Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

This article was downloaded by: [DTU Library]

On: 02 May 2014, At: 08:25


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/ueqe20

AN ENERGY-BASED FORMULATION FOR FIRST-AND


MULTIPLE-MODE NONLINEAR STATIC (PUSHOVER)
ANALYSES
a b c
ENRIQUE HERNÁNDEZ-MONTES , OH-SUNG KWON & MARK A. ASCHHEIM
a
Associate Professor, University of Granada , Campus de Fuentenueva, 18072 Granada, Spain
E-mail:
b
Graduate Research Assistant, University of Illinois , 205 North Mathews, Urbana, Illinois,
61801, USA E-mail:
c
Associate Professor, Civil Engineering Department , Santa Clara University , 500 El Camino
Real, Santa Clara, California, 95053, USA E-mail:
Published online: 03 Jun 2008.

To cite this article: ENRIQUE HERNÁNDEZ-MONTES , OH-SUNG KWON & MARK A. ASCHHEIM (2004) AN ENERGY-BASED
FORMULATION FOR FIRST-AND MULTIPLE-MODE NONLINEAR STATIC (PUSHOVER) ANALYSES, Journal of Earthquake Engineering,
8:1, 69-88, DOI: 10.1080/13632460409350481

To link to this article: http://dx.doi.org/10.1080/13632460409350481

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the
publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations
or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any
opinions and views expressed in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses,
actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever
caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone
is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/
terms-and-conditions
Journal of Earthquake Engineering, Vol. 8, No. 1 (2004) 6%88
@ Imperial College Press $~ r n p College
e ~ PRSS
www.icprass.co.uk

AN ENERGY-BASED FOFtiMULATZON FOR FIRST-


AND MULTIPLEMODE NONLINEAR STATIC
(PUSHOVER) ANALYSES

ENRJQUE HERNANDEZ-MONTES
Associate Professor, Uniuersity of Gtrmudu,
Campzrs de fientenueva, 18072 Gmnadu, Spain
emontes Ougr.es

OH-SUNG KWON
Gmduute Reseanh Assistant, University of Illinois,
Downloaded by [DTU Library] at 08:25 02 May 2014

205 North Muthews, Urbana, Illinois 61801, USA


k5sung@yahoo.corn

MARK A. ASCHHEIM
Associate Pm,/essor, Civil Engineering Department,
Santa Clam Uniue~sity,500 El Camino Real,
Santa Clam, California 95053, USA
mllschheimOscu. edu

Received 13 December 2002


Revised 20 March 2003
Accepted 19 May 2003

Existing nonlinear static (pushover) methods of analysis establish the capacity curve of
a structure with respect t o the roof displacement. Disproportionate increases in the roof
displacement, and even outright reversals in the case of higher mode pushover analyses,
can distort the capacity curve of the "equivalent" SDOF system. Rather than viewing
pushover analyses from the perspective of roof displacement, this paper considers the
energy absorbed (or the work done) in the pushover analysis. Simple relations establish
an energy-based displacement that is equivalent to the spectral displacement obtained
by conventional pus hover analysis methods within the linear elastic domain. Extensions
to the nonlinear domain allow pushover curves to be established that resemble tradi-
tional first mode pushover curves and which correct anomalies observed in some higher
mode pushover curves. An acample illustrates the application of a modified Multimode
Pushover Analysis procedure wing Yield Point Spectra.

Keyworth: Nonlinear static analysis; modal pushover; absorbed energy.

1. Introduction
The use of pushover analysis methods for characterising the predominant mode
of response of structures responding nonlinearly to earthquake ground motions
has become well established in recent years. Approximate and exact first mode
analysis procedures are accepted procedures in documents such as ATC-40 (19971
and FEMA-356 [2000]. Recent proposals have focused on combined mode [Mwafy
70 E. H d n d e z - M o n t e s , 0.3.Kwon tY M. A. Aschheim

and Elnashai, 20011 and multiple mode analysis procedures [Gupta and Kunnath,
2000; Chopra and Goel, 20021. In each analysis method, lateral forces are applied
monotonically in a stepby-step static analysis. In the case of modal analysis pro-
cedures, the applied lateral forces are proportional to the product of the mass and
mode shape amplitude at each level of the structure. In other cases, a non-modal
shape vector may be used in place of the mode shape, where the shape vector may
be an inverted triangular shape or a rectangular shape. In still other cases, the
lateral forces may be proportional to a code lateral force distribution. In the case
of adaptive pushover analyses, the load patterns are modified as the analysis pr+
gresses to reflect changes in structural properties that occur with the development
of nonlinearity in the structural components.
In pushover analysis procedures, the behaviour of the structure is characterbed
by a capacity curve. In nearly all csses, the capacity curve is a plot of the base
shear force versus the displacement of the roof, as determined in the stepby-step
Downloaded by [DTU Library] at 08:25 02 May 2014

