cross-sections and/or with complex strength, or specific geometry, such as “captive” columns.
16

Performance criteria
Within the context of a fibre-based modelling approach for the reinforced concrete frames, the different performance
checks are carried out for each integration section of the selected member. Material strains do usually constitute the best
parameter for identification of the performance state of a given structure, especially for the case of structures with different
configurations and ductility levels. Two limit states are identified at element level:
 Yielding of element (limit state of serviceability) corresponding to the yielding of the steel in tension.
 Crushing of element corresponding to the ultimate concrete compressive strain, given by (Priestley et al. 1996):
1.4  s f yh  su
 cu  0.004  (2.1)
f cc

It is worth to mention that, alternatively, the limit of chord rotation corresponding to the condition of yield rotation and
ultimate rotation, may also be used to model the performance criteria at element level (ASCE 2000, CEN 2004).

2.3.2 Modelling unreinforced masonry infill panels

In general practice, the infill walls are commonly made of masonry bricks or blocks, varying in specific weight, strength and
brittleness depending on age and quality of construction. In the literature, many models of infill panel have been proposed
in an attempt to improve the simulation of the real behavior of infilled frames. Although there is a robust body of work on
developing mezo-modelling for the numerical simulation of infill panels by 2-dimensional finite element (see Ellul 2006,
D’Ayala et al 2009, Ellul and D’Ayala 2012 for thorough literature reference and alternative modeling strategies), currently
the diagonal strut model (see Figure 2.3) is still the most frequently used by researchers as a simplified modeling
approach for bulk analysis, and has been adopted in many documents and new guidelines, such as, CSA-S304.1 (CSA
2004), ASCE/SEI 41-06 (ASCE 2006), NZSEE (2006), MSJC (2010)…etc.

(a) (b) (c)

Figure 2.3 Diagonal strut model for masonry infill panel modelling. (a) Equivalent diagonal strut representation of an infill panel; (b)
Variation of the equivalent strut width as function of the axial strain; (c) Envelope curve in compression.

Modelling element-related parameters

The equivalent strut width (see Error! Reference source not found.a) is the most investigated parameter to assess the
stiffness and strength of an infill panel. As per literature, different formulae have been proposed by several researchers
(Holmes 1963; Liauw and Kwan 1984, Paulay and Priestley 1992). A comparative analysis to show variance associated
with this models is provided in ANNEX B. For the GEM study the model adopted is the one based on the early work of
Mainstone and Weeks (1970) and Mainstone (1971), following the recommendation given by ASCE/SEI 41-06 (ASCE
2006) and by several other provisions and guidelines such as, FEMA-356 (ASCE 2000):
 17

1
 0.4  E t sin 2  4
a  0.175  h  where,    
m inf (2.1)
r ,
 I col  inf I  
 c col inf 
4E I h

 I is coefficient used to determine equivalent width of infill strut; hcol is column height between centrelines of beam; hinf is
height of infill panel; Ec is expected modulus of elasticity of frame material; Em is expected modulus of elasticity of frame

material (taken as Em  550 f m ; where f m is compressive strength of infill material); I col is moment of inertia of column;
rinf is diagonal length of infill panel; t inf is thickness of infill panel and equivalent strut; and  is angle whose tangent is the infill
height-to-length aspect ratio.

Performance criteria
The implementation of diagonal strut model requires the definition of the followings: the reduced strut width “ared” (Figure
2.3.b); and the envelope curve in compression in terms of strain at maximum stress and ultimate strain (Figure 2.3.c).
Within the framework of developing this present document, a parametric analysis was conducted for these parameters,
which show very complex inter-relationship and in general are best calibrated directly from experiment (see Figure 2.4 and
Figure 2.5). The parametric analysis was conducted for the case of one-storey, single bay, RC frame intended to
represent the ground floor of a four--storey masonry infilled concrete building and are representative of older structures
designed using Italian reinforced concrete non-seismic code provisions. Experimental data was collected from Colangelo
(2005).
 Estimation of reduced strut width ared

When the elastic limit of the infill panel is exceeded due to the cracking, the contact length between the frame and the infill
decreases as the lateral and consequently the axial displacement increases, affecting thus the area of equivalent strut. To
take into account this fact the width of the equivalent strut must be reduced. In this work, the strut area is assumed to vary
linearly as function of the axial strain as shown in Figure 2.3.b. This variation takes place between two strains: strut area
reduction strain (ε1) and residual strut area strain (ε2).
According to Al-Chaar (2002), a reduction factor for infill panel damage can takes values of 0.7 and 0.4 for moderate and
severe damage, respectively. From the result of parametric analysis, it has been observed that the strut width reduction
parameter has a significant influence on the peak load leading to differences of up to 38% from the minimum value.
However, this factor does not seem to have a significant effect on the ultimate drift at failure of the infill panel (see Figure
2.4).

Figure 2.4 Parametric analysis for the reduction strut width (ared) parameter.
18

 Estimation of envelop curve in compression

With regards to the envelop curve, it is well known that in general failure of infill panel occurs at small lateral displacement
before the frame reaches its strength. However, the system frame-infill panel is able to resist increasing lateral loads, by
acting as confinement to the cracked panel. This effect leads to a less brittle behavior of the infill and smoother
redistribution between the two components and loss of capacity of the system. According to Crisafulli (1997), the
descending branch of the strength envelope can be described by a parabolic curve as it is shown in Figure 2.3.c. Crisafulli
(1997) also assumed that the expression of strain-stress proposed by Sargin et al. (1971) originally for concrete can
approximately represent the envelope curve for masonry.

The parametric analysis was conducted for the two parameters defining an envelope cuve, i.e. Strain at Maximum Stress
(εm) and Ultimate Strain (εult) considering several ratios εult/εm as shown in Figure 2.5. In addition, εm which should be
calibrated through the consideration of experimental data, may vary from 0.001 to 0.005, as reported in SeismoStruct
(2013). The result of parametric analysis has shown that both parameters, εm and εult, do not seem to have an effect on the
peak load capacity, while it significantly influences the post-peak branch of the capacity curve, hence, influencing the
uncertainty in evaluation of post peak performance points.

For the present study, the assumption that has been considered is that a complete collapse occurs just after appearance
of cracking as it has been widely observed from experiment. The envelope curve model used is εult = 5.5•εm (εm = 0.0012).

Figure 2.5 Effect of envelop curve on the simulation of the capacity functions. (a) Effect of strain at maximum stress, εm; (b) Effect of
ultimate strain, εult.

2.3.3 Definition of global threshold damage states

Regarding the evaluation of different limit states at the level of the structure, there is a lack of clear guidance in the
literature, beyond some qualitative description of observed damage. The existing relations and expressions for the
calculation of capacities (i.e. performance criteria) are in general mostly defined at the element (ass described in the
previous subsection) rather than at the global level. At the global level, the damage thresholds are defined conservatively
on the basis of a minimum number of elements having reached or past a specific damage threshold.
 19

Several definitions have been implemented in guidelines and codes for the estimation of the global damage states,
through the observation of the progression of local damage at elements level (Table 2.2). As per most of documents, the
damage are described by three main levels, as shown in Table 2.2.

For instance, ASCE 41-06 (ASCE 2007) describes damage levels by:
- Immediate Occupancy (IO): level for which the building is expected to sustain minimal or no damage to their
structural elements and only minor damage to their non-structural components;
- Life Safety (LS): level for which the building may experience extensive damage to structural and nonstructural
components; and
- Collapse Prevention (CP): level for which the building is may experience a significant hazard to life safety
resulting from failure of non-structural components.
Similarly, the damage level in a building is defined in Eurocode-8 (CEN 2005) by three limit states:
- Damage Limitation (DL): Building meeting this level is considered as slightly damaged;
- Significant Damage (SD): Building meeting this level is considered as significantly damaged; and
- Near Collapse (NC): Building meeting this level is considered as heavily damage.

From an analytical point of view the challenge is to correlate the qualitative description of damage provided in the above
definitions with specific performance to be identified numerically in terms of strain, drift or attainment of strength thresholds
on a given number of elements, as per indication of the numerical models. For instance, Dolsek and Fajfar (2008) propose
the following correlation: Damage Limitation (DL), Significant Damage (SD), and Near Collapse (NC), as per the
Eurocode-8 (CEN 2005) definitions, where DL is attained for the value of lateral drift causing the last infill in a storey
starting to degrade. For the case of bare frames this threshold is attained at the yield displacement of the idealized
pushover curve; SD limit state is attained when the rotation at one hinge of any column exceeds 75% of the ultimate
rotation; NC limit state is attained when the rotation at one hinge of any column exceeds 100% of the ultimate rotation.
This implies that the structure will become unstable if one of the columns at one storey fails.

