Sunteți pe pagina 1din 30

A Damage Model for Practical Seismic

Design That Accounts for Low


Cycle Fatigue
Amador Teran-Gilmore
a
and James O. Jirsa,
b
M.EERI
The structural properties of a structure deteriorate when deformations
reach the range of inelastic behavior. A possible consequence of deterioration
of the hysteretic behavior of a structure is failure of critical elements at
deformation levels that are signicantly smaller than its ultimate deformation
capacity. Seismic design methodologies that account for low cycle fatigue can
be formulated using the concept of target ductility. The practical use of one
such methodology requires the consideration of simple low cycle fatigue
models that consider the severity of repeated loading through a normalized
plastic energy parameter. The inconsistencies inherent to the use of such
indices can be corrected through simple empirical rules derived from an
understanding of the effect of the history of energy dissipation in the
assessment of the level of structural damage.
DOI: 10.1193/1.1979500
INTRODUCTION
During the last few decades, there has been a considerable increase in understanding
the dynamic response of earthquake-resistant structures. Improvement in modeling and
analysis make it possible to consider a new approach to seismic design. In particular, the
perception that the engineer has to design a structure against given and unchanging con-
ditions imposed by nature has shifted towards educating the engineering profession to
focus on performance-based design.
Current philosophy for seismic design of typical residential or commercial structures
accepts the possibility that signicant inelastic behavior will occur during severe seismic
excitations. The mechanical characteristics of a structure deteriorate when deformations
reach the range of inelastic behavior. A possible consequence of deterioration of the hys-
teretic behavior of a structure is failure of critical elements at deformation levels that are
signicantly smaller than its ultimate deformation capacity under unidirectional loading.
In this paper, this failure mode will be termed low cycle fatigue.
This paper discusses the concept of low cycle fatigue, and some of the tools devel-
oped to account for it during seismic design. Emphasis is placed on the information re-
quired for the design of reinforced concrete structures. Then the convenience of using
energy demands as representation of the severity of cumulative plastic cycling, and the
issues that as a consequence arise for seismic design, are explored. Finally, a simple
a
Universidad Autonoma Metropolitana, Av. San Pablo 180, Col. Reynosa Tamaulipas, Mexico 02200, D.F.
b
University of Texas at Austin, 10100 Burnet Road, Bldg. 177, Austin, TX 78758
803
Earthquake Spectra, Volume 21, No. 3, pages 803832, August 2005; 2005, Earthquake Engineering Research Institute
energy-based low cycle fatigue model is formulated. The applicability and reliability of
the model is assessed through the comparison of design results obtained from the model
and other well-known damage indices.
A PARAMETRICAPPROACHTO SEISMIC DESIGN
Traditionally, earthquake-resistant design has been formulated as a demand-supply
problem. First, all relevant seismic demands in the building have to be estimated, and
then they must be satised with adequate seismic supplies in terms of stiffness, strength,
and maximum and cumulative deformation capacity. Although the demand-supply equa-
tion should be formulated explicitly for different design objectives, this paper will focus
on structures that undergo severe plastic cycling when subjected to intense ground mo-
tion. Within a numerical performance-based methodology, structural properties should
be supplied to the structure so that the response of its structural and nonstructural mem-
bers is limited within response threshold levels established as a function of the required
seismic performance. Teran-Gilmore 1998 has observed that recently proposed design
methodologies contemplate this check at three different steps:
1. Global Predesign. Quick and reasonable estimates of global seismic demands
should be established and checked against global threshold levels. Within this
context, the judicious use of response spectra provides information that allows
the determination of a set of global structural properties base shear, period, de-
formation capacities that can adequately control and accommodate, within
technical and cost constraints, the global response of the structure.
2. Preliminary Local Design. Once the global structural properties have been de-
termined, it is necessary to establish the member properties and detailing at the
local level. This step contemplates the analyses of complex analytical models of
the structure.
3. Revision of the Preliminary Design. Recommendations have been formulated
for the revision of the preliminary design through a series of dynamic structural
analyses that address the global and local performance of the structure.
In this paper, seismic design will be approached from the Global Predesign step. A
parametric approach will be used Teran-Gilmore 1998; that is, each one of the relevant
structural properties will be handled during the design process through a structural pa-
rameter: base shear V
b
and fundamental period of translation T to dene the global
lateral strength and stiffness, respectively; and the ultimate displacement ductility
u

and a fourth structural parameter b to characterize the global maximum and cumula-
tive deformation capacities, respectively. Although the values of these four parameters
are interrelated, the relations between the parameters are complicated and hard to char-
acterize for design purposes in a manner that assessment of low cycle fatigue requires
explicit and independent consideration of each one.
LOW CYCLE FATIGUE
Experimental and eld evidence indicate that the strength, stiffness, and ultimate de-
formation capacity of reinforced concrete elements and structures deteriorate during ex-
804 A. TERAN-GILMORE AND J. JIRSA
cursions into the plastic range of behavior. Excessive hysteretic degradation may lead to
an accumulation of deformation and degradation of the ultimate deformation capacity
that may lead to failure at plastic deformation levels that are signicantly smaller than
the ultimate plastic deformation capacity of the structure under uni-directional loading.
The importance of plastic cycling on the deformation capacity of reinforced concrete
structures has been known for some time. Several experimental studies have been car-
ried out since the 1970s to study the cyclic response of ductile members and beam-
column subassemblages, and even of models of entire buildings Bertero and Popov
1977; Gosain et al. 1977; Yamada 1992; El-Bahy et al. 1999a, 1999b. The results ob-
tained from such studies have emphasized the need to account for the effect of cycling
during performance evaluation and seismic design of earthquake-resistant structures.
Several researchers have established analytical models to capture the degradation of
the structural properties of elements and subassemblages, and to assess the occurrence
of low cycle fatigue. Particularly, efforts have been devoted to characterize the stability
of the hysteretic cycle of the elements, and to study the impact that the number, ampli-
tude, and, in some cases, even the sequence of plastic cycles has on the occurrence of
low cycle fatigue. Among many other researchers, relevant work in this eld has been
carried out by Banon et al. 1981, Park and Ang 1985a, Stephens and Yao 1987,
Chung et al. 1989, Kunnath et al. 1990, Daali and Korol 1996, and Krtzig and
Mescouris 1997. Williams and Sexsmith 1995 and Mehanny and Deierlein 2000 of-
fer excellent discussions on the use of damage models, and offer an integrated study of
a wide collection of damage indices that can be used to assess low cycle fatigue.
Since the 1980s, the engineering profession confronted the need to design structures
with predictable performance. Performance-based seismic design became a fundamental
concept for the formulation of seismic design methodologies. As a consequence, pro-
posals for design against low cycle fatigue began focusing on deformation control rather
than relying exclusively on detailing recommendations to ensure stable hysteretic behav-
ior. A key issue during the development of design methodologies to control low cycle
fatigue was the recognition that the lateral strength of a structure plays an instrumental
role in controlling the seismic demands that eventually induce this type of failure.
The end result of considering low cycle fatigue during seismic design is the need to
design structures to undergo maximum deformation demands that may be considerably
smaller than their ultimate deformation capacity. Several proposals have been made in
this direction Fajfar 1992, Cosenza and Manfredi 1996. To illustrate this, consider that,
while Panagiotakis and Fardis 2001 observe that the deformation at failure of rein-
forced concrete elements subjected to typical load histories applied in laboratory tests
can be estimated as 60% of their ultimate deformation capacity, Bertero 1997 recom-
mends that the maximum ductility demand a structure undergoes during ground motion
should be limited to 50% of its ultimate ductility.
Different levels of evaluation can be considered for the assessment of structural dam-
age and low cycle fatigue. A local evaluation refers to the damage state of a single mem-
ber or subassemblage. Global evaluation deals with the whole structure as a single entity.
Attempts have been made to assess damage at the global level through the use of dam-
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 805
age indices developed at the element level. Although different weighted averages of
damage indices in the different structural elements have been proposed to assess global
damage Park et al. 1985b, Powell and Allahabadi 1987, Mehanny and Deierlein 2000,
signicant inconsistencies may arise Ghobarah et al. 1999.
Depending on ones perspective, a damage index calculated at a story level can be
considered as a global damage indicator when contrasted to the member level Mehanny
and Deierlein 2000. Considering the damage indicator at the critical story as a global
damage index, Teran-Gilmore 2004 found a high correlation between the values of the
Park and Ang damage index Park and Ang 1985a established from equivalent single-
degree-of-freedom SDOF systems, and the global value of that index corresponding to
regular frames. For this purpose, he estimated the damage index at the critical story as
the mean value of the damage index in the beams located at that story. Although Teran-
Gilmore 2004 concludes that a local damage index can be used in a global sense with
the aid of an equivalent SDOF system to adequately conceive and globally predesign
regular frames, he discusses the importance of acknowledging the limitations of the use
of SDOF systems during global predesign, and emphasizes the need for carrying out a
detailed revision of the seismic performance of the preliminary design of the structure.
ENERGYAS DESIGN REPRESENTATION OF CUMULATIVE LOADING
For several decades, researchers have discussed the importance of considering the
effect of repeated plastic cycling during seismic design. An option that has been consid-
ered attractive due to its simplicity has been the characterization of cumulative loading
through energy concepts. Housner 1956 offered one of the earliest discussions regard-
ing the need to consider explicitly the effect of plastic cycling through energy concepts.
Later, several attempts were made to estimate the energy demands in simple systems,
and to offer insights on how to use these demands for design purposes Zahrah and Hall
1984, Akiyama and Takahashi 1992, Leelataviwat et al. 2002.
Design for low cycle fatigue was advanced with the formulation and calibration of
damage indices Powell and Allahabadi 1987, Cosenza et al. 1993, and the formaliza-
tion of an energy balance equation Uang and Bertero 1990. Based on these concepts,
several design methodologies that account for low cycle fatigue have been formulated
Fajfar 1992, Bertero and Bertero 1992, Krawinkler and Nassar 1992, Cosenza and
Manfredi 1996. As for today, there are still signicantly different approaches towards
the formulation of a design representation for the energy demands. Some researchers
suggest that energy spectra could be formulated and used for design purposes Akiyama
and Takahashi 1992, Chou and Uang 2000, Manfredi 2001. Other options include ac-
counting for cumulative loading through indirect measures of the plastic energy Fajfar
1992, Bertero and Bertero 1992, and deriving the plastic energy demands from other
relevant seismic demands Teran-Gilmore 1996, Decanini and Mollaioli 2001.
The plastic energy, E
H
, can be interpreted physically by considering that it is equal
to the total area under all the hysteresis loops a structure undergoes during the ground
motion. Although E
H
provides a rough idea of the cumulative plastic deformations de-
806 A. TERAN-GILMORE AND J. JIRSA
mands, this response parameter by itself does not provide enough information to assess
structural performance so that it is convenient to normalize it as follows:
NE
H
=
E
H
F
y