pushover analysis. This representation is convenient for use in design offices and is
both meaningful to and easily visualised by the engineer. The roof displacement is
used in current procedures not only as an index for the capacity curve, but also to
establish the seismic demands over the height of the structure a t the estimated peak
displacement, or performance point. For example, for the conventional first mode
pushover (e.g. ATC-40) and the Multimode Pushover Analysis [Chopra and Goel,
20021 methods, the seismic demands are determined throughout the structure based
on the peak roof displacement estimated in each of the modal pushover analyses.
It is recognised that the roof displacement was selected for this representation
because it is convenient for use in practice. In the linear elastic domain, the floor
and roof displacements increase proportionately when subjected to proportionately
increasing lateral forces, as is done in a typical pushover analysis. The capacity
curve could just as easily be based on the displacement at any floor, but the roof
displacement has been preferred because it emphasizes the overail response of the
structure and provides better numerical accuracy, particularly when higher modes
are involved. While the roof displacement is useful for characterising the behaviour
of many buildings such as the moment resistant frame of Fig. l(a), it is not clear
that the roof displacement is the most meaningful index for other structures, such
as the braced frames of Figs. l(b) and l(c), even for linear elastic behaviour under
quasi-first mode lateral force patterns.

(b)
Fig. 1. Different type of frames.
First- and Multiple- Mode Nonlinear Static Analyses 71

When nonlinear behaviour develops in the pushover analysis, the displacements


of the floors and roof will increase disproportionately with increasing load, in
general. The arbitrary choice to plot the base shear as a function of the roof
displacement introduces an arbitrariness to the inelastic portion of the capacity
curve. For systems with sharply defined yield points, disproportionate increases
in displacements over the height of the building primarily affects the post-yield
stiffness of the capacity curve. Because small deviations in the post-yield stiffness
of the capacity curve of the "equivalent" single-degree-of-freedom (SDOF) system
typically have only minor effects on the dynamic response statistics, any departures
from theoretically ideal values can be difficult to discern in computational studies.
Where yielding is more gradual, disproportionate increases in the roof displace-
ment may, in addition, affect the effective yield strength that is determined for the
structure: when methods such as those described in ATC-40 [I9971are used.
The arbitrariness in the choice of the roof displacement as the index used for
plotting the capacity curve is readily apparent in pushover curves obtained for
Downloaded by [DTU Library] at 08:25 02 May 2014

the second, third, or higher modes, as has been observed by some of the primary
contributors to multiple mode pushover analysis methods [Goel and Chopra, 20021.
Roof displacements may increase at a decreasing rate or may even reverse, leading to
capacity curves that display unusual behaviour, as illustrated subsequently. A literal
interpretation of the capacity curves obtained in these cases would indicate that the
structure does not always absorb energy in a pushover analysis, but instead, may
be a source of energy for some inelastic regimes. Such an interpretation implies a
violation of the first law of thermodynamics, and points out the degree to which the
use of the roof displacement can be misleading. There is no doubt that external work
is consumed by the deformations of plastic hinges (and any changes in recoverable
strain energy) that take place in a monotonic pushover analysis.. The notion that
the structure maybe a source of energy is a consequence of the arbitrary choice to
use the roof displacement as the index (abscissa) of the capacity curve.
This paper aims to provide a more general characterisation of the resistance
developed by a multi-degree-of-freedom (MDOF)structure under increasing lateral
loads in a pushover analysis by explicit consideration of the work done throughout
the entire structure. The intent of this paper is to retain the utility of both first
mode and multiple mode methods of analysis while improving their theoretical
basis. Rather than relying on the displacements at an arbitrary location, absorbed
energy (or equivalently, external work) is used to characterise the resistance of the
structure to lateral forces. Simple relationships are developed to allow the resistance
function to be expressed in terms of the displacement of an equivalent SDOF
system, thereby providmg an alternative means to generate the capacity curve
of the equivalent SDOF system. For the sake of simplicity, the resistance of the
structure is determined with respect to invariant modal force patterns, although
simple extensions can make the procedure compatible with adaptive force patterns.
The resulting capacity curves appear to be generally compatible with those obtained
from traditional first mode pushover analyses, and thereby provide a stronger
72 E. Hernbndez-Montes, 0.-S.Kwon & M. A. AscMeim

theoretical underpinning for the use of conventional first mode pushover analyses.'
The results obtained for higher mode pushovers avoid the problems associated with
the arbitrary choice of the roof displacement. However, the hypothesis that in-
dependent modal pushover analyses may be combined in the domain of inelastic
response [Chopra and Goel, 20021 is not addressed and may require further study
to be adequately assessed.

