See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/278699178

Time History Seismic Analysis

Chapter · January 2013

DOI: 10.1007/978-3-642-36197-5_134-1

CITATIONS READS

0 181

3 authors:

Nikos Lagaros Chara Ch. Mitropoulou

National Technical University of Athens National Technical University of Athens
182 PUBLICATIONS   2,456 CITATIONS    18 PUBLICATIONS   127 CITATIONS   

SEE PROFILE SEE PROFILE

Papadrakakis Manolis
National Technical University of Athens
319 PUBLICATIONS   4,836 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

OCP: Optimization Computing Platform View project

Research Topic on "Innovative Approaches in Computational Structural Engineering" View project

All content following this page was uploaded by Nikos Lagaros on 21 June 2015.

The user has requested enhancement of the downloaded file.

 Time history seismic analysis

Nikos D. Lagaros, Chara Ch. Mitropoulou and Manolis Papadrakakis

Institute of Structural Analysis & Antiseismic Research,
Department of Structural Engineering,
School of Civil Engineering,
National Technical University of Athens,
9, Iroon Polytechniou Str., Zografou Campus, 157 80 Athens, Greece
e-mail: {nlagaros, chmitrop, mpapadra}@central.ntua.gr

Summary: In the case of design or assessment purposes the seismic capacity of the structural
system should be computed for different seismic hazard levels by means of linear or nonlinear time
history seismic analyses using a number of properly selected ground motions. The selection of the
ground motions plays an important role in the efficiency and accuracy of the design or assessment
procedure. In this chapter, we provide an overview of the time history seismic analysis applied for
design and assessment purposes.
Synonyms: Direct integration methods; seismic loading; natural records; artificial accelerograms;
linear and nonlinear analysis; inertia and damping forces.

1. INTRODUCTION

Most of the current seismic design codes belong to the category of prescriptive design procedures
(or limit-state design procedures), where if a number of checks are satisfied then the structure is
considered safe since it fulfils the safety criterion against collapse. A typical limit-state based
design can be viewed as one (i.e., ultimate strength) or two limit-state approach (i.e., serviceability
and ultimate strength). Existing seismic design procedures are based on the principal that a
structure will avoid collapse if it is designed to absorb and dissipate the kinetic energy that is
induced during a seismic excitation. Most of the modern seismic norms express the ability of the
structure to absorb energy through inelastic deformation using a reduction or behaviour factor that
depends on the material and the construction type of the structure. The concept of performance-
based design (PBD) was introduced a few decades ago, for designing structures subjected to
seismic loading conditions. In PBD more accurate and time-consuming analysis procedures are
employed, to estimate non-linear structural response. The progress that takes place in the area of
computational mechanics, as well as in computer technology, continuously expands the capabilities
and the applicability of PBD procedures. The main objective of this kind of design procedures is
to achieve more predictable and reliable levels of safety and operability against natural hazards.

1
According to PBD procedures, the structures should be able to resist earthquakes in a quantifiable
manner and to present specific target performance levels of possible damages. PBD design criteria
try to define certain levels of structural performance for various levels of seismic hazard.
In construction industry decision making for structural systems situated in seismically active
regions, requires consideration of damage cost and other losses resulting from earthquakes
occurring during the lifespan of a structure. Thus, nowadays life-cycle cost analysis (LCCA)
becomes an essential component of the design process used to control the initial and the future cost
of building ownership. In early 1960s LCCA was applied in the commercial area and in particular
in the design of products considering the total cost of developing, producing, using and retiring.
The introduction of LCCA in construction industry was made in the field of infrastructures as an
investment assessment tool. In particular, in early 1980s it was used in USA as an appraisal tool
for the total cost of ownership over the lifespan of an asset. Later, in view of large losses due to
extreme hazards, like earthquakes and hurricanes, there was a need for new design procedures of
facilities that could lead to life protection and reduction of damage and economical impact of such
hazards to an acceptable level. In this context LCCA was introduced in the field of constructions
as a complex investment appraisal tool incorporating a structural performance criterion.
Over the last decades risk management of structural systems has gained the attention of various
economic and technical decision centers in modern society. The optimal allocation of the public
resources for a sustainable economy requires proper tools for estimating the consequences of
natural hazardous events on the built environment. The risk management addresses this claim
indicating the way for implementing optimal choices. Thus, the main purpose of the risk
management process is to choose among different options relying on technical and economical
considerations. Risk assessment and decision analysis are the main steps of the risk management
concept. It is therefore essential to establish a reliable procedure for assessing the seismic risk of
structural systems. Seismic fragility analysis, which provides a measure of the safety margin for
the structural system, is considered as the core of the risk assessment framework.
The implementation of PBD, LCCA or seismic risk procedures require a reliable computing tool
for estimating the capacity and the demand for any structural system. Some of these computing
tools (Fajfar 2000; Chopra and Goel 2002; Vamvatsikos and Cornell 2002), are based on time
history seismic analyses, and they are considered as analysis methods for obtaining good estimates
of the structural performance in the case of earthquake hazard and therefore they are considered as
appropriate methods to be incorporated into such design and assessment frameworks. In this

2
chapter, we provide an overview of the time history seismic analysis applied for design and
assessment purposes.

