Sunteți pe pagina 1din 60

I.U.S.S.

Istituto Universitario Università degli


di Studi Superiori Studi di Pavia

EUROPEAN SCHOOL OF ADVANCED STUDIES IN


REDUCTION OF SEISMIC RISK

ROSE SCHOOL

A REVIEW OF EXISTING PUSHOVER METHODS FOR 2-D


REINFORCED CONCRETE BUILDINGS

A Individual Study Submitted in Partial


Fulfilment of the Requirements for the PhD Degree in

EARTHQUAKE ENGINEERING

By

MANUEL ALFREDO LÓPEZ MENJIVAR

Supervisor: Dr. RUI PINHO

September 2004
The dissertation entitled “A review of existing pushover methods for 2-D reinforced concrete
buildings”, by Manuel Alfredo López Menjivar, has been approved in partial fulfilment of the
requirements for the Doctor of Philosophy Degree in Earthquake Engineering.

Rui Pinho

Stelios Antoniou
Abstract

ABSTRACT

It is well known, within the Earthquake engineering community, that the most accurate method of
seismic demand prediction and performance evaluation of structures is nonlinear time history
analysis. However, this technique requires the selection and employment of an appropriate set of
ground motions and having a computational tool able to handle the analysis of the data and to produce
ready-to-use results within the time constrains of design offices; clearly, a simpler analysis tool is
desirable. One method that has been gaining ground, as an alternative to time history analysis, is the
nonlinear static pushover analysis. The purpose of the pushover analysis is to assess the structural
performance by estimating the strength and deformation capacities using static, nonlinear analysis and
comparing these capacities with the demands at the corresponding performance levels. Traditionally,
conventional (i.e. non-adaptive) has been used and implemented in design codes. Although it provides
crucial information on response parameters that cannot be obtained with conventional elastic methods
(either static or dynamic), this method is not exempt from some limitations such as the inability to
include higher mode order effects or progressive stiffness degradation. Therefore, the need for a fully
adaptive procedure that overcomes the above deficiencies is readily noted. A revision of the “non-
conventional” pushover procedures recently proposed was carried out, from which advantages and
disadvantages of each approach were identified. It was noted that adaptive pushover methodology
constitute a viable option to nonlinear dynamic analysis since it solves the inherent deficiencies that
conventional pushover analysis possesses. Additionally, the possibility of using an alternative modal
combination to the quadratic modal combination rules (i.e. SRSS and CQC) has been proposed and
assessed. The technique is named Direct Vectorial Addition, DVA. The feasibility of this alternative
modal combination is tested by using case models of actual buildings. To verify the accuracy of the
different pushover schemes the results obtained by these nonlinear static methods are compared to
those from the nonlinear time history analysis and the standard error is used to evaluate the accuracy
of the comparisons.

i
Acknowledgements

ACKNOWLEDGEMENTS

ii
Index

INDEX

ABSTRACT ................................................................................................................................................................I

ACKNOWLEDGEMENTS ............................................................................................................................................II

INDEX...................................................................................................................................................................III

LIST OF TABLES ...................................................................................................................................................... V

LIST OF FIGURES .................................................................................................................................................... VI

1. INTRODUCTION ....................................................................................................................................................1
1.1 Scope ............................................................................................................................................................1
1.2 Objectives.....................................................................................................................................................2
1.3 Outline of this Dissertation...........................................................................................................................2

2. NONLINEAR STATIC ANALYSIS (PUSHOVER) – OVERVIEW ....................................................................................4


2.1 Introduction ..................................................................................................................................................4
2.2 Generalities...................................................................................................................................................5
2.2.1 Description of the traditional pushover analysis................................................................................5
2.3 Development and assessment of the method ................................................................................................9

3. DESCRIPTION OF CASE STUDIES ..........................................................................................................................13


3.1 Introduction ................................................................................................................................................13
3.2 Modelling ...................................................................................................................................................13
3.2.1 Analytical tool .................................................................................................................................13
3.2.2 Modelling of the members ...............................................................................................................13
3.2.3 Case study one: RM15.....................................................................................................................14
3.2.4 Case study two: The ICONS frame .................................................................................................18
3.3 Nonlinear dynamic analysis........................................................................................................................20

iii
Index

3.3.1 Case study one: RM15.....................................................................................................................20


3.3.2 Case study two: The ICONS frame .................................................................................................21

4. DESCRIPTION AND ASSESSMENT OF PUSHOVER METHODOLOGIES ......................................................................24


4.1 Introduction ................................................................................................................................................24
4.2 Measuring the accuracy of the pushover techniques ..................................................................................24
4.3 Non-adaptive non-modal pushover analysis...............................................................................................25
4.4 Non-adaptive modal pushover analysis ......................................................................................................26
4.5 Force-based adaptive pushover procedures ................................................................................................29
4.6 Displacement-based adaptive pushover procedures ...................................................................................34
4.7 Possible drawbacks due to the use of quadratic modal combination rules .................................................36
4.8 The ICONS frame.......................................................................................................................................39
4.8.1 Non-adaptive non-modal pushover analysis ....................................................................................39
4.8.2 Pushover analysis using DVA .........................................................................................................40

5. CONCLUSIONS ....................................................................................................................................................44
5.1 Summary ....................................................................................................................................................44
5.2 Future Research ..........................................................................................................................................45

6. REFERENCES ......................................................................................................................................................46

iv
Index

LIST OF TABLES

Table 3.1 Definition of the considered structural system, after Antoniou and Pinho [2004a]..............................15

Table 3.2 Member cross-sectional dimensions for regular buildings, after Mwafy [2001] ..................................15

Table 3.3 Material properties used in the assessment, after Mwafy [2001]..........................................................16

Table 3.4 Calculated floor gravity loads and masses............................................................................................18

Table 3.5 Characteristics of the records employed by Antoniou and Pinho [2004a]............................................21

Table 4.1 Standard error of the pushover schemes in Figure 4.1..........................................................................25

Table 4.2 Standard error of the pushover schemes in Figure 4.4..........................................................................28

Table 4.3 Standard error of pushover schemes shown in Figure 4.7 ....................................................................34

Table 4.4 Standard error of pushover schemes shown in Figure 4.8 ....................................................................36

Table 4.5 Standard error of pushover schemes shown in Figure 4.7 and Figure 4.9 ............................................38

Table 4.6 Standard error of pushover schemes shown in Figure 4.8 and Figure 4.10 ..........................................39

Table 4.7 Standard error of the pushover schemes in Figure 4.11 and Figure 4.12..............................................40

Table 4.8 Standard errors of the profiles shown in Figure 4.13 to Figure 4.14 ....................................................41

Table 4.9 Standard errors of pushover schemes shown in Figure 4.15 and Figure 4.16.......................................42

v
Index

LIST OF FIGURES

Figure 2.1 Pushover curves of the MDOF and equivalent SDOF systems, after Alvarez-Botero and
López-Menjivar [2004] .......................................................................................................................6

Figure 2.2 Pushover curves of the equivalent SDOF and bilinear systems, after Alvarez-Botero and
López-Menjivar [2004]. ......................................................................................................................6

Figure 2.3 Displacement profiles of the ICONS simplified model under constant-force and constant-
displacement distributions...................................................................................................................8

Figure 3.1 Fibre plasticity discretization in a reinforced concrete section. ........................................................14

Figure 3.2 Location of Gauss points along the member length. .........................................................................14

Figure 3.3 Plane and sectional elevation of the 12-storey regular frame building set, after Mwafy
[2001]................................................................................................................................................15

Figure 3.4 Adopted modelling approach, after Mwafy [2001] ...........................................................................17

Figure 3.5 Elevation and plan views of the frame, after Carvalho et al. [1999] .................................................19

Figure 3.6 Reinforcement detail of the columns, after Carvalho et al. [1999]....................................................19

Figure 3.7 Scheme of vertical loads for nonlinear analysis, after Carvalho et al. [1999]...................................20

Figure 3.8 Elastic response spectra of the four records (5% equivalent viscous damping). ...............................21

Figure 3.9 Respose spectra of input motion: (a) acceleration; (b) displacement. ...............................................22

Figure 3.10 Artificial acceleration time histories for (a) 475 year (Acc-475) and (b) 975 year (Acc-975)
return period......................................................................................................................................22

Figure 3.11 Analytical and experimental top frame displacement: (a ) Acc-475; (b) Acc-975............................23

Figure 3.12 Analytical and experimental drift profiles: (a ) Acc-475; (b) Acc-975 .............................................23

Figure 4.1 Comparison of drift profiles from triangular and uniform pushovers with the envelope of
maximum inter-storey drifts from dynamic analysis for the RM15 frame........................................25

Figure 4.2 The multimodal pushover procedure [after Paret et.al., 1996]..........................................................26

vi
Index

Figure 4.3 Properties of the nth mode inelastic SDOF system derived from the corresponding pushover
curve [after Chopra and Goel, 2001].................................................................................................27

Figure 4.4 Comparison of drift profiles from MPA with the envelope of maximum inter-storey drifts
from dynamic analysis for RM15 frame ...........................................................................................28

Figure 4.5 Graphical representation of loading force vector calculation with incremental updating .................32

Figure 4.6 Graphical interpretation of incremental updating using tangent stiffness .........................................33

Figure 4.7 Comparison of drift profiles from FAP (SRSS) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15 ............................................................................................33

Figure 4.8 Comparison of drift profiles from DAP (SRSS) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15 ............................................................................................36

Figure 4.9 Comparison of drift profiles from FAP (DVA) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15 ............................................................................................38

Figure 4.10 Comparison of drift profiles from DAP (DVA) with the envelope of maximum inter-storey
drifts from dynamic analysis .............................................................................................................39

Figure 4.11 Comparison of drift profiles from triangular and uniform pushovers with the drift profile at
the time step of maximum inter-storey drift from dynamic analysis using the a) Acc-475
record and b) Acc-975 record............................................................................................................40

Figure 4.12 Comparison of drift profiles from triangular and uniform pushovers with the envelope of
maximum inter-storey drifts from dynamic analysis using the a) Acc-475 record and b)
Acc-975 record..................................................................................................................................40

Figure 4.13 Comparison of drift profiles from FAP with the drift profile at the time step of maximum
inter-storey drift from dynamic analysis using the a) Acc-475 record and b) Acc-975 record .........41

Figure 4.14 Comparison of drift profiles from FAP with the envelope of maximum inter-storey drifts
from dynamic analysis using the a) Acc-475 record and b) Acc-975 record ....................................41

Figure 4.15 Comparison of drift profiles from DAP with the drift profile at the time step of maximum
inter-storey drift from dynamic analysis using the a) Acc-475 record and b) Acc-975 record .........42

Figure 4.16 Comparison of drift profiles from DAP with the envelope of maximum inter-storey drifts
from dynamic analysis using the a) Acc-475 record and b) Acc-975 record ....................................42

vii
Chapter 1: Introduction

1. INTRODUCTION

1.1 SCOPE
The best way to assess the performance and to predict the demand on a structure subjected to
earthquake action is nonlinear time history analysis. However, for this technique to be reliable some
parameters need to be clearly defined. A set of characteristic ground motions that may affect the area
[Bommer et al., 2003] and a mathematical tool able to handle all analyses often exceed the
capabilities of a design office which works under tight time constraints. Thus, a simpler yet reliable
method of structural analysis is desirable.

During the last decade, the nonlinear static pushover analysis has been gaining ground among the
structural engineering society as an alternative mean of analysis. The purpose of the pushover analysis
is to assess the structural performance by estimating the strength and deformation capacities using
static, nonlinear analysis and comparing these capacities with the demands at the corresponding
performance levels. The assessment is based on the estimation of important structural parameters,
such as global and inter-storey drift, element deformations and internal forces. The analysis accounts
for the geometrical nonlinearity and material inelasticity, as well as the redistribution of internal
forces. Typically, the traditional procedure is to push the structure with a constant triangular or
uniform distribution of forces. Although it provides crucial information on response parameters that
cannot be obtained with conventional elastic methods (either static or dynamic), this method is not
exempt from some limitations such as the inability to include higher mode order effects or progressive
stiffness degradation. Therefore, the need for a fully adaptive procedure that overcomes the above
deficiencies is readily noted.

An appealing adaptive pushover method is the one proposed by Antoniou and Pinho [2004a; 2004b].
In this adaptive pushover technique, the lateral load distribution is not kept constant but is
continuously updated during the process, according to modal shapes and participation factors derived
by eigenvalue analysis carried out at each analysis step. The method is multimodal and accounts for
softening of the structure, its period elongation, and the modification of the inertial forces due to
spectral amplification. Two variants of the method currently exist: Force-based Adaptive Pushover
(FAP) and Displacement-based Adaptive Pushover (DAP). Moreover, of the two variants, the latter
has already proved to provide results very similar to those obtained from the time history analysis
when the response of seismically design buildings has been analysed [Antoniou and Pinho, 2004b].

Nevertheless, the adaptive pushover may not be exempt of limitations the main of which could be the
way how different modes are combined. The modal combination rules used are SRSS and CQC that
remove the contribution of negative quantities, since they are squared, rendering results that are
always positive. It is known that during an earthquake the deform shape of a building is not always
positive instead it can be composed of a wide range of displacements, with different signs, along the
height of the structure. The possible flaw is evident when the response of 3D structures is studied. In
this case, torsional modes are almost wiped away and their contribution to the overall force vector is
completely modified because the signs of the modal displacements are removed by the present modal
combination rules. This became obvious during the research performed by López-Menjivar [2003] on

1
Chapter 1: Introduction

3D reinforced concrete buildings. Thus, a new approach to combine the contributions of the modes is
clearly needed.

However, it was observed in a previous research [Antoniou and Pinho, 2004b], that DAP might
overcome the flaw mentioned in the precedent paragraph in the case of seismically design 2D building
structures. Therefore, more study is needed before stating a definite conclusion.

1.2 OBJECTIVES
The nonlinear static pushover analysis is an alternative method, with respect to the nonlinear time
history analysis, of demand prediction and performance evaluation that has been used in the last
decade. Nevertheless, it has been made obvious that, despite its efficiency and applicability, it exhibits
significant limitations [Krawinkler and Seneviratna, 1998].

Recently, many attempts have been proposed to improve the capability of the pushover analysis.
Innovations such as considering high mode order effects, stiffness degradation and input motions have
been developed. One of those current methods is an adaptive pushover analysis elaborated by
Antoniou and Pinho [2004a, 2004b]. By making an extensive review of the current pushover methods,
the main aims of this research are the following: The first one is to discuss and assess different
pushover methodologies by employing a typical seismically designed concrete frame structure. The
second purpose is to explore the possibility of using and alternative modal combination ruled, named
DVA, to the usual SRSS and CQC rules to further improve the capabilities of the above mentioned
adaptive approach.

1.3 OUTLINE OF THIS DISSERTATION


This research is organized into five chapters. It covers several aspects ranging from the motivation
behind the selection of the research topic, the literature review, the development of a new modal
combination scheme, to be used in an adaptive pushover analysis, to the testing of different pushover
methods and the validation of this new scheme by applying it to two case studies.

Chapter 1 is devoted to the presentation of the research topic and the identification of the general
scope and specific objectives.