As mentioned earlier, the global damage definition provided in literature (Table 2.2) are very conservative and would
produce significant bias in the fragility curve derivation in relation to empirical fragility curves, derived for instance on the
basis of the EMS ’98 damage state description (Grunthal 1998), for which 5 grades of damage are used: Grade 1:
negligible to slight damage (no structural damage, slight non-structural damage); Grade 2: moderate damage (slight
structural damage, moderate non-structural damage); Grade 3: substantial to heavy damage (moderate structural damage,
heavy non-structural damage); Grade 4: very heavy damage (heavy structural damage, very heavy non-structural
damage); and Grade 5: destruction (very heavy structural damage)

For this reason in the framework of GEM Analytical Vulnerability Method (GEM-AVM), while four global limit states have
been considered in a manner similar to the previous study, Slight Damage, Moderate Damage, Near Collapse, and
Collapse, these have been associated with a more distributed progression of local damage through several structural and
non-structural elements. The choice has been made based on the existing definitions presented above as per different
existing guidelines, and also on field observation and earthquake reconnaissance data that have been reported in many
documents, such EERI report (2000), EEFIT report (2003). For the purpose of the present study of sensitivity, only Slight
Damage, Moderate Damage and Near Collapse are considered, as presented in Table 2.3 (see also Figure 2.6).
20
The choice of relating a global damage thresholds or state to a higher proportion of damaged elements within a structure is
particularly significant for low engineered structures which might present RC columns of variable dimensions and oriented
according to architectural rather than structural criteria, hence showing diverse interstorey drift related performance
Table 2.2 Definition of limit states at structure level as per several guidelines.
ASCE/SEI 41-06 (ASCE 2007); ATC-58-2 (ATC 2003), FEMA-356 (ASCE 2000)
Performance Level Immediate Occupancy (IO)
Minor hairline cracking; limited
yielding possible at a few
Primary
locations; no crushing (strains
Concrete Frames
below 0.003)
Minor spalling in a few places in
ductile columns and beams;
Secondary flexural cracking in beams and
columns; shear cracking in
joints < 1/16" width.
Unreinforced Masonry
Minor (<1/8" width) cracking of
masonry infills and veneers;
Primary
Infill Walls
minor spalling in veneers at a
few corner openings.
Secondary Same as primary.
Performance Level Operational
ATC-58-2 (ATC 2003), Vision 2000
Minor hairline cracking (0.02”);
limited yielding possible at a
Primary RC Elements
(SEAOC 1995)
few locations; no crushing
(strains below 0.003)
Secondary RC
Elements Same as primary
Performance Level Damage Limitation (DL)
Dolsek and Fajfar (2008)
Eurocode-8 (CEN 2004),
For the case of infilled frames:
limit state is attained at the
deformation when the last infill
in a storey starts to degrade.
Observed damage
For the case of bare frames:
limit state is attained at the yield
displacement of the idealized
pushover curve.
Life Safety (LS)
Extensive damage to beams;
spalling of cover and shear
cracking (<1/8" width) for ductile
columns; minor spalling in
nonductile columns; joint cracks
< 1/8" wide.
Extensive cracking and hinge
formation in ductile elements;
limited cracking and/or splice
failure in some nonductile
columns; severe damage in short
columns.
Extensive cracking and some
crushing but wall remains in
place; no falling units. Extensive
crushing and spalling of veneers
at corners of openings.
Same as primary.
Life Safe
Extensive damage to beams;
spalling of cover and shear
cracking (<1/8”) for ductile
columns; minor spalling in
nonductile columns; joints
cracked < 1/8” width.
Extensive cracking and hinge
formation in ductile elements;
limited cracking and/or splice
failure in some nonductile
columns; severe damage in short
columns
Significant Damage (SD)
The most critical column controls
the state of the structure: the limit
state is attained when the
rotation at one hinge of any
column exceeds 75% of the
ultimate rotation
Collapse Prevention (CP)
Extensive cracking and hinge
formation in ductile elements;
limited cracking and/or splice
failure in some nonductile
columns; severe damage in
short columns.
Extensive spalling in columns
(limited shortening) and beams;
severe joint damage; some
reinforcing buckled.
Extensive cracking and
crushing; portions of face course
shed.
Extensive crushing and
shattering; some walls dislodge.
Near Collapse
Extensive cracking and hinge
formation in ductile elements;
limited cracking and/or splice
failure in some nonductile
columns; severe damage in
short columns.
Extensive spalling in columns
(possible shortening) and
beams; severe joint damage;
some reinforcing buckled
Near Collapse (NC)
The most critical column
controls the state of the
structure: the limit state is
attained when the rotation at
one hinge of any column
exceeds 100% of the ultimate
rotation.
 21

Table 2.3 Definition of different damage conditions at global level proposed in GEM AVM(see also Figure 2.6)

Damage State Description

Slight Damage RC members: first yielding at column, no crushing.

Masonry infills: appearance of cracking in masonry infills

Moderate Damage RC members: cover spalling at several locations for columns and beams
Masonry infills: crushing/failing of infills at first story; extensive cracking at other stories

Near Collapse RC members: extensive crushing in some columns/beams

Masonry infills: extensive crushing/failing of infills

Figure 2.6 Force-displacement curve for infilled RC building and definition damage conditions at global level. (a) Force-displacement
relationships at global level; (b) Force-displacement relationships for the infill panels.

2.4 Index building

Low ductility reinforced concrete frame structures constitute one of the largest portions of existing residential building stock
in several earthquake prone countries and in the rest of the world. For the sensitivity analysis, the class of structure
considered is a typical four-storey RC building, built according to the first generation of seismic codes, in the 1970’s and
located in a high-seismically region of Turkey; see Figure 2.7 and Table 2.4. The building is representative of residential
buildings stock designed according to the earlier seismic codes, and which are in general characterized by low strength
concrete, mild steel smooth rebars, relatively high strength infill, and general poor construction details and quality. The
building has four bays with the raster of 4m in the X direction, and four bays with the raster of 3m in the Y direction. The
slab has the thickness of 15 cm. The amount and arrangement of longitudinal reinforcement in columns and beams are
shown in Figure 2.7. In addition of the self-weight of the structure the 2 kN/m2 of permanent load was assumed in order to
represent floor finishing and partitions, and 30% of participating live load (live load = 2 kN/m2) was also adopted.

The analyses were performed by SeismoStruct (Seismosoft 2013), which is a fiber-based finite elements software
framework for simulation applications in earthquake engineering using finite element method. The infill panel is
represented by means of two diagonal struts placed between the beam-column joints able to resist load in compression. In
general, 40% to 60% of masonry infill panels present in the infilled RC building are structurally effective as the remaining
22

portion of the masonry infills are meant for functional purpose such as doors and windows openings (Pauley and Priestley,
1992). In this present document, the buildings were modeled using 50 % masonry infills (external panels), as shown in
Figure 2.7.

3D view of the building Plan view of the building

Reinforcement of columns in storey-1 and 2 Reinforcement of columns in storey-3 and 4

Figure 2.7 Typical four-storey masonry infilled RC building of 1970’s located in a high-seismically region of Turkey.

Table 2.4 Classification of 4-storey RC building according to the GEM Basic Building Taxonomy
GEM Taxonomy 4-Storey RC Building
#
Attribute Attribute Levels Level 1 Level 2
1 Material of the Lateral Load-Resisting System Material type (Level 1) CR CIP
Material technology (Level 2)
Material properties (Level 3)
2 Lateral Load-Resisting System Type of lateral load-resisting system (Level 1) LFINF DU
System ductility (Level 2)
3 Roof Roof material (Level 1) RC RC1
Roof type (Level 2)
4 Floor Floor material (Level 1) FC FC1
Floor type (Level 2)
5 Height Number of stories H:4
6 Date of Construction Date of construction YEP:1975
7 Structural Irregularity Type of irregularity (Level 1) IRN
Irregularity description (Level 2) IRH IROH
IROH IROV
8 Occupancy Building occupancy class - general (Level 1) RES RES2
Building occupancy class - detail (Level 2)
 23

3 Effect of structural characteristics-related parameters

In many situations when there is lack of information, especially, for the case of older structures, where design documents
are generally not available, researchers and engineers attempt to adopt estimated or default values to be assigned for
structural characteristics-related parameters. This section of the present document, aim to provide details for engineers on
the consequences of their different choices, and hence, take into account the expected uncertainty with respect to the
accuracy in predicting the response of the structure.

Considering what has been discussed in Section 2.1 with regard to the selected most probable values range and mean,
associated to each structural characteristics-related parameter, several models have been implemented to evaluate the
effect on the structure response, as shown in Table 3.1. The values chosen within each range for each parameter in Table
3.1 (see also Table 2.1), are considered the most expected values that might be possibility assigned by the assessor, in
case of lack of information (e.g. design documents are not available), to represent the class of low-ductility RC buildings
constructed with earlier seismic codes.

Figure 3.1 shows example of resulted force-displacement capacities, from adaptive pushover analysis, with respect to the
variation in concrete compressive strength’s values. The result clearly shows the effect of this parameter on the
deformation capacity of the structure; however, in terms of load capacity, almost no significant influence has been
observed.

As result of sensitivity analyses, Figure 3.2 to Figure 3.7 show the influence of the variation in values for all structural
characteristics-related parameters, that have been considered in this study, on the structural response, for different
damage condition in terms of roof drift. Table 3.2 summarizes the level of sensitivity of the response to the change for
each parameter in terms of Coefficient of Variation (CV), defined as ratio of standard deviation to mean value, and the
percentage difference (Diff) in deformation capacity for different damage condition.

The result of sensitivity analysis has shown that structural characteristics-related parameters are found to have a
significant effect on the structural response, for different damage condition. Indeed, at the highest level of damage a
remarkable variation (CV reaches a value up to 38%) in terms of deformation capacity (roof drift) has been observed even
for a modest variation in compressive strength of concrete (CV = 12.7%), as shown in Figure 3.2 and Table 3.2; however,
no significant difference in structural response has been found at the lowest level of damage, i.e. Slight Damage.

Figure 3.1 Influence of concrete compressive strength on the deformation capacity of the building.
24

Table 3.1 Implemented models for sensitivity analysis to structural characteristics-related parameters

Concrete Compressive Steel Yield Transverse Reinforcement Story Thickness of Infill Compressive Strength Number of Number of Bays
Variability Dimension
Strength [MPa] Strength [MPa] Spacing [mm] height [m] Panel [cm] of Infill [MPa] Storyes X Direction Y Direction
14 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
Concrete Compressive

15 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

16 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
Strength

17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

18 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
19 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
20 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 200 200 2.8 16 1.25 4 4x4m 4x3m 3-D

Steel Yield Strength

17 220 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 240 200 2.8 16 1.25 4 4x4m 4x3m 3-D
17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
17 280 200 2.8 16 1.25 4 4x4m 4x3m 3-D
17 300 200 2.8 16 1.25 4 4x4m 4x3m 3-D
17 320 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 260 150 2.8 16 1.25 4 4x4m 4x3m 3-D

Reinforcement

17 260 175 2.8 16 1.25 4 4x4m 4x3m 3-D

Transverse

Spacing

17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 260 225 2.8 16 1.25 4 4x4m 4x3m 3-D
17 260 250 2.8 16 1.25 4 4x4m 4x3m 3-D

17 260 200 2.5 16 1.25 4 4x4m 4x3m 3-D

Compressive Thickness of Story height

17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 260 200 3.2 16 1.25 4 4x4m 4x3m 3-D

17 260 200 2.8 13 1.25 4 4x4m 4x3m 3-D

Strength of Infill Infill Panel

17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D

17 260 200 2.8 19 1.25 4 4x4m 4x3m 3-D

17 260 200 2.8 16 1.00 4 4x4m 4x3m 3-D

17 260 200 2.8 16 1.25 4 4x4m 4x3m 3-D
17 260 200 2.8 16 1.50 4 4x4m 4x3m 3-D
 25

Figure 3.2 Sensitivity of the structure response to the variation in compressive strength of concrete

Figure 3.3 Sensitivity of the structure response to the variation in tensile strength of steel

Figure 3.4 Sensitivity of the structure response to the variation in transverse reinforcement spacing
26

Figure 3.5 Sensitivity of the structure response to the variation in storey height.