2. GROUND MOTION EXCITATION

Selecting the seismic loading for design and/or assessment purposes is not an easy task due to the
uncertainties involved in the very nature of seismic excitations. One possible approach for the
treatment of the seismic loading is to assume that the structure is subjected to a set of records that
are more likely to occur in the region where the structure is located.

2.1 Natural records

For the implementation of dynamic seismic analysis a scale factor is calculated for each hazard
level considered and for each one of the records selected. In order to preserve the relative scale of
the two components of the records in the longitudinal and transversal directions, the component of
the record having the highest SA(T1,5%) is scaled first, while the scaling factor that preserves their
relative ratio is assigned to the second component.

2.2 Generation of artificial accelerograms

In order for the artificial accelerograms to be representative they have to match some requirements
of the seismic codes. The most essential one is that the accelerograms have to be compatible with
the elastic design response spectrum of the region. Εach accelerogram corresponds to a single
response spectrum for a given damping ratio. On the other hand, on each response spectrum
corresponds an infinite number of accelerograms. Gasparini and Vanmarke (1976) were the first to
propose the creation of artificial accelerograms based on a specific response spectrum. The mean
(μ) response spectrum, is depicted in Figure 1, along with its dispersion (μ+σ and μ-σ denoted with
dotted lines) for the case of the natural class of records (NAT) selected by PEER (2012) and
artificial accelerograms (ART) generated according to the procedure described below in this
section. The response spectra for the NAT class of records were obtained after scaling the records
to the same SA(T1,5%), where T1=0.628 sec for the 40/50 hazard level, in accordance to the hazard
curve of the city of San Diego, California. Comparing NAT and ART class of accelerograms it can
be seen that the dispersion of the response spectra for the ART class of accelerograms is lower to
that of the NAT class.

3
 8 14

7 12

6
10
SA(T1,5%) (m/sec )

SA(T1,5%) (m/sec2)
2

5
8
4
6
3

4
2

1 2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
T (sec) T (sec)

(a) (b)
Figure 1. Mean response spectra of the records for the class of (a) artificial accelerograms and (b) natural
records (dotted lines represent the μ+σ and μ-σ response spectra)

2.3 Intensity measures

Earthquake engineering is a scientific field strongly connected with the effect of strong ground
motions on people and environment. Strong ground motions are extremely complicated and it is
required a lot of data for their full description. The definition of a number of ground motion
parameters, named intensity measures (IM), simplifies the description of a strong ground motion
and links the seismic hazard with the structural data required for the solution of earthquake
engineering problems. The most significant characteristics of a ground motion from an earthquake
engineering point of view are the frequency content, the amplitude and the motion duration. Some
of the IM are related to amplitude or to frequency content, to duration, or to more than one of the
three essential ground motion characteristics. The IM that are related to the effect of more than one
ground motion characteristics are considered more reliable for the description of a ground motion
and are most suitable to reflect the potential damage that a ground motion can produce.
The most commonly used amplitude IM derived from an accelerogram are the peak ground
acceleration (high frequency component), peak velocity (intermediate frequency component), peak
displacement (low frequency component), sustained maximum acceleration and velocity and the
effective design acceleration. Amplitude IMs are used for the derivation of empirical attenuation
relationships used in probabilistic hazard analysis, because their production is based on IMs’
dependence, on the magnitude of the earthquake and on the site-to-source distance. Frequency
content IMs describe through different types of spectra, how the amplitude of a ground motion is
distributed among different frequencies. IMs related to frequency content are Fourier spectra and
power spectra that correspond to the frequency content of the ground motion itself, and the response
spectra that correspond to the influence of the ground motion on structures with different first

4
eigenperiods. Among this type of IMs are spectral parameters, like the predominant period,
bandwidth, central frequency, shape factor, Kanai-Tajimi parameters and the ratio vmax/amax which
describes the frequency content of a ground motion. The most commonly used duration IM is the
bracketed duration which is defined as the time between the first and the last exceedance of a
threshold acceleration, usually equal to 0.05g.
Arias intensity (IA), characteristic intensity (IC) and cumulative absolute velocity (CAV) are
three IMs that reflect the amplitude, the frequency content and the duration of a strong ground
motion respectively which correlate well with structural damage. Arias intensity is defined as the
time-integral of the square of the ground acceleration

 
IA   a(t )2dt (1)
2g 0

where a(t) is the ground acceleration, g is the acceleration due to gravity (9.81 m/sec2) and is
expressed in units of velocity i.e. meter per second. The symbol of infinity in the time integration
means that IA is calculated over the entire duration and not over the duration of the strong ground
motion Td that is defined through specific methods.
Characteristic intensity is defined as
I C  a1.5
rmsTd
0.5
(2)

Where αrms is the rms acceleration (root mean square acceleration) ground motion parameter and
Td is the duration of the strong motion.
Cumulative absolute velocity is defined as the integral of the absolute acceleration in a time
history and it is obtained through the following equation
Td

CAV   a(t ) dt (3)

0

where |a(t)| is the absolute value of the acceleration time series at time t and Td is the duration of
the strong motion. CAV expresses the absolute area under the absolute accelerogram and
correspond to the cumulative absolute velocity that is well correlated with structural damage
(Kramer 1996). These three IMs are used for investigating their dispersion with reference to the
two classes of seismic excitations considered. In particular, the Characteristic intensity is a
measure that is related to an index that quantifies the damage due to the maximum deformation and
the absorbed hysteretic energy. In order to examine the influence of the class of seismic excitations

5
considered on the values of the three IMs, the box plots depicted in Figure 2 have been generated
for the 40/50 hazard level (denoting 40% probability of exceedance in 50 years).