The nonlinear static pushover analysis procedure, which has recently attracted considerable attention
as a tool for seismic assessment, is described in detail in Chapter 2. Its merits and drawbacks are also
discussed, along with essential issues, such as target displacements and the applied load pattern.

The study employs two reinforced concrete buildings. The first structure, RM15, belongs to a set of
buildings already employed in a by Antoniou [2002] and includes frames that are seismically design.
The second case study is a four-storey bare frame known to have a soft storey at the third floor,
named ICONS frame hereafter. The latter is attributed to a drastic change in strength and stiffness at
this level through a reduction in both the reinforcement content and the section dimensions in the
columns between the second and third storeys, coinciding also with the location of lap-splicing.
Structural characteristics of each building are illustrated in Chapter 3.

A review of recent research carried out worldwide on pushover analysis is presented in Chapter 4 with
particular emphasis placed on attempts to extend its applicability and efficiency. One such successful
attempt, the adaptive pushover analysis proposed by Antoniou and Pinho [2004a, 2004b] is described
and scrutinized here. An attempt is made at identifying a limitation of the above adaptive pushover
method following which a scheme, the Direct Vectorial Addition (DVA) is proposed and expounded
as a solution. Along the development of the chapter, a case study is used to compare the response

2
Chapter 1: Introduction

obtained from different pushover analyses to obtain preliminary conclusions. The chapter closes with
the application of DVA to another case study.

Chapter 5 is a summary of the most important aspects discussed in the dissertation, a thorough
presentation of the main findings and conclusions, and the identification of the possible projects for
future research are also outlined.

3
Chapter 2: Nonlinear static analysis (Pushover) −Overview

2. NONLINEAR STATIC ANALYSIS (PUSHOVER) –


OVERVIEW

2.1 INTRODUCTION
The use of the nonlinear static analysis, named as pushover analysis hereafter, dates back to the
1970’s but only after gaining importance during the last 10-15 years had dedicated publications
started to appear on the subject. Initially the majority of them concentrated on discussing the range of
applicability of the method and its advantages and disadvantages, compared to elastic or non-linear
dynamic procedures [e.g. Lawson et al., 1994; Krawinkler and Seneviratna, 1998].

Only very recently, there have been efforts to extend pushover analysis to take into account higher
mode effects [Paret et al., 1996; Sasaki et al., 1998; Moghadam and Tso, 2002; Chopra and Goel,
2001, 2002]. Furthermore, there have also been some attempts to derive fully adaptive procedures
[Bracci et al., 1997; Gupta and Kunnath, 2000; Requena and Ayala, 2000; Elnashai, 2000; Antoniou
et al., 2002; Aydinoğlu, 2003] with update force distributions that take into account the strength and
stiffness state of the building at each step.

The potential of the pushover analysis has been recognised in the last decade and it has found his way
into seismic guidelines [ATC, 1997; SEAOC, 1995; CEN 1995]. It is expected to gain more
popularity in the future and it is already included in some codes [PCM, 2003].

The purpose of the pushover analysis is to assess the structural performance by estimating the strength
and deformation capacities using a static, non-linear analysis algorithm and comparing these
capacities with the demands at the corresponding performance levels. The assessment is based on the
estimation of important structural parameters, such as global and interstorey drift or element
deformation and forces. The analysis accounts for the geometrical nonlinearities and material
inelasticities, as well as the redistribution of internal forces, Hence it provides crucial information on
response parameters that cannot be obtained with conventional elastic methods (either static or
dynamic).

Response characteristics that can be obtained with the pushover analysis include [Krawinkler and
Seneviratna, 1998]:

- Realistic force demands on potentially brittle elements, such as axial demands on columns,
moment demands on beam-to-column connections or shear forces demands on short, shear
dominated elements.

- Estimates of the deformation demands on elements that have to deform inelastically, in order to
dissipate energy.

- Consequences of the strength deterioration of particular elements on the overall structural


stability.

4
Chapter 2: Nonlinear static analysis (Pushover) −Overview

- Identification of the critical regions, where the inelastic deformations are expected to be high.

- Identification of strength irregularities in plan or elevation that cause changes in the dynamic
characteristics in the inelastic range.

- Estimates of the interstorey drifts, accounting for strength and stiffness discontinuities. In this
way, damage on non-structural elements can be controlled.

- Sequence of the member’s yielding and failure and the progress of the overall capacity curve of
the structure.

- Verification of the adequacy of the load path, considering all the elements of the system, both
structural and non-structural.

Clearly, these benefits come at the cost of additional analysis effort, associating with incorporating
all-important elements and modelling their inelastic load-deformation characteristics.

2.2 GENERALITIES

2.2.1 Description of the traditional pushover analysis


The subsequent description abides by, in a rough manner, proposals contain in the EC-8 [CEN, 1995],
Italian Building code [PCM, 2003] and NEHRP guidelines [ATC,1997]. The Nonlinear Static
Procedure (commonly known as pushover analysis) is described as follows: “A model directly
incorporating inelastic material response is displaced to a target displacement and resulting internal
deformations and forces are determined”. A certain load pattern is selected and the intensity of the
lateral load is monotonically increased. The sequence of cracking, plastic hinging and failure of the
structural components throughout the procedure is observed, until either the target displacement is
exceeded or the building collapses. The target displacement is intended to represent the maximum
displacement likely to be experienced during the expected ground motion.

The static pushover analysis has no robust theoretical background. It is based on the assumption that
the response of the multi-degree-of-freedom (MDOF) structure is directly related to the response of an
equivalent single-degree-of-freedom (SDOF) system with appropriate hysteretic characteristics. This
implies that the dynamic response of the MDOF system is determined by a single mode only and that
the shape {Φ} of that mode is constant, throughout the time-history, regardless of the level of
deformation.

Presuming that the vector {Φ} is known, the transformation factor, Γ can be computed thorough
equation 2.1.

Γ=
{Φ}T ⋅ M .{1}
(2.1)
{Φ}T ⋅ M ⋅ {Φ}
where, M is the mass vector of the MDOF system.

The force-deformation characteristics of the equivalent SDOF system can be determined from the
results of the nonlinear pushover analysis of the MDOF system by employing the equation 2.2

d * = dn Γ F * = Fb Γ (2.2)

5
Chapter 2: Nonlinear static analysis (Pushover) −Overview

where, Fb and dn are the base shear force and control node displacement of the MDOF system. The
procedure derives a base shear vs. top displacement curve for the equivalent SDOF system (Figure
2.1).

PUSHOVER CURVES OF THE MDOF AND EQUIVALENT SDOF SYSTEMS

1200000

1000000

800000
BASE SHEAR (N)

600000

400000

200000

0
0 50 100 150 200 250 300 350 400 450 500
DISPLACEMENT (mm)

MDOF System Equivalent SDOF System

Figure 2.1 Pushover curves of the MDOF and equivalent SDOF systems, after Alvarez-Botero and López-
Menjivar [2004]

From such plot an idealized elasto-perfectly plastic force-displacement relationship can be


constructed, Figure 2.2. The yield force Fy*, which represents the ultimate strength of the idealized
system, is equal to the peak force of the pushover curve of the equivalent SDOF system. The initial
stiffness of the idealized elastic-perfectly plastic system is determined in such a way that that the areas
under the actual and the idealized system force-deformation curve are equal, see Figure 2.2. Based on
this assumption, the yield displacement of the idealized SDOF system dy* is given by:

⎛ E* ⎞
d *y = 2⎜ d m
*
− m⎟ (2.3)
⎜ F y* ⎟⎠

where, Em* is the area under the equivalent SDOF system pushover curve.

EQUIVALENT SDOF SYSTEM AND BILINEAR SYSTEM

1000000

750000
Base Shear [N]

500000

250000

0
0 100 200 300 400
Displacement [mm]

Equivalent SDOF System

Figure 2.2 Pushover curves of the equivalent SDOF and bilinear systems, after Alvarez-Botero and
López-Menjivar [2004].

Although these considerations are apparently incorrect, sensitivity studies have shown that the
modification factor Γ can be considered constant for small to moderate changes in {Φ} and that rather
accurate predictions can be attained, if the structural response is dominated by the fundamental mode
[Krawinkler and Seneviratna, 1998; Lawson et al., 1994; Fajfar and Gaspersic, 1996].

6
Chapter 2: Nonlinear static analysis (Pushover) −Overview

The maximum displacement of the equivalent SDOF system, dt*, subjected to the expected ground
motion can now be found by means of elastic spectra, inelastic spectra or time-history analysis. The
expected deformation level of the MDOF structure, dt, can be estimated by the equation:

d t = Γd t* (2.4)

Several critical parameters of the procedure are worthy of consideration, namely the target
displacement, the shape of the load distribution, as well as its nature (forces or displacements).

2.2.1.1 Target displacement


The target displacement of pushover analysis should approximate the maximum level of deformation
that is expected during the design earthquake. It can be calculated by any procedure that accounts for
the effects of non-linear response on displacement amplitude.

It was explained in the previous section that it is assumed that the target displacement for the MDOF
structure can be estimated from the displacement demand of the equivalent SDOF system, through the
use of the selected shape vector {Φ} (usually corresponding to that of the fundamental mode) and
equation (2.4). Therefore, a method is sought to determine the target displacement of the SDOF
system.

For increased sophistication, dynamic time-history analysis of the SDOF model can be used,
assuming simple hysteretic rules. However, several other approaches exist. The N2 method [Fajfar
and Gaspersic, 1996] utilises the strength reduction factor R, the period T of the SDOF model and R-
µ-T relationships to calculate the ductility demand µ = dt*/d*y and , consequently, the displacement
demand dt*. Similar procedures have been presented by Krawinkler and Seneviratna [1998]. The
Capacity Spectrum Method estimates the displacement demand by comparing graphically the lateral
capacity of the system with highly damped spectra (or inelastic spectra as modified by Fajfar [1999])
in the Acceleration-Displacement Response Spectrum format.

2.2.1.2 Applied forces vs. applied displacements


Considering earthquake loading as a set of imposed energy input, ground displacements and
deformations of the structural members rather than a set of lateral forces seems a much more rational
approach. After all, the fact that earthquake input has been modelled as forces rather than
displacements can only be explained by historical reasons, related to the developments of
contemporary engineering methods in countries of low seismic hazards, like England and Germany,
where the most significant actions are the vertical gravity loads. Had the modern engineering made its
initial step in earthquake-prone regions like New Zealand, California or Southern Europe, today’s
code provisions would probably be based on deformations. Therefore, applying displacements rather
than force patterns in the pushover procedures appears to be more appropriate and theoretically
correct [Priestley, 1993].

However, displacement-based pushover analysis suffers from significant inherent deficiencies. Due to
the constant nature of the applied patterns, it can be conceal important structural characteristics, such
as strength irregularities. This is illustrated by means of an example in Figure 2.3, where the ICONS
frame model, which will be described in section 3.2.4, has been pushed to the same target
displacement with constant displacement and constant force patterns (triangular distributions).
Although the interstorey drift at the soft-storey during an earthquake is expected to be larger than the
other storeys, the displacement-based pushover yields equal drifts for all the storeys.

7
Chapter 2: Nonlinear static analysis (Pushover) −Overview

Storey
2

0
0 50 100 150 200 250
Displacement (mm)

Constant force profile Constant displacement profile

Figure 2.3 Displacement profiles of the ICONS simplified model under constant-force and constant-
displacement distributions.

Hence, to apply displacements, rather than forces, requires adaptiveness meaning to update the
displacement patterns, according to the structural properties of the analysed model, such as the
stiffness of the mass distribution. The applied displacements at every step would be determined by
modal analysis or any other method that explicitly accounts for the structural characteristics at the
current level of inelasticity, in a way that approximates the expected dynamic deformations. On one
side, such procedure would be theoretically more rigorous and match the new trends for displacement-
based design and assessment, and, alternatively, it would expose the structural weakness that are
concealed with fixed-displacement patterns and yield accurate results both at the local and the global
level [Antoniou and Pinho, 2004b].

2.2.1.3 Lateral load patterns


The lateral load patterns should approximate the inertial forces expected in the building during an
earthquake. Although, clearly, the inertia force distributions will vary with the severity of the
earthquake and with time, usually an invariant load pattern is used. This approximation is likely to
yield adequate predictions of the element deformation demands for low to medium-rise framed
structures, where the structure behaviour is dominates by a single mode. However, pushover analysis
can be grossly inaccurate for structure of larger periods, where higher mode effects tend to be
important. Moreover, Mwafy and Elnashai [2000] and Lawson et al. [1994] observed that pushover
procedures are particularly poor in predicting the response of frame-wall structures, probably due to
significant period shift and change of inertia force distribution upon yielding of the wall base.

Several investigations [Mwafy and Elnashai, 2000; Gupta and Kunnath, 2000] have found that,
whereas in the elastic range force distributions of a triangular or trapezoidal shape provide a better fit
to dynamic analysis results, at large deformations the dynamic envelopes are closer to the uniformly
distributed force solutions. The above happens after the structure has sustained significant damage at a
particular storey level, favouring a pattern akin to a SDOF system.

Since the constant distribution methods are incapable of capturing such variations in characteristics of
the structural behaviour under earthquake loading, the use of at least two different patterns has been
proposed. Various codes and guidelines [CEN, 1995; PCM, 2003; ATC, 1997] suggest that use of a
“uniform” pattern, where the lateral forces are proportional to the local masses at each floor level, and
a “modal” pattern, which is determined by a modal combination using a sufficient number of modes
and an appropriate spectral shape. Alternatively, the “triangular”, in which the accelerations are
proportional to the storey heights, rather than the “modal” pattern, may be utilized.

8
Chapter 2: Nonlinear static analysis (Pushover) −Overview

Different suggestions have been made in the past. These include the use of lateral loads proportional
to the deflected shape of the structure [Fajfar and Fichinger, 1988] or proportional to the storey shear
resistances at the previous step [Bracci et al., 1997], whereas Gupta and Kunnath [2000]. Similarly,
Requena and Ayala [2000] suggested the derivation of the forces through modal combinations using
the square root of the sum of squares (SRSS) method, taking into account a predefined number of
modes of interest.

2.3 DEVELOPMENT AND ASSESSMENT OF THE METHOD


Inelastic static analysis in Earthquake engineering has been first employed by Gulkan and Sozen
[1977] and Saiidi and Sozen [1981]. In their Q-model, Saiidi and Sozen suggested the use of the
moment-curvature relationships of the individual members to derive the top-level displacement vs.
shape moment curve of the MDOF (as opposed to the base shear that is normally used nowadays).
The curve is idealized with a bilinear curve to derive the force displacement characteristics of the
SDOF system. The deflected shape at yield is assumed to describe the characteristic vibration shape
of the building {Φ}. A similar procedure is followed by the N2 method [Fajfar and Fischinger, 1998;
Fajfar and Gaspersic, 1996] which after carrying out the pushover analysis relates the quantities Q* of
the SDOF system with the quantities Q of the MDOF structure through the equation (2.5).

Q* = Q Γ =
{Φ}T ⋅ M ⋅ {Φ} ⋅ Q = ∑ mi ⋅ Φ i2 ⋅ Q
(2.5)
{Φ}T ⋅ M ⋅ {1} ∑ mi ⋅ Φ i
and it uses R-T-µ relationships to calculate the displacement demands.

