Estimating Inelastic Displacements

for Design: Extended Pile-Shaft-Supported

Bridge Structures

T. C. Hutchinson,a) M.EERI, Y. H. Chai,b) M.EERI, R. W. Boulanger,b) M.EERI,

and I. M. Idriss,b) M.EERI

Accurate estimation of inelastic displacements is important for the evalu-

ation of the seismic performance of structures with desired ductile response.
In this paper, nonlinear dynamic analyses results from a companion numeri-
cal study investigating the response of ductile-designed bridge structures,
were compared with a commonly applied inelastic displacement estimation
approach and an alternative approach. The extended pile-shaft-supported
bridge structures considered are susceptible to amplified response under
long-period velocity pulses, and hence an evaluation of design methods for
estimating inelastic displacement demands is warranted. In this case,
force-reduction–displacement-ductility–period (R⫺␮⌬⫺T) relations and a
mean spectral displacement approach are investigated. The alternative ap-
proach estimates inelastic displacement demand using the mean elastic spec-
tral displacement between two spectral periods that are important for the
structure’s response. Results support the conceptual merits of using the mean
spectral displacement method, indicating that the approach is capable of re-
ducing the uncertainty in predicting inelastic displacement demands for the
types of structures considered when subjected to near-fault ground motions.
[DOI: 10.1193/1.1811613]

INTRODUCTION
Current simplified seismic design procedures often use nonlinear static methods to
compare system capacity with imposed demand. These methods generally couple non-
linear pushover (capacity) curves with a presumed (typically smoothed) earthquake
ground surface spectrum (demand). Methods commonly used to estimate demand in-
clude force-reduction–displacement-ductility–period (R⫺␮⌬⫺T) and substitute struc-
ture approaches. However, previous researchers have noted the difficulty and limitations
associated with estimating inelastic displacement demands using such techniques ap-
plied to shorter-period structures, particularly for structures subjected to near-fault mo-
tions. Bertero et al. (1978) were perhaps the first to illustrate the damaging effects near-
fault motions have on structures with their analysis of the Olive View Hospital following
the 1971 San Fernando earthquake. It was concluded that inelastic response could not be
predicted with reasonable accuracy using methods that modify elastic response spectra.
More recently, in evaluating the sensitivity of bridge structures to near-fault motions,

a)
Department of Civil and Environmental Engineering, University of California, Irvine, CA 92697
b)
Department of Civil and Environmental Engineering, University of California, Davis, CA 95616

1081
Earthquake Spectra, Volume 20, No. 4, pages 1081–1094, November 2004; © 2004, Earthquake Engineering Research Institute
1082 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Mahin and Hachem (1998) performed dynamic analyses of single-degree-of-freedom

(SDOF) systems and identified trends relating the displacement demands to the funda-
mental period of the structure and the predominant period of the pulse in the motion.
Regression analysis of their results indicated that for intermediate and long-period struc-
tures current practice using the equal displacement principle provided reasonable esti-
mates of inelastic displacement demands. However, they noted the inaccuracy of using
current spectral-based techniques (such as R⫺␮⌬⫺T) for shorter-period structures.
Similarly, Baez and Miranda (2000), using mean values of 82 near-fault ground motions
concluded that for structures with periods less than about 1.3 seconds, inelastic demands
were underestimated using current spectral-based techniques.
In this paper, inelastic displacement demands from the nonlinear finite element (FE)
results presented in a companion study (Hutchinson et al. 2004) were compared with
displacement demands predicted using a force-reduction–displacement-ductility–period
(R⫺␮⌬⫺T) relation and an alternative mean spectral displacement approach. Given the
long-period nature of these structures, they can be vulnerable to long-period velocity
pulses in near-field motions and hence the motivation to evaluate design methods for
estimating inelastic displacement demands. The alternative approach estimates inelastic
displacement demand using the mean elastic spectral displacement between two spectral
periods that are important for the structure’s response. Results indicate that the mean
spectral displacement approach is capable of reducing the uncertainty in predicting in-
elastic displacement demands for these types of structures and conditions considered
during near-fault ground motions.