Figure 3.6 Sensitivity of the structure response to the variation in thickness of infill panels

Figure 3.7 Sensitivity of the structure response to the variation in compressive strength of infill panel
 27

Table 3.2 Effect of the variation in the structural characteristics-related parameters’ values on the structure response
Parameters Parameters’ values Slight Damage Moderate Damage Near Collapse
Range CV [%] CV [%] Diff. [%] CV [%] Diff. [%] CV [%] Diff. [%]
Compressive strength of concrete (fc) 14MPa ~ 20MPa 12.71 4.13 11.11 37.47 197.56 32.28 139.08
Tensile strength of steel (fy) 200MPa ~ 320MPa 16.62 6.24 12.50 9.22 30.80 7.12 22.27
Transverse reinforcement spacing (S) 150mm ~ 250mm 19.76 0.00 0.00 17.01 48.39 13.24 39.50
floor-to-floor Story height (h) 2.5m ~ 3.2m 12.39 19.16 45.09 17.96 42.32 10.59 22.93
Thickness of infill walls (tw) 13cm ~ 19cm 18.75 31.49 66.67 19.85 49.23 16.65 39.68
Compressive strength of infill walls (fw) 1.0MPa ~ 1.5MPa 20.00 44.30 128.57 33.17 102.08 14.90 34.35

For the tensile strength of steel, as shown in Figure 3.3, the effect has been found to be pretty different comparing to the
compressive strength. The effect is almost insignificant. For a CV= 16.1% of tensile strength, the CV in deformation
capacity increases very slightly from Slight Damage to Moderate Dam-age and attains a value of only 9.2%, and then de-
creases to 7% at Near Collapse.

Increased ductility is accounted for by transverse reinforcement spacing, adopting a range of values obtained from
literature on structural characteristics assessment in existing buildings (EERI 2000, Inel & Ozmen 2006). According to the
results, the structural response has been found to be moderately affected by the full range of variation in transversal
reinforcement spacing (s=150 to 250mm), as shown in Figure 3.4. For a variation of spacing CV = 19.76%, the CV in
structural response attained a value of 18% and 10.6% for Moderate Damage and Near Collapse, respectively. At Slight
Damage level, no difference was observed in the structural response, as this parameter, although improves the concrete
confinement has no effect on the onset of yielding in the concrete elements or cracking of the infills.

Floor-to-floor story height also shows a moderate effect on the seismic performance of the structure (see Figure 3.5). The
full range of variation CV = 12.4% leads to comparable difference in roof drift at different damage condition (CV in
deformation capacity reaches value from 10.6 to 19.2%).

The effect of strength and stiffness of infill walls was examined in terms of compressive strength and the thickness of infill
walls. The sensitivity analysis was conducted for values from 13cm to 19cm for thickness and 1.0MPa to 1.5MPa for
compressive strength of masonry infill walls, as shown in Figure 3.6 and Figure 3.7, respectively. According to the result
of analyses, the two parameters have shown significant effect on the structural performance, at Slight to Moderate
Damage condition. For a total variation of the thickness of infill walls, 18.75%, the structural response has been found to
be CV=31.5% at Slight Damage and decrease to 19.85% at Moderate Dam-age. For CV=20% in compressive strength of
infill walls, the variation in structural response has been found to be CV=44.3% at Slight Damage and de-crease to 33.2%
at Moderate Damage. Both parameters show a significantly reduced effect on the structural response at Near Collapse,
CV=16.65% due to variation of compressive strength and 14.9% due to variation of thickness of infills, as the contribution
of infill past the peak capacity is significantly reduced. This can be explained by the fact that the damage in infills in
general occurs at an early stage comparing to the RC members (see Figure 3.1, softening branch of the curve); hence,
the infills will start to have less effect with increasing damage.

It is evident from the results shown above that for low-ductility buildings characterized by poor quality of materials,
workmanship and detailing, structural characteristics-related parameters variation might have a significant effect on
estimating realistic structural response. Most importantly it should be noted that the relationship between parameters and
28

response is non-linear, non-monotonic and non-correlated for the three damage thresholds, indicating that values for
epistemic uncertainty cannot be interpolated or extrapolated linearly from one state to the next.

4 Effect of mathematical modelling

The effect of model completeness is investigated by performing: (1) a comparative analysis of structure response and
performance generated with and without considering the contribution of masonry infill panels (comparing the extracted
results from infilled frames system with those exacted from bare frames system); (2) a comparative analysis between
using two-dimensional (2-D) modeling and three-dimensional (3-D) modeling in the evaluation of building response.

4.1 The use of bare frames model to represent masonry infilled RC building

Extensive literature review (see D’Ayala and Meslem 2012) has shown that often, researchers and engineers attempt to
conduct vulnerability assessment of infilled RC buildings using bare frame models without considering infills, in order to
reduce the calculation efforts. Depending on the level of stiffness and structural capacity of the infills and their connection
to the main structural system , such simplifying assumption may render the capacity curve and fragility curves obtained
wherein totally unrepresentative of the assumed building class.

To clearly demonstrate this, a comparative analysis is performed between two modelling hypotheses, i.e. modeling without
inclusion of infills (bare frames system) and modeling with inclusion of infills as described in Section 3 (infilled frames
system). Figure 4.1 shows an example of comparison of resulted adaptive pushover curves using bare frame model with
the resulted from infilled frame model, with respect to variation in concrete compressive strength.

Figure 4.1 Comparison of the resulted adaptive pushover curves from infilled frame and bare frame models, considering concrete
compressive strength as variable parameter

On comparing the behaviour of two modelling hypotheses, and for the same structural characteristics configuration, a
remarkable difference in estimating the response of the building is observed as it can be seen from Figure 4.1. This
 29

important difference, shows how much the uncertainty can be significant in evaluation of seismic performance if infills are
not considered in modelling. Indeed, the stiffness of structures increased with the presence of infills. However, at small
value of displacement (value of top drift 0.25%) a first crush of infill was observed for the infilled structures. This first crush
occurred for the infill panels located at ground floor and then second floor, as shown in Figure 2.6.b. On the other hand,
the presence of infill panels have caused the occurrence of first crush of concrete members (Moderate Damage level), and
then the ultimate capacity at the global level (Near Collapse level), at earlier stage (at top drift of 0.36~1.09% and
0.77~1.85%, respectively, for the considered values range of concrete compressive strength). However, bare frame
structures show more flexibility regarding the occurrence of the first crush of concrete member and then the ultimate
capacity (at top drift of 1.25~1.62% and 1.34~2.88%, respectively, for the considered values range of concrete
compressive strength).

In addition to concrete compression strength parameter, the comparison between the two modeling hypotheses has been
also conducted considering values ranges for tensile strength of steel, transverse reinforcement spacing, and floor-to-floor
story height. In total nineteen infilled frame models, and nineteen bare frame models are considered, as shown in Figure
4.2. As shown in Table 4.1, the computed mean value of deformation capacity at Slight Damage level for the whole range
of variation of all parameters has been found to be 6 times greater for bare frame models than the one calculated for
infilled frame models. At Moderate Damage level, the difference in the structural response between infilled frame and bare
frame structures is 2.2 times greater for bare frame models than the value for infilled frame models. For Near Collapse,
this factor, in terms of mean value, is estimated to have a value of 1.8.

Figure 4.2 Comparison between the use of infilled frame and bare frame models for different structural characteristics configuration

Table 4.1 Sensitivity of structural response to the contribution of masonry infill panels.
Roof Drift
Slight Damage Moderate Damage Near Collapse
Mean [%] CV [%] Mean [%] CV [%] Mean [%] CV [%]
Infilled Frame 0.08 8.71 0.69 24.88 1.30 20.35
Bare Frame 0.48 7.54 1.53 5.96 2.29 18.10
Factor 6 2.2 1.8
30

On comparing the ability of capturing the sensitivity, the bare frame models seem to be no sensitive to change in structural
characteristics-related parameters, comparing to the case of infilled frame models (see Figure 4.2). Indeed, at the
moderate damage bare frame models have shown less sensitivity, with respect to the variation in structural characteristics
parameters’ values. At Near Collapse damage, the non-sensitivity of bare frame models is mostly remarkable with respect
to the change in the value of steel tensile strength.

In the literature, some authors (e.g. Pasticier et al. 2008) have argued that uncertainty and variations in structural
characteristics-related parameters might be considered not as critical as the uncertainty in the seismic record, for instance.
However, it is important to mention that this statement could be based on a specific result of sensitivity that was conducted
using bare frame models. It is worth to recall that in the work by Liel and Deierlein (2008), the sensitivity study, was
conducted using bare frame structures as numerical models; however, the parameters considered as variable are number
of storey and framing system.

4.2 The use of two-dimensional (planar) model

The employment of two-dimensional (planar) model for the derivation of either vulnerability functions/fragility curves or
capacity curves have been widely employed in literature, for instance, Erberik & Elnashai (2003, 2004) for 5-storey RC
building, UTCB (2006) for 13-storey RC building in Bucharest, Barbat et al. (2006), for a range of 2-8 stories for RC and
masonry buildings in Barcelona, Howary and Mehanny (2011) for 8-storey RC buildings in Cairo. The aim is to reduce the
computational effort, especially when using nonlinear dynamic analysis. However it should be kept in mind that the
epistemic uncertainty associated with this modelling strategy can be significant, especially for buildings with irregular
geometries or with buildings with non-uniform distribution of infills.

Figure 4.3 shows a comparison of capacity curves obtained by 3-D with those obtained by superposition of 2-D models. It
is clearly seen that there is a remarkable difference between the two procedures. By using 2-D models, the displacement
corresponding to first crush of concrete member seems to be overestimated. The first crush is estimated to be at top drift
of 1.28%. However, for 3-D model the first crush of concrete is estimated to be at top drift of 0.45%. In addition to that, the
peak loading capacity is underestimated by using 2-D models.