A number of publications has recently reviewed the merits and deficiencies of the method. Lawson et
al. [1994] discuss in some detail the range of applicability and the expected realism for various
structural systems, and highlight the encountered difficulties. Four steel structures, which heights vary
from two to 15 storeys, have been analysed using DRAIN-2DX under three different patterns of static
loading (uniform, triangular and modal using the SRSS method to combine the modal shapes and their
spectral amplifications). The results were compared against dynamic analysis using seven different
earthquake records. The correlation between static and dynamic responses was good for the two and
five storey buildings. However, the predictions of the static analysis were inadequate for the 10 and
15 storey buildings. The authors concluded that pushover analysis might be grossly inaccurate for tall
buildings, where high mode effects are important. Surprisingly, the pushover analyses with the modal
distribution yielded unacceptably poor results, which has been attributed to an exaggeration of higher-
mode effects by the SRSS combination method. Further, unsuccessful attempts to correlate the
hysteretic energy of the dynamic analysis with the area under the pushover curve showed that this
area is a poor indicator of cumulative damage effects. The paper also discusses the issue of the target
displacement, at which the building should be pushed. The procedure proposed by Qi and Moehle
[1991] and Miranda [1991] is presented together with a simple method, whereby the target
displacement of the MDOF system is calculated from its fundamental period and the spectral ordinate
corresponding to it. Both methods provided good estimates of the target displacements. The authors
close their paper with some recommendations on why, when and how to use pushover analysis. It is
advisable to use the method, when there is doubt about the efficiency of simple elastic code-based
procedures, for the seismic evaluation of existing structures and the design of retrofit schemes, but
always with great care and good engineering judgment for the interpretation of results.

On the same basis, Faella [1996] compares the response of three, six and nine storey buildings
subjected to artificial and real earthquakes with pushover analysis, and concludes that static analysis
can, indeed, identify collapse mechanisms and critical regions, yielding reasonable estimates for the
interstorey drifts. Although the effect of different load patterns is not investigated, confining the
observations to the triangular distribution, the importance of accurate determination of the target
displacement is stressed. It is also suggested that, since the sum of the maximum dynamic drifts is

9
Chapter 2: Nonlinear static analysis (Pushover) −Overview

larger than the maximum dynamic roof displacement, due to the random nature of earthquake loading,
“it is advisable to compute the static interstorey drift for a maximum roof displacement higher than
the dynamic one”. Finally, the author points out difficulties with static-dynamic comparison when the
strong-motion input is rich in long period frequencies.

In the course of describing recent trends in seismic design methodology, Krawinkler [1995] presents
pushover analysis as a simplified performance evaluation method that can also be used as a design
tool. He discusses the theoretical limitations of the method, as well as the procedures for the
estimation of the targer displacement, and emphasises the fact that pushover analysis “cannot disclose
performance problems cused by changes in the inelastic dynamic characteristics, due to higher mode
effects”. The latter is one of the most significant problems of conventional static procedures, as
discussed in previous sections, but can be solved or mitigated with the application of adaptive
patterns.

Furthermore, Krawinkler and Seneviratna [1998] summarise the basic concepts of the method
stressing that its theoretical background is not rigorous, being based on the assumption that the
MDOF response is related to the response of a SDOF oscillator. The conditions, under which
pushover analysis can provide adequate information, are identified, and the important issues of the
target displacement and the lateral load pattern are discussed. Like Lawson et al. [1994] the authors
suggest that use or more than one invariant load pattern, and consider the application of adaptive
pattern rather attractive. Moreover, they mention that the selection of the load pattern is more
important than that of the target displacement and that to their belief “the load pattern issue is the
weak point of the pushover analysis procedure”. Presenting an example of a successful pushover
analysis they discuss the limitations of the method, cases in which the pushover predictions are
inadequate or even misleading, and suggest areas of future development. Not surprisingly, “the higher
mode effects, once a local mechanism has formed,” is considered the most important problem to be
solved.

Ken and D’Amore [1999] assess the accuracy of the method, in comparison with inelastic time-
history procedures using DRAIN-2DX to analyse an instrumented six storey steel building built in
1977. They conclude, as expected, that not all dynamic analyses of the same structure under a set of
different earthquake records are predicted by pushover analysis. The roof displacement vs. base shear
curve is considered too simplistic and inadequate, as it cannot describe the dynamic nature of the
response during earthquakes, especially when a predefined and fixed transverse load vector is used.
However, the superiority of the method compared to the code-based procedures is recognised.
Moreover, as an improvement the authors suggest the use of pushover analysis supplemented by the
use of 3-D nonlinear dynamic analysis.

Finally, Naeim and Lobo [1999] present the common mistakes committed during a pushover analysis.
Amongst other issues, the authors discuss the importance of the loading shape function, the selected
performance objectives, P-∆ effects and gravity loading, shear failure mechanisms and the post peak
behaviour. In particular, for the latter, they note that “if an analysis package cannot model structural
failure, then either the push has to be stopped at the onset of the first hinging, or extreme care must be
exerted on the interpretation of the post-failure behaviour as reported by the program”, mentioning
that most of the programs are not capable to model adequately the post peak response.

As already mentioned, there are good reasons for using pushover analysis rather than simplified
elastic methods for estimating the deformation demands. Moreover, the simplicity of the method
makes it a more attractive approach for everyday practice than nonlinear time-history analysis.
Traditional pushover analysis can be extremely useful tool, if used with caution and acute engineering
judgment, but, as discussed in the previous sections, it exhibits significant shortcomings and
limitations, which are summarised below:

10
Chapter 2: Nonlinear static analysis (Pushover) −Overview

1. The theoretical background of the method is not robust and it is difficult to defend. As mentioned
earlier, an important implicit assumption behind pushover analysis is that the response of a multi-
degree-of-freedom structure is directly related to an equivalent single-degree-of-freedom system.
Although in several cases the response is dominated by the fundamental mode, this can by not
means be a generalised statement. Moreover, in dynamic time-history analysis the shape of the
fundamental mode itself may vary significantly depending on the level of inelasticity and the
location(s) of damage.

2. As a consequence of point 1, the deformation estimates obtained from a pushover analysis may be
highly inaccurate for structures where higher mode effects are significant. The method, as
currently prescribed in codes and seismic guidelines, explicitly ignores the contribution of the
higher modes to the total response. In the cases where this contribution is significant, the
pushover estimates may be totally misleading.

3. It is difficult to model three-dimensional and torsional effects. Pushover analysis is very well
established and has been extensively used with 2-D models. However, little work has been carried
out for problems that apply specifically to asymmetric 3-D systems, with stiffness or mass
irregularities. Therefore, it is by no means clear how to derive the load distributions and how to
calculate the target displacement for the different frames of the building. Moreover, there is no
consensus regarding the application of the lateral force in one or both horizontal directions.

4. Traditional pushover analysis is still force-based, due to the inability of displacement-based


pushover to capture important structural weakness, such as strength irregularities. However, as
discussed in 2.2.1.2, it would be theoretical more rigorous to find a way to apply displacement
distributions that are appropriately updated at different deformation levels since displacement are
better related to structural damage.

5. The progressive stiffness degradation that occurs during the cyclic non-linear earthquake loading
of the structure is not taken into account. This degradation leads to changes in the periods and the
modal characteristics of the structure that affect the loading attracted during earthquake ground
motion.

6. Being a static method, pushover analysis concentrates on the strain energy of the structure,
neglecting other sources of energy dissipation, which are associated with the dynamic response,
such as the kinetic and the viscous damping energy. Moreover, it neglects duration effects and
cumulative energy dissipation demand.

7. Only horizontal earthquake load is considered. The vertical component of the earthquake loading,
which can be in some cases of great importance is ignored, since no method has been proposed up
to now on how to combine pushover analysis with actions that account for the vertical ground
motion.

8. A separation between the supply and the demand is implicitly in the method. This is clearly
incorrect, as the inelastic structural response is load-path dependent and the structural capacity is
always associated to the earthquake demand.

Obviously, pushover analysis lacks many important features of dynamic non-linear analysis and it
will never be a substitute as the most accurate tool for structural analysis and assessment.
Nevertheless, several possible developments can considerably improve the efficiency of the method.
Indeed, from the summary of pitfalls short listed above, points 1 to 5 can be overcome with the
derivation of a fully adaptive procedure, which accounts for both higher mode contributions and the
alterations of the local resistance and the modal characteristics, once a local mechanism has formed.
In this way, the stiffness degradation, the period elongation and the higher mode effects can be

11
Chapter 2: Nonlinear static analysis (Pushover) −Overview

explicitly considered. Moreover, finding a way to incorporate, somehow, the expected ground motion
in the analysis will provide site-specific results that apply to certain areas with particular seismic
hazard characteristics. This can be achieved with the utilisation of spectra representative of these
areas. Finally, the application of adaptive displacement rather than force patterns will provide a both
conceptually appealing and accurate tool to replace “traditional” force-based approaches.

12
Chapter 3: Description of case studies

3. DESCRIPTION OF CASE STUDIES

3.1 INTRODUCTION
In order to assess different pushover methodologies two buildings have been selected, RM15 and the
ICONS frame. The former has been already used in previous research [Antoniou, 2002] and the latter
is a model of a full scaled structure constructed for pseudo dynamic testing [Carvalho et al., 1999].
Subsequently, a detailed description of the analytical tool, the modelling approach as well as the
nonlinear dynamic analyses used is presented.

3.2 MODELLING

3.2.1 Analytical tool


The finite element analysis program SeismoStruct [SeismoSoft, 2004] is utilized to run all analysis.
The structural package is able to predict the large displacement behaviour of space frames under static
or dynamic loading, taking into account both geometric nonlinearities and material inelasticity.
SeismoStruct accepts static loads (either forces or displacements) as well as dynamic (accelerations)
actions and has the ability to perform eigenvalues, nonlinear static pushover (conventional and
adaptive), nonlinear static time-history analysis, nonlinear dynamic analysis and incremental dynamic
analysis.

3.2.2 Modelling of the members


Structural members have been discretised by using a beam-column model based on distributed
plasticity-fibre element approach. The model takes into account geometrical nonlinearity and material
inelasticity. Sources of geometrical nonlinearity considered are both local (beam-column effect) and
global (large displacement/rotation effects). Since a constant generalized axial strain shape function is
assumed in the adopted cubic formulation of the element, it results that its application is only fully
valid to model the nonlinear response of relatively short members and hence a number of elements
(usually three to four per structural member) is required to accurate model the structural frame
members. Material inelasticity is explicitly represented through the employment of a fibre modelling
approach which allows for the accurate estimation of structural damage distribution, the spread of
material inelasticity across the section area and along the members length. In the fibre model the
sectional stress-strain state or beam-column elements is obtained through the integration of the
nonlinear uniaxial stress-strain response of the individual fibres in which the section has been
subdivided. If a sufficient number of fibres is employed, the distribution of material nonlinearity
across the section area is accurately modelled, even in the highly elastic range, see Figure 3.1

13
Chapter 3: Description of case studies

R C S ectio n U n co n fin ed C o n fin ed S te el F ib res


C o n crete F ib res C o n crete F ib res

Figure 3.1 Fibre discretisation in a reinforced concrete section.

The spread of inelasticity along member length then comes as a product of the inelastic cubic
formulation suggested by Izzudin [2001]. Two integration Gauss points per element are used for the
numerical integration of the governing equations of the cubic formulation, Figure 3.2. If a sufficient
number of elements is used the plastic hinge length of structural members subjected to high levels of
material inelasticity can be accurately estimated.

Figure 3.2 Location of Gauss points along the member length.

It is worth mentioning that shear strains across the element cross section are not modelled; in addition,
warping strains and warping effects are not considered in the current formulation, either. Additionally,
the elastic torsional rigidity is used in the formulation of the nonlinear frame elements; this clearly
involves some degree of approximation for the case of reinforced concrete sections.

No viscous damping was considered in any dynamic analysis, since energy dissipation through
hysteresis is already implicitly included within the nonlinear fibre model formulation of the inelastic
frame elements, and non-hysteretic type damping was assumed to be negligible within the scope of
the present endeavour.

3.2.3 Case study one: RM15


The description of RM15 has mainly been taken from Antoniou and Pinho [2004a] and Mwafy
[2001]. The RM15 model is part of a set of structures that three different structural configurations
were employed: a 12 storey regular frame, an eight storey irregular frame and a dual wall-frame
system. Additionally, different ductility classes and design ground acceleration were considered,
resulting in a total of 12 models. The latter represent common reinforced concrete structures and are
based on buildings designed and detailed at the University of Patras [Fardis, 1994], seemingly
according to the 1995 version Eurocode 8 [CEN, 1995]. Subsequently, they were modelled by Mwafy
[2001] under a framework of different project, and were then adapted by Rovithakis [2001]. The
Rm15 belongs to the group of regular buildings and their general characteristics are defined in Table
3.1.

14
Chapter 3: Description of case studies

Table 3.1 Definition of the considered structural system, after Antoniou and Pinho [2004a]

Structural Number of Structure Ductility Design Behavior Tuncracked


System Storeys Reference Level PGA (g) Factor (q) (s)
RH30 High 5.00 0.697
0.30
Regular RM30 Medium 3.75 0.719
12
Frame RM15 Medium 3.75 0.745
0.15
RL15 Low 2.50 0.740

The geometric characteristics of the structural system is illustrated in Figure 3.3. The overall plan
dimensions is 15m by 20m. The total height is 36 metres, with equal storey heights of 3m. The lateral
force resisting system is a moment frame, wall-frame sub-set has both a central core extending over
the full height and moment frames on the perimeter. The floor system is solid slab. The member
cross-section sizes are given in Table 3.2.

Table 3.2 Member cross-sectional dimensions for regular buildings, after Mwafy [2001]
Columns Beams b × h
SR X-dir. X-dir. Slab
Internal External Corner Z-dir.
(1st storey) (Long.)
RH30 .80×.80 .70×.70 .70×.70 .35×.65 .35×.65 .35×.65 0.14
RM30 .80×.80 .70×.70 .70×.70 .35×.60 .35×.60 .35×.60 0.14
RM15 .80×.80 .70×.70 .70×.70 .30×.60 .30×.60 .30×.60 0.14
RL15 .80×.80 .70×.70 .70×.70 .30×.60 .30×.60 .30×.60 0.14
12 x 3.0 = 36.0 m

5 x 4.0 = 20.0 m

Z
3 x 5.0 = 15.0 m

Solid slabs

Figure 3.3 Plane and sectional elevation of the 12-storey regular frame building set, after Mwafy [2001]

Mean values of material strengths are utilised in the analyses rather than the characteristics values
used in the design, an approach consistent with ‘assessment’. These values are presented in Table 3.3.