y
1
where NE
H
is the normalized plastic energy, and F
y
and
y
are the yield strength and
yield displacement, respectively. For an elasto-perfectly-plastic system subjected to a
single plastic excursion Figure 1a:
E
H
=
c

y
F
y
=

y
1
y
F
y
=
c
1
y
F
y
2
where
c
is the cyclic displacement, and
c
, equal to
c
/
y
, is the cyclic ductility. The
NE
H
for the plastic excursion is a direct measure of the amplitude of the plastic dis-
placement:
NE
H
=
E
H

y
F
y
=
c
1 3
For an elasto-perfectly-plastic system subjected to multiple plastic excursions, NE
H
is the sum of all plastic displacements reached in the different cycles normalized by
y
,
in such way that
NE
H
=

i=1
Nexc

ci

y

y
=

i=1
Nexc

ci
1 4
where
ci
and
ci
are the cyclic displacement and ductility, respectively, associated with
the ith excursion, and Nexc is the total number of plastic excursions during the ground
motion. Note that NE
H
is a direct measure of the cumulative plastic displacement de-
mands. For a system with degrading hysteretic behavior, NE
H
could be dened to in-
Figure 1. Denitions of strength and deformation quantities.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 807
clude all plastic excursions for which the capacity does not degrade to a value less than
a specied fraction of F
y
say 0.75. Such a denition allows for rational evaluation of
structural damage in reinforced concrete structures.
Several researchers have used NE
H
to develop recommendations for the design and
detailing of ductile reinforced concrete elements Gosain et al. 1977, Scribner and Wight
1980, Darwin and Nmai 1985. Krawinkler and Nassar 1992 formulated a design
methodology that explicitly considers the effect of plastic cycling through NE
H
.
IMPLICATIONS OF USE OF PLASTIC ENERGY DURING SEISMIC DESIGN
Although using energy-derived parameters as a representation of repeated cyclic
loading allows the formulation of relatively simple seismic design methodologies that
account for low cycle fatigue, this approach should be carefully assessed. Particularly,
the plastic energy dissipating capacity of a structure does not depend exclusively on its
mechanical characteristics. It has been repeatedly observed that the plastic energy dis-
sipated up to failure by an element or structure can change signicantly as a function of
the amplitude of the plastic cycles, in such way that the plastic energy dissipated by a
large number of small amplitude cycles can signicantly exceed that dissipated up to
failure through the application of a few large amplitude cycles Chen and Gong 1986,
Chung et al. 1989, Teran-Gilmore et al. 2003. A second issue, which will not be dis-
cussed herein, is the feasibility of deriving a simple design representation for the plastic
energy demands. In the previous section, references were cited to document the work
carried out around this issue.
GROUND MOTIONS
Four sets of ground motions were established to study the impact of considering
NE
H
as the design representation of the severity of repeated cyclic loading. Three of
these sets correspond to the Los Angeles LA urban area and one corresponds to the
lake zone of Mexico City. The ground motions for LA, established as part of the FEMA/
SAC Steel Project Somerville et al. 1997, were grouped in sets of 20 motions as fol-
lows: design earthquake for rm soil with 10% exceedance in 50 years LA 10in50,
design earthquake for rm soil with 50% exceedance in 50 years LA 50in50, and de-
sign earthquake for soft soil with 10% exceedance in 50 years LA Soft. The set of
Mexican motions Mexico Soft was formed of seven narrow-banded, long-duration
ground motions recorded in the Lake Zone of Mexico City. The Mexico Soft motions
were scaled up in such way that their peak ground velocity was equal to that correspond-
ing to the EW component of the motion recorded at SCT during 1985. Figure 2 shows
mean strength spectra for the four sets of motions. All spectra shown were obtained for
elasto-perfectly-plastic behavior and 5% of critical damping.
In Figure 2, the circles identify the corner period, dened as the period at which the
strength spectra decreases after peaking either at a single point or at a plateau. As a ref-
erence, the corner period is denoted T
s
in FEMA-273 ATC 1997. Note that LA 10in50
has a corner or dominant period around 0.3 sec, while those of LA 50in50, LA Soft, and
Mexico Soft are around 0.4, 1.0, and 2.0 sec, respectively. Figure 3 shows mean relative
input energy E
I
spectra. Mexico Soft has the largest input energy demands, followed, in
808 A. TERAN-GILMORE AND J. JIRSA
order, by LA Soft, LA10in50, and LA50in50. With the exception of Mexico Soft, the cor-
ner periods dened according to Figure 2 do not correspond to the period at which the
input energy for 5% damping peaks.
Figure 4 shows mean NE
H
spectra. For constant ductility, Mexico Soft has the larg-
est plastic cumulative demands, followed, in order, by LA 50in50, LA Soft, and LA
10in50. Note the signicant difference in the relative damage potential established for
the different sets of motions from Figures 3 and 4. If, as suggested before, NE
H
is as-
sumed to be a direct measure of the plastic cumulative deformation demands in a struc-
ture, it follows that input energy and even plastic energy without normalization spectra
should be used carefully when assessing the energy content of a ground motion. The
maximum NE
H
demands for Mexico Soft are about two to three times larger than those
corresponding to the LA motions.
There is a distinctive feature in the NE
H
spectra corresponding to the sets of LA
motions: starting from very small T, the NE
H
demand tends to increase until T reaches
the value of the corner period, after which it remains fairly constant. For the Mexico Soft
set, NE
H
tends to increase until T reaches the value of the corner period. After that, it
tends to decrease with a further increase in T. Note that the corner period dened ac-
cording to Figure 2 delimits two distinctive zones in the NE
H
spectra.
Figure 5 shows the coefcient of variation COV associated with mean spectra
shown in Figures 2 and 4. The COV is presented for two purposes: 1 to provide an idea
of the uncertainty and variability involved in establishing mean spectra, and 2 to pro-
Figure 2. Mean strength spectra, 5% damping.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 809
vide reference values against which the COV associated to the use of the low cycle fa-
tigue model developed here can be assessed. While the COV of the strength spectra cor-
responding to the four different motion sets does not seem to follow a well-established
pattern, the COV of the NE
H
spectra does show a surprising similitude for all sets of
motion, and is characterized by values usually ranging from 0.3 to 0.7.
DISTRIBUTIONS OF EXCURSIONSAND DAMAGE
To clarify the possible effect of using NE
H
to characterize the severity of cumulative
cycling during the assessment of low cycle fatigue, the distribution of plastic excursions
according to their amplitude was studied. For this purpose, the response of elasto-
perfectly-plastic SDOF systems having 5% of critical damping was studied when sub-
jected to the different sets of motions.
The plots included in Figures 68 are divided into 10 intervals along the horizontal
axis. These correspond to 10 intervals that equally divide the range of possible values of
the cyclic ductility demand. Figure 6 shows, for LA 10in50, the mean number of plastic
excursions n as a function of their amplitude for different values of
max
and T. The
following trends are observed:
The number of excursions increases as their amplitude decreases.
As T increases departs from the value of the corner period, the number of small
excursions intervals 13 decreases with respect to that of large excursions.
Figure 3. Mean relative input energy spectra, 5% damping.
810 A. TERAN-GILMORE AND J. JIRSA
As the value of
max
increases, the number of small excursions increases with
respect to that of large excursions.
There is a bump in the distributions at intervals 5 and 6, which correspond to
excursions with cyclic ductility demands close to
max
. The bump tends to de-
crease and even disappear as the value of
max
increases.
Figure 6 also shows damage distributions for LA10in50. Particularly, the fraction of
overall damage induced by excursions of different amplitude is plotted. Damage was es-
timated using Equation 10 discussed in detail later with b equal to 1.5. The damage
distributions tend to peak at intervals 5 and 6, and the peak value tends to increase as