2. Theoretical Basis of Pushover Methods


The differential equation of the dynamic response of a linear elastic multi- degree
of freedom structure subjected to a horizontal base excitation iig isa:

In the case of a multistorey building, u is a vector of N components that represents


Downloaded by [DTU Library] at 08:25 02 May 2014

the lateral displacements of the floors relative to the base, and m, c and k are the
mass, damping and stiffness matrices of the structure. The vector 1 is a column
vector with each component equal to 1, s is a vector equal to ml and represents
the shape of the effective forces pea(t).
The displacement vector, u, can be decomposed into components expressed in
terms of the free vibration mode shapes ($,), where qn is the nth modal coordinate.
N
u= C
n=l
#n~n(t) (2)

The expression of the displacement vector in terms of the mode shapes [Eq. (2)]
allows the system of N coupled equations represented by Eq. (1)to be uncoupled in
terms of the modal coordinates. Substitution of Eq. (2) into Eq. (1)and application
of the properties of orthogonality of the free vibration mode shapei with respect to
m, c and k result in:
iin( t ) + i ~ n w qn
n ( t ) + wEqn ( t ) = -rnGg ( t ) (3)
where C, is the damping ratio, w, is the natural vibration frequency and r, is
the modal participation factor. A further simplification can be achieved by setting
qn(t) = r,D,(t), resulting in the following differential equation of motion for the
SDOF system:

The solution of Eq. (4) for the D,(t) corresponding to each mode is the basis of
modal response history analysis (MRHA), for which the vector u is given by:

aFor the sake of simplicity, the notation used by Chopra and Goel [2002] is adopted, where
applicable.
First- and Multiple-Mode Nonlinear Static Analyses 73

In order to address the forces that act on the structure for each modal response,
the equation of motion for the response of the SDOF system representing the nth
mode is presented. Displacements are proportional to the nth mode shape:

allowing Eq. (1) to be expressed as

The effective force, pefi( t ) can be decomposed taking note of the orthogonality
of the mode shapes with respect to the mass matrix:

Substituting Eq. (8) into Eq. (7), and multiplying both sides by #: results in
Downloaded by [DTU Library] at 08:25 02 May 2014

which indicates that only the s, component of pee results in a non-zero response
in the nth mode, for which a, is equal to the participation factor .,'I Thus, sn and
pea,n can be expressed as:

Thus, it is apparent that only p , ~ , ,causes response in the nth mode.


An equivalent static force can b e associated with the nth mode displacement
u,(t). The equivalent static force f, is the statically applied force that results in a
displacement equal to u,(t):

fn ( t ) = ku, ( t )= k&q,(t) = w ~ r n & r , ~ ,( t ) = s,A, ( t ), (11)


where A, ( t ) is the pseudo-acceleration [A, ( t ) = wjf ~ , ( t ) ] For
. the elastic response,
any response quantity r ( t ) (e.g. displacements, internal element forces, or moments)
may be calculated as a combination of each of the modal responses r,(t) in MRHA:

r(t) = C r, ( t )= C ritA, ( t ),
where r z is the static response of quantity T due to the external force s,.
The peak value of r for the nth mode is called T,,. Thus, r,. = rit - An,
where A, is the ordinate of the pseudeacceleration design (or response) spectra
corresponding to the nth modal period. The peak value of the total response of the
quantity r, given by r,, can be estimated according to combination rules such as
CQC or SRSS. These rules combine the peak values obtained for each mode.
74 E. H m d n d e z - M o n t e s , 0.3.K w m & M. A. Aschheim

Also of interest is the static force that produces the maximum displacements
for each mode, f,, ,

where s, and :s are vectors having different lengths but the same shape.

3. Conventional Pushover Analysis Methods


The ATC-40 document refers to several pushover methods involving various quasi-
first mode approaches and multiple mode approaches. The Multimode Pushover
Analysis (MPA) procedure of Chopra and Goel [2002] combines quantities deter-
mined in independent modal pushover analyses. The capacity curve determined
for the "equivalent" SDOF system for the first mode load pattern of the MPA
procedure is identical to that determined in the first mode, pushover method of
Downloaded by [DTU Library] at 08:25 02 May 2014

ATC-40. Thus, the following discussion will refer to the nth mode pushover of the
MPA procedure, recognising that the specialisation for n = 1 applies equally to the
ATC-40 first mode pushover.
The MPA procedure proposes to estimate peak dynamic response quantities of
inelastic structures based on a combination of possibly nonlinear responses obtained
independently for each mode. The structure is subjected to a static force distributed
over the height of the building according to s i , with amplitude increasing until the
roof displacement equals or exceeds the maximum displacement, u,,,, expected
in each mode (u,,, = I',#,,D,), where &, is the component of 4, at the roof
level. The peak modal responses r,,, each determined in an independent modal
pushover analysis, are combined according to the SRSS method to obtain an esti-
mate of the peak value r , of the total response, where the r,, values may refer to
floor displacements, interstorey drifts, storey shears, overturning moments, or other
response quantities.
When applied to structures responding inelastically, the method neglects the
influence of the modal forces s, on the response of other modes. That is, superpw
sition is assumed, or alternatively, interaction among the modes is neglected, just
as in elastic modal analysis.
An algebraic mapping is used in order to relate the capacity curves obtained
in modal pushover analyses to a common design or response spectrum for each of
the SDOF systems. To establish the algebraic mapping, the equations of motion for
the MDOF system are developed, and from these, the expressions for the SDOF
system may be extracted.
The maximum displacement of the SDOF system is plotted on the abscissa of the
common capacity spectrum representation (hereafter referred to as the "common"
representation). The nth mode capacity curve is plotted with respect to one of
the lateral degrees of freedom, typically the roof displacement urn. According to
First- and Multiple-Mode Nonlinear Static Analyses 75