15
Chapter 3: Description of case studies

Table 3.3 Material properties used in the assessment, after Mwafy [2001]
Material parameter Values used in analysis
Mean compressive strength, fcm 33 N/mm2
Concrete grade
Mean tensile strength, fct 2.6 N/mm2
C25/30
Crushing strain, εc 0.0022
Modulus of elasticity, Ec 30.5 kN/mm2

Yield strength, fy 585 N/mm2


Steel S500

Ultimate strength, fu 680 N/mm2


Ultimate strain, εsu 0.094
Young’s modulus, Es 200 kN/mm2

Reinforced concrete column-section and T-section are used for modelling of columns and beams,
respectively. The contribution of the slab width to the beam has a significant effect on the stiffness
and hence on the overall response of the buildings. Several values for the flange width are
recommended by seismic codes. Due to the fact that the participation of the slab to the beam stiffness
is less than the participation to the flexural strength, as a result of the moment reversal and the low
contribution of the flange in tension [Paulay and Priestley, 1992]. It is suggested [ACI 318, 1995] to
use an effective slab width equal to one half of that recommended for gravity load design. It is also
recommended by EC8 to reduce the effective slab width for buildings subjected to seismic forces due
to inelastic effects. The beam width plus 7% of the clear span of the beam on either side of the web is
the effective flange width that is adopted in the model employed in the design of the buildings [Fardis,
1994] as well as in the current analysis. This provides values between the conservative flange width of
EC8, which is intended for design purposes, and the full slab or the width recommended for gravity
load design.

On the structure level 1064 elements are used to model the 12-storey regular frame. This includes the
number of cubic elasto-plastic elements, the joint elements connecting the frames in the orthogonal
direction, the cubic elastic elements and the shear spring elements, as subsequently discussed. On the
member level, seven elements are used to model each beam. Three of them are elasto-plastic elements
representing the beam between the faces of the columns. The lengths of these elements are determined
in accordance with the distribution of transverse and longitudinal reinforcement specified in the
design. On the basis of the arrangement of the transverse reinforcements, the confinement factors for
the cross-section of these elements are evaluated using the mean values of material strengths. Two
rigid elements are utilised to connect the beam ends with the framing columns (the length between the
face and the centreline of the vertical elements, as shown in Figure 3.4(a).

16
Chapter 3: Description of case studies

Zero length joint Beam critical


Shear spring Column lengths Column
elements connecting a
the frames in with zero
the orthogonal length
direction
External Frames
Rigid arms

(a) Beam column connection


Art-rec1 Internal Frames
0 10 sec
b
Kobe Gauss sections
0 15 sec
Concrete and
steel layers
Loma Prieta
0 15 sec

Regular frame building 0.21 L 0.58 L 0.21 L


L
(b) Cubic elasto-plastic element

c
B1 Unconfined concrete fibre
Global Y B2
Global Z d2 d1
Monitoring
Steel
point
D1D2 fibres
Confined
Global X concrete
b2 fibre
b1

(c) Decomposition of beam T-section into fibres


Irregular frame building

Figure 3.4 Adopted modelling approach, after Mwafy [2001]

Under earthquake load, beam-column joints are subjected to high shear stresses that could lead to
diagonal cracking and significant shear deformation. The total joint deformations can be considerable
as a consequence of this cyclic load. Typically, 20% of the interstorey drift due to earthquake forces
may be originate from joint deformations [Paulay and Priestley, 1992]. Towards this end, two shear
spring elements are introduced in the present study to represent the shear stiffness of the beam-column
connection. A simple linear elastic force-deformation relationship is utilised to calculate the shear
stiffness of the joint. On the other hand, the structural mesh employs three cubic elasto-plastic
elements for modelling of each vertical element, with the exception of the core. The method adapted
to model the core is described below.

Gravity loads are applied as point loads at beam nodes. To account for inertia effects during dynamic
analysis, masses are distributed in the same pattern adopted for the gravity loads and are represented
by lumped 2D mass elements. The numerically-dissipative Hilber-Hughes-Taylor α-integration
scheme is utilised to integrate the equations of motion.

According to the data used in the design [Fardis, 1994], the following static loads per unit area are
considered to calculate the total gravity loads on the frames,
-
Slab self-weight, Wf = 2.0 kN/m2
- Live load, Q = 2.0 kN/m
Using the appropriate coefficients, Ψ2 and φ, from the design code, the vertical loads are combined
with seismic actions by applying the following rules:
- At top floor: 1.0 G + 0.3 Q + EL
- For other storeys: 1.0 G + 0.15 Q + EL
where, G is the dead load and EL is the seismic load. Gravity loads are applied as point loads at nodes.
To account for inertia effects during dynamic analysis, masses are calculated in a manner consistent

17
Chapter 3: Description of case studies

with the gravity loading combinations and represented by a lumped 2D mass element. Table 3.4
summarises the total gravity loads and mass values for different storeys of the building.

Table 3.4 Calculated floor gravity loads and masses

Structure Gravity load (kN)


Group Total mass
reference st
1 floor Other Top floor Total
RH30 3199.96 3199.96 2900.04 38099.60 3883.8
RM30 3147.96 3147.96 2847.92 37475.48 3820.1
1
RM15 3058.24 3058.24 2758.24 36398.88 3710.4
RL15 3058.24 3058.24 2758.24 36398.88 3710.4

3.2.4 Case study two: The ICONS frame


The second case study is a four-storey bare frame known to have a soft storey at the third floor. The
latter is attributed to a drastic change in strength and stiffness at this level through a reduction in both
the reinforcement content and the section dimensions in the columns between the second and third
storeys, coinciding also with the location of lap-splicing. Such characteristics are common in
buildings designed predominantly for gravity-loading and the failure of a storey mid-way up a
building has been observed in past devastating earthquakes such as the Kobe earthquake of 1995 (e.g.
EERI, 1997).

The four-storey, three-bay reinforced concrete bare frame, Figure 3.5, was design and built at the
European Laboratory for Structural Assessment (ELSA) of the Joint Research Centre (JRC) at Ispra,
Italy. The full-scale model was constructed for pseudo-dynamic testing, under the auspices of the EU-
funded ECOEST/ICONS programme. The frame was designed essentially for gravity loads and a
nominal lateral load of 8% of its weight. The reinforcement details attempted to mirror the
construction practices used in southern European countries in the 1950’s and 1960’s.

The columns were non ductile, smooth reinforcing bars were used, capacity design principles were
ignored and lap splicing occurred in critical regions, Figure 3.6. Detailed information about the frame,
testing and set up of the experimental rig can be found at Carvalho et al. [1999] and Pinho and
Elnashai [2000].

18
Chapter 3: Description of case studies

Figure 3.5 Elevation and plan views of the frame, after Carvalho et al. [1999]

Figure 3.6 Reinforcement detail of the columns, after Carvalho et al. [1999]

The materials considered at the design phase were a low strength concrete class C16/20 (CEN, 1991)
and smooth reinforcement steel class Fe B22k (Italian standards). The latter refers to smooth bars with
a yield stress of 235 MPa and ultimate strength of 365 MPa.

The vertical loads considered in the design consisted of the self-weight of the slab and transverse
beams, finishes, infill walls and the quasi-permanent static load. Figure 3.7 shows the scheme of
vertical loads applied to the structure.

19
Chapter 3: Description of case studies

Figure 3.7 Scheme of vertical loads for nonlinear analysis, after Carvalho et al. [1999]

In total, 112 inelastic frame elements, capable of representing progressive cracking and spread of
inelasticity, are used to model the bare frame. All member sections are represented explicitly and, for
the purpose of strain/stress evaluation, are subdivided into a number of fibres (200-300) varying
according to the section size. The length of the elements varies according to their location, with
smaller elements being used in the vicinity of beam-column connections where large levels of
inelastic deformation are expected.

All connections are assumed rigid and fully-fixed boundary conditions are adopted at the base of the
building. Effective slab widths of 1.0 m and 0.65 m were adopted for the long and short spans,
respectively, following the formulae presented in Eurocode 2 [CEN, 1991].

Vertical loads and masses are applied at each beam node and at the beam-column joints mirroring the
load and mass distribution presented previously in Figure 3.7.

Concrete is represented by a uniaxial constant confinement model (Mander et al., 1988) and is
calibrated using the concrete characteristics values obtained during testing. As for the reinforcement
bars, the Menegotto-Pinto [1973] steel model, with an isotropic hardening constitutive relationship
(Filippou et al. 1983), was adopted.

3.3 NONLINEAR DYNAMIC ANALYSIS

3.3.1 Case study one: RM15


Four input time-histories, consisting of one-artificially generated accelerogram [Campos-Costa and
Pinto, 1999] and three natural records (Loma Prieta earthquake, USA, 1989), were employed for the
dynamic analysis of the study carried out by Antoniou [2002]. The selection of these four records
aimed at guaranteeing that all models would be subjected to a wide-ranging type of earthquake action,
in terms of frequency content, peak ground acceleration, duration and number of high amplitude
cycles. Indeed, the original PGA of the records varies between 0.12g and 0.93g, the spectral shapes
are markedly distinct and provide high amplifications at different periods, Figure 3.8, and the ratio
between the significant duration (defined as the interval between the build up of 5% and 95% of the
total Arias Intensity [Bommer and Martinez-Pereira, 1999]) and the total duration ranges from 22% to
72%. The characteristics of the records are summarised in Table 3.5, while their elastic response
spectra, for an equivalent viscous damping of 5%, are shown in Figure 3.8.

20
Chapter 3: Description of case studies

Table 3.5 Characteristics of the records employed by Antoniou and Pinho [2004a]
Total Significant
PGA PRA 5% AI 95% AI
Record Duration Duration teff/ttotal
(g) (g) Threshold Threshold
ttotal teff
A975 0.30 1.28 2.32 s 12.75 s 15.0 s 10.43 s 69.5%
Emeryville 0.25 0.90 11.23 s 20.16 s 40.0 s 8.93 s 22.3%
Hollister 0.12 0.50 1.02 s 9.52 s 33.2 s 8.50 s 25.6%
Loma 0.93 4.25 1.44 s 8.68 s 10.0 s 7.24 s 72.4%

1 1

0.8 0.8
Acceleration (g)

Acceleration (g)
0.6 0.6

0.4 0.4

0.2 0.2

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

(a) Record A975 (b) Record Emeryville

0.5 3

2.5
0.4

2
Acceleration (g)
Acceleration (g)

0.3

1.5

0.2
1

0.1
0.5

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

(c) Record Hollister (d) Record Loma


Figure 3.8 Elastic response spectra of the four records (5% equivalent viscous damping).

3.3.2 Case study two: The ICONS frame

3.3.2.1 Earthquake input


A collection of artificial records was available for use in the pseudo-dynamic experimental
programme. The records were derived following a probabilistic seismic hazard analysis carried out at
JRC (Campos-Costa and Pinto, 1999). For the purpose of the experimental programme, the severity
class “Moderate-High”, typical of Southern European countries, was chosen. A set of hazard-
consistent time histories was artificially generated to fit the uniform risk spectra (URS) for return
periods of 100, 475, 975 and 2000 years.

In Figure 3.9, the acceleration and displacement elastic response spectra for all accelerograms,
computed for an equivalent viscous damping of 5%, are shown. These indicate peak acceleration
response for periods of vibration of up to 0.5 seconds, with values ranging from 0.3g to 1.1g. Figure
3.10 shows the artificial Acc-475 (475 years return period) and Acc-975 (975 years return period)
accelerograms; only these two records will be considered in the nonlinear dynamic and pushover

21
Chapter 3: Description of case studies

analyses carried out herein. As is common with artificial records, a wide range of frequencies is
present in the accelerograms.
1.20 300
2000 years
975 years
1.00 250 475 years
100 years

Displacement [mm]
0.80
Acceleration [g]

200
2000 years
0.60 975 years 150
475 years
100 years
0.40 100

0.20 50

0.00 0
0 1 2 3 4 0 1 2 3 4
Period [s] Period [s]

(a) (b)
Figure 3.9 Respose spectra of input motion: (a) acceleration; (b) displacement.

0.3 0.4

ACC-475 ACC-975
0.3
0.2

0.2
0.1
Acceleration (g)
Acceleration (g)

0.1
0.0
0.0

-0.1
-0.1

-0.2
-0.2

-0.3 -0.3
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Time (s) Time (s)

(a) (b)
Figure 3.10 Artificial acceleration time histories for (a) 475 year (Acc-475) and (b) 975 year (Acc-975)
return period

3.3.2.2 Comparison of experimental and analytical results


The bare frame specimen of Figure 3.5 was subjected to a pseudo-dynamic test using the Acc-475
input motion and subsequently to a second test carried out with the Acc-975 input motion. The results
of the tests showed that the deformation demand concentrated in the 3rd storey for the Acc-475 test
and that collapse of the 3rd storey was almost reached for the Acc-975 test (Pinho and Elnashai, 2000).

Nonlinear dynamic analyses of the model described in Section 3.2.4 have been carried out for the two
records by placing them consecutively in order to reproduce the testing sequence and better predict
the behaviour of the frame. The top frame displacements, obtained for the experimental test, for the
Acc-475 record and the Acc-975 record are shown in Figure 3.11a and Figure 3.11b, respectively. In
addition, the top displacement response obtained by numerical simulation of each test has been
superimposed. It may be concluded that the analytical prediction agrees with the results obtained
through testing.

Inter-storey drift is a crucial parameter in terms of structural response since it is closely related to the
damage sustained by buildings during seismic action. The soft-storey in the test case frame can only
be identified through observation of the inter-storey drift profile. Figure 3.12 shows the drift profiles
at the peak displacement from both the analytical and experimental time-histories; it is clear that the
analytical model is able to predict the soft-storey at the third floor. Further refinement of the
analytical model could perhaps produce a closer match between the analytical and experimental drift
profiles but this is not within the scope of the present exercise. It suffices that the nonlinear fibre

22
Chapter 3: Description of case studies

analysis can predict the soft-storey at the third floor and thus this will be the reference to which all
other analytical analyses will be compared. Note that no post-test calibration has been carried out. The
structure was modelled as it is. The differences between Analysis versus experiment are likely to be
due to the fact the third storey developed a shear failure mechanism, not yet incorporated in the
program used.
TOP LEVEL DISPLACEMENTS TOP LEVEL DISPLACEMENTS

100 150

100

50

DISPLACEMENTS (mm)
DISPLACEMENTS (mm)

50

0 0

-50

-50
-100

-100 -150
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
TIME (S) TIME (S)
EXPERIMENTAL ANALYTICAL EXPERIMENTAL ANALYTICAL

(a) (b)
Figure 3.11 Analytical and experimental top frame displacement: (a ) Acc-475; (b) Acc-975

4 4

3 3

Analytical Analytical
Storey

Storey

2 2
Experimental Experimental

1 1

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6

Drift (%) Drift (%)

(a) (b)
Figure 3.12 Analytical and experimental drift profiles: (a ) Acc-475; (b) Acc-975

There are two drift profiles, which may be obtained from the dynamic analysis: the profiles
corresponding to the time step when the maximum inter-storey drift at any level is reached and the
envelope of maximum drift in all storeys. The former profile corresponds to a real state of the
structure, structural equilibrium is thus conserved and so comparison of this profile with a pushover
would seem more logical. The latter envelope is a combination of drifts occurring at different time
steps and so is not a state within which the structure ever exists, however, it is more useful from a
design or assessment point of view and so it would be desirable to obtain a pushover that could
capture such a profile giving the maximum drift at all storeys.

23
Chapter 4: Description and assessment of pushover methodologies

4. DESCRIPTION AND ASSESSMENT OF PUSHOVER


METHODOLOGIES

4.1 INTRODUCTION
In the following sections different pushover approaches, ranging from traditional to adaptive, will be
presented. At the same time, the performance of selected pushover methodologies will be assessed.
The evaluation will be performed by using the RM15 and the ICONS frames. Additionally, both case
studies will be employed to examine the performance of the direct vectorial addition, DVA, when
applied to the adaptive pushover method followed by Antoniou and Pinho.