I (m/sec)
4

A
2

0
ART NAT
Seismic Excitation

4
I (m1.5/sec 2.5)

1
C

0
ART NAT
Seismic Excitation

20
CAV (m/sec)

15

10

0
ART NAT
Seismic Excitation

Figure 2. Mean and 25%, 75% confidence bounds for the values of IA, IC and CAV obtained for the classes
of records considered

For each IM considered and for each class of seismic excitation a box plot was created. On each
box, the central mark denoted with a bold black line is the median value of the IM in question. The
edges of the box represent the 25% and 75% percentiles while the whiskers denoted with black
lines extend to the most extreme data points i.e. represent the range of the IM values. For all three
IMs, ART class of accelerograms, has the narrower range and confidence bounds. On the other
hand, the median values of ART are close to each other while NAT varies significantly for the case
of IC and CAV. For the case of IA the three median values are quite close. For the 25% and 75%
percentiles, IC and CAV, NAT class of seismic excitations have similar box sizes denoting denotes
that is related to the length of the range of values.

3. ANALYSIS METHODS

Having an appropriate model for the structure and its earthquake loads, structural analysis is the
next step. The choice of linear elastic or nonlinear modeling of the element force-deformation
behavior and the consideration, or not, of geometric nonlinearities, essentially dictate the selection
of the appropriate analysis method. Modeling of structural mass distribution and the dynamic

6
nature of earthquake loads allows the use of dynamic rather than static methods of analysis. Thus,
broadly speaking, four categories of analysis methods are available, each with its own modeling
and solution algorithm requirements: (a) nonlinear dynamic, (b) nonlinear static, (c) elastic
dynamic (d) elastic static (ASCE 41-06 2007).

3.1 Linear time history analysis methods

Although simple elastic linear methods may seem antiquated by modern research standards, they
define, by overwhelming majority, the current standard-of-practice for seismic design. Despite
many efforts to offer build a design methodology on a nonlinear static or dynamic basis (Fib 2012),
the vast majority of practical earthquake engineering work is still done at the elastic level using
either: (a) Equivalent linear static, whereby a fixed lateral load pattern is applied to an elastic
structural model and (b) linear time history analysis and (c) modal response spectrum analysis,
where the modal responses are combined to estimate the peak MDOF response (e.g., EN1998
2005). Inelastic response is universally taken care by dividing the elastic seismic loads (i.e., design
spectral acceleration values) by the appropriate reduction R (or behavior q) factor that is meant to
represent the ductility and overstrength of a yielding system. Although little recent research has
been directed in the way of elastic methods, recent advancements in nonlinear analysis have helped
to shed some light into the premise of using elastic results to capture nonlinear behavior.
The equilibrium equations for a system in motion can be written as:
Mu(t)  Cu(t)  Ku(t)  R(t) (4)

where M, C, and K are the mass, damping and stiffness matrices; R(t) is the external load vector,
while u(t), u(t) and u(t) are the displacement, velocity and acceleration vectors of the finite
element assemblage, respectively. The choice for a static or dynamic analysis (i.e., for including
or neglecting velocity and acceleration-dependent forces in the analysis) is usually decided by
engineering judgment. Mathematically, Eq. (4) represents a system of linear differential equations
of second order and the solution of this system can be obtained by standard procedures for the
solution of differential equations. In practical finite element analysis, we are mainly interested in a
few effective methods and we will concentrate in the next sections on the presentation of those
techniques and in particular on the direct integration ones. In direct integration the system of linear
differential equations in Eq. (4) are integrated using a numerical step-by-step procedure, the term
“direct” meaning that no transformation of the equations is carried out, prior to the numerical
integration.

7
3.1.1 Central difference method

One procedure that can be very effective in the solution of certain problems is the central difference
method where it is assumed:
1
u(t)  [u(t  t)  2u(t)  u(t  t)] (5)
t 2
and the velocity expansion is defined by:
1
u(t)  [u(t  t)  u(t  t)] (6)
2t
The displacement expression for time t+Δt is obtained by substituting Eqs. (5) and (6) into Eq. (4):
 1 1   2   1 1 
 2 M+ C  u(t  t)  R(t)   K  2 M  u(t)   2 M  C  u(t  t) (7)
 t 2t   t   t 2t 

from which we can solve for u(t+Δt). It should be noted that the solution of u(t+Δt) is based on
using the equilibrium conditions of Eq. (4) at time t. For this reason the integration procedure is
called an explicit integration method. On the other hand, the Houbolt, Wilson, and Newmark
methods, considered in the next sections, use the equilibrium conditions at time t+Δt and are called
implicit integration methods.