Figure 4.3 Comparison of capacity curves obtained by 3-D model with those obtained by superposition of 2-D models.
 31

5 Derivation of fragility curves

Fragility functions express the probability of a damage state, dsi , sustained by an examined building class, being reached
or exceeded given a level of ground motion intensity measure, IM . The fragility functions correspond to n number of
discrete descriptive damage states, which are typically correlated with a threshold of a selected response variable (e.g.
roof drift, interstorey drift, ultimate capacity), estimated from the structural analysis of the simulated building class. This
threshold represents the capacity of the building class, i.e. the ability of their structures to withstand a given level of
damage or to be in a state of damage given a measure of ground motion intensity IM . Therefore, the fragility functions
are constructed by estimating the probability of the generic structural response, D , reaching or exceeding a specified
threshold d i , conditioned to a range of ground motion intensity measure IM . As commonly done in most of seismic
fragility studies, the functions are assumed to take the form of lognormal cumulative distribution functions having a median
value and logarithmic standard deviation, or dispersion. The mathematical form for such a fragility functions is:

 lnIM    
     D | IM 
F  DS  ds | IM   F  D  d | IM      (5.1)
 i   i   
ds 
 i 
Where  is the standard normal cumulative distribution function;  D|IM is the lognormal mean of D conditioned on the

ground motion intensity, IM ; and ds is the lognormal standard deviation of D | IM .

i

In literature a variety of intensity measures have been used to define fragility curves. The result of investigation has shown
that Spectral Acceleration ( S a T  ), Spectral Displacement ( S d T  ), and Peak Ground Acceleration (PGA) are the most
widely used as intensity measures for analytical fragility curves, for different building typologies (D’Ayala and Meslem
2012). It is worth to mention that S a T  and S d T  are most easily correlated with the structural response and hence
are considered as the most suitable variable for damage functions, especially for nonlinear static or dynamic-based
methodologies. Methods based on limit state analysis and simplified methods tend to use PGA, also for ease of correlation
with empirical vulnerability analyses. For instance, considering the case of developing S d -based fragility curves, the
Equation 5.1 becomes:

  S T   
 1  
F  DS  ds | S    ln d  (5.2)
 i d
  ds  S d , ds  
 i  i 

where S d ,dsi is the median value of spectral displacement at which the building reaches the threshold of damage state,

dsi .

5.1 Median values S d ,dsi and dispersion  dsi

For nonlinear static-based fragility curves, S d ,dsi is obtained by the transformation of adaptive pushovers curves, defined
in terms of base shear vs. top displacement, into the equivalent Single Degree of Freedom (SDoF) capacity curves,
defined in terms of pseudo spectral acceleration ( S a ) vs. spectral displacement ( S d ), and which can be carried out using
32

the standard approach documented in many codes of practice, e.g. ATC-40 (ATC 1996); HAZUS-MH MR3 (FEMA-NIBS
2003).

In the following, fragility curves are derived for the two modelling hypotheses: infilled frame and bare frame models,
considering the expected range and central values of each structural characteristics-related parameter, as shown in Table
2.1. Nineteen infilled frame models and nineteen bare frame models are selected. See Figure 4.2 and Table 3.1 for the
considered parameters and variation of values, which are considered the most expected values that might be possibility
assigned, in case of lack of information, to represent the class of low-ductility RC buildings constructed with earlier seismic
codes (see Section 3).

Procedure 1: Distribution parameters computed from sensitivity analysis

The median S d ,dsi is obtained as the central values (or the expected mean values) of the full range of variations of the

structural characteristics-related parameters considered in this study, as described in Section 2.3.3. The variability ds
i

associated to each damage state threshold, is obtained by calculating, for the full range of variations, the lognormal of
each structure response and the variance.

Figure 5.1 shows the fragility curves derived for each set of modelling choices, i.e., infilled frame and bare frame models.
The corresponding values of spectral displacement and dispersion for each damage state threshold are presented in
Table 5.1. It is evident form Figure 5.1 the role played by the inclusion or exclusion of the masonry infill in the modelling.
The exclusion of infills’ contribution leads to a significant bias in fragility curves. The median displacement capacity varies
from infilled frame models to bare frame models by a factor of 6.2, 2.2, and 1.8 for Slight Damage, Moderate Damage, and
Near Collapse, respectively, as shown in Table 5.1. Indeed, when masonry infilled RC building are modelled as bare
frame structure, the resulted fragility curves show greater lateral displacement capacity for all damage levels; whereas, the
building is found to be more vulnerable when the infilled frame model is used. In fact, the result of adaptive pushover
analysis has shown that the first-storey mechanism is the most recurring, and modelling the infills leads to the occurrence
of this mechanism for smaller drifts compared to the case of bare frame assumption. It should be noted that difficulties
have been encountered to predict shear failure in columns. It is to recall that shear column failure might have a significant
effect on the structure performance, especially for structures designed without considering horizontal actions, or building
with low concrete strength.

With regards to the level of dispersion, this latter has found to be almost less different at Slight Damage (by factor of 0.7)
and Near Collapse Damage (by a factor of 0.9), and remarkable at Moderate Damage (by a factor of 0.3), between the two
modelling hypotheses. On the other hand, the results have clearly showed that the value of dispersion varies from one
state of damage to the next, and this variation is neither linear nor monotonic, for the two modelling hypotheses. The value
of dispersion is found to be 0.32, 0.54, and 0.44 at Slight Damage, Moderate Damage, and Near Collapse, respectively,
for set of infilled frame model; while for set of bare frame models the value is found to be 0.24, 0.17, and 0.40 at Slight
Damage, Moderate Damage, and Near Collapse.

In the literature, the uncertainty in the structural capacity is accounted for by assuming an average values, constant for all
damage states. For instance, Kappos et al. (2006) constructed fragility curves by assuming an average value of 0.3 and
0.25 for the uncertainty in the capacity for low and high code buildings, respectively, for all building types and constant all
damage states. These values have been suggested by FEMA-NIBS (2003) and HAZUS-MH (FEMA 2003). Throughout the
study by Shahzada et al. (2011) a same default value of 0.3 was assigned for the uncertainty associated with the capacity
curve of buildings for all damage states, as it is proposed in Wen et al. (2004). Satter and Liel (2010) and Raghunandan et
 33

al. (2012) have used a default value of 0.5, which has been suggested based on previous research work by Liel et al.
(2009), to account for uncertainty due to the structural modelling, for Collapse level only.

The different values of ds associated to damage state thresholds that have been obtained from numerical analysis are
i

found to be within the range of values used from various sources in literature.

Table 5.1 Median value and dispersion calculated for each threshold of damage state, and used in Figure 5.1
Slight Damage (SD) Moderate Damage (MD) Near Collapse (NC)
System
Median [mm] β Median [mm] β Median [mm] β
Set of infilled frame models 7 0.32 61 0.54 114 0.44
Set of bare frame models 43 0.24 132 0.17 206 0.40
Ratio 6.2 0.7 2.2 0.3 1.8 0.9

Figure 5.1 Comparisons of fragility curves of the structures with and without considering the contribution of masonry infill walls.

Procedure 2: First-Order Second-Moment (FOSM)

Recently, the First-Order Second-Moment (FOSM) method has been suggested by Vamvatsikos and Fragiadakis (2010)
for the estimation of the median value S d ,dsi and the dispersion  ds i for every limit state. Actually, this method has been
already implemented in some previous studies, such as by Lee and Mosalam (2005) and Baker and Cornell (2008,), to
estimate uncertainty associated with structural response. For the implementation of this method, the number of simulations
required is 2 K  1 , where K is number of random variables. For a given limit state, the log of the median spectral
displacement threshold is considered as a function of the random parameters. As four random parameters have been
used, this can be written as follows:

ln S d  f  X   f  fc , fy , s , h  (5.2)

where X ( X   fc, fy,s,h ) is the vector of random modelling parameters with mean:

X   fc ,  fy ,  S , h  (5.3)
34

The random parameters are set equal to their mean  X k (where, k = 1,..., 4) to evaluate ln S d . The remain 2 K
0

simulations are obtained by considering in turn the parameter’ values ( X k   X k ) while all other variables remain equal

to their mean  X j (with, j  k ), obtaining ln S dX

max min
and ln S dX .

Using the second derivative of f with respect to X k , the median- is estimated as:

max min
1 K 2 f 1 K ln S dX k  2 ln S d0  ln S dX k
 ln S  
0
 2
 ln S   0
 X2 k (5.4)
 
mln S
2 k 1 X k2
d Xk d 2
2 k 1 X k   Xk
d max
Xk

and by assuming lognormality, the median displacement threshold can be computed simply as:

S d  exp mln S  d
 (5.5)

The standard deviation of the logs (dispersion) is estimated using a first-order derivative of f with respect to X k :
2
 ln S dX k  ln S dX k  2
2
 f 
max min
K K
   
2
  2
   X (5.6)
k 1  X k 
Xk
 X max  X min
k 1 
 k

Xk
k k 

In Equations 5.4. and 5.6, the standard deviations  X k is determined simply on the basis of the sensitivity analysis range,
having assumed a uniform distribution for each parameter (see Section 3), considering Low/Upper bounds and central
values (see Table 2.1). The assumption followed in the process is that any of the values in the range has the same
probability associated to it, of being the central value of an aleatoric distribution, for either an index building or an entire
class, for a given parameter. The variability in response determined with the sensitivity analysis hence defines the
epistemic uncertainty associated with the attribution of any specific value in that range as default value.

Using the aforementioned procedure, Figure 5.2 shows the fragility curves that have been derived for each set of
modelling choices, i.e., infilled frame and bare frame structures. The corresponding values of spectral displacement and
dispersion for each damage state threshold are presented in Table 5.2. Either for set of infilled frame or bare frame
models, the comparison of the results from procedure 1 (distribution parameters) or procedure 2 (FOSM) shows an
agreement and matching, especially in estimating the median value at each damage state threshold (factor range between
1.0 and 1.1). Moderate difference is observed in the calculated dispersion (factor range between 1.4 and 1.7).