4.2 MEASURING THE ACCURACY OF THE PUSHOVER TECHNIQUES


Note that, as showed by Antoniou and Pinho [2004a], it is the drift profile, rather than the so-called
pushover curve, that should be employed in comparison analyses. Because structural damage is
directly related to local deformations, rather than to forces, the interstorey drift has been selected as
the parameter to compare the different analyses. The inter-storey drift profiles obtained from each
pushover analysis are to be compared to the drift profiles from the nonlinear dynamic analysis and the
standard error of the pushover results, with respect to the dynamic, calculated.

The standard error of the profile is found using the following formula:

1 n ⎛ ∆ iD − ∆ iP ⎞
Error (%) = 100 ∑⎜ ⎟ (4.1)
n i =1⎜⎝ ∆ iD ⎟

where, ∆iD is the inter-storey drift at a given level i from the dynamic analysis, ∆iP is the
corresponding inter-storey drift from the pushover analysis and n is the number of storeys. The closer
the standard error is to zero, the closer the pushover response is to the nonlinear dynamic response.

For the case of RM15, the inter-storey drift profiles obtained from each pushover analysis are to be
compared to the drift profiles from the nonlinear dynamic analysis and the standard error of the
pushover results, with respect to the dynamic, calculated. The envelope of maximum drift in all
storeys will be the interstorey drift profile to be employed. This envelope is a combination of drifts
occurring at different time steps and so is not a state within which the structure ever exists; however,
it is more useful from a design or assessment point of view. The profile has been obtained at 2.5%
total drift since at this percentage of global drift the structure is already deep within the inelastic
range. The ground motion record employed to perform the dynamic analysis is Hollister. The selected
record was deliberately chosen as that for which all the force-based variants, adaptive or fixed, failed
to provide acceptable response predictions. Due to the effects of higher modes, highly amplified by
the selected ground motion, the dynamic drifts tend to increase from the structure’s mid-height
upwards, a feature that all force-based static techniques struggle to reproduce

24
Chapter 4: Description and assessment of pushover methodologies

Additionally, for the ICONS frame case, the results of selected pushover analyses presented herein
will be compared with both of the profiles described in section 3.3.2. For the first profile mentioned
previously, the drift at each storey will be found from the pushover analysis at a roof displacement
level corresponding to the roof displacement at the time step when the maximum inter-storey drift in
the dynamic analysis is reached. For the second type of drift profile, the drift at each storey will be
obtained from the pushover analysis at a roof displacement corresponding to the maximum roof
displacement obtained during the dynamic analysis.

4.3 NON-ADAPTIVE NON-MODAL PUSHOVER ANALYSIS


The current state of the practice e.g. FEMA 356 (ASCE, 2000) requires a pushover curve to be
computed by applying monotonically increasing lateral loads to the structure with an invariant vertical
profile. The shape of this profile might follow a triangular or uniform distribution. Such pushover
analyses have been carried out on the RM15 frame and are compared with the drift profile described
previously, Figure 4.1. Immediately, it is apparent that both the triangular and uniform pushover
analyses fail to estimate dynamic drift profile. The drift at the lower storeys are largely overestimated
while the upper ones are underestimated. The standard errors are presented in Table 4.1 wherein it
may be noted that the triangular pushover analysis produces lower standard errors, however, they are
still of an unacceptable magnitude.

2.50%
12
DYNAMIC
TRIANGULAR
10 UNIFORM

8
STOREY

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.1 Comparison of drift profiles from triangular and uniform pushovers with the envelope of
maximum inter-storey drifts from dynamic analysis for the RM15 frame

Table 4.1 Standard error of the pushover schemes in Figure 4.1

Standard error

Uniform 51.07

Triangular 91.60

Considering the problems and limitations of constant force distributions, the use of enhanced adaptive
patterns, which are updated at every step to account for the progressive stiffness degradation that
occurs during the actual earthquake loading, seems an attractive alternative. In this way, the changes
in the modal characteristics, the period elongation and the different spectral amplifications can be
considered.

25
Chapter 4: Description and assessment of pushover methodologies

4.4 NON-ADAPTIVE MODAL PUSHOVER ANALYSIS


In an attempt to consider higher mode effects, Paret et al. [1996] and Sasaki et al. [1998] suggested
the simple, yet efficient, multi-mode pushover procedure (MMP) illustrated in Figure 4.2. This
method comprises several pushover analyses under forcing vectors representing the various modes
deemed to be excited in the dynamic response.

The individual pushover curves are converted to the Acceleration-Displacement Response Spectrum
(ADRS) format and the Capacity Spectrum Method is utilised to compare the structural capacity with
the earthquake demand. In this way, it becomes apparent which mode is more critical and where
damage is likely to occur. The procedure is intuitive, and does indeed provide qualitative information
and identify potential problems due to higher modes that conventional single mode pushover analysis
fails to highlight. However, the effects of higher modes cannot be easily quantified, since the method
does not provide estimation of the response.

Figure 4.2 The multimodal pushover procedure [after Paret et.al., 1996]

A refinement of the multi-mode pushover procedure is the PRC method (Pushover Results
Combination), which was proposed by Moghadam [2002]. According to this method, the maximum
seismic response is again estimated by combining the results of several pushover analyses, which are
carried out using load patterns that match the modal shapes of a predefined number of modes. The
final structural response is calculated form the pushover analyses results, according to the following
combination rule:

N
R = ∑ β i ⋅ Ri (4.2)
i =1

where R is the final estimation of the response at a particular location, βi is the modal participation
factor of mode i and Ri is the value of the corresponding response from the pushover analysis that
utilises a load pattern with the shape of mode i. Usually, the first three or four modes are considered.

A similar procedure, also based on MMP, is the MPA (Modal Pushover Analysis) method suggested
by Chopra and Goel [2002]. The procedure consists of the following steps:

1. Perform eigenvalue analyses to compute the natural frequencies and modes φn.

2. For a selected number of modes, derive the top displacement vs. base shear curve, employing
force distributions sn* according to sn*=m.φn, where m is the mass matrix of the structure.

26
Chapter 4: Description and assessment of pushover methodologies

3. Idealise the pushover curve as bilinear curve (Figure 4.3)

4. Convert the idealized curves to the Fsn/Ln – Dn relation; where,

Fsny V bny u rny


= , and Dny = .
Ln M n* Γnϕ rn

Mn* = LnΓn is the effective modal mass


N
∑ mk ϕ kn N
k =1
Γn =
N
, and Ln = ∑ mk ϕ kn
2 k =1
∑ mk ϕ kn
k =1

5. Compute the peak deformation of the nth inelastic SDOF system through response history or
spectral analysis. The elastic period of vibration of the nth system is:

Ln Dny
Tn = 2π
Fsny

6. Calculate the peak roof displacement associated to the nth mode.

7. From the pushover analysis extract values of the desired response rn (floor displacements, storey
drifts, plastic hinge rotations, etc.) at the peak roof displacement.

8. Determine the total demand by combining the peak modal response using the SRSS rule. Use as
many modes as required for sufficient accuracy. Typically, two or three modes are enough,
according to the authors.

Figure 4.3 Properties of the nth mode inelastic SDOF system derived from the corresponding pushover
curve [after Chopra and Goel, 2001]

The researches further stated that when compared to inelastic dynamic analysis the MPA gave rather
good estimates for global response parameters, such as interstorey drifts or floor displacements. The
results were superior to the results of the pushover analysis with the fixed force distributions
suggested by FEMA, which greatly underestimated drift demands and led to unacceptable large
errors. However, it was found inadequate to compute local response quantities, such as plastic
rotations, although this has been a characteristic problem of all the pushover procedures employed and
not specifically of the suggested multi-modal variation.

27
Chapter 4: Description and assessment of pushover methodologies

It has been shown in the previous section the failure of traditional pushover analysis in predicting the
dynamic drift profile of the RM15. Such outcome calls for an analysis technique which is an
improvement to the non-adaptive non-modal pushover analysis, using a triangular or uniform
distribution of forces. Note that the Modal Pushover Analysis (MPA) is actually a complete
procedure, incorporating already estimation of the seismic demand. It is certainly rigorous and the
relationship between ductility and damping has been thoroughly dealt with; thus, it is gaining
attention from the seismic engineering community. Since it is a such well developed method seems to
be a reasonable choice, among the non-adaptive modal pushover analyses, to be employed to explore
the improvement that this analysis might have over conventional pushover methods. Thus, the drift
profiles computed by MPA (with the first three modes considered) are compared to those from
nonlinear dynamic analysis Figure 4.4 and Table 4.2 presents the standard error for the
aforementioned analysis.

The drift profile obtained from MPA predicts very well the dynamic drift shape for the lower storeys;
however, completely underestimates the drifts at the top storeys failing to capture the largest drifts
that are developed in the frame. The standard error is smaller than those from the uniform and
triangular pushovers; therefore, a relative improvement for this particular case study has obtained.
However, it is stressed that only a single frame has been used herein and so general conclusions on the
accuracy of any single method of pushover analysis should not be made. Note, also, that the current
application follows the author’s interpretation of the MPA with the possibility of miscalculation not
being excluded.

2.50%
12
MPA
DYNAMIC
TRIANGULAR
10 UNIFORM

8
STOREY

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.4 Comparison of drift profiles from MPA with the envelope of maximum inter-storey drifts from
dynamic analysis for RM15 frame

Table 4.2 Standard error of the pushover schemes in Figure 4.4

Standard error

MPA 36.21

Triangular 51.07

Uniform 91.60

28
Chapter 4: Description and assessment of pushover methodologies

Overall, the above multi-modal procedures are a significant improvement of conventional pushover
analysis. They are theoretically more robust and conceptually more attractive, since they explicitly
consider the response of mode than one mode and the expected ground motion, and thus yield results
that are closer to rigorous inelastic time history analysis. In particular MPA has been extensively
developed and verified, and provides already a full approach for assessment of structures, since
procedures to estimate demand are included. However, the need to carry out three or four different
analyses, as well as calculations to attain satisfactory accuracy may be a significant drawback that
undermines the main objective of pushover analysis, which is to be simple and efficient replacement
of detailed dynamic analysis.

4.5 FORCE-BASED ADAPTIVE PUSHOVER PROCEDURES


When multiple modes are taken into account, it is customary, and rational, to combine the effects,
rather than the applied forces. However, this is only valid when one consider elastic systems (e.g.
elastic modal analysis and MPA), where the principle of superposition holds. For the case of nonlinear
systems, the combination of the effects obtained from different pushover analyses, each of which
corresponding to a given mode, leads to internal force which do not correspond to an equilibrium
state. Nonetheless, Gupta and Kunnath [2000] proposed a methodology in which the applied load is
constantly updated, depending on the instantaneous dynamic characteristics of the structure; in
addition, a site-specific spectrum can be used to define the loading pattern. According to the method,
eigenvalue analysis is carried out before each load increment, utilising the current structural stiffness
state. The number of modes of interest that will be taken into account is predefined and the storey
forces for each mode are estimated by:

Fij = Γ j ϕ ijWi S a ( j ) (4.3)


where, i is the storey number, j is the mode number, N is the number of the modes considered in the
analysis, Φij is the mass normalised mode shape value for the ith storey and the jth mode, Wi is the
weight of the ith storey, Sa(j) is the spectral amplification of the jth mode and Γj is the modal
participation factor for the jth mode computed by using
N N
2
Γ j = ∑ m k φ kj ∑ m k φ kj
k =1 k =1

The modal base shears Vj are calculated and combined using SRSS to compute the building base shear
V. The storey forces are then uniformly scaled using the ratio Sn between the predetermined base shear
increment ∆V and the base shear calculated from the modal analysis V, as represented by equations
4.4 and 4.5.

∆V
V j' = S nV j = Vj (4.4)
V

∆V
F j' = S n Fij = Fij (4.5)
V
A static analysis is then carried out for each mode independently. The calculated action effects for
each mode are combined with SRSS and added to corresponding values form the previous step. At the
end of the step, the structural stiffness state is assessed, in order to be used in the eigenvalue analysis
of the next step. The estimates of interstorey drifts and the sequence of the formation of local or
global collapse mechanism presented in the paper were satisfactory. However, the utilisation of SRSS
to combine the responses, derived for each mode, implies that equilibrium is not satisfied at the end of
each step.

29
Chapter 4: Description and assessment of pushover methodologies

More conceptually sound is instead to carry out single nonlinear adaptive pushover analysis, whereby
the modal combination is applied to the forces, rather than the responses. Along this line of thinking,
Bracci et al. [1997] were the first to introduce such procedure that utilises fully adaptive patterns. The
analysis starts by assuming an initial lateral load pattern {Fi}, usually triangular, whereas the
additional loads {∆Fi} imposed in subsequent increments are calculated from the instantaneous base
shear and storey resistances of the previous load step, according to the equation:

⎛ F k F k −1 ⎞ ⎛ k ⎞
∆Fik +1 = V j ⎜ i − i ⎟ + ∆V k +1 ⎜ Fi ⎟
⎜ V k V k −1 ⎟ b ⎜V k ⎟ (4.6)
⎝ b b ⎠ ⎝ b ⎠
where, i is the storey number, k is the increment number, Vb is the base shear and ∆Vb is the increment
of the base shear.

The procedure was implemented in the dynamic analysis package IDARC [Valles et al., 1996]
leading to the attainment of apparently promising results. Lefort [2000] implemented an extended
version of this method, which employed an additional force scaling equation, equation 4.7, to account
for higher mode contributions, obtaining, however, limited-accurate response predictions.

∑ (ϕ i⋅ j Γ j )
m
2
Wi
j =1
Fi = ⋅ ∆Vb + Fiold (4.7)
∑ W L ∑ (ϕ L⋅ j Γ j )
N m
2
L =1 j =1

where, i is the storey number, j is the mode number, Γj is the modal participation factor of mode j, Φij
is the mass normalised mode shape value for the ith storey and the jth mode, m is the number of modes
considering in the analysis, N is the number of storey in the building and ∆Vb is the increment of the
base shear.

Further, Requena and Ayala [2000] discussed two variations of adaptive pushover (referred as
approaches 2-A and 2-B) and compared them with a modal fixed pattern scheme (approach 1).
According to approach 2-A the force applied to the different storey levels is derived by the
combination of the contributions of the considered modes using SRSS and equals to:

2
⎛⎡ N ⎞
N ⎜⎢
(
⎜ ∑ mk φ kj )⎤⎥ ⎟
Fi = ∑ ⎜ ⎢ k =N1 ⎥ ⋅ φij Sα j mi ⎟ (4.8)

j =1 ⎜ ⎢ 2 ⎥

⎜ ⎢ ∑ mk φ kj ⎥ ⎟
⎝ ⎣ K =1 ⎦ ⎠
where, økj is the modal shape for the kth storey and the jth mode and Sαj is the pseudo-spectral
acceleration mode j. The alternative approach 2-B accepts the existence of an “equivalent”
fundamental mode φ i , which is determined through a combination of vibration modes using the SRSS
rule:

∑ (φij Γ j )
N
2
φi = (4.9)
j =1

N N
2
Γj is the participation factor of mode j defined as Γ j = ∑ m k φ kj ∑ m k φ kj
k =1 k =1

The distribution of the lateral loads is defined according to:

30
Chapter 4: Description and assessment of pushover methodologies

mi φ i
Fi =
N (4.10)
∑ mk φ k
k =1

It is possible, when the structural stiffness changes, due to the formation of plastic hinges, to calculate
a new distribution of the lateral loads that reflects the current state of inelasticity. The new load
distribution may be used for the subsequent steps of the analysis, until the stiffness state of the
structure changes again. These two alternative adaptive methods are appealing indeed. They are
theoretically rigorous and they explicitly account for higher modes and spectral contributions.
However, and unfortunately, the analytical results presented by the authors were only limited and the
accuracy procedures could not be effectively assessed or judged.