FORCE-REDUCTION–DISPLACEMENT-DUCTILITY–PERIOD
(R−µ⌬−T) RELATION
Seismic design methodologies that rely on reducing the elastic lateral-force demand
through force reduction coefficients have been proposed and studied by many research-
ers. Miranda and Bertero (1994) provide a comprehensive review of some of these ap-
proaches. Force reduction factors implemented in design codes are intended to account
for damping, energy dissipation capacity, and overstrength. However, several researchers
have expressed their concern regarding the lack of physical basis for force-reduction–
displacement-ductility–period relations. These and other limitations of the approach are
well known (e.g., Uang 1991, Tso and Naumoski 1991, Priestley 1993). Nonetheless, the
current design approach for bridge structures in seismically active regions may rely on
the use of force reduction factors when simplified nonlinear static methods are applied.
In use of an R⫺␮⌬⫺T relation, the force reduction factor R is defined as the ratio of
the elastic lateral force demand to the design lateral strength of the system. The elastic
lateral force demand is obtained from the 5% damped elastic acceleration response spec-
trum at the ground surface using the equivalent elastic period Te of the system, as shown
in Figure 1. The displacement ductility factor is defined as ␮⌬⬅⌬inelastic /⌬y , where ⌬y
⫽elasto-plastic (EP) yield displacement and ⌬inelastic⫽maximum inelastic displacement
of a structure, which may be determined from a finite element (FE) analysis, for ex-
ample. The nonlinear pushover analyses (as schematically shown in Figure 2a) may be
used to define an equivalent EP response of the system and determine the required pa-
rameters, including; the design lateral strength (which may be taken as the yield strength
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1083

Figure 1. Estimating the force reduction factor R from a ground surface acceleration spectrum
based on the definitions used in this study.

Voy ), ⌬y , and Te . The definition used to define the EP response in this case is described
in Hutchinson et al. (2004—in this issue). The generation of the nonlinear pushover re-
sponse for the systems considered was accomplished using the finite element model dis-
cretization illustrated in Figure 2b.
The relation between the force reduction factor R and the displacement ductility fac-
tor ␮⌬ for analyses of the bridge structures described in the companion study and sup-
ported on 3.0-m-diameter pile shafts is plotted in Figure 3. The data in Figure 3 includes
a range of equivalent elastic periods (Te⫽1.12 to 2.84 sec) and lateral strengths, which
are a result of the different above-ground pile extension heights (2D, 4D, and 6D, where
D⫽pile diameter). Ground motions used represent a range of characteristics and are
noted in Table 1 of Hutchinson et al. (2004—in this issue). As expected, the displace-
ment ductility factor generally increases with increasing force reduction factor. For com-
parison purposes, two well-known R⫺␮⌬ relations proposed by Veletsos and Newmark
(1960) are shown in Figure 3. The equal displacement observation, which implies that
R⫽␮⌬ , is generally assumed to be applicable for long-period structures. The equal en-
ergy observation, which implies R⫽冑(2␮⌬⫺1), is generally assumed to be applicable for
medium-short-period structures.
For displacement ductility factors ␮⌬⬎3.0, 86% of the cases shown in Figure 3 fall
below the equal displacement observation and nearly all of these results are associated
with near-fault or long duration motions. Although these structures might generally be
classified as ‘long-period’ systems, the equal energy relation has been shown to be a
lower bound when estimating the inelastic response of systems, particularly when the
peak deformation demand is incurred during the early cycles of loading (Ye and Otani
1999). Based on energy balance concepts, Ye and Otani (1999) suggest simplified rela-
tions for estimating maximum inelastic displacements applicable for long-period sys-
tems. The lower bound of their relation is equivalent to the equal energy relation.
Although many R⫺␮⌬⫺T expressions exist in the literature, the analysis results for
near-fault motions were compared to a general bilinear relation that assumes equal dis-
1084 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Figure 2. Schematic illustrations of (a) nonlinear pushover analysis and idealization into elasto-
plastic (EP) response (with P-⌬ effects), and (b) system considered and corresponding finite
element discretization.

placement in the long-period range and a linear relation between the force reduction fac-
tor R and displacement ductility factor ␮⌬ in the short-period range.