Table 5.2 Comparison of median value and dispersion calculated using distribution parameters and FOSM procedure
Slight Damage (SD) Moderate Damage (MD) Near Collapse (NC)
System Procedure
Mean [mm] β Mean [mm] β Mean [mm] β
Distribution parameters 7 0.32 61 0.54 114 0.44
Set of infilled frame
FOSM 7 0.18 55 0.38 107 0.30
models
Ratio 1.0 1.7 1.1 1.4 1.1 1.4

Distribution parameters 43 0.24 132 0.17 206 0.40

Set of bare frame
FOSM 40 0.15 130 0.11 187 0.26
models
Ratio 1.1 1.6 1.0 1.5 1.1 1.5
 35

Figure 5.2 Comparisons of fragility curves of the structures using distribution parameters and FOSM method. (a) Resulted fragility
curves from set of infilled frame models; (b) Resulted fragility curves from set of bare frame models.

5.2 Effect of structural characteristics parameters and mathematical modeling-based dispersion

Actually, the parameter lognormal standard deviation, ds , in Equations 5.1 and 5.2, describes the total uncertainty of the
i

fragility curves, and which in general should consider three primary sources of uncertainty (see Section 1.1), namely the
uncertainty associated with the capacity curve ( C ), the uncertainty in the demand imposed on the structure by the
earthquake ground motion (  D ), and uncertainty associated with the damage state threshold (  T ,dsi ). The total

uncertainty or dispersion, Total (i.e. ds ), in the fragility curve is calculated using the following expression (NIBE-FEMA
i

2003, FEMA 2003):

 dsi   C2   D2   T2,dsi (5.7)

The values of dispersion used in the derivation of fragility curves in previous subsection, i.e. in Figure 5.1and Figure 5.2,
were computed considering structural capacity variability (i.e. structural characteristics and mathematical modelling) only
(i.e., for  dsi   C ), without considering structural demand variability, i.e. the record-to-record dispersion in ground
motion, and/or uncertainty associated with the damage state threshold .

In order to account for record-to-record variability (  D ), the dispersion calculated from the numerical analysis (  C ) is

combined with a value of lognormal distribution dispersion of 0.45 (i.e.,  C2  0.45 2 ) as suggested in ATC-58 (ATC
2011) in absence of specific data. The resulted dispersion in capacity and record-to-record, and fragility curves are shown
in Table 5.3 and Error! Reference source not found., respectively, considering both modelling hypotheses.

For the set of infilled frame models, the shift between structural capacity (structural characteristics-related parameters)
uncertainty (  C2 )-based fragility curves and structural capacity and record-to-record uncertainty (  C2  0.45 2 )-
based fragility curves is slightly/moderately significant, expressing the remarkable effect of structural characteristics-
36

related parameters dispersion on the fragility curve. The bigger increase in uncertainty due to the adding of record-to-
record dispersion is 42%, at Slight Damage.

However, for the set of bare frame models, the shift has been found to be remarkable with a biggest increase of 65%, at
Moderate Damage, due to the adding of record-to-record dispersion. This difference expresses the non-significant effect of
structural characteristics-related parameters dispersion on fragility curve, comparing to the effect of record-to-record
dispersion, when using set of bare frame models. Hence, the completeness of model (exclusion of masonry infill
contribution) might be remarkable and significantly lead to the non-capturing accurately the sensitivity of fragility curves to
structural characteristics-related parameters.

With regard to the total uncertainty in Equation 5.7, FEMA-NIBS (2003) have provided average values which are assumed
to be constant for all damage state. For buildings designed to old codes, the average value is estimated to be
 dsi  0.75 . By using this value, fragility curves have been derived and presented in Figure 5.3, for both modelling
hypotheses. The result shows that for set of infilled frame models, the difference in the shift in fragility curve resulted from
different uncertainty modelling (i.e.,  dsi   C2 ,  C2   D2 ,  C2   D2   T2,dsi ) is less significant, with total
biggest increase of 57%; while the difference is very significant, with total biggest increase of 77%, when using set of bare
frame models.

Table 5.3 Comparison of structural characteristics and mathematical modelling dispersion with record-to-record dispersion, and
average value of total dispersion (as per literature)
Set of infilled frame models Set bare frame models
Dispersion
Slight Damage Moderate Damage Near Collapse Slight Damage Moderate Damage Near Collapse
β 0.32 0.54 0.44 0.24 0.17 0.40
42% 23% 30% 53% 65% 33%
β & βD 0.55 0.70 0.63 0.51 0.48 0.60
26% 7% 16% 32% 36% 20%
βTotal 0.75 0.75 0.75 0.75 0.75 0.75

Figure 5.3 Comparisons of structural capacity (structural characteristics and mathematical modelling) uncertainty-based fragility curves
with total uncertainty-based fragility curves. (a) Considering set of infilled frame models; (b) Considering set of bare frame models.
 37

6 Basic elements and parameters for modeling and analysis requirement

Table 6.1 and Table 6.2 presented the basic elements and parameters, respectively, for modelling and analysis
requirement for reinforced concrete buildings. In Table 6.1, attributes (elements) are divided into three classes of
relevance: Essential, if no meaningful result can be obtained without it; Qualifying, if relevant to discriminate behaviour;
and Desirable, for results refinement. In Table 6.2, attributes (parameters) are presented in terms of buildings
configuration, mechanical characteristics, geometric characteristics, and structural detailing requirements.

Table 6.1 Basic attributes (elements) for modelling and analysis requirement for reinforced concrete buildings
Modelling Requirement
Basic Attributes Source of Information
Essential Qualifying Desirable
Frame Elements X
Masonry infills X
Structural and Non- design documentation, on site
Diaphragm Elements X
Structural Elements observation, literature reference,
Roof X X
code reference
Claddings X X
Loads Live and Dead loads X
Retrofitting X X on site observation, literature
Modifications
Damage X X reference

Table 6.2 Basic attributes (parameters) for modelling and analysis requirement for reinforced concrete buildings
Basic Attributes
Building Configuration and Dimension
Mechanical characteristics
Geometry Characteristics
and Structural Detailing
Source of Information
Number of stories Design documentation,
Story heights (floor-to-floor height for the ground floor, and for other floors) on site observation,
Number of lines (number of bays) and spacing in x-direction literature reference,
Number of lines (number of frames) and spacing in y-direction code reference
Concrete Compressive strength
Modulus of elasticity
Strain at peak stress
Specific weight
Reinforcing Bar Modulus of elasticity
Yield stress
Ultimate stress
Strain hardening parameter
Specific weight
Masonry infill Compressive strength
Modulus of elasticity
Shear strength
Specific weight
Reinforced concrete Cross-section dimensions for columns and beams
elements Transversal reinforcement: type and spacing
Longitudinal reinforcement: type and number
Thickness of slabs
Masonry infill panel Dimensions and thickness of walls
Dimension of opening: windows and doors
38

7 Final Comment
This present document was devoted to provide, for the GEM end-user, details on the effect and the choice of building
capacity-related parameters with regard to the expected uncertainties that might have on the results of analytical fragility
and vulnerability assessment and, hence, help user to decide with more efficient way for a better quantification of the
uncertainties. The sensitivity analyses were conducted with regards to the variation in structural characteristics-related
parameters’ values (i.e. in terms of mechanical properties, geometric configuration and dimension, structural details) and
in mathematical modelling (i.e. completeness of models). The classes of structures considered are low-ductility RC
buildings designed according to earlier seismic codes, and which are in general characterized by poor quality of materials,
workmanship and detailing. This class of building constitutes one of the largest portions of existing residential building
stock in earthquake prone countries.

The sensitivity analyses were performed based on nonlinear adaptive pushover for forty-two (42) 3-D models. It was
clearly observed that special care should be given when assigning values to represent the structural characteristics,
especially, material characteristics-related values. Concrete and masonry strength-related variation values have shown a
significant effect on the building capacity, and this effect increase with the progress of damage condition for the concrete.
The comparison of fragility curves between the two modelling hypotheses, infilled and bare frames has shown a
remarkable bias, leading to a significant difference in predicting the seismic performance of the building. Modelling building
as bare frame structure leads to underestimate the risk of damage.

It has found that the exclusion of infill’s contribution might negatively affect the ability of the model to capture the sensitivity
of fragility curves to structural characteristics-related parameters. Furthermore, the dispersions tend to have less effect on
the fragility curve derivation when using infilled frame models, while their effect can highly increase if bare frame model is
used.
With regard to 2-D (planar) model, when used a significant epistemic uncertainty should be expected, especially for
buildings with irregular geometries or with buildings with non-uniform distribution of infills.
 39