A methodology conceptually identical to the 2-A method proposed by Requena and Ayala, was also
suggested by Elnashai [2001]. The difference being that its implementation within a fibre analysis
framework allowed for a continuous, rather than discrete, force distribution update to be carried out.
The work of Elnashai, however, was confined to the application to simplified, not-necessarily
realistic, “stick-models”. Antoniou and Pinho [2004a, 2004b] extended and verified the method to the
case of actual reinforced concrete frames and implemented it in a finite element package that is freely
available from the internet [SeismoSoft, 2004].

In the adaptive pushover approach followed by Antoniou and Pinho, the lateral load distribution is not
kept constant but is continuously updated during the process, according to modal shapes and
participation factors derived by eigenvalue analysis carried out at each analysis step. The method is
multimodal and accounts for softening of the structure, its period elongation, and the modification of
the inertial forces due to spectral amplification. Two variants of the method exist: Force-based
adaptive pushover (FAP) and Displacement-based adaptive pushover (DAP).

The algorithm can be structured in four main stages; (i) definition of the nominal load vector and
inertia mass, (ii) computation of the load factor, (iii) calculation of the normalised scaling vector and
(iv) updating of the loading force vector. Whilst the first step is carried out only once, at the start of
the analysis, its three remaining counterparts are repeated at every equilibrium stage of the nonlinear
static analysis procedure, described as it follows.

The loading vector shape is automatically defined and updated by the solution algorithm at each
analysis step, for which reason the nominal vector of forces, P0, must always feature a uniform
(rectangular) distribution shape, in height, so as not to distort the load vector configuration determined
in correspondence to the dynamic response characteristics of the structure at each analysis step. In
addition, it is noteworthy that the adaptive pushover requires the inertia mass M of the structure to be
modelled, so that the eigenvalue analysis, employed in updating the load vector shape, may be carried
out.

The magnitude of the load vector P at any given analysis step is given by the product of its nominal
counterpart P0,, defined above, and the load factor λ at that step (see Equation 2.15). The latter is
automatically increased, by means of load control or response control strategy, until a predefined
analysis target, or numerical failure, is reached.

P = λ.P0 (4.11)
The normalized modal scaling vector, F , used to determine the shape of the load vector (or load
increment vector) at each step, is computed at the start of each load increment. In order for such
scaling vector to reflect the actual stiffness state of the structure, as obtained at the end of the previous
load increment, an eigenvalue analysis is carried out. To this end, the Lanczos algorithm [Hughes,

31
Chapter 4: Description and assessment of pushover methodologies

1987] is employed to determine the modal shape and participation factors of any given predefined
number of modes, after which the total storey forces Fij can be determined as:

Fij = Γj Φij Mi (4.12)


where, i is the storey number and j is the mode number, Γj is the modal participation factor for the jth
mode, Φij is the mass normalized mode shape value for the ith storey and jth mode, and Mi is the mass
of the ith storey.

As defined in the above equation, the lateral load pattern is spectrum independent and is only
determined by the modal properties of the system. If a spectral shape is considered, Equation 4.12
becomes:

Fij = Γj Φij Mi Sa(j) (4.13)


th
where, Sa(j) is the spectral amplification of the j mode.

The lateral load profiles of each vibration mode are then combined by using either the Square Root of
the Sum of the Squares (SRSS), if the modes can be assumed as fully uncoupled, the Complete
Quadratic Combination (CQC) method, if cross-coupling of the modes and respective viscous
damping is to be considered, or a weighted vectorial addition. Further, since only the relative values
of storey forces (Fi) are of interest in the determination of the normalised modal scaling vector, F ,
which defines the shape, not the magnitude, of the load or load increment vector, the forces obtained
from the modal combination rules, presented above, are normalized with respect to the total value

Fi
Fi = (4.14)
∑ Fi
Once the normalised scaling vector and load factor have been determined, and knowing also the value
of the initial nominal load vector, the loading vector Pt at a given analysis step t is obtained by adding
to the load vector of the previous step, Pt-1 (existing balance loads), a newly derived load vector
increment, computed as the product between the current load factor increment ∆λt, the current modal
scaling vector F and the nominal vector P0:

Pi = Pt-1+∆ λt F i P0 (4.15)
The procedure, named as incremental updating by Antoniou and Pinho [2004a], is schematically
represented in Figure 4.5.
normalised nominal new increment
shape at step t load vector of forces

∆Pt = ∆λt × × =
∑=1
new increment
existing forces of forces new forces applied at step t

Pt=Pt-1+∆Pt= + =

Figure 4.5 Graphical representation of loading force vector calculation with incremental updating

In incremental updating, the tangent stiffness is used not to calculate the whole load P, but the
incremental quantity ∆P to increase P of the previous analysis step. As in nonlinear time-history
analysis, the analysis is run in incremental fashion, through piecewise linearization, as schematically
shown in Figure 4.6. This approach can be viewed like “shifting” the origin of the system to the actual
state of force and displacement, as a sort of pre-loaded state. Starting from that point, and from the

32
Chapter 4: Description and assessment of pushover methodologies

actual stiffness state of the system (tangent in “global coordinates”, but also secant in the “shifted
coordinates”), the increment of load ∆P is calculated according to the spectrum scaling. The load on
the structure is not dependent as whole on the actual stiffness state, but it is increased according to the
actual stiffness state.

F t+ ∆ F t
Ft
O 't Kt

F k+ ∆ F k Kk
Fk
O 'k

O d k d k+ ∆ d k dt d t+ ∆ d t
Figure 4.6 Graphical interpretation of incremental updating using tangent stiffness

It is noted how the computation of the response in terms of piece-wise linear increments renders fully
valid the use of tangent stiffness to compute the vibration characteristics of the system, as well as
spectral amplification.

Since it is a conceptually sound method, because the modal combination is applied to the loads rather
than to responses, exhibits a displacement-based variant and is implemented in a freely available finite
element package, the adaptive pushover adopted by Antoniou and Pinho seems to be a good selection
to be applied to the example case. The force-based adaptive pushover analysis has been employed in
the present section. The loads from all modes have been combined by using the SRSS rule. Analysis
with spectral amplification has been undertaken by employing the elastic response spectra for the
Hollister record, record used in the dynamic nonlinear analysis. The results of the analysis with FAP
are presented in Figure 4.7 and the standard errors are shown in Table 4.3.

2.50%
12
FAP-SRSS
DYNAMIC
TRIANGULAR
10 UNIFORM

8
STOREY

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.7 Comparison of drift profiles from FAP (SRSS) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15

33
Chapter 4: Description and assessment of pushover methodologies

Table 4.3 Standard error of pushover schemes shown in Figure 4.7

Standard error

FAP-SRSS 93.32

Triangular 51.07

Uniform 91.60

The results obtained through FAP appear to be similar to those from conventional pushover analysis,
and the standard errors presented in Table 4.3 are comparable. The reason for the disappointing results
could be the use of the SRSS rule, which always leads to a positive load profile that pushes the
structure. A possible solution to this problem is explored in section 4.7.

4.6 DISPLACEMENT-BASED ADAPTIVE PUSHOVER PROCEDURES


In non-adaptive pushover, practical problems arise when one pushes the structure with a fixed pattern
of displacements: the relative displacement between consecutive floor levels is fixed, thus structural
characteristics such as strength irregularities and soft-storeys are concealed and the results are
misleading. Thus, if displacement loading is to be employed in pushover runs, a method to
realistically update the deformation patterns, as the stiffness of the structure changes, is required The
employment of displacements rather than forces is conceptually more desirable and follows the
present drive towards deformation-based design and assessment methods. Hence, an innovative
algorithm has been proposed that will allow displacement-based pushover analysis to be carried out
without the aforementioned shortcomings [Antoniou and Pinho, 2004b].

The implementation of the proposed displacement-based algorithm can be structured in four main
stages; (i) definition of the nominal load vector and inertia mass, (ii) computation of the load factor,
(iii) calculation of the normalised scaling vector and (iv) updating of the loading displacement vector.
Whilst the first step is carried out only once, at the start of the analysis, its three remaining
counterparts are repeated at every equilibrium stage of the nonlinear static analysis procedure, as
described in the following subsections.

Steps (i) and (ii) above are identical to those described for the Force-based adaptive pushover
algorithm (FAP), with the only difference consisting in the fact that the load vector now consist of
displacements (U), rather than forces (P). In general terms, the former is obtained, at each step of
analysis, through Equation 2.20, where λ represents the load factor, determined by means of load
control algorithm and U0 is the nominal vector.

U = λ.U0 (4.16)
It is noted that in DAP, applied loads and response deformations can be considered as effectively
coincident, since both consist of displacement. Hence, the employment of the slightly more elaborate
response control algorithm (see Antoniou and Pinho, 2004) is not required here, in view of the fact
that load and response control types lead to the attainment of equal results. For this reason, the simpler
load control algorithm (see Antoniou and Pinho, 2004) is used.

The normalized modal scaling vector, D , used to determine the shape of the load vector (or load
increment vector) at each step, is computed at the start of each load increment. In order for such
scaling vector to reflect the actual stiffness state of the structure, as obtained at the end of the previous
load increment, an eigenvalue analysis is carried out. To this end, the Lanczos algorithm [Hughes,

34
Chapter 4: Description and assessment of pushover methodologies

1987] is employed to determine the modal shape and participation factors of any given predefined
number of modes. Modal loads can be combined by using either the Square Root of the Sum of the
Squares (SRSS) or the Complete Quadratic Combination (CQC) method. The procedure for the
computation of the scaling displacement vector, D, through displacement- or drift-based approach is
described next.

Displacement-based scaling refers to the case whereby storey displacement patterns Di are obtained
directly from the eigenvalue vectors, as described in Equation 4.17, where i is the storey number and j
is the mode number, Γj is the modal participation factor of the jth mode, Φij is the mass normalized
mode shape value for the ith storey and the jth mode, and n stands for the total number of modes. The
first approach is analogous to the force-based adaptive procedure, making use of the maximum storey
displacements calculated directly by modal analysis to determine the scaling displacement vector:

∑ (Γ jφij )
n n
2
Di = ∑ Dij2 = (4.17)
j =1 j =1

The maximum displacement of a particular floor level, however, being essentially the relative
displacement between that floor and the ground, provides insufficient insight into the actual level of
damage incurred by buildings subject to earthquake loading. On the contrary, inter-storey drifts
feature a much clearer and direct relationship to horizontal deformation demand on buildings. Hence,
an alternative scaling scheme, whereby maximum inter-storey drift values obtained directly from
modal analysis, rather than from the difference between not-necessarily simultaneous maximum floor
displacement values, are used to compute the scaling displacement vector, has also been developed.

In such an interstorey drift-based scaling technique, the eigenvalue vectors are thus employed to
determine the inter-storey drifts for each mode ∆ij, as shown in Equation 4.18, while the displacement
pattern Di at the ith storey is obtained through the summation of the modal-combined inter-storey
drifts of the storeys below that level, i.e. drifts ∆1 to ∆i:

with ∆i = ∑ ∆2ij = ∑ [Γ j (φi , j − φi −1, j )]2


i n n
Di = ∑ ∆ k (4.18)
k =1 j =1 j =1

It is nonetheless noted that, although the summation of modal inter-storey drifts constitutes an
improvement over direct combination of displacements, it is still approximate, since it assumes that
all drift maxima at different storeys occur at the same time, which is of course unrealistic. Work
carried out by Antoniou and Pinho [2004b] indicates that interstorey drift scaling is better; thus, it is
the scaling technique that have been adopted.

Since only the relative values of storey displacements (Di) are of interest in the determination of the
normalised modal scaling vector, D , which defines the shape, not the magnitude, of the load or load
increment vector, the displacements obtained by Equation 4.18 are normalised so that the maximum
displacement remains proportional to the load factor, as required within a load control framework:

Di
Di = (4.19)
max Di

Once the normalised scaling vector, Dt , and load factor Γt or load factor increment ∆Γt have been
determined, and knowing also the value of the initial nominal load vector U0, the loading vector Ut at
a given analysis step t can be updated using the alternative previously introduced, incremental
updating, as give in Equation 4.20:

Ut = Ut-1+∆ λt Dt U0 (4.20)

35
Chapter 4: Description and assessment of pushover methodologies

Because of the reasons exposed in the previous section, the Displacement-based variant of the
adaptive pushover method adopted by Antoniou and Pinho is applied to the example case. As it was
for FAP, the loads from all modes have been combined using the SRSS rule. Spectral amplification
has been utilized in the present example. As it was stated above the interstorey drift scaling has been
adopted. The results DAP are presented in Figure 4.8 and the standard errors are shown in Table 4.4.

2.50%

12
DAP-SRSS
DYNAMIC
TRIANGULAR
10
UNIFORM

8
STOREY

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.8 Comparison of drift profiles from DAP (SRSS) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15

Table 4.4 Standard error of pushover schemes shown in Figure 4.8

Standard error

DAP-SRSS 25.58

Triangular 51.07

Uniform 91.60

It is clearly observed that the displacement-based adaptive pushover did manage to provide much
improved approximation of the highly irregular dynamic deformation profile envelope. In effect, even
though the attainment of fully accurate predictions was not observed, DAP was capable of predicting
the trend of the dynamic drift envelopes, a feature beyond the capabilities of force-based pushovers.
At the same time, the standard error of DAP is the smallest obtained so far from the studied pushover
methods. The reason for this satisfactory outcome may lies in that although SRSS is used to combine
the contributions of each mode, thus implying permanently positive storey drift profiles, DAP drift
profiles do feature changes of their respective gradient (i.e. the trend with which drift values change
from one storey to the next), introduced by the contributions the higher modes. When such gradient
variations imply a reduction of the drift of a given storey with respect to its two adjacent floor levels,
then the corresponding applied storey horizontal force must also be reduced, in some cases to the
extend of sign inversion.

4.7 POSSIBLE DRAWBACKS DUE TO THE USE OF QUADRATIC MODAL COMBINATION RULES
Analysis methods, which employ mode shapes to compute actions or responses, invariable they use
SRSS and CQC modal rules to combine the contributions from different structural modes. Modal
Pushover Analysis [Chopra and Goel, 2001] and the methodology proposed by Antoniou and Pinho

36
Chapter 4: Description and assessment of pushover methodologies

[2004a, 2004b] are examples where the modal combination is performed by using any of mentioned
rules.