3.1.2 Houbolt method

The Houbolt integration scheme is somewhat related to the central difference method in the sense
that standard finite difference expressions are used to approximate the acceleration and velocity
components in terms of displacement. The following expansions are employed in the Houbolt
(1950) integration method:
1
u(t  t)  [2u(t  t)  5u(t)  4u(t  t)  u(t  2t)] (8)
t 2
and
1
u(t  t)  [11u(t  t)  18u(t)  9u(t  t)  2u(t  2t)] (9)
6t
In order to obtain the displacement u at time t+Δt, we now consider Eq. (4) at time t+Δt, which
gives:
Mu(t  t)  Cu(t  t)  Ku(t  t)  R(t  t) (10)

Substituting Eqs. (8) and (9) into Eq. (10) and arranging all known vectors on the right-hand side

8
of the equations, we obtain:
 2 11   5 3 
 2 M+ C + K  u(t  t)  R (t  t)   2 M  C  u(t)
 t 6t   t t 
(11)
 4 3   1 1 
 2 M  C  u(t  t)   2 M  C  u(t  2t)
 t 2t   t 3t 

As shown in Eq. (11), the solution of u(t+Δt) requires knowledge of u(t), u(t-Δt) and u(t-2Δt).

3.1.3 Wilson θ method

The Wilson θ method is essentially an extension of the linear acceleration method, where linear
variation of the acceleration from time t to time t+Δt is assumed. In the Wilson θ method the
acceleration is assumed to be linear from time t to time t+ θΔt, where θ > 1 (Wilson et al. 1973).
Let τ denote the increase in time, where 0≤ τ ≤θΔt; then for the time interval t to t+θΔt, it is assumed:

u(t  )  u(t)  [u(t  t)  u(t)] (12)
t
Integrating Eq. (12), we obtain:
2
u(t  )  u(t)  u(t)  [u(t  t)  u(t)] (13)
2t

and
1 3
u(t  )  u(t)  u(t)  u(t) 
2
[u(t  t)  u(t)] (14)
2 6t

Using Eqs. (13) and (14), we obtain for the time t+θΔt,
t
u(t  t)  u(t)  [u(t  t)  u(t)] (15)
2

2 t 2
u(t  t)  u(t)  u(t)t  [u(t  t)  2u(t)] (16)
6

from which we can solve for u(t  t) and u(t  t) in terms of u(t  t) :
3 t
u(t  t)  [u(t  t)  u(t)]  2u(t)  u(t) (17)
t 2

6 6
u(t  t)  [u(t  t)  u(t)]  u(t)  2u(t) (18)
 t
2 2
t
To obtain the solution for the displacements, velocities, and accelerations at time t+Δt, the

9
equilibrium equations given in Eq. (4) are considered at time t+θΔt. However, since accelerations
are assumed to vary linearly, a linearly extrapolated load vector is used:
Mu(t  t)  Cu(t  t)  Ku(t  t)  R(t  t) (19)

where
R(t  t)  R(t)  [R(t  t)  R(t)] (20)

Substituting Eqs. (17) and (18) into Eq. (19), an equation is obtained from which u(t+θΔt) can be
solved. Then substituting u(t+θΔt) into Eq. (18), we obtain u(t  t) , which is used in Eqs. (12),
(13) and (14), all evaluated at time τ = Δt to calculate u(t  t), u(t  t) and u(t  t) .

3.1.4 Newmark method

The Newmark integration scheme is an extension of the linear acceleration method. Under this
scheme the variation of velocity and displacement are given by (Newmark, 1959):
u(t  t)  u(t)  [(1  )u(t)  u(t  t)]t (21)

u(t  t)  u(t)  u(t)t  [(0.5  )u(t)  u(t  t)]t 2 (22)

where α and δ are parameters that can be determined to obtain integration accuracy and stability.
When δ = 1/2 and α = 1/6, relations (21) and (22) correspond to the linear acceleration method. In
addition to (21) and (22), for calculating the displacements, velocities, and accelerations at time
t+Δt, the equilibrium equations (4) are also considered at time t+Δt:
Mu(t  t)  Cu(t  t)  Ku(t  t)  R(t  t) (23)

Solving from (22) for u(t  t) in terms of u(t+Δt) and then substituting for u(t  t) into (21),
we obtain equations for u(t  t) and u(t  t) each in terms of the unknown displacements
u(t+Δt). These two relations for u(t  t) and u(t  t) are substituted into Eq. (23) to solve for
u(t+Δt) after using (21) and (22), u(t  t) and u(t  t) can be also be calculated. As a result of
this substitution the following well-known equilibrium equation is obtained at each Δt:
K eff u(t  t)  R eff (t  t) (24)
where
K eff  K  0M  1C (25)

R eff (t  t)  R(t  t)  M[0u(t)   2u(t)  3u(t)]  C[1u(t)   4u(t)  5u(t)] (26)

with δ≥0.50; δ  0.25(0.5 + δ)2, while

10
 1  1 1 
0  , 1  , 2  , 3   1,  4   1,
t 2
t t 2 
t   
5    2  ,  6  t(1  ),  7  t.
2  