max
decreases and T increases.
The excursion distributions for LA 50in50, shown in Figure 7, exhibit similar ten-
dencies to those of LA 10in50. Nevertheless, the former distributions exhibit a larger
number of small amplitude excursions. As the plastic energy content of the ground mo-
tion increases, a larger percentage of the plastic energy tends to be dissipated in excur-
sions of smaller amplitude. Although the damage distributions for LA 10in50 and LA
50in50 are similar, the peaks at intervals 5 and 6 tend to be signicantly smaller for the
latter set. As a consequence, the damage distributions corresponding to LA 50in50 tend,
as
max
increases and T approaches the value of the corner period, to be fairly constant
with respect to the amplitude of the plastic excursions.
The excursion distributions for Mexico Soft, shown in Figure 8, are similar to those
established for the LA motions. The different NE
H
content and the different dependence
Figure 4. Mean normalized plastic energy spectra, 5% damping.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 811
Figure 5. COV of strength and normalized plastic energy spectra, 5% critical damping.
Figure 6. Distribution of plastic excursions and damage, LA 10in50.
812 A. TERAN-GILMORE AND J. JIRSA
that NE
H
exhibits with respect to T for the Mexico and LA motions, help explain the
following particularities for Mexico Soft:
As T approaches the value of the corner period, the number of small excursions
increases signicantly with respect to that of large excursions.
The bump located at intervals 5 and 6 tends to disappear, even for small values
of
max
.
As the NE
H
demands increase from LA 50in50 to Mexico Soft, a slightly larger
percentage of the plastic energy tends to be dissipated in smaller excursions.
Figure 7. Distributions of plastic excursions, LA 50in50.
Figure 8. Distributions of plastic excursions and damage, Mexico Soft.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 813
From the results summarized in Figures 68, it can be concluded that as the NE
H
demand on a structure increases product of a larger energy content of the ground mo-
tion, a larger
max
demand and/or the similarity between the value of T and the corner
period, its relative number of small excursions increases. Because the plastic energy
dissipating capacity of a structure increases as the amplitude of the plastic cycles de-
creases, it can be said in qualitative terms that using NE
H
to characterize the severity of
cumulative cycling during the assessment of low cycle fatigue implies, in general terms,
an increased level of conservatism as the energy content of the motion increases.
LOW CYCLE FATIGUE MODELS
Damage indices that account for low cycle fatigue should explicitly consider the ef-
fect of cumulative loading. Two signicantly different approaches have been used. On
the one hand, some damage indices consider plastic energy as a measure of the severity
of plastic cycling. On the other hand, some damage indices consider the number and
amplitude of the different plastic cycles, and in some cases, even their sequence.
In this section two damage indices, considered representative of each one of the two
approaches mentioned before, will be discussed. Then an energy-based low cycle fatigue
model that uses NE
H
to characterize the severity of cumulative cycling is derived. These
three damage indices will be used to dene, in quantitative terms, the impact of using
NE
H
as sole representation of ground motion intensity.
PARKANDANG DAMAGE INDEX
Park and Ang 1985a have formulated a damage index to estimate the level of dam-
age in reinforced concrete elements and structures subjected to cyclic loading:
DMI
PA
=

max

u
+
E
H
F
y

u
5
where
max
is the maximum deformation demand during the ground motion,
u
is the
ultimate deformation capacity, and is a structural parameter. DMI
PA
less than 0.4 im-
plies repairable damage; from 0.4 to 1.0, irreparable damage; and greater than 1.0, fail-
ure of the element. Under the presence of repeated cyclic loading, 1.0 represents the
threshold value at which low cycle fatigue is expected to occur. While of 0.15 corre-
sponds to systems that exhibit fairly stable hysteretic behavior, values of ranging from
0.2 to 0.4 should be used for systems exhibiting substantial strength and stiffness dete-
rioration Stephens and Yao 1987, Cosenza et al. 1993, Williams and Sexsmith 1997,
Silva and Lopez 2001. DMI
PA
can be written as
DMI
PA
=

max

u
+
NE
H

u
6
where
max
, equal to
max
/
y
, is the maximum ductility demand during the ground mo-
tion, and
u
, equal to
u
/
y
, is the ultimate ductility. Within a seismic design methodol-
ogy that accounts for low cycle fatigue, DMI
PA
can be used as follows DMI
PA
=1:
814 A. TERAN-GILMORE AND J. JIRSA

max
=
u
NE
H
or
u
=
max
+ NE
H
7
Within Equation 7, NE
H
becomes the ground motion parameter that quanties the
severity of cumulative loading, and the structural parameter that characterizes the cy-
cling capacity of the structure. An increase in NE
H
implies larger plastic demands,
while a decrease in implies a more stable hysteretic behavior better detailing.
LINEAR CUMULATIVE DAMAGETHEORY
A damage index that takes into account the change in energy dissipating capacity of
a structure as a function of its displacement history can be formulated from Miners hy-
pothesis, which considers that damage induced by each plastic excursion is independent
of the damage produced by any other excursion. Since the deformation history is un-
likely to consist of regular, complete cycles, this history is usually divided in half-cycles
termed excursions herein rather than full cycles, using methods such as the Rainow
Counting Method Powell and Allahabadi 1987.
Once the displacement history is separated into Nexc plastic excursions, the linear
cumulative damage theory requires these excursions to be classied into intervals ac-
cording to their amplitude. In this paper, Ndif denotes the number of different intervals
into which all plastic excursions are classied, and
ci
is the cyclic displacement am-
plitude associated with the ith interval. For earthquake loading, the linear cumulative
damage theory can be formulated as
DMI
MH
=