Two shortcomings of the modal combination rules can be pointed out: the first one is that the result
obtained does not fulfil equilibrium; the second limitation is that signs are lost during the combination
process eliminating the contribution of negative quantities. The latter, it is believed, played an
important role during the research performed by López-Menjivar on 3D reinforced concrete buildings.
Such problem has been previously identified by Priestley [2003] and by Antoniou and Pinho [2004a].
Therefore, an alternative modal combination is needed. The proposal developed in this research is to
add the participation of each mode directly, in an algebraic fashion, without removing any sign; thus,
the contribution from each mode is fully taken with no modification. Such procedure will be named,
hereafter, Direct Vectorial Addition (DVA).

The procedure can be expressed in the following way. The modal storey loads are computing by using
the approach proposed by Antoniou and Pinho [2004a, 2004b]:

Ψij = Γ jφij M i S a, j (4.21)


where, i is the storey number, j is the mode number, Ψij is the modal storey loads which either could
be forces or displacements, Γj is the modal participation factor of mode j, øij is the mass normalised
mode shape value for the ith storey and the jth mode, Mi is the mass of the ith storey, Sa,j is the response
spectrum ordinate, either acceleration or displacement, corresponding to the period of vibration of the
jth mode.

The lateral load profiles of each vibration mode are then combined using the Direct Vectorial
Addition, DVA, in the following fashion:

n
Ψi = ∑ η j Ψij (4.22)
j =1

where, Ψi is the storey loads which either could be forces or displacements, ηj is the participation
factor of each mode which could be either positive or negative. This participation factor, ηj, can take a
value ranging from zero, no mode participation, to 1, full mode participation.

The idea of using modal addition, to compute responses of structures, is not strange to Earthquake
Engineering field. Priestley [2003] sketched some ideas about vectorial addition by using a finite
number of modes to obtain response quantities of wall buildings.

More recently, Kunnath [2004] proposed to perform various pushover analyses. Each pushover would
have an invariant load distribution vector obtained from the addition of load patterns which have been
computed from the deform shape of each individual mode. The contribution of each modal load
pattern can be scaled and shift the sign as well. It is showed that by this approach, envelopes that
contain the dynamic response of structures can be constructed. However, the author does not express a
definite way or rule to combine the modal load vectors and points out the need for more research to
achieve a unique set of modal factor adequate to match the dynamic response in any case.

A manner to test the suitability of the proposed method, The Direct Vectorial Addition (DVA), is to
try it on the example case comparing results with both nonlinear time history analysis and non-
adaptive pushovers. This process will be exposed and described next.

The direct vectorial addition of the first three modes will be applied to combine the loads/response, be
they forces or displacements, in all the following analyses. For the force-based pushover a weighting
of 100% to the first three modes will be applied (positive in the first and third modes and negative in

37
Chapter 4: Description and assessment of pushover methodologies

the second mode), whilst for the displacement-based pushover 100% weighting to the first mode, -
10% to the second mode and 10% to the third mode will be considered.

The force-based adaptive algorithm has been applied to the RM15 frame with a vectorial addition of
the modal loads at each adaptive step. The inter-storey drift profile obtained from FAP, with a
vectorial sum of the modes, is compared to the drift profiles from nonlinear dynamic analysis and
traditional pushovers in Figure 4.9. A clear improvement is observed when DVA is used in FAP since
the adaptive pushover drift profile is now closer to the dynamic drift envelope than its non-adaptive
counterparts. The enhancement is also noticed by comparing the standard errors presented in Table
4.5.

2.50%
12
FAP-DVA
DYNAMIC
TRIANGULAR
10 UNIFORM

8
STOREY

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.9 Comparison of drift profiles from FAP (DVA) with the envelope of maximum inter-storey
drifts from dynamic analysis for RM15

Table 4.5 Standard error of pushover schemes shown in Figure 4.7 and Figure 4.9

Standard error

FAP-DVA 38.00

FAP-SRSS 93.32

Triangular 51.07

Uniform 91.60

The displacement-based adaptive pushover has again been applied to the case study frame using the
inter-storey drift scaling technique but this time with a vectorial addition of the modal loads, as
opposed to using the SRSS rule. As was the case of FAP, DAP has managed to improve the prediction
of the dynamic interstorey drift when is DVA is used. The dynamic profile envelope is closely
matched by the drifts obtained from DAP, Figure 4.10. The superior performance, in predicting the
dynamic drifts, by employing DAP is further reinforced by the standard error calculations presented in
Table 4.6.

38
Chapter 4: Description and assessment of pushover methodologies

2.50%

12
DAP-DVA
DYNAMIC
TRIANGULAR
10
UNIFORM

STOREY
6

0
0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00%
INTERSTOREY DRIFT

Figure 4.10 Comparison of drift profiles from DAP (DVA) with the envelope of maximum inter-storey
drifts from dynamic analysis

Table 4.6 Standard error of pushover schemes shown in Figure 4.8 and Figure 4.10

Standard error

DAP-DVA 14.81
DAP-SRSS 25.58
Triangular 51.07
Uniform 91.60

To further test the applicability of the DVA in predicting the response obtained from nonlinear time
history analysis the new methodology will be applied and assessed by using the ICONS frame. The
idea to use the ICONS frame is to examine the performance of DVA, when applied to the adaptive
pushover method followed by Antoniou and Pinho, for a single case study building known to have
structural characteristics that may not be reproduced using conventional pushover approaches.

4.8 THE ICONS FRAME

4.8.1 Non-adaptive non-modal pushover analysis


The triangular and uniform loading profile pushovers been carried out on the case study frame and are
compared with the two types of dynamic drift profile, described previously, as presented in Figure
4.11 and Figure 4.12. Immediately it is apparent that both the triangular and uniform pushover
analyses fail to estimate the soft-storey in the third floor of the frame. The drifts at the lower storeys
are also largely overestimated with both force distributions. The standard errors are presented in
Table 4.7 wherein it may be noted that the triangular pushover analysis produces lower standard
errors, however, they are still of an unacceptable magnitude.

39
Chapter 4: Description and assessment of pushover methodologies

4 4

3 3

Dynamic Dynamic
Storey

Storey
2 2
Triangular Triangular

Uniform Uniform

1 1

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Drift (%) Drift (%)

(a) (b)
Figure 4.11 Comparison of drift profiles from triangular and uniform pushovers with the drift profile at
the time step of maximum inter-storey drift from dynamic analysis using the a) Acc-475 record and b)
Acc-975 record

4 4

3 3

Dynamic Dynamic
Storey

Storey

2 2
Triangular Triangular

Uniform Uniform

1 1

0 0
0 0.2 0.4 0.6 0.8 1 0 0.5 1 1.5 2

Drift (%) Drift (%)

(a) (b)
Figure 4.12 Comparison of drift profiles from triangular and uniform pushovers with the envelope of
maximum inter-storey drifts from dynamic analysis using the a) Acc-475 record and b) Acc-975 record

Table 4.7 Standard error of the pushover schemes in Figure 4.11 and Figure 4.12
Profile at time step of
Envelope of max. inter-
maximum inter-storey
storey drifts
drift

Acc-475 Acc-975 Acc-475 Acc-975

Uniform 50 % 52 % 36 % 47 %
Triangular 23 % 30 % 29 % 25 %

4.8.2 Pushover analysis using DVA


The direct vectorial addition of the first three modes will be applied to combine the loads/response, be
they forces or displacements, in all the following analyses. For the force-based pushovers a weighting
of 100% to the first three modes will be applied (positive in the first mode and negative in the higher
modes), whilst for the displacement-based pushovers 90% weighting to the first mode, -10% to the
second mode and -2.5% to the third mode will be considered.

The force-based adaptive pushover (FAP) has been applied to the case study frame with a vectorial
addition of the modal loads at each adaptive step. The inter-storey drift profile obtained from FAP,
with a vectorial sum of the modes, is compared to the drift profiles from nonlinear dynamic analysis
in Figure 4.13 and Figure 4.14. It can be observed that the agreement between the profiles is very
close both in both shape and magnitude. The force-based adaptive pushover, with direct modal
combination, manages to capture the soft-storey mechanism in the third floor for both records, and

40
Chapter 4: Description and assessment of pushover methodologies

matches closely the drift levels in all other storeys. The superior behaviour predicted by FAP is
further reinforced by the standard error calculations presented in Table 4.8. An improvement to the
non-adaptive pushover results is observed through the consideration of the changing stiffness
characteristics of the structure with adaptive pushover.
4 4

3 3

Dynamic Dynamic
Storey

Storey
2 2
Adaptive Adaptive

1 1

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Drift (%) Drift (%)

(a) (b)
Figure 4.13 Comparison of drift profiles from FAP with the drift profile at the time step of maximum
inter-storey drift from dynamic analysis using the a) Acc-475 record and b) Acc-975 record

4 4

3 3

Dynamic Dynamic
Storey

Storey

2 2
Adaptive Adaptive

1 1

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Drift (%) Drift (%)

(a) (b)
Figure 4.14 Comparison of drift profiles from FAP with the envelope of maximum inter-storey drifts
from dynamic analysis using the a) Acc-475 record and b) Acc-975 record

Table 4.8 Standard errors of the profiles shown in Figure 4.13 to Figure 4.14
Profile at time step of Envelope of max. inter-storey
maximum inter-storey drift drifts
Acc- 475 Acc-975 Acc-475 Acc-975
Triangular 23 % 30 % 29 % 25 %
Uniform 50 % 52 % 36 % 47 %
FAP with DVA 11 % 10 % 14 % 6%

Subsequently, the displacement-based adaptive pushover (DAP) has been applied to the case study
frame using the inter-storey drift scaling (IDAP) technique with a vectorial addition of the modal
loads as the modal combination rule.

Figure 4.15 show the inter-storey drift profiles from DAP as well as those from dynamic analysis at
the time step of maximum inter-storey drift. The adaptive pushovers have again managed to predict
the drift shape of the dynamic analysis for both records. The soft-storey is anticipated with DAP the
drift ratios in all other floors are closely predicted. The comparison of the results obtained by DAP
with the maximum envelope of inter-storey drift from dynamic analysis is presented in Figure 4.16.

41
Chapter 4: Description and assessment of pushover methodologies

The standard errors are presented in Table 4.9 for DAP. A significant improvement to the results is
noted by the use of vectorial addition of the modes rather than the SRSS rule. By referring back to
Table 4.8, it can be observed that slightly higher errors have been obtained with displacement-based
adaptive pushover than force-based adaptive pushover but they are now much more comparable.
4 4

3 3

Dynamic Dynamic
Storey

Storey
2 2
Adaptive Adaptive

1 1

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Drift (%) Drift (%)

(a) (b)
Figure 4.15 Comparison of drift profiles from DAP with the drift profile at the time step of maximum
inter-storey drift from dynamic analysis using the a) Acc-475 record and b) Acc-975 record
4 4

3 3

Dynamic Dynamic
Storey

Storey

2 2
Adaptive Adaptive

1 1

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Drift (%) Drift (%)

(a) (b)
Figure 4.16 Comparison of drift profiles from DAP with the envelope of maximum inter-storey drifts
from dynamic analysis using the a) Acc-475 record and b) Acc-975 record

Table 4.9 Standard errors of pushover schemes shown in Figure 4.15 and Figure 4.16
Profile at time step of Envelope of max. inter-
maximum inter-storey drift storey drifts
Acc- 475 Acc-975 Acc-475 Acc-975
Triangular 23 % 30 % 29 % 25 %
Uniform 50 % 52 % 36 % 47 %
DAP with DVA 20 % 8% 18 % 8%

It has been observed that all adaptive pushover variants with vectorial modal addition, as the modal
combination employed, were able to predict the third floor soft-storey and produced much lower
standard errors than the traditional pushover analyses. These conventional pushover methods failed to
reproduce both the local and global dynamic response of the frame, falling short to predict the soft-
storey at the third floor.

Finally, in order to conclude the present chapter, it must be highlighted that general conclusions on
the accuracy of any method cannot be made from a couple of buildings and that this would require
statistical studies to be carried out. Therefore, the next step of the research is to apply the newly
proposed methodology to a group of buildings that display different characteristics such as height,
number of storeys and bays subjected to several ground motions. The response obtained by using the

42
Chapter 4: Description and assessment of pushover methodologies

proposed method of modal combination will be compared to results obtained through a rigorous
nonlinear time history analysis.

43
Chapter 5: Conclusions

5. CONCLUSIONS

5.1 SUMMARY
The main objectives of this research were (i) to discuss and assess different pushover methodologies
and (ii) to explore the advantages that the use of an alternative modal combination rule could bring
when applied to adaptive pushover methodology proposed by Antoniou and Pinho. The first aim was
carried out by employing a typical seismically designed concrete frame structure in order to evaluate
the results obtained by using different pushover approaches. The second objective has been attain by
proposing a modal combination scheme, named Direct Vectorial Addition (DVA), which attempts to
overcome the problem presented by the current quadratic modal rules, which is to change the sign of
modal quantities involved turning them into positive values

The initial phase of this research reiterated that currently the most accurate procedure of demand
prediction and performance evaluation of structures is the non-linear time history analysis.
Nevertheless, the reliability of this method hinges on the unambiguous definition of certain
indispensable parameters such as, a set of site-specific ground motions [Bommer et al., 2003],
appropriate and accurate force-deformation relationships of various structural components and a
robust mathematical tool capable of handling all analysis data and producing results within the time
constrains of a design office. Consequently, a simpler yet reliable technique of structural analysis is
desirable.

An alternative to the nonlinear time history analysis is the nonlinear static pushover analysis. The
purpose of the pushover analysis is to assess the structural performance by estimating the strength and
deformation capacities using static, nonlinear analysis and comparing these capacities with the
demands at the corresponding performance levels. Although it provides crucial information on
response parameters that cannot be obtained with conventional elastic methods (either static or
dynamic), this method is not exempt from some limitations such as the inability to include higher
mode order effects or progressive stiffness degradation. Therefore, the need for a procedure that
overcomes the above deficiencies is readily noted.

A revision of the state-of-the-art pushover procedures was carried out, from which advantages and
disadvantages of each approach were identified. At the same time, the accuracy of traditional
pushover analyses, modal pushover analysis and adaptive pushover algorithm in predicting the
seismic response characteristics of a non-seismically designed reinforced concrete frame model has
been studied. Comparisons between drift profiles obtained through static (pushover) and dynamic
nonlinear fibre analysis have been made and the standard error has been computed to test the accuracy
of each method for a test case.

From the above comparative study, it was noted that the adaptive pushover methodology constitute a
viable alternative to nonlinear dynamic analysis since solve the inherent deficiencies that conventional
pushover analysis possesses. One methodology is particularly appealing which is the adaptive

44
Chapter 5: Conclusions

pushover approach proposed by Antoniou and Pinho [2004a, 2004b]. In this method, the lateral load
distribution is not kept constant but is continuously updated during the process, according to modal
shapes and participation factors derived by eigenvalue analysis carried out at each analysis step. The
method is multimodal and accounts for softening of the structure, its period elongation, and the
modification of the inertial forces due to spectral amplification. The above methodology has proved to
give similar results to those obtained from nonlinear dynamic analysis, for 2D frame systems,
especially when its displacement-based variant has been employed. Nevertheless, it was expressed
that the methodology might have a drawback, due to the way the modal combination is carried out, to
compute and update the load vector.