REFERENCES
Abbasnia, R., Davoudi, A.T., and Maddah, M.M., 2013. An adaptive pushover procedure based on effective modal mass
combination rule, Engineering Structures 52: 654-666.
Al-Chaar, G. 2002. Evaluating Strength and Stiffness of Unreinforced Masonry Infill Structures. Engineer Research and
Development Center. ERDC/CERL TR-02-1.
American Society of Civil Engineers (ASCE). 2000. FEMA-356: Prestandard and commentary for the seismic rehabili-tation of
buildings. Federal Emergency Management Agency, Washington, DC.
American Society of Civil Engineers (ASCE), 2007. Seismic rehabilitation of existing buildings. ASCE Standard ASCE/SEI 41-06,
Southpointe, Canonsburg, PA.
Applied Technology Council (ATC). 1996. Seismic evaluation and retrofit of concrete buildings, ATC-40. Redwood City, CA.
Applied Technology Council (ATC), 2003. Preliminary Evaluation of Methods for Defining Performance, ATC-58-2, Redwood
City, CA.
Applied Technology Council (ATC), 2007. Recommended Methodology for Quantification of Building System Performance and
Response Parameters, ATC-63 90% Draft, Redwood City, CA.
Applied Technology Council (ATC), 2011. Seismic perfor-mance assessment of buildings, Volume 1 - Methodology, ATC-58-1
75% Draft. Redwood City, CA.
Ay, B.O., 2006. Fragility based assessment of low-rise and mid-rise reinforced concrete frame buildings in Turkey, Middle East
Technical University, MSc. Thesis.
Ay, B.O. & Erberik, A. 2008. Vulnerability of Turkish low-rise and mid-rise reinforced concrete frame structures. Journal of
Earthquake Engineering12: 2-11.
Bakhshi, A., and Karimi, K., 2006. Method of developing fragility curves a case study for seismic assessment of masonry
buildings in Iran. 7th International Congress in Civil Engineering, Tehran, Iran.
Bal, I.E., Crowley, H. and Pinho, R., 2008. Detail assessment of structural characteristics of Turkish RC buildings stock for loss
assessment models. Soil Dynamic and Earthquake Engineering, 28, 914-932.
Bal, I.E., Crowley, H. and Pinho, R., 2010. Displacement-based earthquake loss assessment: method development and
application to Turkish building stock, Rose School, Research Report No. ROSE-2010/02.
Barbat, A.H., Yépez 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.
Beker, J.W. & Cornell, C.A. 2008. Uncertainty propagation in probabilistics seismic loss estimation. Structural Safety 30: 236-
252.
Bertero, V., Aktan, A.E., Charney, F. and Sause, R., 1984 Earthquake simulator tests and associated experimental, analytical
and correlation studies of one-fifth scale model. Earthquake Effects on Reinforced Concrete Structures, 4: 375-424,
American Concrete Institut.
Canadian Standards Association (CSA), 2004. Design of masonry structures (S304.1), Ontario, Canada.
Charney, F., and Bertero, V., 1982. An evaluation of the design and analytical seismic response of a seven story reinforced
concrete frame-wall structure. Report 82/08, Earthquake Engineering Research Centre, University of California, Berkeley.
Chaulagain, H., Rodrigues, H., Jara, J., Spacone, E. and Varum, H., 2013. Seismic response of current RC buildings in Nepal: A
comparative analysis of different design/construction, Engineering Structures 49, 284-294.
Colangelo. F. (2005). Pseudo-dynamic seismic response of reinforced concrete frames in lled with non-structural brick masonry.
Earthquake Engineering and Structural Dynamics. 34,1219-1241.
40

Computers & Structures Inc (CSI), 2009. SAP2000, Integrated Software for Structural Analysis & Design, Berkeley, CA.
Crisafulli, F.J. (1997). Seismic Behaviour of Reinforced Concrete Structures with Masonry Infills. PhD Thesis, University of
Canterbury, New Zealand.
Crowley, H. and Pinho, R., 2006. Simplified equations for estimating the period of vibration of existing buildings, First European
Conference on Earthquake Engineering and Sesimology, Geneva, Switzerland, 3-8 September, paper number 1122.
D’Ayala, D., 2005. Force and displacement based vulnerability assessment for traditional buildings. Bulletin of Earthquake
Engineering, 3, 235-265.
D’Ayala, D., Meslem, A., 2012. DRAFT Guide for selection of existing analytical fragility curves and Compilation of the Database,
GEM Technical Report 2012-X, GEM Foundation, Pavia, Italy.
D’Ayala, D., Spence, R., Oliveira, O., and Pomonis, A., 1997. Earthquake loss estimation for Europe’s historic town centres.
Earthquake Spectra, 13, 773-793.
D’Ayala, D., and Paganoni, S., 2010. Assessment and analysis of damage in L’Aquila historic city center after 6th April 2009.
Bulletin of Earthquake Engineering, 9, 81-104.
Dimova, S.L., and Negro, P., 2006. Assessment of seismic fragility of structures with consideration of the quality of construction,
Earthquake Spectra, Vol. 22, No. 4, 909-936.
Dolsek, M., 2012. Simplified method for seismic risk assessment of buildings with consideration of alreatory and epistemic
uncertainty. Structure and Infrastructure Engineering, Vol 8, No. 10, 939-953.
Dolsek, M., Fajfar, P., 2008. The effect of masonry infills on the seismic reponse of a four-storey reinforced concrete frame – a
deterministic assessment, Engineering Structures, 30, 1991-2001.
Dymiotis, C., Kappos, A.J. and Chryssanthopoulos, M.K., 1999. Seismic Reliability of RC Frames with Uncertain Drift and
Member Capacity, Journal of Structural Engineering, ASCE, Vol.125, No.9, pp. 1038–1047.
Earthquake Engineering Field Investigation Team (EEFIT), 2003a. The Kocaeli, Turkey earthquake of 17 August 1999.
Institution of Structural Engineers, a Field Report by EEFIT.
Earthquake Engineering Field Investigation Team (EEFIT), 2003b. The Bingol, Turkey earthquake of the 1st of May 2003, Field
Report.
Earthquake Engineering Research Institute (EERI). 2000. Kocaeli. Tyurkey, earthquake of August 17, 1999: recon-naissance
report. Earthquake Spectra.
Ellul, F.L. 2006. Static nonlinear finite element analysis of low engineered masonry infilled reinforced concrete frames for seismic
assessment, PhD Thesis, University of Bath.
Ellul, F.L. and D’Ayala, D., 2012. Realistic FE models to enable push-over non linear analysis of masonry infilled frames, The
Open Construction and building Technology Journal, 6: 213-235.
Erol, G., Yuksel, E., Saruhan, H., Sagbas, G., Tuga, P.T., Karadogan, H.F., 2004. A complementary experimental work on brittle
partitioning walls and strengthening by carbon fibers, In: Proceedings of the 13th World Conference on Earthquake
Engineering, Vancouver, Canada.
European Committee for Standardization (CEN). 2004. Design of structures for earthquake resistance, Part 1: General rules,
seismic actions and rules for buildings, Eurocode-8, ENV 1998-1-1, Brussels, Belgium.
European Committee for Standardization, (2005). Design of structures for earthquake resistance, Part 3: Strengthening and
repair of buildings, Eurocode 8, EN 1998-3, Brussels, Belgium.
Federal Emergency Management Agency (FEMA), 1999. HAZUS 99 - Technical and user’s manual, Washington DC.
Federal Emergency Management Agency (FEMA), 2003. HAZUS-MH - Technical and user’s manual, Washington DC.
Federal Emergency Management Agency – National Institute of Building Sciences (FEMA-NIBS), 2003. Multi-hazard loss
estimation methodology – earthquake model: HAZUS®MH Advanced engineering building module - Technical and user’s
manual, Washington DC.
Ghosh, A.K., and Made, A.M., 2002. Finite element analysis of infilled frames. Journal of Structural Engineering, 128, 881-889.
 41

Gulkan, P., Aschhein, M. and Spence, R., 2002. Reinforced concrete frame building with masonry infills, World Housing
Encyclopedia, Report No. 64.
Haldar, P., Singh, Y., and Paul, D.K., 2012. Effect of URM infills on seismic vulnerability of Indian code designed RC frame
buildings, Earthquake Engineering and Engineering Vibration 11: 233-241.
Holmes, M. (1963). Combined Loading on Infilled Frames. Proceeding Of The Institution Of Civil Engineers. Vol 25:6621, 31-38.
Howary, H.A. & Mehanny, S.S.F. 2011. Seismic vulnerability evaluation of RC moment frame buildings in moderate seismic
zones. Earthquake Engineering and Structural Dynamics 40: 215-235.
Iervolino, I., Manfredi, G., Polese, M., Verderame, G.M., and Fabbrocino, G., 2007. Seismic risk of RC building classes.
Engineering Structures, 29, 813-820.
Inel, M. & Ozmen, H.B. 2006. Effects of plastic hinge proper-ties in nonlinear analysis of re-inforced concrete buildings.
Engineering Structures 28: 1494-1502.
Jiang, H., Lu, H., and Chen, L., 2012. Seismic fragility assessment of RC moment-resisting frames designed according to the
current Chinese seismic design code. Journal of Asian Architecture and Building Engineering, 11, 153-160.
Kappos, A.J., Panagopoulos, G. & Penelis, G. 2006. A hybrid method for the vulnerability assessment of R/C and URM
buildings. Bulletin of Earthquake Engineering 4: 391-413.
Khan, R.A. & Naqvi, T. 2012. Reliability analysis of steel building frame under earthquake forces. International Journal of
Emerging Technology and Advanced Engineer-ing 2: 356-361.
Kircil, M.S. & Polat, Z. 2006. Fragility analysis of mid-rise R/C frame buildings. Engineer-ing Structures 28: 1335-1345.
Lee, T.H. and Mosalam, K.M. 2004. Probabilistic Fiber Element Modeling of Reinforced Concrete Structures, Computers and
Structures, 2004, Vol. 82, No. 27, pp. 2285–2299.
Lee, T.H. & Mosalam, K.M. 2005. Seismic demand sensitivity of reinforced concrete shear-wall building using FOSM method,
Earthquake Engineering and Structural Dynamics 34: 1719-1736.
Liauw, T.C. & Kawn, K.H. (1984). New Development in Research of Infilled Frames. 8th World Conf. on Earthq.Engng, San
Francisko, 4, 623-630
Liel, A.B., Haselton, C.B., Deierlein, G.G. & Baker, J.W. 2009. Incorporating modelling uncertainties in the assessment of
seismic collapse risk of buildings. Structural Safety 31: 197-211.
Liel, A.B. & Deierlein, G.G. 2008. Assessment the collapse risk of California’s existing reinforced concrete frame structures:
Metrics for seismic safety decisions. Blume Center Technical Report No. 166.
Lourdes, L.A., Lopez, R.R., Saffar, A., 2007. Development of fragility curves for medium rise reinforced concrete shear wall
residential buildings in Puerto Rico. S.A. Elaskar, E.A. Pilotta, G.A. Torres eds. Mecanica Computacional Vol XXVI,
Cordoba, Argentina, October.
Mainstone, R. J. (1971). On the stiffness and strengths of infilled frame. Proceedings, Institution of Civil Engineers, Supplement
IV, 57–90.
Mainstone, R. J. and Weeks, G. A. (1970). The influence of bounding frame on the racking stiffness and strength of brick walls.
2nd International Brick Masonry Conference, Stoke-on-Trent, UK.
Masonry Standards Joint Committee (MSJC), 2010. Building Code Requirements for Masonry Structures / Commentary Working
Draft. Building Code Requirements for Masonry Structures
Monteiro, R., Ribeiro, R., Marques, M., Delgado, R., and Costa, A., 2008. Pushover analysis of RC bridges using fiber models or
plastic hinges. 14th World Conference on Earthquake Engineering, Beijing, China, October 12-17.
Mulgund, G.V., and Kulkarni, A.B., 2011. Seismic assessment of rc frame buildings with brick masonry infills. International
Journal of Advanced Engineering Sciences and Technol-ogies, 2, 140-147.
New Zealand Society for Earthquake Engineering (NZSEE). 2006. Assessment and Improvement of the Structural Performance
of Buildings in Earthquakes. New Zealand.
Turkish Standards Institute (TSE), 1985. TS-500, Building code requirements for reinforced concrete. Ankara, Turkey.
42