Eq. (5):

Equation (14) gives the relation between the displacement of the capacity curve of
the "equivalent" SDOF system (D,)and the roof displacement for the nth mode
(u,,). Thus, the common representation for the nth mode pushover is obtained by
plotting on the abscissa the following quantity:

where urn is the obtained from the pushover analysis with forces applied propor-
tional to s,.
In Chopra and Goel's description of the nth mode capacity curve (Vh versus
urn)the term k, is used to represent the ratio of Vh and u,, in the elastic domain.
Downloaded by [DTU Library] at 08:25 02 May 2014

Because this may be misinterpreted as suggesting that the ratio is independent of


the locations where the displacements and shears are measured, we prefer to use
the term k+, for this ratio, resulting in Vh = b, u,, in the elastic domain.
The ordinate of the common representation is the restoring force of the unit
mass SDOF system, given by the third term of the left side of Eq. (4)) which has
maximum value w ~ D , , . The equation for the base shear, Vh,of the MDOF system
is developed below, in order to identify the values to be plotted on the ordinate of
the common representation.

An expression for krn can be obtained from:

The eigenproblem used to determine the undamped free vibration mode shapes
establishes that

Thus, the value to plot on the ordinate of the common representation is obtained
from Eqs. (16),(17),and (18)as:

. where a, is the modal mass coefficient for the nth mode.


76 E. Hedndez-Montes, 0.3.Kwon & M. A. Aschheim

4. Example Illustration
The example considered is the three-storey steel frame designed for Los Angeles
as part of the SAC project. The building consists of four bays in the north-south
direction and six bays in the east-west direction. The building is 120 ft (36.58 m)
by 180 ft (54.86 m) in plan and 39 ft (11.89 m) in elevation, with s 2 ft (0.61 m)
extension horn the perimeter column lines to the building edge. Typical floor-
tefloor heights are 13 ft (3.96 m). Figure 2 shows an elevation of north-south
lateral force-resisting system of the building. For the frame presented in Fig. 2,
all connections are moment-resistant for the first three bays. The interior bays
consist of frames with simple connections. The last column line consists of "dummy"
columns used to model P-Delta effects associated with the gravity framing, although
PiDelta effects were not modeled in the analyses reported herein. Gravity loads are
provided in Table 1. The M I model of the frame (FEMA-355C, 2000) was used, for
which beam and column framing is modelled along centreline dimensions, without
Downloaded by [DTU Library] at 08:25 02 May 2014

representing the panel zones.


The conventional MPA procedure requires that the capacity curves be deter-
mined in a number of independent modal pushover analyses. Figure 3 presents the
capacity curves for modes 1, 2 and 3'as plots of base shear versus roof displacement.
Results were computed both with SAP2000NL (CSI,2000) and Drain-2DXVersion
1.10 [Prakash et al., 19931.

Roof

3rd

2nd

Ill

Fig. 2. Elevation, framing members, and element and nodal numbering for M1 model of SAC
t hree-storey steel frame.

Table 1. Dead and live loads applied to the three-storey frame


example.

Floor dead load for weight calculations 96 psf (4.60 kN/m2)


Floor dead load for mass calculations 83 psf (3.98kPJ/m2)
Roof dead load excluding penthouse 83 psf (3.98 kN/m2)
Penthouse dead load 116 psf (5.56 k ~ / r n ~ )
Reduced live load per floor and for roof 20 psf (0.96 kPI/m2)
First- and Multiple-Mode Nonlinear Static Andyses 77

The capacity curves for the first and second mode pushover analyses of Fig. 3
display softening behaviour, and sensible estimates of peak displacement response
can be obtained using the Capacity Spectrum and Yield Point Spectra [Aschheim
and Black, 20001 approaches. However, the third mode capacity curve displays
. reversal behaviour after yielding. Clearly, within the linear elastic domain, the
Downloaded by [DTU Library] at 08:25 02 May 2014

2 3 4
Roof Displacement/tleight (%)

Fig. 3. Comparison between pushover curves obtained from DRAIN-2DX and those obtained
from ~ ~ ~ 2 (three-storey
0 0 0 st& frame, without P-Delta).

Floor

-1 -0.8 4.6 -0.4 -0.2 0 0.2


Floor DisplacernentRteight (%)

Fig. 4. Floor displacement in the third mode pushover analysis (locations identified in Fig. 3).
78 E. H d n d e r - Montes, 0.3.
Kwon & M. A. AscMeirn

displ+cements over the height of the structure will remain proportional to the force
distribution in the pushover analysis [ s i in Eq. (13)l. This indicates that the d i s
placements at any level may be used interchangeably for plotting the capacity curve
for the linear elastic response.
The displaced shapes corresponding to the points indicated by letters A-E on
the third mode capacity curve of Fig. 3 are shown in Fig. 4. The non-proportional
increases in floor displacements, including the reversal of the roof displacement, are
apparent and illustrate the arbitrariness of the selection of the roof displacement
in the MPA method, for this conventional three-storey structure.