The adaptive pushover of Antoniou and Pinho uses the SRSS, or CQC, as modal combination rule.
These rules transform the contribution of negative quantities turning them into positive values;
therefore, this particular characteristic may mislead the response when flexible and tall buildings are
analysed. The above fact affects mainly the force-based variant of the method for reasons already
discussed (Section 4.5). Thus, it is logical to assess if an alternative method will lead to better results.

The Direct Vectorial Addition, DVA, was presented as an alternative to the usage of SRSS. The DVA
directly combine the modes in algebraic fashion without changing the sign of the modal contributors.
The proposed modal combination was tested by using the seismically designed reinforced concrete
frame model, employed in the evaluation of different pushover schemes. In addition, the DVA was
tested with a well-known case study, the ICONS frame where it was observed that the failure
behaviour of such structure could be fully captured by using DVA.

5.2 FUTURE RESEARCH


It is highlighted that general conclusions on the accuracy of any of the methods presented in previous
chapters cannot be made from two case studies and that this would require statistical studies to be
carried out. Therefore, the next step of the research is to apply the newly proposed methodology to a
group of buildings that display different characteristics such as height, number of storeys and bays
subjected to several ground motions. The response obtained by using the proposed method of modal
combination will be compared to results obtained through a rigorous nonlinear time history analysis.
As it was done for the case of RM15 and ICONS frame, the similarity will actually be quantified by
using the Standard error of each pushover respect to the Time history analysis. It is noted that each
pushover result will be compared to Nonlinear Dynamic Analysis output obtained for each single
accelerogram, as opposed to the statistical average of all dynamic cases. In doing it so, it is believed
that a much more demanding and precise assessment of the static procedures is effectively carried out,
since structural response peculiarities introduced by individual input motions are not smoothed out
through results averaging.

45
Chapter 6: References

6. REFERENCES
ACI318 [1995]. “Building code requirements for structural concrete ACI318-95, and commentary ACI318R-
95,” American Concrete Institure, Detroit, MI, USA.

Albanesi, T., Biondi, S. and Petrangeli, M. [2002]. “Pushover analysis: An energy based approach.”
Proceedings of the 12th European Conference on Earthquake Engineering, Paper 605. Elsevier Science
Ltd.

Alvarez-Botero J.C. and López-Menjivar M.A. [2004] “Assessment of the ‘Sede comunale Piazza Europa’
building, Comune di Vagli Sotto, Provincia di Luca, Regione Toscana (Technical Report),” EUCentre,
Universitá degli Studi di Pavia. 49 pag.

American Concrete Institute (ACI) [1984]. “Earthquake effects on reinforced concrete structures,” Publication
SP-84, US-Japan research, American Concrete Institute, Detroit, MI, USA.

Antoniou, S. [2001]. “Pushover analysis for seismic assessment of RC structures,” Transfer Report, Department
of Civil Engineering, imperial College London, United Kingdom.

Antoniou, S. [2002]. “Advanced inelastic static analysis for seismic assessment of structures,” PhD Thesis,
Imperial College London, University of London, United Kingdom.

Antoniou S., Rovithakis A. and Pinho R. [2002] “Development and verification of a fully adaptive pushover
procedure,” Proceedings Twelfth European Conference on Earthquake Engineering, London, UK, Paper
No. 822.

Antoniou, S. and Pinho, R. [2004a]. “Advantages and limitations of adaptive and non-adaptive force-based
pushover procedures.” Journal of Earthquake Engineering, 8(4), pp. 497-522.

Antoniou, S. and Pinho, R [2004b]. “Development and verification of a displacement-based adaptive pushover
procedure,” Journal of Earthquake Engineering, Vol. 8, No. 5, pp. 643-661.

ATC [1997]. NEHRP “Guidelines for the seismic rehabilitation of buildings.” FEMA Report No. 273, Federal
Emergency Management Agency, Applied Technology Council, Washington D.C.

Aydinoğlu, M.N. [2003]. “An incremental response spectrum analysis procedure based on inelastic spectral
displacements for multi-mode seismic performance evaluation,” Bulletin of Earthquake Engineering, No.
1, pp. 3-36.

Bommer, J.J. and Martinez-Pereira, A. [1999] “The effective duration of earthquake strong motion,” Journal of
Earthquake Engineering, 3(2), pp 127-172.

46
Chapter 6: References

Bommer, J.J., Acevedo, A.B. and Douglas, J. [2003]. “The selection and scaling of real earthquake
accelerograms for use in seismic design and assessment,” Proceedings of the ACI International Conference
on seismic bridge design and retrofit, American Concrete Institute. USA.

Bracci, J.M., Kunnath, S. K. and Reinhorn, A.M. [1997]. “Seismic performance and retrofit evaluation of
reinforced concrete structures.” Journal of Structural Engineering, 123(1), pp. 3-10.

Campos-Costa, A. and Pinto A.V. [1999]. “European seismic hazards scenarios – An approach to the definition
of input motion for testing and reliability assessment of Civil Engineering structures,” JRC Special
publication No.X.99.XX, Joint Research Centre, Ispra, Italy.

Carvalho, E.C., Coelho, E. and Campos-Costa, A. [1999] “Preparation of the full-scale tests on reinforced
concrete frames,” ICONS Report, Innovative Seismic Design Concepts for New and Existing Structures,
European TMR Network, LNEC.

CEN [1991]. “European Prestandard ENV 1992-1-1: Eurocode 2 – Design of concrete structures, Part 1:
General rules and rules for buildings”, Comite Europeen de Normalisation, Brussels, Belgium.

CEN [1995]. European Prestandard ENV1998-1-1: Eurocode 8 – Earthquake resistant design of structures, Part
1: General rules and rules for buildings. Comite Europeen de Normalisation, Brussels, Belgium.

Chopra, A.K. and Goel, R.K. [2001]. “A modal pushover analysis procedure to estimate seismic demands for
buildings: Theory and preliminary evaluation.” PEER Report 2001/03, Pacific Earthquake Engineering
Research Center, College of Engineering, University of California, Berkeley, January.

Chopra, A.K. and Goel, R.K. [2002]. “A modal pushover analysis procedure for estimating seismic demands
for buildings.” Earthquake Engineering and Structural Dynamics, Vol. 31, pp. 561-582.

Earthquake Engineering Research Institute [1997] “The EERI Annotated Slide Collection”, EERI, Oakland.

Faella, G. and Kilar, V. [1998]. “Asymmetric multi-storey R/C frame structures: push-over versus nonlinear
dynamic analysis.” Proceedings of the Eleventh European Conference on Earthquake Engineering, A.A.,
Ballkema, Rotterdam.

Fajfar, P. and Gasperic, P. [1996]. “The N2 method for the seismic damage analysis of RC buildings.”
Earthquake Engineering and Structural Dynamics, Vol 25:31-46.

Fajfar, P. [1999]. “Seismic assessment and retrofit of RC structures.” Proceedings of the Eleventh European
Conference on Earthquake Engineering – Invited Lectures, Paris, France, pp. 237-249.

Fardis, M.N. [1994] “Analysis and design of reinforced concrete buildings according to Eurocode 2 and 8.
Configuration 3, 5 and 6,” Reports on Prenormative Research in Support of Eurocode 8.

Fardis, M.N. and Panagiotakos, T.B. [1997]. “Seismic design and response of bare and masonry-infilled
reinforced concrete buildings. Part I: Bare structures,” Journal of Earthquake Engineering, Vol. 1, No. 1,
pp 219-256.

Filippou F.C., Popov E.P. and Bertero V.V. [1983] "Modelling of R/C joints under cyclic excitations," Journal
of Structural Engineering, Vol. 109, No. 11, pp. 2666-2684.

Gupta, B. and Kunnath, S.K. [2000]. “Adaptive spectra-based pushover procedure for seismic evaluation of
structures.” Earthquake Spectra, 16(2), pp. 367-391.

47
Chapter 6: References

Hughes, T.J.R. [1987]. The Finite Element Method, Linear Static and Dynamic Finite Element Analysis,
Prentice-Hall, Inc.

Izzuddin, B.A. [2001] “Conceptual issues in geometrically nonlinear analysis of 3-D frame structures,”
Computer Methods in Applied Mechanics and Engineering, Vol. 191, pp. 1029-1053.

Kilar, V. and Fajfar, P. [1996]. “Simplified push-over analysis of building structure.” Proceedings of the
Eleventh World Conference on Earthquake Engineering, Paper 1011, Pergamon, Elsevier Science Ltd,
Acapulco, México.

Kilar, V. and Fajfar, P. [1997]. “Simple push-over analysis of asymmetric buildings.” Earthquake Engineering
and Structural Dynamics, Vol. 26, pp.233-249.

Kilar, V. and Fajfar, P. [2001]. “On the applicability of pushover analysis to the seismic performance
evaluation of asymmetric buildings” European Earthquake Engineering, XV(1), pp. 20-31, Jan-Apr.

Kilar, V. and Fajfar, P. [2001]. “On the applicability of pushover analysis to the seismic performance
evaluation of asymmetric buildings,” European Earthquake Engineering, Vol.. XV, No. 1, pp. 20-31, Jan-
Apr.

Krawinkler, H. and Seneviratna, G.D.P.K. [1998]. “Pros and cons of a pushover analysis of seismic
performance evaluation.” Engineering Structures, 20(4-6), pp 452-464.

Kunnath, S.K. [2004]. “Identification of modal combination for nonlinear static analysis of building structures,”
Computer-Aided Civil and Infrastructure Engineering, v. 19, pp 246-259.

Lawson, R.S., Vance, V. and Krawinkler, H. [1994]. “Nonlinear static push-over analysis - why, when, and
how?.” Proceedings of the Fifth U.S. National Conference on Earthquake Engineering, Earthquake
Engineering Research Institute, Oakland, California, I, Vol I:283-292.

Lefort, T. [2000]. “Advance pushover analysis of RC multi-storey buildings.” Engineering Seismology and
Earthquake Engineering Section, Imperial College of Science, Tehnology and Medicine. Report No. 91/2,
March 1991.

López-Menjivar, M.A. [2003] “3D Pushover of irregular reinforced concrete buildings,” Master’s Thesis. ROSE
School, Universitá degli Studi di Pavia, pp. 72.

Mander J.B., Priestley M.J.N. and Park R. [1988] "Theoretical stress-strain model for confined concrete,"
Journal of Structural Engineering, Vol. 114, No. 8, pp. 1804-1826

Menegotto M. and Pinto P.E. [1973] "Method of analysis for cyclically loaded R.C. plane frames including
changes in geometry and non-elastic behaviour of elements under combined normal force and bending,"
Symposium on the Resistance and Ultimate Deformability of Structures Acted on by Well Defined Repeated
Loads, International Association for Bridge and Structural Engineering, Zurich, Switzerland, pp. 15-22.

Moghadam, A.S. and Tso, W.K. [1996]. “Damage assessment of eccentric multistory buildings using 3-D
pushover analysis.” Proceedings of the Eleventh World Conference on Earthquake Engineering, Paper
997, Pergamon, Elsevier Science Ltd, Acapulco, México.

Moghadam, A.S. and Tso, W.K. [1998]. “Pushover analysis for asymmetrical multistory buildings.”
Proceedings of the Sixth U.S. National Conference on Earthquake Engineering, Earthquake Engineering
Research Institute, Oakland, California, 13 pag.

48
Chapter 6: References

Moghadam, A.S. and Tso, W.K. [2000a]. “Pushover analysis for asymmetric and set-back multistory
buildings.” Proceedings of the Twelfth World Conference on Earthquake Engineering, Paper 1093. New
Zealand Society for Earthquake Engineering, Upper Hutt, New Zealand.

Moghadam, A.S. and Tso, W.K. [2000b]. “3-D pushover analysis for damage assessment of buildings.” Journal
of Seismology and Earthquake Engineering, 2(3), pp.23-3.

Moghadam, A.S. and Tso, W.K. [2002]. “A pushover procedure for tall buildings.” Proceedings of the 12th
European Conference on Earthquake Engineering, Paper 395. Elsevier Science Ltd.

Mwafy, A.M. and Elnashai, S.A. [2000]. “Static pushover versus dynamic-to-collapse analysis of RC
buildings.” Engineering Seismology and Earthquake Engineering Section, Imperial College of Science,
Technology and Medicine. Report No. 00/1, January.

Mwafy, A.M. [2001]. “Seismic performance of code-designed RC buildings,” PhD Thesis. Department of Civil
and Environmental Engineering, Imperial College of Science, Technology and Medicine, London.

Paulay, T and Priestley, M.J.N. [1992]. “Seismic design of reinforced concrete and masonry buildings,” John
Wiley and Sons, New York, 744 pag.

Paret, T.F., Sasaki, K.K., Elibeck, D.H. and Freeman, S.A. [1996]. “Approximate inelastic procedures to
identify failure mechanism from higher mode effects” Proceedings of the Eleventh World Conference on
Earthquake Engineering, Paper 966, Pergamon, Elsevier Science Ltd, Acapulco, México.

Penelis, G. and Kappos, A.J. [2002]. “3-D pushover analysis: The issue of torsion” Proceedings of the 12th
European Conference on Earthquake Engineering, Paper 015. Elsevier Science Ltd.

Pinho, R. and Elnashai, A.S. [2000]. “Dynamic collapse testing of a full-scale four storey RC frame”, ISET
Journal of Earthquake Technology, Paper No. 406; 37(4), pp. 143-164

Presidente del Consiglio dei Ministri [2003]. Norme Tecniche per il Progetto, la Valutazione e l'Adeguamento
Sismico Degli Edifici, Ordinanza no. 3274.

Priestley, M.J.N. [1993]. “Myths and fallacies in earthquake engineering-conflicts between design and reality,”
Bulletin of New Zealand National Society for Earthquake Engineering, Vol. 26, No. 3, pp. 329-341.

Pristley, M.J.N. [2003]. Myth and Fallacies in Earthquake Engineering, Revisited, IUSS Press, Pavia, Italy.

Requena, M. and Ayala, G. [2000]. “Evaluation of a simplified method for the determination of the non-linear
seismic response of RC frames.” Proceedings of the Twelfth World Conference on Earthquake
Engineering, Paper 2109. New Zealand Society for Earthquake Engineering, Upper Hutt, New Zealand.

Rovitahkis, A. [2001]. “Verification of adaptive pushover analysis procedures,” MSc. Dissertation, Department
of Civil and Environmental Engineering, Imperial College London, United Kingdom.

Sasaki, K.K., Freeman, S.A. and Paret, T.F. [1998]. “Multi-mode pushover procedure (MMP) – a method to
identify the effects of higher modes in a pushover analysis.” Proceedings of the Sixth U.S. National
Conference on Earthquake Engineering, Earthquake Engineering Research Institute, Oakland, California,
12 pages.

SEAOC [1995]. “Performance based seismic engineering of buildings.” Vision 2000 Committee Structural
Engineering Association of California, Sacramento, California.

49
Chapter 6: References

SeismoSoft [2004]. “SeismoStruct- A computer program for Static and Dynamic Non-linear Analysis of
Framed Structures [online].” Available from URL: http://www.seismosoft.com

Valles, R.E., Reinhorn, A.M., Kunnath, S.K., Li, C. and Madan, A. [1996]. “IDARC2D Version 4.0: A
computer program for the inelastic analysis of buildings.” Technical Report NCEER 96-0010, State
University of New York at Buffalo, Buffalo, NY.

50

S-ar putea să vă placă și