3.2 Nonlinear time history analysis

Nonlinear dynamic analysis is undoubtedly the most realistic and accurate analysis method
available. It is also referred as “nonlinear time history analysis”, “nonlinear response history
analysis”, or according to ASCE 41-06 (2007) as “nonlinear dynamic procedure” (NDP).
Earthquake loading is taken into consideration as a natural or a synthetic ground motion on a
structural model that incorporates elements with inelastic (inelastic and nonlinear are typically used
interchangeably for seismic applications) force-deformation relationships and at least a first-order
approximation of geometric nonlinearities (P-Delta effects). The propagation of the ground motion
throughout the structure generates complete response histories for any quantity of interest (e.g.
displacements, stress resultants) leading to a wealth of data. While different levels of complexity
are possible due to modeling choices, different ground motion records will produce demands that
vary considerably. This record-to-record variation dominates the application of dynamic methods.
Therefore, in order to get reliable response estimates an appropriately selected set of several ground
motion records is necessary, since a single time history analysis is of limited practical use.
Furthermore, one or more levels of seismic intensity, typically corresponding to one or more levels
of the IM may need to be employed to investigate structural behavior at different regimes of
response or damage (e.g. elastic, post-yield or near-collapse). Thus, nonlinear dynamic analysis
procedures may be categorized into the narrow- and broad-range assessment categories.
In most practical design/assessment situations, only a narrow, single-point estimate of
structural response is required. This is consistent with current seismic codes that only provide a
design hazard spectrum (typically at an exceedance probability of 10% in 50yrs) and require
checking that a structure shall not sustain significant or life-threatening damage at such a level of
intensity. Thus, seismic codes (e.g. ASCE 7-10 2010; EN1998 2005) prescribe using ground
motion records that match or exceed the design spectrum in the period range of interest. If 3 to 6
records are employed, the overall maximum of the recorded peak responses is taken as the
structural demand, while for 7 records and above, the mean of the peak responses can be employed.

11
3.3 Incremental analysis procedures

In the framework of seismic assessment of structures a wide range of seismic records and more
than one hazard levels should be considered in order to take into account the uncertainties inherent
in any seismic hazard for assessing the performance of a structure under seismic actions. The most
reliable, but also computational intensive methodology for the performance-based assessment
implements non-linear dynamic analyses based on a single or a multiple hazard level approach. In
fragility analysis of structural systems or life-cycle cost analysis, the two most appropriate methods
for performing this task are the multiple-stripe dynamic analysis and incremental dynamic analysis
(Vamvatsikos and Cornell 2002). To be consistent with the current technical literature IDA
abbreviation is used for both methods. The main objective of an IDA study is to define a curve
through the relation of the intensity level with the maximum seismic response of the structural
system. The intensity level and the seismic response are described through an intensity measure
(IM) and an engineering demand parameter (EDP), respectively.

Generate N pairs of
Conduct MIDA for each pair Obtain the 16%, 50% and 84%
record-incident angle α using LHS
median curves

1st pair α1 MIDA

.
.
.
Nth pair αN MIDA

Figure 3. The MIDA procedure

Selecting IM and EDP is one of the most important steps of IDA. In Giovenale et al. (2004) the
significance of selecting an efficient IM is discussed. The IM should be a monotonically scalable
ground motion intensity measure, like the peak ground acceleration (PGA), peak ground velocity
(PGV), the ξ=5% damped spectral acceleration at the structure’s first-mode period (SA(T1,5%)) as
well as others. On the other hand, the damage may be quantified by using any of the EDPs whose
values can be related to particular structural damage states. A number of available response-based
EDPs were discussed and critically evaluated in the past for their applicability in seismic damage
evaluation. In the work by Ghobarah et al. (1999) the EDPs are classified into four categories based
on: maximum deformation, cumulative damage and combinations of maximum deformation and
cumulative damage. According to the multicomponent incremental dynamic analysis (MIDA)

12
proposed by Lagaros (2010) which is an extension of IDA, a set of natural records, each one
represented by its longitudinal and transverse components, are applied to the structure in order to
account for the randomness on the seismic excitation. The difference of the MIDA framework from
the original one component version of the IDA stems from the fact that for each record a number
of MIDA representative curves can be defined depending on the incident angle selected, while in
most cases of the one component version of IDA only one IDA representative curve is obtained. A
schematic representation of MIDA procedure can be seen in Figure 3.

(a) (b)
Figure 4. Eight storey steel-RC composite: (a) plan view and (b) front view

4. USE OF TIME HISTORY SEISMIC ANALYSES

4.1 Structural models and numerical simulation

Nonlinear analyses need detailed simulation of the structure in the regions where inelastic
deformations are expected to develop. Either the plastic-hinge or the fibre approach can be adopted
for this cause. Given that the plastic hinge approach has limitations in terms of accuracy fibre beam-
column elements are preferable (Fragiadakis and Papadrakakis 2008). According to the fibre
approach, every beam-column element has a number of integration sections, each divided into
fibres. Each fibre in the section can be assigned concrete, structural steel, or reinforcing bar material
properties. The sections are located either at the centre of the element or at its Gaussian integration
points. The main advantage of the fibre approach is that every fibre has a simple uniaxial material

13
model allowing an easy and efficient implementation of the inelastic behaviour. This approach is
considered to be suitable for inelastic beam-column elements under dynamic loading and provides
a reliable solution compared to other formulations. However, it results to higher computational
demands in terms of memory storage and CPU time. When a displacement-based formulation is
adopted the discretization should be adaptive with a dense mesh at the joints and a single elastic
element for the remaining part of the member. On the other hand, force-based fibre elements allow
modelling a member with a single beam-column element.