5 . A n Energy-Based Formulation of Modal Pushover Analysis


At its core, the capacity curve of a structure represents the development of resistance
to lateral forces as a function of increasing lateral displacements. The capacity curve
Downloaded by [DTU Library] at 08:25 02 May 2014

has great value in characterising the degree of nonlinearity that may develop in a
first or predominant "mode", recognising, of course, that the onset of nonlinearity
causes changes in modal properties and invalidates modal superposition. Because
floor displacements over the height of the building generally increase disproportion-
ately as the response becomes increasingly nonlinear, one cannot rigorously justify
the use of the displacement at any one location for the abscissa of the capacity curve,
since the apparent post-yield stiffness of the capacity curve depends on the location
selected. As shown in the previous example, outright reversals in the capacity curve
may result in some cases. Rather than relying on the roof displacement, we propose
to use the en&&'ibsorbed by the structure in each modal pushover analysis to
determine the corresponding capacity curve of the equivalent SDOF system, recog-
nising that the behaviour of the MDOF system and its analogous SDOF system can
be appreciated from both conventional and energy-based perspectives. The energy-
based formulation developed below avcids the arbitrary selection of a single floor
(or roof) location as the parameter for representing the capacity curve, and may
be used with single or multimode analysis procedures.
The equation of motion is often expressed as the dynamic equilibrium of force
quantities [Eq. (I)],but can equivalently be expressed in terms of energy quantities.
The "absolute" energy form of Eq. (1)) expressed in terms of the energy developed
from the time that the excitation starts, can be obtained by integrating Eq. (1)
with respect to displacement, as described by Usng and Bertero [1988]:

where mi is the lumped mass associated with the ith story and iiti is the absolute
(or total) acceleration at the ith story, and f, is the restoring force.
In both the "absolute" and "relative" energy formulations of the equation of
motion, the absorbed energy, E, is
First- and Multiple- Mode Nonlinear Static Analyses 79

which is the third term of Eq. (20). The absorbed energy is composed of the
recoverable elastic strain energy and nonrecoverable energy associated with energy
dissipated by the hysteretic response of the structural components.
As pointed out in the development of Eq. ( l l ) ,the static force associated with
the nth mode is f,(t). The restoring force is assumed to be equal to sum of the
modal components fn(t). Following this assumption, the restoring force f, can be
represented in terms of its modal components:

fb ( t )= fn (t) = ugm&J',~, ( t ).

Due to the orthogonality of modes with respect to k (q5T . k - 4j # 0 for i = j


and 0 otherwise) the force fn does work only for displacements in the nth mode.
The work done by this force on the other modal displacements is zero.
In the elastic domain, the absorbed energy associated with the static force f,
going through an elastic displacement from 0 to u, may be computed by substi-
Downloaded by [DTU Library] at 08:25 02 May 2014

tuting Eq. (22) for f,, and Eq. (5) for u,:

The corresponding base shear associated with the nth mode pushover is:

Substituting Eq. (24) into Eq. (23), we obtain

for the response in the elastic domain.


Equation (25) can be interpreted graphically as the area beneath the curve in
a plot of Vi,, with respect to D,, in the elastic domain (Fig. 5). Therefore, we
define the energy-based displacement, D,,,,to be equal to 2En/Vb,, in order to

y
.----+ D,

r I

Elastic Domain

Fig. 5. Extension of D,,, definition to inelastic domain. .


80 Kwon & M. A. Aschheim
E. H d n d e z - M o n t e s , 0.3.

assure that D,,,= D, in the elastic domain. More generally, for both the elastic
and inelastic response, the work done by VbInin a differential displacement dD,,,
is d E , :

which is necessarily equal to the work done by the static force f, in a differential
displacement of the structure in this mode. Using an incremental formulation, the
terms AE, and Vbncan be computed for each step in the pushover analysis. Then,
the corresponding increment in the energy-based displacement, AD,,,, may be
calculated as

The value of D.,, corresponding to the base shear is determined by summation.


Equation (27) is consistent with Eq. (25) in the elastic domain.
Downloaded by [DTU Library] at 08:25 02 May 2014

The possible influence of -changes in the deformed shape from static forces
associated with modes other than the nth mode is neglected in this formulation,
because orthogonality of the load vector and the elastic mode shapes is assumed,
as described earlier.
As with conventional pushover approaches, the mapping for the ordinate of the
common representation can be obtained by solving Eq. (24) for the term w I f ~ , ( t ) :

(28)

In this case, the values plotted on the ordinate of the common representation are
determined as before, as:

6. Example ~llustration

The three-storey moment-resistant frame of the previous example was subjected


to the 1940 NS El Centro ground motion. The amplitude of the ground acceler-
ation was scaled in order to obtain a peak roof displacement equal to 2% of the
height of the building in a nonlinear dynamic analysis, performed using the program
Drain-2DX.
Figures 6, 7, and 8 plot the capacity curves obtained by nonlinear static
(pushover) analysis for the first, second, and third modes. In these figures, the
base shear is plotted with respect to the roof displacement and with respect to
the energy-based displacement D,,,[determined using Eq. (27)). In these plots, the
values expressed by Eqs. (15) and (19) are used in the case of the conventional
MPA procedure (based on the roof displacement), while the values of Eqs. (27)
and (29) are used in the case of the energy-based approach, for the abcissas
and ordinates, respectively. Figure 6 illustrates that the energy-based represen-
tation of the first mode response nearly coincides with the conventional first mode
First- and Multiple-Mode Nonlinear Static Analyses 81