共␮⌬兲Formula⫽ 再 共R⫺1兲•
R
TLP
Te
⫹1
for
Te⭐TLP
Te⬎TLP
(1)

where TLP is the period of the long-period pulse in the near-fault ground motions, for the
ground motions considered in this study (Hutchinson et al. 2004). The period of the
pulse TLP is defined as the period associated with the peak in the elastic (5% damped)
acceleration response spectra that would most affect the response of this class of struc-
tures (i.e., in the long-period range). It is important to note that the near-fault ground
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1085

Figure 3. R⫺␮⌬ relation for bridge structures supported on extended pile shafts—results from
3.0-m-diameter pile shaft FE analyses with 0.05f c⬘Ag (with P-⌬ effects), where f c⬘⫽unconfined
compressive strength of the concrete and Ag⫽gross area of the pile shaft.

motions selected for this study result in a ratio of maximum velocity to acceleration
(vmax /amax) from 0.11 to 0.56 seconds (over a factor of five). Similarly, the range of
maximum displacements to velocities (⌬max /vmax) is from 0.24 to 2.52 seconds (a factor
of ten), indicating a broad range of frequency content for these motions. It is well known
that the motions’ frequency content in the range of the system’s fundamental frequency
has a large impact on the relation between inelastic and elastic displacement demands
(Veletsos 1969, Veletsos and Vann 1971). Further, subsequent R⫺␮⌬⫺T definitions have
adopted alternative (and perhaps more rigorous) period range definitions than that of
Equation 1 to account for such frequency dependency of the response (e.g., Newmark
and Hall 1973, Vidic et al. 1992). However, it is important to note that the equal dis-
placement and equal energy concepts are widely used in practice, in many cases with
little regard for more sophisticated definitions of frequency content of the ground motion
relative to the natural frequency of the system. This is done, although early work by Ve-
letsos and Newmark (1960) distinguished seven period regions, whereby the equal dis-
placement, equal energy, and elastic response would be observed. In contrast, FEMA-
356 (ASCE 2000) uses only one period range, defined by a characteristic period. In their
case, the characteristic period is defined as the period of transition between the constant
acceleration and constant velocity regions of the spectrum.
The applicability of the R⫺␮⌬⫺T relation to bridge structures supported on ex-
tended pile shafts is studied through a comparison of the actual displacement ductility
factor ␮⌬ , as obtained from the FE analysis, with the displacement ductility factor
(␮⌬)Formula calculated using Equation 1. The set of data in Figure 3 from the near-fault
ground motions is plotted against the period ratio Te /TLP in Figure 4, in terms of the
ratio of displacement ductility factors C␮ where

␮⌬
C␮⬅ (2)
共␮⌬兲Formula
The data in Figure 4 correspond to the ratio of displacement ductility factors C␮ with
1086 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Figure 4. Ratio of displacement ductility factors C␮ versus period ratio Te /TLP for structures
supported on 1.5-m- and 3.0-m-diameter pile shafts and subjected to the near-fault motions
used in this study.