(a) (b)
Figure 5. Eight storey RC building: (a) plan view and (b) front view

The two multi-storey buildings shown in Figures 4 and 5 are considered for the numerical tests
presented herein. The first one is an eight storey steel-RC composite building and the second test
example is an eight storey RC building having symmetrical plan view, while the second one is a
five storey RC building with non-symmetrical plan view. The building has been designed for
minimum initial cost (Mitropoulou et al. 2010). Concrete of class C20/25 (nominal cylindrical
strength of 20 MPa) and steel of class S500 (nominal yield stress of 500 MPa) are assumed. Each
member is modelled with a single force-based, fibre beam-column element. The modified Kent
and Park (1971) model is employed for the simulation of the concrete fibres. This model was
chosen because it allows for an accurate prediction of the structural demand for flexure-dominated
RC members despite its relatively simple formulation. The inelastic behaviour of the reinforcing

14
bars was simulated with the Menegotto and Pinto (1973) model. A bilinear material model with
pure kinematic hardening is adopted for the steel fibres, while geometric nonlinearity is explicitly
taken into consideration.

4.2 Critical orientation of the seismic incidence

A structure subjected to the simultaneous action of two orthogonal horizontal ground accelerations
along the directions Ow and Op is illustrated in Figure 6. The orthogonal system Oxyz defines the
reference axes of the structure (structural axes). The angle defined with a counter clock wise
rotation of the structural axis Ox to coincide with the ground motion axis Ow is called as incident
angle of the record.

Figure 6. Definition of the incident angle α

According to Penzien and Watabe (1975) the orthogonal directions of a ground motion can be
considered uncorrelated in the principal directions of the structure. This finding was the basis for
many researchers in order to define the orientation that yields the maximum response when the
response spectrum dynamic analysis was applied. In the work by Wilson et al. (1995) it was
proposed an alternative code approved method that results in structural designs which have equal
resistance to seismic motions from all directions. Lopez and Torres (1997) proposed a simple
method which can be employed by the seismic codes to determine the critical angle of seismic
incidence and the corresponding peak response of structures subjected to two horizontal
components applied along any arbitrary directions and to the vertical component of earthquake
ground motion. The CQC3 response spectrum rule for combining the three orthogonal components
of ground motion to the maximum value of a response quantity is presented in the work by Menun
and Der Kiureghian (1998). In the work by Lopez et al. (2000) it is proposed an explicit formula,
convenient for code applications, in order to calculate the critical value of the structural response
to the two principal horizontal components acting along any incident angle with respect to the

15
structural axes. In the work by Menun and Der Kiureghian (2000) a response spectrum based
procedure for predicting the envelope that bounds two or more responses in a linear structure is
developed. In the work by Anastassiadis et al. (2002) a seismic design procedure for structures is
proposed where the three translational ground motion components on a specific point of the ground
are statistically noncorrelated.
There are only few studies in the literature where the case of the critical incident angle is
examined when time history analysis is employed. In these studies it was found that in the most
general case, where nonlinear behaviour is encountered, it is not an easy task to define the critical
angle. In the work by MacRae and Mattheis (2000) it is shown the ability of the 30% SRSS rule
and the sum of absolute values methods to assess building drifts for bidirectional shaking effects,
while it is also shown that the response is dependent on the reference axes chosen. MacRae and
Tagawa (2001) have found that design level shaking caused the structure to exceed story yield
drifts in both directions simultaneously and significant column yielding occurred above the base.
Shaking a structure in the direction orthogonal to the main shaking direction increased drifts in the
main shaking direction, indicating that 2D analyses would not estimate the 3D response well.
Ghersi and Rossi (2001) examined the influence of bi-directional seismic excitations on the
inelastic behaviour of in-plan irregular systems having one symmetry axis where it was found that
in most cases the adoption of Eurocode 8 provisions to combine the effects of the two seismic
components allows the limitation of the orthogonal elements ductility demand. In the work by
Athanatopoulou (2005) analytical formulae were developed for determining the critical incident
angle and the corresponding maximum value of a response quantity of structures subjected to three
seismic correlated components. Rigato and Medina (2007) studied a number of symmetrical and
asymmetrical structures having fundamental periods ranging from 0.2 to 2.0 seconds where it was
examined the influence that the incident angle of the ground motion has on several engineering
demand parameters.
In order to examine the influence of the incident angle on the seismic response of the structure,
three records have been selected at random and are applied to the steel-RC composite building of
Figure 4. The three records considered are the Loma Prieta, the Leona Valley and the Lake Huge.
The three records have been applied considering a varying incident angle in the range of 0 to 360
degrees with a step of 5 degrees. In order to examine the influence of the incident angle on the
maximum interstorey drift to different intensity levels, the three records have been scaled with
respect to the 5% damped spectral acceleration at the structure’s first mode period to 0.05g, 0.15g,

16
0.30g and 0.50g, and the maximum interstorey drift has been recorded for all the incident angles
and the intensity levels considered. The variation of the maximum interstorey drift with respect to
the incident angle and the intensity level for the three records is depicted in Figure 7. As it can be
seen from the seismic response when the incident angle varies in the range of 0 and 180 degrees
almost coincides with the seismic response corresponding to incident angle varying in the range of
185 to 360 degrees. This is because the relative ratio of the two horizontal components of the
records is close to one, thus the two components are scaled to almost the same intensity level.
Hazard level 1: PGA=0.05g
0.325
Loma Prieta

0.3 Leona Valley

Lake Huge
0.275
θmax (%)

0.25

0.225

0.2

0.175

0.15
0 30 60 90 120 150 180 210 240 270 300 330 360
Incident angle (Degrees)
(a)
Hazard level 2: PGA=0.15g
0.9
Loma Prieta
Leona Valley
0.8
Lake Huge

0.7
θmax (%)