-Energy based i1
- Conventional 1

I!
s

0 200 400 600 800 1000 1200


Spectral displacement (mm)
Fig. 6. Capacity curve obtained from first-mode pushover analysis,plotted using current ATC-40
met hod and using the energy-based displacement.

g 6-
Downloaded by [DTU Library] at 08:25 02 May 2014

a 5-
-
a 4:
-P 2-
-Energy based
=s' 1- - Conventional

Spectral displacement (mm)


(a) Entire curve

!
u

-Energy based
-Conventional
I I

Spectral displacement (mm)

(b) Close-up of initial portion


''

Fig. 7.Capacity curve obtained fromsecond-mode pushover analysis,plotted using current MPA
method and using the energy-based displacement.

pushover capacity curve that is determined using the MPA and ATC-40 procedures.
Figure 7(a) illustrates some departure of the conventional and proposed approaches
for the nonlinear portion of the second mode response. The use of the roof displace
ment in t h e conventional MPA approach leads to an apparent stiffening in the
post-yield response, while the energy-based approach shows monotonic softening
5i
Y

-
C
.-0
m
-2
Q)
0
0
-0 -Energy based
2 5- - Conventional
C
0 n
Q) r
V r i

5 -4500 -2500 -500 1500 3500


Spectral displacement (rnm)
(a) Entire curve
Downloaded by [DTU Library] at 08:25 02 May 2014

-a
b

8 -Energy based
-EO
s
c.
0
- Conventional
a
A
I

Spectral displacement (mm)


(b) Close-up of initial portion
Fig. 8. Capacity curve obtained in third-mode pushover analysis, plotted using current MPA
method and using the energy-based displacement.

with increasing displacement. Because each plastic hinge has a bilinear moment-
rotation relationship, the apparent stiffening after initial yielding is not physically
reasonable, but instead must be viewed as a consequence of the arbitrary choice to
index the capacity curve by a displacement that increases, disproportionately in
the nonlinear regime. Figure 7(b) shows a close-up view of the elastic portion of
the second mode capacity curve, illustrating the equivalence of the energy-based
and conventional capacity curves in the elastic domain. Figure 8(a) illustrates that
the reversal apparent with the conventional roof-displacement approach is rectified
when the energy-based approach is used. A close-up of the elastic portion of the
third mode capacity curve is shown in Fig. 8(b). In all cases, the energy-based and
conventional capacity curves are the same in the linear elastic domain.
The rate at which the floor displacements increase in each modal pushover
analysis is shown in Figs. 9, 10,and 11. Plotted against the energy-based displace-
ment, D,,,,are the displacements of the floors (or roof) for each of the modal
pushover analyses. The floor displacements are normalised by rn#i ,*, where # i , n
is the amplitude of the nth mode shape at the ith floor level. Each plot illustrates
First- and Multipte-Mode Nonlinear Static Analyses 83

0 . 200 400 600 800 1000 1200


Energy-based displacement (mm)
I -+First floor 1
-

++ Second floor ++Roof +Energy-based

Fig. 9. Comparison of floor displacements during first-mode pushover andysis.


Downloaded by [DTU Library] at 08:25 02 May 2014

0 100 200 300 400 500 600 700


Energy-based displacement (mm)
I +- First floor *Second floor +Roof *Energy-based I
Fig. 10. Comparison of Aoor displacements during second-mode pushover analysis.

Energy-based djsplacement (mm)


I 4-First floor +Second floor +Roof *Energy-based I
Fig. 11. Comparison of floor displacements during third-mode pushover analysis. .

that any floor level can be used interchangeably to index the capacity curve for
the elastic response, but the floor displacements increase disproportionately in the
nonlinear regime of response. The roof displacement can be observed to reverse in
Fig. 11. Thus, the choice to use any one displacement to index the capacity curve is
seen to introduce an arbitrariness to the post-yield behaviour of the capacity curve.
84 E. Hedndez-Montes, 0.4.
Kwon M. A . AscMeim

The energy-based displacement is based on the overall behaviour of the structure


and appears to properly convey the softening of the MDOF system into the capacity
curves of the equivalent SDOF systems.
The SDOF capacity curves for each modal pushover analysis (Figs. 6, 7, and 8)
can be plotted together with curves representing the spectral demands in order to
estimate the peak responses of each mode, following the basic idea of the ATC-40
and MPA procedures. Figure 12 plots the capacity curves together with the Yield

-Demand spectrum (p =1.68)


- 1st mode capacity curve
0 Effectiveyield point
Perfamanee point
Downloaded by [DTU Library] at 08:25 02 May 2014

Spectral displacement (mm)