i=1
Ndif
n
i
N
i
8
where N
i
is the number of plastic excursions the structure can actually undergo before
failure when cycled to excursions with amplitude
ci
, and n
i
is the number of plastic
excursions of amplitude
ci
resulting from the ground motion demands on the structure.
DMI
MH
equal to one implies incipient failure.
There is a difference between the traditional denitions of deformation and the ones
needed for application of Equation 8. Figure 1b illustrates the concepts of maximum
ductility and maximum cyclic ductility demand. While
max
is measured with respect to
the underformed position, the maximum cyclic ductility demand
maxc
is measured
with respect to the point in which the largest half-cycle initiates. Note that the largest
possible cyclic ductility demand the structure can undergo is
maxc
=2
max
1. Similar
concepts as those discussed before should be applied to the ultimate deformation capac-
ity of a structure. In this sense,
uc
is dened as the ultimate cyclic displacement. The
normalization of
uc
by
y
yields
uc
ultimate cyclic ductility. Equation 8 can be re-
formulated as follows Cosenza and Manfredi 1996:
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 815
DMI
MH
=

i=1
Nexc

ci

y

uc

y

b
9
where
ci
is now the cyclic displacement of the ith excursion, and b is a structural pa-
rameter. Within seismic design against low cycle fatigue DMI
MH
=1,

uc

y

b
=

i=1
Nexc

ci

y

uc
1
b
=

i=1
Nexc

ci
1
b
10
Within Equation 10,
i=1
Nexc

ci

b
characterizes the severity of cyclic loading, and
b the stability of the hysteresis loops. While typical values of b range from 1.6 to 1.8, b
of 1.5 is a reasonable conservative value to be used for seismic design and damage
analysis Powell and Allahabadi 1987, Baik et al. 1988, Cosenza and Manfredi 1996.
An increase in b implies a more stable hysteretic behavior. Regarding its limitations,
DMI
MH
does not take into account the sequence of plastic excursions, and inconsisten-
cies arise when applied to systems that develop permanent plastic deformations.
AN ENERGY-BASED MODELTO PREDICT LOW CYCLE FATIGUE
Consider the case in which n
i
and N
i
can be related, for all i in Equation 8, through
the same proportionality constant .
n
i
= N
i
11
If Equation 11 is substituted into Equation 8, the value of N
i
cancels out for each
term in the summation. Under the assumption of proportionality, the level of damage in
a structure depends exclusively on its NE
H
demand Teran-Gilmore et al. 2003. In this
section, a simple low cycle fatigue model is developed. Basically, this model represents
a simplication of the linear cumulative damage theory through the assumption of a
xed shape for the distribution of plastic excursions. If Equation 11 holds up to failure
a structure can dissipate normalized plastic energy equal to
NE
H
=

i=1
Ndif
n
i

ci
1 =

i=1
Ndif
N
i

ci
1 =

i=1
Ndif

uc
1

ci
1

ci
1 12
Equation 12 can be formulated in closed form as follows:
NE
H
=
0

uc
1
n
c
1d
c
=
0

uc
1

uc
1

c
1

c
1d
c
=

uc
1
2
2 b
13
where n is the number of plastic excursions of amplitude
c
demanded by the ground
motion, and
c
, equal to
c
/
y
, is the plastic cyclic ductility. The value of DMI
MH
cor-
responding to Equation 13 can be estimated according to the closed form of Equation 8:
DMI
MH
=
0

uc
1
n
N
d
c
=
0

uc
1
d
c
=
uc
1 14
816 A. TERAN-GILMORE AND J. JIRSA
By considering the right-hand sides of Equations 13 and 14, a simplied estimate of
DMI
MH
can be obtained:
DMI
MH
S
= 2 b
NE
H

uc
1
15
In Equation 15, NE
H
quanties the severity of ground motion, and
uc
and b the
ultimate and cumulative deformation capacities of the structure. Although the analytical
upper limit for the value of
uc
is given by 2
u
1, the physical upper limit of
uc
will be
somewhat less than this because a plastic excursion close to
u
will damage signicantly
the capacity of a structure to accommodate plastic deformation in the opposite direction:

uc
1 = 2r
u
1 16
where r is a reduction factor less than one. For incipient collapse, Equation 15 can be
reformulated in terms of
u
as DMI
MH
=1:

u
=
2 bNE
H
2r
+ 1 17
Figure 9a compares, for LA 50in50, damage estimates derived from Equations 9 and
15 b=1.5 and
uc
=8.5. The value of
uc
was established from Equation 16 by assum-
ing
u
of 6 and r equal to 0.75. The discontinuous lines correspond to Equation 9. Equa-
Figure 9. Estimates of damage from Equations 9 and 15, LA 50in50.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 817
tion 15 yields higher estimates of damage for
max
of 2, slightly higher estimates for
max
of 3, and slightly lower estimates for
max
of 4.
The energy dissipating capacity of a structure increases as the amplitude of its plas-
tic excursions decreases. In the case of
max
of 2, the amplitude of the majority of the
plastic excursions is small with respect to
uc
. While Equation 9 accounts for an in-
creased energy dissipation capacity, Equation 15 does not, so that the latter yields higher
estimates of damage. As the value of
max
increases, the mean amplitude of the plastic
excursions increases with respect to the ultimate deformation capacity. Because the en-
ergy dissipating capacity of a system will tend to decrease under these circumstances,
Equation 15 yields similar estimates of damage than Equation 9 for
max
of 3 and 4.
As illustrated in Figure 9b, the COV of the damage estimates obtained from both
equations is practically equal. If the structural parameters involved in Equations 9 and
15 are considered deterministic, the uncertainty in the estimation of the level of damage
is similar to that involved in the determination of the energy demands see Figure 5c.
Figure 9c shows the mean ratio of the damage estimates obtained from Equations 9
and 15. The ratio shows a strong dependence on
max
and a weak variation with respect
to T. Figure 9d shows the COV associated with the damage ratio is very small. Similar
results as those shown in Figures 9c and 9d were obtained for the other sets of ground
motions.
As the amplitude of the plastic cycles decreases, a structure is able to accommodate
larger plastic energy demands before failure due to low cycle fatigue. Within seismic
design, this implies that the amplitude of the plastic excursions should decrease with
respect to the ultimate deformation capacity of the structure as the energy content of
motion increases. Under these circumstances, Equation 15 yields with respect to Equa-
tion 9 the following:
Slightly lower estimates of damage when applied to structures subjected to mo-
tions with low energy content.
Similar or slightly higher estimates of damage when applied to motions with
moderate and large energy content.
Higher estimates of damage when applied to motions with very large energy con-
tent.
Because of the above, it was considered convenient to adjust Equation 15 by intro-
ducing a parameter a that accounts for the energy content of the motion and thus, in-
directly, for the manner in which energy is expected to be dissipated:
DMI
MH
S
= 2 b
aNE
H

uc
1
18
For incipient collapse, Equation 18 can be reformulated in terms of
u
as DMI
MH
=1:

u
=
2 baNE
H
2r
+ 1 19
818 A. TERAN-GILMORE AND J. JIRSA
CONSIDERATIONS FORTHE USE OF DAMAGE INDICES IN PRACTICAL
SEISMIC DESIGN
Urgent issues that need to be addressed to improve the application of damage indices
within practical seismic design are 1 the harmonization of denitions used and results
obtained by different researchers, and 2 a clear understanding of the implications of the
application of a damage index to the design of a particular structure.
Regarding the rst issue, the practical use of damage indices requires a precise and
reliable denition of failure. After identifying important differences in the reported
DMI
PA
threshold values corresponding to failure, Williams and Sexsmith 1997 suggest
that a source of such differences is the different denition of failure used by different
researchers. Iemura 1980 observes that the linear damage cumulative theory yields dif-
ferent results depending on the criteria used to dene failure.
The second issue emphasizes the need to have a clear understanding of the expected
behavioral and failure modes of a given structure subjected to ground motion. These
modes, which can change from structure to structure, can even change in a particular
structure as a function of the characteristics of the seismic excitation Krawinkler and
Zohrei 1983, Chung et al. 1989, Manfredi and Pecce 1997. In this respect, the charac-
terization of the structural parameter b, and the uncertainty involved in the use of a
damage index does not only vary according to the structural properties of the structure,
but also according to the type of loading it is subjected to.
PRACTICAL CONSIDERATION OF LOW CYCLE FATIGUE IN
SEISMIC DESIGN
The impact of using NE
H
to quantify the severity of ground motion will be assessed
through the comparison of the results obtained from the three damage models consid-
ered in this paper. The results obtained from Equation 10 will be considered benchmark
values against which the reliability of Equations 7 and 19 will be assessed.
TARGET DUCTILITY
Target ductility is dened herein as the maximum ductility
max
the structure can
reach during the design ground motion to prevent failure due to low cycle fatigue. Even
though the concept of target ductility has obtained acceptance from researchers and
practicing engineers, the direct use of energy demands to establish the severity of seis-
mic cyclic loading is still an unresolved issue. While Fajfar 1992 suggests that the tar-
get ductility can be reasonably estimated by using the plastic energy demand, Cosenza
and Manfredi 1996 discuss the importance of the manner in which this energy has
been dissipated. In general, it has been agreed that as the plastic energy demand in-
creases,
max
should decrease with respect to
u
. How much smaller
max
should be with
respect to
u
or how much bigger
u
with respect to
max
depends on three variables:
the value of the known ductility either
max
or
u
, a ground motion parameter that
quanties the severity of the plastic demands, and a structural parameter that character-
izes the cycling capacity of the structure.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 819
Using the concept of target ductility, two approaches can be considered for the for-
mulation of a performance-based design methodology:
1. Approach A requires estimating
max
given that
u
is known. The second column
of Table 1 summarizes the steps involved in Approach A. First, a decision needs
to be made about the type of detailing to be used for the structure ductile vs.
nonductile, and values of
u
and b should be established according to it. Next,
the value of T is established. The determination of T within the context of
performance-based design is discussed elsewhere Bertero and Bertero 1992,
Priestley 2000. Then the value of
max
is established as a function, among other
things, of
u
and b. Once
max
is known, it is possible to establish the design
base shear that will allow the structure to control its maximum plastic demand
within this threshold. Approach A has been used by Fajfar 1992, Bertero and
Bertero 1992, and Cosenza and Manfredi 1996 to formulate design method-
ologies.
2. Approach B requires estimating
u
given that
max
is known. The third column
of Table 1 summarizes the steps involved in Approach B. First, a decision needs
to be made about the type of detailing to be used for the structure, and a value
of b established accordingly. Next, the value of T is established. Once a prelimi-
nary value of
max
is assumed, the values of
u
and V
b
can be established as a
function of it. Approach B has been used by Arroyo and Teran-Gilmore 2002
and Ridell and Garcia 2002.
Recommendations for the use of Equations 18 and 19 in practical seismic design are
summarized in Table 2. These recommendations were obtained from extensive studies of
the seismic performance of SDOF systems designed according to Equation 19 and sub-
jected to the sets of ground motions considered in this paper. Two different values of a
are suggested for practical application: 0.75 for the use of Approach B with motions
Table 1. Target ductility-based design approaches
Step Approach A Approach B
1 Assume
u
, b=fdetailing Assume
max
=fjudgment, and b=fdetailing
2 Determine T Determine T
3 Estimate
max
=fT,
u
, b Estimate
u
=fT,
max
, b
4 Estimate V
b
=fT,
max
Estimate V
b
=fT,
max

Table 2. Considerations for the practical use of Equations 18 and 19


Energy Content Approach A Approach B
Low a=1,
max

u
a=1,
u

max
Moderate or High a=1 a=1
Very High Mexico City a=1 a=0.75
820 A. TERAN-GILMORE AND J. JIRSA
with very large energy content e.g., Lake Zone of Mexico City, and 1.0 for any other
case. The unsafe nature of Equations 18 and 19 with a=1 for motions with low energy
content can be taken into consideration by establishing maximum or minimum values,
respectively, for
max
and
u
such as those indicated in Table 2.
APPROACHA
Table 3 shows implementation details for the use of Approach A with the three dam-
age indices considered herein. Once the values for b,
u
, and T are established, design
against low cycle fatigue implies estimating the design base shear. The determination of
V
b
is carried out in two steps: 1
max
is determined, and 2 the design V
b
corresponds
to the minimum strength required to control the global plastic response of the structure
within the threshold established by
max
.
Approach A requires the target ductility to satisfy the conditions formulated in the
last column of Table 3. Due to inconsistencies in their formulation, the following con-
dition was imposed on the value of
max
derived from DMI
MH
and DMI
MH
S
:
max

u
.
Figure 10a shows, for LA 50in50 and
u
of 5, a general comparison of the values of

max
obtained with DMI
MH
and DMI
PA
. The three values of used with DMI
PA
are con-
sidered to characterize a wide range of structural behavior. For DMI
MH
, r was set equal
to 0.75 and b was set equal to 1.2 and 1.8. There is similarity between the values of
max
obtained from DMI
PA
with of 0.05 and DMI
MH
with b of 1.8, and those obtained from
of 0.15 and b of 1.2. Figure 10b shows values of
max
corresponding to structures with
stable hysteretic behavior =0.15 and b=1.5. Note the similitude of the results de-
rived from the three models.
For rm soil motions, the value of
max
is not particularly sensitive to the values of
and b. The results suggest that
max
should be limited to about 0.65
u
in structures
that exhibit signicant deterioration of their hysteresis loop b=1.2, and to about
0.75
u
for structures with stable hysteretic behavior b=1.5. The conservatism in-
volved in the deformation thresholds suggested for displacement-control methodologies,
such as FEMA-273, appears enough to protect stable structures from low cycle fatigue.
Figures 10c and 10d show the minimum strength required to control, for LA 50in50,
the maximum ductility demand within the values shown in Figures 10a and 10b, respec-
tively. Considering that the lateral strength is the actual structural property to be de-
signed within Approach A, Figure 10 suggests that the impact of using one or another
low cycle fatigue model during seismic design would be minimal.
Table 3. Use of Approach A with the three different models
Model Known Unknown Target ductility should satisfy:
DMI
MH
b,
uc
1=2r
u
1
i=1
Nexc