0.6

0.5

0.4

0.3
0 30 60 90 120 150 180 210 240 270 300 330 360
Incident angle (Degrees)
(b)

17
 Hazard level 3: PGA=0.30g
1.6
Loma Prieta
1.5 Leona Valley
1.4 Lake Huge
1.3
θmax (%)

1.2

1.1

0.9

0.8

0.7

0.6
0 30 60 90 120 150 180 210 240 270 300 330 360
Incident angle (Degrees)
(c)
Hazard level 4: PGA=0.50g
2.25
Loma Prieta
Leona Valley
2 Lake Huge
θmax (%)

1.75

1.5

1.25

1
0 30 60 90 120 150 180 210 240 270 300 330 360
Incident angle (Degrees)
(d)
Figure 7. Eight-storey steel-RC composite test example- maximum interstorey drift with respect to the
incident angle of the record scaled to (a) 0.05g, (b) 0.15g, (c) 0.30g and (d) 0.50g intensity levels

4.3 Fragility analysis

In the second part of this numerical investigation four limit-state fragility curves are obtained for
the RC building of Figure 5. The limit-states considered are defined by means of maximum drift
values and cover the whole range of structural damage from serviceability, to life safety and finally
to the onset of collapse. The following θmax values are chosen, according to HAZUS (2003), for
each of the four limit-states: 0.25%, 0.50%, 1.50% and 4.0%. For each limit state the IDA and
“vertical” statistics are implemented for computing the statistical parameters μ and β. “Vertical”
statistics is performed for all four limit states described above. For the purposes of the present
investigation five cases have been examined for the calculation of the two parameters (μk , βk ,
k=1,..4): IDA20, IDA40, IDA60, IDA80 and IDA100 , where 20 to 100 records are implemented.

18
 Figure 8 depicts the fragility curves corresponding to limit-states that cover the whole range of
structural damage developed for two buildings considered. The fragility curves corresponding to
IDA100 are considered as the “correct” ones. For the case of the eight storey symmetric building, it
can be seen that for all limit states IDA20 overestimate the structural capacity while in the cases of
IDA40 and IDA60 the structural capacity is underestimated. Furthermore, it can be seen that 80
records provide a good estimate of the two parameters μk and βk, since the fragility curves of IDA80
almost coincide with those of IDA100 case. Therefore, it can be said that IDA80 case is a good
compromise between robustness and efficiency.
Fragility Curve (θ=0.25%) Fragility Curve (θ=0.5%)
1 1
IDA100 IDA100
0.9 0.9
Probability of exceeding the limit state

Probability of exceeding the limit state

IDA80 IDA80
0.8 IDA60 0.8 IDA60
IDA40 IDA40
0.7 0.7
IDA20 IDA20
0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
SA(T1,5%) (g) SA(T1,5%) (g)

(a) (b)
Fragility Curve (θ=1.5%) Fragility Curve (θ=4.0%)
1 1
IDA100 IDA100
0.9 0.9
Probability of exceeding the limit state

Probability of exceeding the limit state

IDA80 IDA80
0.8 IDA60 0.8 IDA60
IDA40 IDA40
0.7 0.7
IDA20 IDA20
0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
SA(T1,5%) (g) SA(T1,5%) (g)

(c) (d)
Figure 8. Eight storey test example-Fragility curves for four limit-states using alternative number of records

5. SOME REMARKS

The choice for a static or dynamic analysis is usually based on engineering judgment. The use of
static analysis, however, should be justified since otherwise the analysis results are meaningless.
In particular, when nonlinearities are taken into account the assumption of neglecting inertia and

19
damping forces may be severe. In the last decade many efforts to formulate a design methodology
based on a nonlinear static or dynamic basis have been done. In the seismic design and assessment
of structures a wide range of seismic records and more than one performance levels should be
considered in order to take into account the uncertainties that the seismic hazard introduces into a
performance-based seismic assessment or design problem. In construction industry decision
making for structural systems situated in seismically active regions, requires consideration of the
cost of damage and other losses resulting from earthquakes which may occur during the lifespan
of the structures. In this chapter an overview of the time history seismic analysis procedures is
provided along with indicative application of such procedures.

6. CROSS-REFERENCES

Analytic Fragility and Limit States [P(EDP|IM)]: Nonlinear Dynamic ProcedEussraeys (374557).
Assessment of Existing Structures using Response History Analysis (370160).
Empirical Fragility [P(DM|IM)] (374559).
Incremental Dynamic Analysis (368982).
Nonlinear Dynamic Seismic Analysis (368986).
Seismic hazards: Ground Shaking (383916).
Selection of Ground Motions for Time-History Analyses (368960).