(4

13mrnode capacity curve I ~erfomancemint 1I

Spectral displacement (mm)


(b)
Fig. 12. Determination of modal displacement demands using Yield Point Spectra for (a) first
rnode response and (b) second and third rnode responses.
First- and Multiple-Mode Nonlinear Static Analyses 85

Point Spectra (YPS), as described by Aschheim and Black (20001 for the scaled El
Centro ground motion that was used in the dynamic analysis. In Fig. 12(a), the only
YPS curve shown is the one that passes through the effective yield point of the first
mode SDOF system. In this case, the constant ductility curve that passes through
the effective yield point has a ductility of 1.68; thus the peak displacement of the
first mode SDOF system is 1.68 times the effective yield displacement of the SDOF
system. Because the second and third modes are determined to respond elastically
according to this procedure, the peak displacements are at the intersection of the
elastic spectralxurve and the second and third mode capacity curves, as shown in
Fig. 12(b).
Several approaches may be taken to estimate the peak displacements of the
MDOF system based on the estimated SDOF responses. If the displacements are
proportional to the elastic mode shape, conventional relationships between the peak
displacement (and ductility) of the equivalent SDOF system and the MDOF system
may be used. Because the displacements may cease to be proportional, the estimated
Downloaded by [DTU Library] at 08:25 02 May 2014

peak displacement of the SDOF system should be determined based on the energy-
based capacity curve and the demands associated with the ground motion hazard.
The base shear or absorbed energy corresponding to this displacement may then be
determined from the capacity curve. The load step in the pushover analysis asso-
ciated with this base shear or absorbed energy may then be identified, along with
the corresponding roof displacement. This load step may also be used to establish
internal forces and other quantities associated with the peak displacement of the
SDOF system. Alternatively, if the structure is to be pushed to a target displace-
ment, the pushover analysis is needed in order to identify the relationship between
the spectral displacement (D,,,) and the roof displacement. If the lateral forces
are applied in proportion to the initial mode shape, an inconsistency may develop
between the forces applied and the deformed shape at the target displacement. In
many cases this inconsistency is not a significant source of error. As an alternative,
adaptive load vectors may be used together with the energy-based displacement,
to accommodate changes to the deformed shape that may occur during the pushover
analyses.
Displacements were estimated for this example at the load step in the pushover
analysis corresponding to the estimated peak SDOF displacement. Results are
provided for each mode in Table 2. The relative accuracy of the estimates are
compared with the results from the nonlinear dynamic analysis in Table 3. In this
case, errors in floor displacement estimates were on the order of lo%, relative to the

Table 2. Spectral displacements obtained for each modal


pushover analysis.

Spectral Displacement for Each Mode


-- - --

1 s t mode 2nd mode 3rd mode


7.07 in (180 mm) 1.81 in (46.1 rnm) 0.487 in (12.4 mm)
86 E. Hmdndez-Montes, 0.-S. Kwon & M. A. Aschheim

Table 3. Modal contributions to estimated Boor displacements and comparison to dynamic


response d u e s .

Estimated Peak Floor Displacement


1st storey 2nd storey 3rd storey
- - --

1st mode 2.45 in (62.1 mm) 5.88 in (150 mm) 8.96 in ( Z 8 mm)
2nd mode -0.7'71 in (-19.6 mm) -0.753 in (-19.1 mm) 0.465 in (11.8 mm)
3rd mode 0.125 in (3.2 mm) -0.101 in (-2.6 mm) 0.0329 in (0.8mm)
SFM 2.57 in (65.2 mm) 5.93 in (151 mm) 8.97 in (228 mrn)
Dynamic Analysis ' 2.87 in (72.8 mm) 6.44 in (164 mm) 9.79 in (249 mm)
Error (%) - 10.4 -7.82 -8.42 .

values determined in the dynamic analysis. The higher mode contributions to the
SRSS displacement estimates are seen to be relatively small.However, substantially
larger errors were determined for this structure at 4% roof drift, raising some
Downloaded by [DTU Library] at 08:25 02 May 2014

concern about the validity of the basic assumption of independence that underlies
the MPA method.

7. Concluding Remarks
Nonlinear static (pushover) methods of analysis have proven to be very usehi in the
short time that has elapsed since their introduction. The capacity curve determined
in a pushover analysis reveals the rate a t which a structure develops resistance to
lateral forces of a given pattern. This resistance can be viewed conventionally from
the point of view of roof displacement or alternatively from the perspective of the
energy absorbed in the lateral load analysis. The roof displacement remains a useful
index for the first mode response of many structures, including structures for which
displacements over the height of the structure cease to increase proportionately.
However, for some structures, the roof displacement may be a poor index, even for
elastic response. For yielding systems, disproportionate increases in displacements
affect the slope of the capacity curve. The response of SDOF systems with negative
post-yield stiffness is known to be sensitive to the post-yield stifbess, and therefore,
estimates made using analogous SDOF systems may be sensitive to the displacement
index used for the capacity curve. For higher mode pushover analyses, the roof
displacements may reverse, leading to capacity curves that imply a violation of the
first law of thermodynamics.
The energy absorbed by the MDOF structure in the pushover analysis may
be used to derive an energy-based displacement that characterises the work done
by the equivalent SDOF system. Thus, in contrast to the conventional view of
pushover analysis [Fig. 13(a)], one may view pushover analysis equivalently in
terms of the work done (or absorbed energy) versus base shear response (illustrated
schematically in Fig. 13(b), with axes rotated). The data associated with Fig. 13(b)
may be used to determine the energy-based displacement, D,,,,according to
Eq. (27). The capacity curve of the equivalent SDOF system may then be obtained
First- and Multiple- Mode Nonlinear Static Analyses 87