ci

max
1
b
Equation 10
DMI
PA
,
u

max
, NE
H

max
Equation 7
DMI
MH
S
b,
uc
1=2r
u
1 NE
H

max
Equation 19
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 821
Figure 11 shows that the COV of the strength demands is fairly insensitive to the
values of b and , and that this COV is similar to that shown in Figure 5a. If the struc-
tural parameters involved in the three models are considered deterministic, the uncer-
tainty involved in determining the minimum strength required to avoid failure due to low
cycle fatigue is similar to that involved in the determination of constant ductility
strength spectra. Note also that the values of COV associated with the estimation of
max
are very small, particularly in the case of DMI
PA
.
Although not shown, results obtained for other values of
u
3, 4, and 6 and for LA
10in50 and LA Soft are similar to those summarized in Figures 10 and 11. Nevertheless,
Figure 10. Design values obtained from three damage models and LA 50in50,
u
=5.
Figure 11. COV of S
a
and
max
obtained from three damage models and LA 50in50,
u
=5.
822 A. TERAN-GILMORE AND J. JIRSA
it was observed that as the energy content of the ground motions decreases, DMI
PA
be-
comes more conservative with respect to DMI
MH
.
Figure 12 shows values of
max
obtained for Mexico Soft. Note that the correspon-
dence between the values of and b not only changes with respect to that observed in
rm soil, but becomes sensitive to the value of T. This can be explained by the larger
NE
H
demands of Mexico Soft, and by the dependence of these demands with respect to
T Figure 4d. While Figure 12a shows that of 0.30 yields results similar to b of 1.2 in
a wide period range, Figures 12b and 12d suggest that DMI
PA
remains conservative with
respect to DMI
MH
in period intervals where the NE
H
demand is small, and becomes
slightly unsafe around the corner period. In contrast, DMI
MH
S
yields slightly higher
strength requirements than DMI
MH
at the corner period and slightly lower strength re-
quirements as T departs from it. As the value of
u
increases, the strength requirements
derived from DMI
PA
around the corner period become progressively smaller than those
obtained from DMI
MH
, while the opposite occurs, under the same circumstances, to the
strength requirements derived from DMI
MH
S
.
For structures subjected to Mexico Soft,
max
should be limited under certain circum-
stances to about 0.40
u
for rapidly degrading structures and to about 0.50
u
for struc-
tures with stable hysteretic behavior. Under these circumstances, the conservatism in-
volved in the deformation thresholds suggested for displacement-control methodologies
would not seem enough to protect adequately structures having T close to the corner
period from the occurrence of low cycle fatigue.
Considering that within Approach A the base shear of the structure is the structural
Figure 12. Design values obtained from three damage models and Mexico Soft,
u
=5.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 823
property to be designed, the results shown in this section suggest that, independently of
the energy content of the ground motion, the impact of using one or another low cycle
fatigue model would be minimal.
APPROACH B
Table 4 shows the implementation details for Approach B. Once values of T, b, and

max
are established, design against low cycle fatigue implies the determination of the
design values for
u
and V
b
. This determination is carried out in two independent steps:
while V
b
corresponds to the minimum strength required to control the global plastic re-
sponse of the structure within the threshold established by
max
,
u
should satisfy the
conditions formulated in the last column of Table 4. Due to the unsafe nature of DMI
MH
S
as the NE
H
content of the motion decreases, the following condition was imposed on
the value of
u
derived from this index:
u

max
.
Figure 13 shows, for LA 50in50, values of
u
obtained from Approach B. Systems
with rapidly deteriorating structural properties =0.30, b=1.2 and with stable hyster-
esis =0.15, b=1.5 are considered. An r of 0.75 was considered to evaluate DMI
MH
and DMI
MH
S
. Note that the estimates of
u
derived from the three models are very simi-
lar.
Figure 13 shows the COV corresponding to
u
. If the structural parameter used to
characterize the stability of the hysteretic behavior is considered deterministic, the COV
is comparable or smaller than that involved in estimating the NE
H
demands for constant
ductility Figure 5c The COV associated with the use of DMI
PA
tends to be smaller than
that of DMI
MH
, which in turn is smaller than that of DMI
MH
S
. Results obtained for LA
10in50 and LA Soft suggest that, as the NE
H
content of the motion decreases, the esti-
mates obtained from DMI
PA
tend to become slightly higher than those derived from
DMI
MH
; while those obtained from DMI
MH
S
, tend to become slightly smaller.
Figure 14 shows values of
u
for Mexico Soft. As specied in Table 2, the results
corresponding to DMI
MH
S
and Mexico Soft were obtained for a=0.75. Considering that
within Approach B the ultimate deformation capacity of the structure is the structural
property to be designed, the results that have been obtained Figures 13 and 14 suggest
that the impact of using one or another low cycle fatigue model would be minimal.
Table 4. Use of Approach B with the three different models
Model Known Unknown Ultimate ductility should satisfy:
DMI
MH
b,
max
,
i=1
Nexc