P-⌬ effects. Results without P-⌬ effects are very similar and lead to the same general
observations. For Te /TLP⭐1.0, 70% of the displacement ductility factors were underes-
timated using Equation 1 (i.e., C␮⬎1.0). The mean of these analyses with Te /TLP⭐1.0 is
C␮⫽1.43, with a coefficient of variation (COV) of 41%. When Te /TLP is less than 1.0,
yielding of the structure causes its secant period to lengthen and become closer to the
period of the pulse in the ground motion, resulting in increased inelastic displacements.
Conversely, if the ratio Te /TLP is greater than 1.0, yielding of the structure causes its
secant period to lengthen and move further away from the period of the pulse in the
ground motion without significant increase in displacements. For the data with Te /TLP
⭓1.0, the ratio of displacement ductility factors C␮ is closer to 1.0 except for three cases
where C␮⬎1.5 occurred for the Taiwan motions with peak outcrop accelerations of
amax⫽0.3 and 0.5 g. These motions had a wide long-period band of strong spectral or-
dinates that descended fairly slowly in the spectrum. Excluding these three data points,
the mean of the data where Te /TLP⬎1.0 was C␮⫽1.05 with a COV⫽23%, suggesting
that the assumption of equal displacement is reasonable for these long-period structures,
provided the elastic period of the structure Te is greater than the period of the pulse in
the motion TLP .
The C␮ versus Te /TLP results in Figure 4 show that Equation 1 generally underesti-
mates displacement ductility demands from near-fault ground motions if the elastic pe-
riod of the structure is less than the ground motion’s pulse period. The scatter in the
analysis results is understandable given the complexities of inelastic response, including
the facts that the ratio Te /TLP provides no information on the intensity of the pulse in the
ground motion and that it is difficult to define the period of the pulse TLP in practice.

MEAN SPECTRAL DISPLACEMENT METHOD

Force reduction approaches appear to have limited accuracy in predicting inelastic
displacement demands imposed on these types of structures by ground motions with
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1087

Figure 5. Schematic illustrating the difficulty in estimating demands from near-fault motions
using a single spectral period.

long-period pulses. One possible reason is that they generally use a single response spec-
trum ordinate as an input to the relation, e.g., the equivalent elastic period Te of the
structure and an elastic response spectrum are often used to estimate displacement de-
mand (or force demand through the force reduction factor). In the example on the left
side of Figure 5, the three motions have identical elastic response spectral values for the
given elastic structural period (Te), but have very different spectral values at longer pe-
riods (such as might be introduced by a near-fault pulse). Inelastic deformations may be
interpreted as a process of structural stiffness degradation with lengthening of the effec-
tive period of the system. The secant stiffness at the peak superstructure displacement
can be used to define a secant period Tsec that represents the longest effective period of
the system. The three motions on the left side of Figure 5 have very different spectral
values at Tsec despite having the same value at Te . The example on the right side of Fig-
ure 5 illustrates the same concept, except that the three spectra have very different spec-
tral values at Te and the same spectral value at Tsec . From these schematic examples, it
seems reasonable to expect that the structure’s inelastic displacement may be better re-
lated to the spectral content between Te and Tsec , and not just to the spectral value at any
single value of T.
An alternative approach for the prediction of inelastic displacements is described
herein. This approach uses the mean spectral displacement between two periods that are
considered most relevant for the structure. Prior to describing the approach, a review of
the literature indicated that a similar approach, using two frequencies to estimate force
demands imposed on a system, was proposed by Kennedy et al. (1984). The approach by
Kennedy et al. (1984) uses two frequencies to estimate the force demands for the seis-
mic response of nuclear power plants subjected to a range of ground motions. In this
case, hysteretic models with degrading stiffness and strength and pinching behavior were
1088 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