7. REFERENCES

Anastassiadis, K., Avramidis, I., Panetsos, P. (2002). Concurrent design forces in structures under three-
component orthotropic seismic excitation. Earthquake Spectra; 18, 1-17.
ASCE, (2007). Seismic rehabilitation of existing buildings, ASCE/SEI Standard 41-06, American Society
of Civil Engineers, Reston, VA.
ASCE/SEI, (2010). Minimum design loads for buildings and other structures, ASCE/SEI Standard 07-10,
American Society of Civil Engineers, Reston, VA.
Athanatopoulou, A.M. (2005). Critical orientation of three correlated seismic components. Engineering
Structures; 27(2), 301-312.
CEN, (2005) Eurocode 8: Design of structures for earthquake resistance, EN 1998, European Committee
for Standardisation, Brussels.
Chopra, A.K., Goel, R.K. (2002). A modal pushover analysis procedure for estimating seismic demands for
buildings, Earthquake Engineering and Structural Dynamics; 31(3), 561-582.
Fajfar, P. (2000). A nonlinear analysis method for performance-based seismic design, Earthquake Spectra;
16(3), 573-592.
FEMA, (2003). National Institute of Building Sciences. HAZUS-MH MR1, Multi-hazard Loss Estimation
Methodology Earthquake Model, Washington, DC.
Fib, (2012). Probabilistic performance-based seismic design, Bulletin 68, International Federation of
Structural Concrete, Lausanne, CH.
Fragiadakis, M., Papadrakakis, M. (2008). Modelling, analysis and reliability of seismically excited
structures: Computational issues. International Journal of Computational Methods, 5(4), 483-511.

20
 Gasparini, D.A., Vanmarke, E.H. (1976) Simulated earthquake motions compatible with prescribed
response spectra, Massachusetts Institute of Technology (MIT), Department of Civil Engineering,
Publication No. R76-4, Cambridge (MA).
Ghersi, A., Rossi, P.P. (2001). Influence of bi-directional ground motions on the inelastic response of one-
storey in-plan irregular systems, Engineering Structures, 23(6), 579-591.
Ghobarah, A., Abou-Elfath, H., Biddah, A. (1999). Response-based damage assessment of structures,
Earthquake Engineering and Structural Dynamics; 28(1), 79-104.
Giovenale, P., Cornell, C.A., Esteva, L. (2004). Comparing the adequacy of alternative ground motion
intensity measures for the estimation of structural responses, Earthquake Engineering and Structural
Dynamics; 33(8), 951-979.
Houbolt, J.C. (1950). A recurrence matrix solution for the dynamic response of elastic aircraft, Journal of
the Aeronautical Sciences; 17, 540-550.
Kent, D.C., Park, R. (1971). Flexural members with confined concrete. Journal of Structural Division;
97(7), 1969-1990.
Kramer, S.L. (1996). Geotechnical Earthquake Engineering, Prentice Hall.
Lagaros, N.D. (2010). Multicomponent incremental dynamic analysis considering variable incident angle,
Journal of Structure and Infrastructure Engineering; 6(1-2), 77-94.
Lopez, O.A., Chopra, A.K., Hernandez, J.J. (2000). Critical response of structures to multicomponent
earthquake excitation. Earthquake Engineering and Structural Dynamics; 29, 1759-1778.
Lopez, O.A., Torres, R. (1997). The critical angle of seismic incidence and the maximum structural
response, Earthquake Engineering and Structural Dynamics; 26, 881-894.
MacRae, G.A., Mattheis, J. (2000) Three-dimensional steel building response to near-fault motions. Journal
of Structural Engineering; 126(1), 117-126.
MacRae, G.A., Tagawa, H. (2001). Seismic behaviour of 3D steel moment frame with biaxial columns.
Journal of Structural Engineering; 127(5), 490-497.
Menegotto, M., Pinto, P.E. (1973) Method of analysis for cyclically loaded reinforced concrete plane frames
including changes in geometry and non-elastic behaviour of elements under combined normal force and
bending. Proceedings, IABSE Symposium on Resistance and Ultimate Deformability of Structures Acted
on by Well Defined Repeated Loads: 15-22.
Menun, C., Der Kiureghian, A. (1998). A replacement for the 30%, 40% and SRSS rules for multicomponent
seismic analysis. Earthquake Spectra; 14(1), 153-163.
Menun, C., Der Kiureghian, A. (2000). Envelopes for seismic response vectors. I: theory. Journal of
Structural Engineering; 126: 467-473.
Mitropoulou, Ch.Ch., Lagaros, N.D., Papadrakakis, M. (2010). Economic building design based on energy
dissipation: a critical assessment, Bulletin of Earthquake Engineering; 8(6), 1375-1396.
Newmark, N.M. (1959). A method of computation for structural dynamics, ASCE Journal of Engineering
Mechanics Division; 85, 67-94.
Pacific Earthquake Engineering Research (PEER), (2012). http://peer.berkeley.edu/smcat/search.html/ (last
accessed November 2012).
Penzien, J., Watabe, M. (1975). Characteristics of 3-dimensional earthquake ground motions, Earthquake
Engineering and Structural Dynamics; 3(4), 365-373.
Rigato, A.B., Medina, R.A. (2007). Influence of angle of incidence on seismic demands for inelastic single-
storey structures subjected to bi-directional ground motions, Engineering Structures, 29(10), 2593-2601.
Vamvatsikos, D., Cornell, C,A. (2002). Incremental dynamic analysis, Earthquake Engineering and
Structural Dynamics; 31(3), 491-514.
Wen, Y.K., Song, S.-H. (2003). Structural reliability/redundancy under earthquakes. Journal of Structural
Engineering; 129(1), 56-67.
Wilson, E.L., Suharwardy, A., Habibullah, A. (1995). A clarification of the orthogonal effects in a tree-
dimensional seismic analysis, Earthquake Spectra; 11(4), 659-666.
Wilson, E.L., Farhoomand, I., Bathe, K.J. (1973). Nonlinear dynamic analysis of complex structures,
Earthquake Engineering and Structural Dynamics, 1, 241-252.

21

View publication stats