ci

max
1
b

u
Equation 10
DMI
PA
,
max
, NE
H

max

u
Equation 7
DMI
MH
S
b,
max
, NE
H

max

u
Equation 19
824 A. TERAN-GILMORE AND J. JIRSA
FINAL CONSIDERATIONS
Empirical considerations have been made in this paper to establish a simple energy-
based model, DMI
MH
S
, to assess the occurrence of low cycle fatigue. First, it has been
considered that the n
i
and N
i
curves for typical earthquake-resistant structures are pro-
portional. In this respect, previous work by Teran-Gilmore et al. 2003 has shown that n
i
curves, such as those shown in Figures 68, have shapes that are similar to those of N
i
curves derived from eld and experimental research. From the results obtained by Teran-
Gilmore et al., it can be concluded that in general, the assumption of proportionality
Figure 13. Values of
u
obtained from three damage models and LA 50in50.
Figure 14. Values of
u
obtained from three damage models and Mexico Soft.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 825
between the n
i
and N
i
curves has a small impact in strength design relative to the un-
certainty introduced to the design process from the denition of the design input and the
determination of the deformation capacity of the structure.
In general terms, quantifying the severity of ground motion solely through the use of
NE
H
proportionality between the n
i
and N
i
curves implies 1 slightly unsafe estimates
of damage for motions with low energy content, 2 reasonable estimates for motions
with moderate and large energy content, and 3 conservative estimates for motions with
very large energy content. Although this conclusion can be considered general, some of
the details discussed herein could be related to the specics of the sets of ground mo-
tions. Of particular concern is the fact that the EW component of the motion recorded at
SCT during 1985 is the only available large intensity motion recorded in the Lake Zone
of Mexico City. With the exception of SCT EW, all motions in Mexico Soft were re-
corded during low-intensity events and were scaled up an order of magnitude to conform
this set. As a consequence, the Mexico Soft set shows, on average, a larger energy con-
tent and larger content of small cycles than SCT EW.
The values of some parameters involved in DMI
MH
S
, such as a and r, have been cali-
brated in such way that, for the sets of ground motions considered in this paper, DMI
MH
S
yields similar assessment of low cycle fatigue as other well-known damage indices. Al-
though on the one hand, extensive calibration of DMI
MH
S
is required before it can be used
for practical seismic design, on the other hand, the analytical results derived from it for
motions with very different frequency and energy content are encouraging.
Simple seismic design methodologies, based on avoiding low cycle fatigue through
solely controlling the energy demands in the earthquake-resistant structure, can be for-
mulated with the aid of DMI
MH
S
. In particular, a three-step methodology is currently be-
ing developed and calibrated for seismic design:
1. Determine T.
2. Establish the maximum NE
H
demand the structure can accommodate before
failure due to low cycle fatigue. The threshold value of NE
H
is established from
Equation 19 once the ultimate and cumulative deformation capacities of the
structure
u
and b, respectively are established according to the detailing to be
provided to the structural elements.
3. Establish, as a function of T, the lateral strength required to control the NE
H
demands within the threshold established in step 2.
Preliminary results derived from the use of the three-step methodology suggest that
seismic design of structures with stable hysteretic behavior and located in rm soil
should focus on controlling their maximum ductility demand. It should be mentioned
that this methodology has been successfully applied to the seismic design of simple sys-
tems located in the Lake Zone of Mexico City, and that the values of a and r recom-
mended in this paper have worked ne so far for this purpose within the limitations
discussed before for the conformation of sets of motions representative of large intensity
seismic events.
The value of b affects the level of conservatism involved in the use DMI
MH
S
. Particu-
826 A. TERAN-GILMORE AND J. JIRSA
larly, as b increases, the level of conservatism of its damage estimates tends to decrease,
in such way that the use of DMI
MH
S
with values of b larger than 1.6 seems unsafe for
design purposes. Although DMI
PA
neglects the way in which the plastic energy has been
dissipated, it can not be considered equivalent to DMI
MH
S
. In fact, the conservatism of
the damage estimates derived from DMI
PA
exhibit opposite tendencies than those ob-
tained from DMI
MH
S
.
The design level derived from the use of DMI
MH
S
and DMI
PA
does not exhibit a high
sensitivity with respect to the values of b and . As suggested before by Cosenza et al.
1993 and the results obtained herein, the uncertainty in the determination of these val-
ues would appear not to impact signicantly the design results obtained from Ap-
proaches A and B. Within this context, b of 1.5 and of 0.15 seem to yield reasonable
results for the seismic design of systems exhibiting fairly stable hysteretic behavior.
Conservative values, such as b of 1.2 and of 0.30, may be considered for structures
exhibiting rapid deterioration of their hysteresis loop.
Although design against low cycle fatigue has been approached in global terms in
this paper, design considerations should also be made at the local level. Detailing con-
siderations of structural members against low cycle fatigue, such as those discussed by
Dutta and Mander 2001, require a clear understanding of the relationship existing be-
tween the plastic deformation demands at the local and global levels.
An issue that has not been considered in this paper is the effect of the hysteretic be-
havior on the seismic demands and capacities of an earthquake-resistant structure. In
some cases, the response of a structure becomes sensitive to the specics of its hysteretic
behavior, particularly for systems that exhibit pinching. Another issue not considered ex-
plicitly is the multi-degree-of-freedom effect. Results obtained by several researchers
suggest that the response of an SDOF system can be used to obtain reasonable estimates
of displacement, energy dissipation, and structural damage in regular structures Qi and
Moehle 1991, Tso et al. 1993, Teran-Gilmore 2004. Nevertheless, there are still issues
to be addressed within this context, such as the effect of higher modes and of layout and
structural irregularities.
Finally, one of the challenges involved in formulating a damage control methodology
that accounts for low cycle fatigue is the estimation of the energy demands in the
earthquake-resistant structure. In particular, it is of interest to formulate methodologies
that can be easily adapted to current seismic design formats.
CONCLUSIONS
Low cycle fatigue is, in many cases of practical interest, an issue during seismic de-
sign. Displacement-control seismic design methodologies, in particular, seem to provide
adequate level of safety for the design of structures with stable hysteretic behavior and
subjected to typical rm soil motions. Nevertheless, the use of low cycle fatigue mod-
els should be considered for the design of structures exhibiting rapid and excessive de-
terioration of their hysteresis loop, and for any type of structure subjected to long-
duration, narrow-banded ground motion.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 827
The concept of target ductility complemented with the use of simple damage indices
seems to provide a robust set of tools for seismic design against low cycle fatigue. This
is particularly true for well-conceived regular structures that exhibit stable hysteretic be-
havior and controlled response during severe ground motion. The application of the prin-
ciples of capacity design and performance-based design are instrumental to achieve this
type of behavior. As for structures that exhibit irregularities and/or exhibit rapidly dete-
riorating hysteretic behavior, this set of tools becomes sensitive to the specics of the
local and global hysteretic behavior, and thus its application becomes less reliable. As
has been done in other contexts, the use of the tools discussed herein can be applied to
determine the strength and ultimate deformation requirements of ductile structures with
stable hysteretic behavior, while a more stringent application should be considered for
structures with erratic seismic behavior.
Urgent issues that need to be addressed to make possible the use of damage indices
in practical seismic design are the harmonization of denitions used and results obtained
by different researchers, and the development of a clear understanding of the implica-
tions of the application of a damage index to the design of a particular structure.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the support of Universidad Autonoma Metro-
politana, University of Texas, Fulbright Scholar Program and Consejo Nacional de Cien-
cias y Tecnologia CONACyT, during Dr. Teran-Gilmores stay at the University of
Texas at Austin as a visiting researcher.
NOTATION
The following symbols are used in this paper:
a structural parameter that accounts for the energy content of the ground
motion
b structural parameter that characterizes the stability of the hysteretic cycle
COV coefcient of variation
DMI
MH
damage index based on linear cumulative damage theory
DMI
MH
N
DMI
MH
S
/ DMI
MH
DMI
MH
S
Teran and Jirsa damage index
DMI
PA
Park and Ang damage index
E
H
plastic energy demand
E
I
relative input energy
F
y
strength at yield
n, n
i
number of plastic excursions of amplitude
c
,
ci
the ground motion demands
from a system
828 A. TERAN-GILMORE AND J. JIRSA
N, N
i
number of plastic excursions of amplitude
c
,
ci
a system is able to undergo
before failure
Ndif number of intervals into which all plastic excursions are classied according
to amplitude
Nexc total number of plastic excursions
NE
H
normalized plastic energy
r reduction factor that characterizes the cyclic deformation capacity of a
system
S
a
spectral acceleration pseudoacceleration
T fundamental period of vibration
T
s
corner period
V
b
base shear
proportionality constant between n and N
constant in Park and Ang damage index that characterizes the stability of the
hysteretic cycle

c
,
ci
cyclic displacement, subscript indicates ith excursion or ith amplitude
interval

max
maximum displacement demand during a ground motion

u
ultimate displacement capacity

uc
ultimate cyclic displacement capacity

y
displacement at yield
maximum ductility demand associated to a spectral value

c
,
ci
cyclic ductility,
c
/
y
, subscript indicates ith excursion or ith amplitude
interval