Pagnini, L., Vicente, R., Lagomarsino, S., and Varum, H., 2008. A mechanical method for the vulnerability assessment of
masonry buildings. 14th World Conference on Earthquake Engineering, October 12-17, Beijing, China.
Papanikolaou, V.K., and A.S. Elnashai, A.S., 2005. Evaluation of conventional and adaptive pushover analysis I: methodology.
Journal of Earthquake Engineering, 10, 923-941.
Papanikolaou, V.K., Elnashai, A.S. and Pareja, J.F., 2006. Evaluation of conventional and adaptive pushover analysis II:
comparative results. Journal of Earthquake Engineering, 10, 127-151.
Pasticier, L. Amadio, C. & Fragiacomo, M. 2008. Non-linear seismic analysis and vulnera-bility evaluation of a masonry building
by means of the SAP2000 V.10 code. Earthquake Engineering and Structural Dynamics 37, 467-485.
Paulay, T. & Priestley, M.J.N. 1992. Seismic Design of Rein-forced Concrete and Masonry Buildings. John Wiley & Sons, Inc.
Polese, M., Verderame, G.M., Mariniello, C., Iervolino, I., and Manferedi, G., 2008. Vulnerability analysis for gravity load
designed RC buildings in Naples – Italy. Journal of Earthquake Engineering, 12, 234-245.
Priestley, M.J.N. Seible, F. & Calvi, G.M.S. 1996. Seismic de-sign and retrofit of bridges. New York: John Wiley & Sons.
Raghunandan, M., Liel, A.B. , Ryu, H., Luco, N. & Uma, S.R. 2012. Aftershock fragility curves and tagging assessment for a
mainshock-damaged building. 15th World Conference on Earthquake Engineering, Lisbon, Portugal.
Rajeev P., and Tesfamariam, S., 2011. Effect of construction quality variability on seismic fragility of reinforced concrete building.
Proceedings of the Ninth Pacific Conference on Earthquake Engineering Building and Earthquake-Resilient Society,
Auckland, New Zealand, April 14-16.
Riddington, J.R., and Stafford-Smith, B., 1977. Analysis of infilled frames subject to racking with design recommendations. The
Structural Engineer, 55, 263-268.
Rivero, C.E., and Walker, W.H., 1984. An analytical study on the interaction of frames and infill masonry walls. 8th World
Conference on Earthquake Engineering, Proceddings, Vol. IV, 591-598.
Rossetto, T., and Elnashai, A., 2005. A new analytical procedure for the derivation of dis-placement-based vulnerability curves
for populations of RC structures. Engineering Structures, 27, 397-409.
Salvador, S., Monica, A., Michael, S., Cristian, J. & Luis, E. 2008. Seismic response of rein-forced concrete buildings in Caracas,
Venezuela. The 14th World Conference on Earth-quake Engineering, Beijin, China, October 12-17.
Sargin, M, Ghosh, S. K. and Handa, V. K. (1971). Effects of lateral reinforcement upon the strength and deformation properties
of concrete. Magazine of Concrete Research, 23:75-76, 99-110
Sattar, S. & Liel, A.B. 2010. Seismic performance of reinforced concrete frame structures with and without masonry infill walls.
9th U.S. National and 10th Canadian Conference on Earthquake Engineering, Toronto, Canada.
SeismoSoft Ltd, 2013. SeismoStruct V. 6.0, Software applications for analysis of structures subjected to seismic actions, Pavia,
Italy.
Shahzada, K., Gencturk, B., Khan, A.N., Naseer, A., Javed, M., & Fahad, M. 2011. Vulnerabil-ity assessment of typical buildings
in Pakistan. International Journal of Earth Sci-ences and Engineering 4: 208-211.
Shakeri, K., Tarbali, K. and Mohebbi, M., 2012. An adaptive modal pushover procedure for asymmetric-plan buildings,
Engineering Structures 36, 160-172.
Stafford-Smith, B. S. (1967). Methods for predicting the lateral stiffness and strength of multi-storey infilled frames. Building
Science. Vol II, 247-257.
Turkish Standards Institute (TSE). 1985. TS-500, Building code requirements for reinforced concrete. Ankara, Turkey.
Uma, S.R., Ryu, H., Luco, N., Liel, A.B., and Raghunandan, M., 2011. Comparison of main-shock and aftershock fragility curves
developed for New Zealand and US buildings. Proceedings of the Ninth Pacific Conference on Earthquake Engineering
Building and Earthquake-Resilient Society, Auckland, New Zealand, April 14-16.
Vacareanu, R., Chesca, A.B., Georgescu, B., and Seki, M., 2007. Case study on the expected seismic losses of soft and weak
groundfloor buildings. International Symposium on Strong Vrancea Earthquake and Risk Mitigation, Bucharest, Romania,
October.
 43

Vamvatsikos, D. & Fragiadakis, M. 2010. Incremental dynamic analysis for estimating seismic performance sensitivity and
uncertainty, Earthquake Engineering and Structural Dynamics 39: 141-163.
Wen, Y.K., Ellingwood, B.R. & Bracci, J. 2004. Vulnerability function framework for consequence-based engineering. Mid-
America Earthquake (MAE) Center, Project DS-4 Re-port
44

ANNEX A Quality classification for index building analysis

Parameters for quality classification/index building

In most of previous studies, the construction quality for the existing buildings is categorized according to specific
parameters that are in general affected by the quality of workmanship. These parameters are constructional, and in terms
of material properties and detailing of the structural elements. For instance, in the studies by Dymiotis et al. (1999), Lee
and Mosalam (2004), and Ay (2006), the parameters that were considered for subclassifying the quality of building are
those related to the strength: concrete strength (fc) and steel yield strength (fy). In the study by Dimova and Negro (2006),
the quality of construction was assessed in terms of spacing of stirrups.

Number of subclass for index building

Within GEM project, the building quality has been categorized into two subclasses: Poor and Good. For each subclass, a
most probable/expected range of values and mean value have been assigned, considering the followings sources from
literature.
- results from statistical structural assessment (e.g. Gulkan et al. 2002; Ay 2006; Crowley and Pinho 2006; Bal et
al. 2008, 2010);
- results from post-earthquakes surveys (EEFIT 1999; EERI 2000, 2002; Ellul 2006); and
- the requirement and recommendation from previous/old seismic codes up-to version TS-500 (TSE 1985) and
recent ones such as Eurocod-2 (CEN 1993) and Eurocode-8 (CEN 2001).
For the case of Turkish building stock, the quality classification for index building is highly related to the period of
construction. Poor subclass, defines in general buildings that were designed according to earlier seismic codes, and which
are characterized by poor quality of material, as highlighted in many results from post-earthquakes survey (EEFIT 1999;
EERI 2000, 2002; Ellul 2006). Good subclass defines buildings that were designed according to newest seismic codes.

Quality classification for index building

Table A- 1, shows the different range of values and mean value that have been considered the most probable/expected
for the parameters that are associated to the strength of reinforced concrete, spacing of transvers reinforcement, strength
and thickness of infills, and story height.
The differentiation between Poor and Good subclasses is mainly based on the concrete compressive strength, steel yield
strength, and transverse reinforcement spacing. According to the results of statistical structural assessment, poste-
earthquakes surveys, and requirement from different versions of seismic codes, there are a significant difference in terms
of expected range of values between buildings designed according to the earlier seismic codes with those designed
 45

according to newest versions. For instance, very low values for both concrete compressive strength and steel yield
strength are expected for older buildings (the expected mean values are 17MPa for fc, and 260MPa for fy) comparing to
those expected for newest buildings (the expected mean values are 24MPa for fc, and 450MPa for fy). In fact, the previous
seismic codes have required a value of 20MPa for concrete compressive strength. In addition to that, it has been usually
mentioned that the older buildings were constructed with less control for the quality of workmanships.

Another parameter that is, usually, affected by the quality of workmanships, is the transverse reinforcement spacing,
especially for the case of older buildings. For this subclass, i.e. Poor subclass, results from post-earthquakes surveys have
shown that a range from 150mm to 250mm (maybe up to 300mm) can be expected. For the newest buildings, i.e., Good
subclass, considering what has been required by Eurocode-2 (CEN 1993) and Eurocode-8 (CEN 2001) with taking into
account the quality of workmanship, the expected values range from 50mm up to 150mm. Accordingly, the assigned mean
value for Poor and Good subclass are 200 mm and 100 mm, respectively

With respect to the parameter story height, this later has been found to be associated with the period of construction. For
the older buildings (usually associated to the Poor subclass as mentioned above), the story height is mostly found to be in
the range of 2.5 to 3.2m, whilst a range of 2.8 to 3.5m is expected for the newest buildings (usually associated to the Good
subclass). Accordingly, the most probable mean value for Poor and Good subclass is estimated as 2.8m and 3.0m,
respectively.

Regarding the parameters related to the masonry infill panel, i.e. the compressive strength and thickness, they do not
differentiate much with respect to the period of construction. Hence, same expected range and mean value have been
assigned for both, Poor and Good subclasses.