used to model the response of these generally short-period structures. The frequency
range under consideration was between 1.8 to 10 hertz. Two global displacement duc-
tility’s representative of the onset of minor and major structural damage of these systems
were considered (␮⌬⫽1.85 and 4.3).
The approach described herein differs from that suggested by Kennedy et al. (1984)
in several ways. In general, the structural systems considered are very different, the sys-
tems used herein are of a long-period nature, while nuclear power plants are typically
very stiff, short-period structures. The approach of Kennedy et al. (1984) also estimates
demands based on forces, whereas the approach used herein provides a direct estimation
of displacements, which is more consistent with the current seismic design philosophy
for new structures and evaluation procedures for existing structures. In addition, the pe-
riods (or frequencies) suggested by Kennedy et al., which bound the averaging tech-
nique, consist of a period greater than the elastic period (a secant period) and an upper-
bound period that is greater than the secant period. The mean spectral displacement
method suggests the use of two bounding periods also, with the first period taken as the
system’s elastic period and the second period best taken as the secant period. The ap-
proach described herein also allows for the evaluation of demands for a wide range of
displacement ductility levels. Kennedy et al. focused on two specific displacement duc-
tility levels indicative of performance states of interest for nuclear power plants. Finally,
Kennedy et al. proposed the use of weighted averages applied to an interval of elastic
spectral acceleration ordinates that have been reduced based on anticipated (changing)
damping levels. The weighting scheme utilized is tied to the number of strong nonlinear
cycles anticipated, therefore, this must be known in advanced. In the context of the
method used herein, where the approach is primarily targeted at reducing the variation in
predictions associated with near-fault motions, it may be difficult to estimate hysteretic
damping dissipated by the system, without performing a nonlinear dynamic analysis.
In the mean spectral displacement method, the inelastic displacement is calculated
from the elastic displacement response spectrum using:

⌬mean⫽
1
T2⫺T1
冕 S 共T兲•dT
T2

T1
e
d (3)

where Sed(T)⫽elastic response displacement spectrum, and T1 and T2 define the period
interval considered most important to the structure. Several possibilities for defining the
‘‘period interval’’ for the integral in Equation 3 are discussed below. Each definition of
the period interval was evaluated by its effect on the ability of Equation 3 to predict the
dynamic analysis results. As was previously suggested and will be shown below, a rea-
sonable choice for defining the period interval is to assume T1⫽Te⫽elastic period of the
structure, and T2⫽Tsec⫽secant period of the structure defined using the maximum in-
elastic displacement of the structure. The mean spectral displacement approach is shown
schematically in Figure 6 for the case where Te and Tsec are used to define the period
interval for Equation 3.
The relative merits of this approach were evaluated in Figure 7 and 8 by comparing
several different choices for the period interval. It should be noted that the damping ratio
was taken as ␰⫽5% for all cases shown in Figures 7 and 8, except Figure 8c where ␰
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1089

Figure 6. Mean spectral displacement method using the elastic displacement spectra at the
ground surface and a period interval between T1 and T2 .

⫽15%. The cases shown in both of these figures are for structures supported on 3.0-m-
diameter pile shafts with axial loads of 0.05•f ⬘c Ag and including P-⌬ effects (similar re-
sults were obtained without P-⌬ effects). In all cases, displacement ratios C⌬ (Equation
4) were used to compare the dynamic analysis results to the mean displacement estimate.

⌬inelastic
C⌬⬅ (4)
⌬mean
where ⌬inelastic⫽inelastic displacement from the dynamic FE analysis, and ⌬mean
⫽mean elastic displacement demand as determined by Equation 3. It follows that C⌬
values less than 1.0 indicate that Equation 3 produced a conservative (high) estimate of
inelastic displacement. Figure 7a shows a case where the two periods are both taken as
Te (i.e., as if only one period was used). In this case, a displacement ratio C⌬⫽1.0 would
correspond to the equal displacement assumption.
Figures 7b and 7c show cases where the two periods are both taken as Tsec (i.e., as if
only one period was used). For Figure 7b, Tsec was defined at the peak inelastic displace-
ment from the dynamic analysis, which assumes that the correct inelastic displacement
is known. While this is clearly never the case, this approach was nonetheless used as a
means of conceptually evaluating the method. For Figure 7c, Tsec was defined at the peak
displacement predicted by the intersection of the nonlinear pushover response and the
elastic displacement spectrum, as illustrated in Figure 9. There is a slight loss of accu-
racy (increase in Sx/y , the standard error of the estimate) in going from Figure 7b to 7c,
which is understandable given that the approach in Figure 7b assumes the correct inelas-
tic displacement is known.
Figures 8a, 8b, and 8c show cases where the period interval is defined by T1⫽Te and
T2⫽Tsec . Figures 8a and 8b show cases where the method is applied using a 5% damped
spectra (as was done for the results shown in Figure 7). Figure 8c uses a 15% damped
spectra to determine C⌬ . The larger value of damping ratio of ␰⫽15% is more compat-
ible with the level of hysteretic damping anticipated for these structures between dis-
placement ductility factors of ␮⌬⫽2⫺5 (Hutchinson et al. 2002). It is important to note,
however, that a higher level of damping (particularly to attain ␰⫽15%) implies stable
1090 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Figure 7. Displacement ratios C⌬(⫽⌬inelastic /⌬mean) as a function of ␮⌬ for 3.0-m-diameter pile