max
maximum ductility demand during a ground motion,
max
/
y

maxc
maximum cyclic ductility demand

u
ultimate ductility capacity,
u
/
y

uc
ultimate cyclic ductility capacity,
uc
/
y
REFERENCES
Akiyama, H., and Takahashi, M., 1992. Response of reinforced concrete moment frames to
strong earthquake ground motions, Nonlinear Seismic Analysis and Design of Reinforced
Concrete Buildings, edited by H. Krawinkler and P. Fajfar, Elsevier Applied Science, pp.
105114.
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 829
Applied Technology Council ATC, 1997. NEHRP Guidelines for the Seismic Rehabilitation of
Buildings, prepared for the Building Seismic Safety Council, published by the Federal Emer-
gency Management Agency, Report No. FEMA-273, Washington, D.C.
Arroyo-Espinoza, D., and Teran-Gilmore, A., 2002. Use of strength reduction factors to account
for low cycle fatigue, Proceedings, 7th U.S. National Conference on Earthquake Engineer-
ing, Boston, Mass. CD-ROM
Baik, S. W., Lee, D. G., and Krawinkler, H., 1988. A simplied model for seismic response
prediction of steel frame structures, Proceedings, IX World Conference on Earthquake En-
gineering, Tokyo, Japan, V, pp. 29.
Banon, H., Biggs, J. M., and Irvine, H. M., 1981. Seismic damage in reinforced concrete
frames, J. Struct. Eng. 107 9, 17131729.
Bertero, R. D., and Bertero, V. V., 1992. Tall Reinforced Concrete Buildings: Conceptual
Earthquake-Resistant Design Methodology, Report No. UCB/EERC-92/16, University of
California, Berkeley.
Bertero, V. V., 1997. Performance-based seismic engineering: A critical review of proposed
guidelines, Proceedings, Seismic Design Methodologies for the Next Generation of Codes,
Bled, Slovenia, pp. 131.
Bertero, V. V., and Popov, E. P., 1977. Seismic behavior of ductile moment-resisting reinforced
concrete frames, Reinforced Concrete Structures in Seismic Zone, ACI SP-53, Detroit, MI,
pp. 247291.
Chen Y., and Gong, S., 1986. Double control damage index of structural ductility and dissi-
pated energy during earthquake, Journal of Building Structures in Chinese, 7 1, 3548.
Chou, C. C., and Uang, C. M., 2000. Establishing absorbed energy spectraAn attenuation
approach, Earthquake Eng. Struct. Dyn. 29, 14411455.
Chung, Y. S., Meyer, C., and Shinozuka, M., 1989. Modeling of concrete damage, ACI Struct.
J. 86 3, 259271.
Cosenza, E., and Manfredi, G., 1996. Seismic design based on low cycle fatigue criteria, Pro-
ceedings, XI World Conference on Earthquake Engineering, Acapulco, Mexico. CD-ROM
Cosenza, E., Manfredi, G., and Ramasco, R., 1993. The use of damage functionals in earth-
quake engineering: A comparison between different methods, Earthquake Eng. Struct. Dyn.
22, 855868.
Daali, M. L., and Korol, R. M., 1996. Adequate ductility in steel beams under earthquake-type
loading, Eng. Struct. 18 2, 179189.
Darwin, D., and Nmai, C. K., 1985. Energy dissipation in RC beams under cyclic loading, J.
Struct. Eng. 112 8, 18291846.
Decanini, L. D., and Mollaioli, F., 2001. An energy-based methodology for the assessment of
seismic demand, Soil Dyn. Earthquake Eng. 21, 113137.
Dutta, A., and Mander, J. B., 2001. Energy based methodology for ductile design of concrete
columns, J. Struct. Eng. 127 12, 13741381.
El-Bahy, A., Kunnath, S. K., Stone, W., and Taylor, A. W., 1999a. Cumulative seismic damage
of circular bridge columns: Benchmark and low-cycle fatigue tests, ACI Struct. J. 96 4,
633641.
El-Bahy, A., Kunnath, S. K., Stone, W., and Taylor, A. W., 1999b. Cumulative seismic damage
of circular bridge columns: Variable amplitude tests, ACI Struct. J. 96 5, 711719.
Fajfar, P., 1992. Equivalent ductility factors taking into account low-cycle fatigue, Earthquake
Eng. Struct. Dyn. 21, 837848.
Ghobarah, A., Abou-Elfath, H., and Biddah, A., 1999. Response-based damage assessment of
structures, Earthquake Eng. Struct. Dyn. 28, 79104.
830 A. TERAN-GILMORE AND J. JIRSA
Gosain, N. K., Brown, R. H., and Jirsa, J. O., 1977. Shear requirements for load reversals on RC
members, J. Struct. Eng. 103 ST7, 14611476.
Housner, G. W., 1956. Limit design of structures to resist earthquakes, Proceedings of the World
Conference on Earthquake Engineering, pp. 5-1 to 5-13.
Iemura, H., 1980. Earthquake failure criteria of deteriorating hysteretic structures, Proceedings,
VII World Conference on Earthquake Engineering, Istanbul, Turkey, 5, pp. 8188.
Krtzig, W. B., and Meskouris, K., 1997. Seismic damage evaluation treated as a low-cycle fa-
tigue process, Proceedings, Seismic Design Methodologies for the Next Generation of
Codes, Bled, Slovenia, pp. 139149.
Krawinkler, H., and Nassar, A., 1992. Seismic design based on ductility and cumulative dam-
age demands and capacities, Nonlinear Seismic Analysis and Design of Reinforced Concrete
Buildings, edited by H. Krawinkler and P. Fajfar, Elsevier Applied Science, pp. 95104.
Krawinkler, H., and Zohrei, M., 1983. Cumulative damage in steel structures subjected to earth-
quake ground motions, Computers and Structures 16, 531541.
Kunnath, S. K., Reinhorn, A. M., and Park, Y. J., 1990. Analytical modeling of inelastic seismic
response of R/C structures, J. Struct. Eng. 116 4, 9961017.
Leelataviwat, S., Goel, S. C., and Stojadinovic, B., 2002. Energy-based seismic design of struc-
tures using yield mechanism and target drift, J. Struct. Eng. 128 8, 10461054.
Manfredi, G., 2001. Evaluation of seismic energy demand, Earthquake Eng. Struct. Dyn. 30,
485499.
Manfredi, G., and Pecce, M., 1997. Low cycle fatigue of RC beams in NSC and HSC, Eng.
Struct. 19 3, 217223.
Mehanny, S. S. F., and Deierlein, G. G., 2000. Modeling of Assessment of Seismic Performance
of Composite Frames with Reinforced Concrete Columns and Steel Beams, Report No. 135,
The John A. Blume Earthquake Engineering Center, Stanford University.
Panagiotakos, T. B., and Fardis, M. N., 2001. Deformations of reinforced concrete members at
yielding and ultimate, ACI Struct. J. 98 2, 135148.
Park, Y. J., and Ang, A. H., 1985a. Mechanistic seismic damage model for reinforced concrete,
J. Struct. Eng. 111 4, 722739.
Park, Y. J., Ang, A. H., and Wen, Y. K., 1985b. Seismic damage analysis of reinforced concrete
buildings, J. Struct. Eng. 111 4, 740757.
Powell, G. H., and Allahabadi, R., 1987. Seismic damage prediction by deterministic methods:
Concepts and procedures, Earthquake Eng. Struct. Dyn. 16, 719734.
Priestley, M. J. N., 2000. Performance based seismic design, Proceedings, 12th World Confer-
ence on Earthquake Engineering, Canterbury, New Zealand. CD-ROM
Qi, X., and Moehle, J. P., 1991. Displacement Design Approach for Reinforced Concrete Struc-
tures Subjected to Earthquakes, Report No. UCB/EERC-91/02, University of California,
Berkeley.
Riddell, R., and Garcia, J. E., 2002. Hysteretic energy spectrum and earthquake damage, Pro-
ceedings, 7th U.S. National Conference on Earthquake Engineering, Boston, Mass. CD-
ROM
Scribner, C. F., and Wight, J. K., 1980. Strength decay in R/C beams under load reversals, J.
Struct. Eng. 106 ST4, 861876.
Silva-Olivera, H., and Lopez-Batiz, O., 2001. Estudio experimental sobre ndices de dao en
estructuras de concreto reforzado sujetas a cargas laterales, Proceedings, XIII Mexican Con-
ference on Earthquake Engineering, Guadalajara, Mexico in Spanish. CD-ROM
A DAMAGE MODEL FOR PRACTICAL SEISMIC DESIGNTHAT ACCOUNTS FOR LOW CYCLE FATIGUE 831
Somerville, P. G., Smith, N., Punyamurthula, S., and Sun, J., 1997. Development of Ground
Motion Time Histories for Phase 2 of the FEMA/SAC Steel Project, Report SAC/BD-97/04,
prepared for the SAC Joint Venture, published by the Structural Engineers Association of
California, Sacramento, CA.
Stephens, J. E., and Yao, J. T. P., 1987. Damage assessment using response measurements, J.
Struct. Eng. 113 4, 787801.
Teran-Gilmore, A., 1996. Performance-Based Earthquake-Resistant Design of Framed Build-
ings Using Energy Concepts, Ph.D. thesis, University of California at Berkeley.
Teran-Gilmore, A., 1998. A parametric approach to performance-based numerical seismic de-
sign, Earthquake Spectra 14 3, 501520.
Teran-Gilmore, A., 2004. On the use of spectra to establish damage control in regular frames
during global predesign, Earthquake Spectra 20 3, 9951020.
Teran-Gilmore, A., Avila, E., and Rangel G., 2003. On the use of plastic energy to establish
strength requirements in ductile structures, Eng. Struct. 25 7, 965980.
Tso, W. K., Zhu, T., and Heidebrecht, A. C., 1993. Seismic energy demands on reinforced con-
crete moment-resisting frames, Earthquake Eng. Struct. Dyn. 22, 533545.
Uang, C. M., and Bertero, V. V., 1990. Evaluation of seismic energy in structures, Earthquake
Eng. Struct. Dyn. 19, 7790.
Williams, M. S., and Sexsmith, R. G., 1995. Seismic damage indices for concrete structures: A
state-of-the-art review, Earthquake Spectra 11 2, 319349.
Williams, M. S., and Sexsmith, R. G., 1997, Seismic assessment of concrete bridges using in-
elastic damage analysis, Eng. Struct. 19 3, 208216.
Yamada, M., 1992, Practical ductility assurance of structures for aseismic design, Proceedings,
10th World Conference on Earthquake Engineering, Madrid, Spain, 6, pp. 34093414.
Zahrah, T. F., and Hall, W. J., 1984, Earthquake energy absorption in SDOF structures, J. Struct.
Eng. 110 8, 17571772.
Received 5 May 2003; accepted 10 April 2004
832 A. TERAN-GILMORE AND J. JIRSA