45
46

Table A- 1 Expected value range and average value associated to structural characteristics-related parameters quality classification and index buildings analysis
GEM-AVM Quality Classification for Index Building
Parameter Information from Literature Analysis
Poor Good
Concrete [Bal et al. 2008, 2010] [Previous Earthquake Codes TS500] most probable values: most probable values:
compressive Older part of building stock: Require a value of 20MPa as Lower bound: 14 MPa Lower bound: 20 MPa
strength 6076 samples ; Gamma dist. the lowest concrete strength Upper bound: 20 MPa Upper bound: 28 MPa
Mean=16.73MPa; COV=51% Mean value: 17 MPa Mean value: 24 MPa
Newer part of building stock: [Gulkan et al. 2002]
433 samples; log-normal dist. 10 - 20 Mpa
Mean=24.9MPa; COV=8.5%
[EERI 2000]
[Ay 2006] Rarely exceed 20MPa
Superior: mean=20MPa; COV=16%; Norm. dist Values as low as 10MPa has been reported
Typical: mean=15MPa; COV=18%; Norm. dist
Poor: mean=10MPa; COV=20%; Norm. dist [Crowley and Pinho 2006]
Ranging from 15 to 29 Mpa were considered

Yield strength [Bal et al. 2008, 2010] [EEFIT 2003] most probable values: most probable values:
of steel S220-type steel; Normal dist Typical value of 240 MPa in ordinary buildings Lower bound: 200 MPa Lower bound: 320 MPa
Mean=371MPa; COV=24% Typical value of 420 MPa in governmental buildings Upper bound: 320 MPa Upper bound: 500 MPa
S420-type steel; before 1990 Mean value: 260 MPa Mean value: 450 MPa
Mean=439MPa; COV=14% [Ay 2006]
S420-type steel; after 1990 Superior: mean=480MPa; COV=10%; Norm. dist
Mean=473MPa; COV=15% Typical: mean=365MPa; COV=11%; Norm. dist
Poor: mean=250MPa; COV=12%; Norm. dist
[Crowley and Pinho 2006]
Ranging from 200 to 380 Mpa were considered

Transverse [EERI 2002] [CEN 1993, 2001] most probable values: most probable values:
reinforcement considered values: 75mm, 120mm Required spacing: 50mm to 150mm. Lower bound: 250 MPa Lower bound: 150 MPa
spacing Upper bound: 150 MPa Upper bound: 50 MPa
Mean value: 200 MPa Mean value: 100 MPa

Floor-to-floor [Bal et al. 2008, 2010] [EERI 2000, 2002; EEFIT 2003] most probable values most probable values
storey height Lognormal dis. Between 2.5 and 3.30m floor-to-floor height : 2.7 to 3.2 m for uper floors Lower bound: 2.5 m Lower bound: 2.8 m
Mean=2.84m; COV=8% floor-to-floor height : 4.0 to 4.5 m for ground floor Upper bound: 3.2 m Upper bound: 3.5 m
Mean value: 2.8 m Mean value: 3.0 m
 47

Thickness of [Bal et al. 2008, 2010] [EERI 2000; Ellul 2006] most probable values: most probable values:
infills Outer walls: 20 cm of thickness constructed with 13.5 cm thick clay bricks Lower bound: 13 cm Lower bound: 13 cm
constructed with 13.5 cm thick clay bricks plus cement-lime plaster up to 30 mm Upper bound: 19 cm Upper bound: 19 cm
plus plaster on either side on either side Mean value: 16 cm Mean value: 16 cm
Inner walls: 15 cm of thickness
constructed with 8.5 cm thick clay bricks [Crowley and Pinho, 2006]
considered values: 10cm and 25cm

Compressive [Bal et al. 2008, 2010] [Kappos et al. 2006] most probable values: most probable values:
strength Lower bound: 1.9 MPa 1.5 and 3.0 MPa have been adopted Lower bound: 1.0 MPa Lower bound: 1.0 MPa
of infills Upper bound: 3.2 MPa Upper bound: 1.5 MPa Upper bound: 1.5 MPa
[Calvi et al. 2004; Bal et al. 2006] Mean value: 1.25 MPa Mean value: 1.25 MPa
Mean value: 1.5 MPa

47
48

ANNEX B Modelling of masonry infill panels

In the literature, many models were proposed in an attempt to provide a better understanding of infill panels’ behaviours
and define potential infills failure mechanisms. Table B- 1 summarizes a large number of different equations, available in
literature, for calculation of the width of equivalent strut “a”. Table B- 2 shows the range of values of equivalent strut width
computed from different relationships for the infill panels implemented in one-storey, single bay RC frame. The results of
the parametric study have shown that these different formulae proposed by several researchers lead, given the same
geometry, to large variability in results, with up to 81% differences (see Figure B- 1). Furthermore, the results have shown
that varying the dimensions and geometry of the infill panel, the use of some formulae for the equivalent masonry strut’s
effective width presented in Figure B- 2 might lead to either under-estimation or over-estimation of the reduction factor,
beyond the recommended range 0.40a to 0.70a (Al-Chaar 2002). For instance, the use of Holmes (1961), Paulay &
Priestley (1992), and MSJC (2010) formulae has conducted to adopt a value for reduction strut width of ared ≤ 20%.a,
ared ≤ 25%.a, and ared ≥ 80%a, respectively, to obtain a better correlation with experimental results. In addition to that,
this under-estimation or over estimation of the reduction factor might lead, as a consequence, to the appearance of
numerical errors in processing due to the sharp decrease in the strut width (Figure B- 2).

Figure B- 1 Comparison of capacity functions obtained using different formulae for the effective width of equivalent strut. The RC frame
is one-storey, single bay, intended to represent the ground floor of a four--storey masonry infilled concrete building and are
representative of older structures designed using Italian reinforced concrete non-seismic code provisions. Experimental data was
collected from Colangelo (2005).
 49

Table B- 1 Formulae of equivalent masonry strut's effective width as per literature

Width of Strut Researchers / Guidelines

1
ܽ ൌ  ‫ݎ‬௜௡௙ Holmes (1961)
3
ି଴Ǥସସହ ଴Ǥ଴଺ସ
௅೔೙೑ ௅೔೙೑
ܽ ൌ0.58 ߣூℎ ௖௢௟ ଴Ǥଷଷହ‫ݎ‬௜௡௙ Stafford-Smith and Carter (1969)
௛೔೙೑ ௛೔೙೑

ܽ ൌ ͲǤͳ͸ߣூℎ௖௢௟ ି଴Ǥଷ‫ݎ‬௜௡௙ Mainstone and Weeks (1970)

ି଴Ǥସ Mainstone (1971), included in FEMA-273, -306,-356, ASCE/SEI

ܽ ൌ ͲǤͳ͹ͷߣூℎ ௖௢௟ ‫ݎ‬௜௡௙
Standard 41-06 (ASCE 2006)

1 Hendry (1981), adopted by the Canadian Standard (CSA S304.1-

ܽൌ ∝ ଶ௛ + ∝ ଶ௅ 04)
2
଴Ǥଽହ௛೔೙೑ ୡ୭ୱ ఏ
ܽ ൌ Liauw and Kwan (1984)
ఒ಺௛೎೚೗

‫ܧ‬௖‫ܣ‬௖ ா೎஺೎
ܽ ؆ ͲǤʹͲ•‹ߠ ‫ݎ‬ ( if 1 < < 5 ) Tassios (1984)
‫ܩ‬௠ ‫ܣ‬௠ ௜௡௙ ீ೘ ஺೘

ߨ
ܽൌ Flanagan and Bennet (1999, 2001)
‫ߣܥ‬ூ…‘•ߠ

1 Paulay and Priestley (1992), adopted in New Zealand Code

ܽ ൌ  ‫ݎ‬௜௡௙
4 NZS4230:1990 (SANZ 1990)

ି଴Ǥଵ
௛ర
೎೚೗
ா೘ ௧೔೙೑
ܽ ൌ ͲǤ͵ʹ •‹ʹߠ •‹ʹߠ Durrani and Luo (1994)
௠ ா೎ூ೎೚೗௛೔೙೑

0.3
ܽൌ recommended by MSJC (2010)
ߣூ…‘•ߠ

ଵ ℎ ௜௡௙
‫ܧ‬௠ ‫ݐ‬௜௡௙‫ ߠʹ݊݅ݏ‬ସ ߠ ൌ –ƒିଵ
ߣூ = ‫ܮ‬௜௡௙
Ͷ‫ܧ‬௙ ‫ܫ‬௖௢௟ℎ௜௡௙

ଵ ଵ
ߨ Ͷ‫ܧ‬௖‫ܫ‬௖௢௟ ℎ ௜௡௙ ସ ߨ Ͷ‫ܧ‬௖‫ܫ‬௕௘௔௠ ℎ ௜௡௙ ସ
∝௛ = ∝ ௅=
2 ‫ܧ‬௠ ‫ݐ‬௜௡௙ •‹ʹߠ 2 ‫ܧ‬௠ ‫ݐ‬௜௡௙ •‹ʹߠ

6‫ܧ‬௕ ‫ܫ‬௕௘௔௠ ℎ ௖௢௟

݉ = 6 1+
ߨ‫ܧ‬௖‫ܫ‬௖௢௟‫ܮ‬

Table B- 2 Range of values of equivalent strut width computed from different relationships for the infill panels.

Dimension Relative stiffness Equivalent Strut Width (mm)

Height (mm) x Length

Holmes Mainstone and Mainstone Liauw and Paulay and MSJC
(mm) x Thikness ʎ.Hcol
(1961) Weeks (1970) (1971) Kwan (1984) Priestley (1992) (2010)
(mm)
1300x2300x160 2.98 881 305 299 623 660 165

49
50

Figure B- 2 Variation of reduction factor for different equations for the effective width of equivalent strut.

The results of parametric study, on different equations for the calculation of the equivalent strut width, show that the
formula based on the early work of Mainstone (1971) leads to reasonable values of strut width reduction, while obtaining a
good agreement with the experimental capacity curve, as shown in Figure B- 1. The formula of Mainstone (1971) is the
most widely used by researchers. This formula has been included in several guidelines, such as, ASCE/SEI Standard 41-
06 (ASCE 2006), FEMA-356 (ASCE 2000).

Finally, it should be recalled that the different equivalent strut models presented above (and from which the Mainstone’s
model was selected to be used in this present document for modelling masonry infills) belong to the macro model
approach, which considered as simple, involving relatively modest calculations effort and computing time. However, there
also numerous models which have been proposed in the literature and which are classified as belonging to the micro
model approach. Models from this later approach are based on a finite element representation (Riddington and Stafford-
Smith 1977, Rivero and Walker 1984; Ghosh and Made 2002).