Base Shear, Vb.n Base Shear, V4n V~.n/cxn


g m ,k JZJmr.k pm=7

Roof Displacement, ~ r , n Absorbed Energy, Ea Energy-based Displacement, Dean

Fig. 13. (a) Conventional capacity curve; (b) alternate view of pushover analyses; (c) capacity
curve of "equivalent" SDOF system.

using conventional transformations of base shear together with the energy-based


displacement, as illustrated schematically in Fig. 13(c). The energy-based capacity
curve was defined to match the capacity curves obtained using conventional
approaches for first and higher mode analyses in the elastic domain. The energy-
Downloaded by [DTU Library] at 08:25 02 May 2014

based capacity curves are able to correct the anomalies observed for higher mode
pushover curves. In rectifying the problems evident with the use of the roof displace-
ment in higher mode pushover curves, we provide an alternative formulation that is
also applicable to first mode pushover analysis. Consequently, it is concluded that
the energy-based formulation provides a stronger theoretical basis for establishing
the capacity curves of the first and higher mode equivalent SDOF systems.

Acknowledgements
We extend our .sincere appreciation to Professors Chopra and Goel for their
insightful contributions to modal pushover analysis, which have clarified many
aspects of the method. Much of the present work was stimulated by the discussions
and developments that took place under the auspices of the ongoing ATC-55 project,
which seeks to improve the use of inelastic analysis procedures for design and
rehabilitation. The erudite contributions of Professor Elnashai are also gratefully
acknowledged. This work made use of the Earthquake Engineering Research Centres
Shared Facilities supported by the National Science Foundation under Award
under EEC-9701785, and was supported in part by an NSF Career Award to the
third author (Grant No. CMS-9984830). Support provided to the first author by
the Spanish Ministry of Education (grant PR2001-0358)is also appreciated. The
original pushover analyses of the three-storey frame and Figs. 2-4 were prepared for
the ATC-55 project by Tjen Tjhin, a doctoral student at the University of Illinois.
His assistance is gratefully acknowledged.

References
Aschheim, M.and Black, E. [2000] "Yield point spectra for seismic design and rehabilita-
tion," Earthquake Spectm, Earthquake Engineering Research Institute 16(2),317-335.
ATC-40 (1997) Seismic Evaluation and Retrofit of Concrete Buildings (Applied Technolog
Council, Redwood City, California).
Chopra, A. K. (20011 Dynamic of Stnrctures: Thmry and Applicntiom to Earthquake
Engineering, 2nd Edition (Prentice Hall, New Jersey).
Chopra, A. K. and Goel, R. K. (2002)"A modal pushover analysis procedure for estimating
seismic demands for buildings," Earthquake Engng. Struct. Dyn. 31, 561-582.
CSI [2000] Computer & Structures, Incorporated, SAP 2000 NL, Berkeley, California.
FEMA-355C (20001 State oj Art Report on Systems Performance of Steel Moment Frames
Subject to Earthquake Gmund Shuking, prepared by the SAC Joint Venture for the
Federal Emergency Management Agency, Washington, DC.
FEMA-356 [ZOO01 Prestandard and Commentary for the Seismic Rehabilihtion of Build-
ings, prepared by American Society of Civil Engineers for the Federal Emergency
Management Agency, Washington, DC.
Goel, R. K. and Chopra, A. K. [2002]"Reversal in higher mode pushover curves and its
implications on implementat ion of the MPA procedure," Report distributed to par-
Downloaded by [DTU Library] at 08:25 02 May 2014

ticipants in project ATG55, Applied Technology Council, Redwood City, California,


December 2, 19 pp.
Gupta, B. and Kunnath, S. K. [2000] "Adaptive spectra-based pushover procedure for
seismic evaluation of structures," Earthquake Spectra l6(2), pp. 367-391.
Mwafy, A. M. and Elnashai, A. S.[2001]"Static pushover versus dynamic collapse analysis
of RC buildings," Engineering Structures 23(5), 407-424.
Prakash, V., Powell, G. H. and Campbell, S. [I9931 Dmin 2DX Base Program Descrip-
tion and User Guide Version 1.10,Report No. UCB/SEMM-93/17, University of
California, Berkeley.
Uang, C.-M. and Bertero, V. V . [1988] Use of Energy as a Design Criterion in
Earthquuake-Resistant Design,Report No. UCB/EERC-88/18, Earthquake Engineer-
ing Research Centre, University of California, Berkeley, November.

You might also like