shafts with P-⌬ effects: (a) C⌬ based on T1⫽T2⫽Te ; (b) C⌬ based on T1⫽T2⫽Tsec at peak ␮⌬
from dynamic analysis; and (c) C⌬ based on T1⫽T2⫽Tsec at intersection of pushover and elastic
displacement spectra (5% damped spectra used for [a], [b], and [c]).

hysteresis has been attained over a number of maximum amplitude cycles. For Figures
8a and 8c, Tsec was defined at the peak inelastic displacement from the dynamic analysis,
while for Figure 8b, Tsec was defined at the peak displacement predicted by the intersec-
tion of the nonlinear static pushover response and the elastic displacement spectra (as
illustrated in Figure 9). There is a slight loss of accuracy going from Figures 8a to 8b, as
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1091

Figure 8. Displacement ratios C⌬ (⫽⌬inelastic /⌬mean) as a function of ␮⌬ for 3.0-m-diameter pile

shafts with P-⌬ effects: (a) C⌬ based on T1⫽Te and T2⫽Tsec at peak ␮⌬ from dynamic analysis
(5% damped spectra); (b) C⌬ based on T1⫽Te and T2⫽Tsec at intersection of pushover and elas-
tic displacement spectra (5% damped spectra); and (c) C⌬ based on T1⫽Te and T2⫽Tsec at peak
␮⌬ from dynamic analysis (15% damped spectra).

was seen from Figures 7b to 7c, due to the fact that Figure 8a assumes the correct in-
elastic displacement is known. Regardless of how Tsec was defined, the use of a mean
spectral displacement between T1⫽Te and T2⫽Tsec resulted in a smaller standard error
than was obtained using only a single period (i.e., using only the elastic period 关T1
⫽T2⫽Te兴 or only the secant period 关T1⫽T2⫽Tsec兴). This can be seen by comparing either
1092 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Figure 9. Estimating the secant period Tsec of a structure by the intersection of its nonlinear
static pushover response and the elastic displacement spectrum.

Figures 7a, 7b, and Figure 8a, or Figures 7a, 7c, and Figure 8b. Although the regression
of the data shown in Figure 8c shows a larger mean value of C⌬ , the standard error of
these data is still lower than the cases using a single period. In addition, the use of a
mean spectral displacement resulted in a C⌬ that had virtually no dependence on ␮⌬ .
The effect of damping ␰ used in the mean spectral displacement method is illustrated in
comparing Figures 8a and 8c. Using the lower value of ␰⫽5% in the approach tends to
be conservative (Figure 8a), while applying the approach using a higher damping ratio of
␰⫽15% is slightly unconservative (Figure 8c). It is reasonable to assume that for these
cases, using either ␰⫽10% and/or some form of weighted average, the displacement ra-
tio C⌬ may have been closer to 1.0.
The results in Figures 7 and 8 represent an initial evaluation of the conceptual merits
of using a mean spectral displacement method (Equation 3), and as such suggest that the
method has promise for reducing uncertainty in predicting inelastic displacements for
these types of structures. The demonstration of the method, in combination with nonlin-
ear static pushover analyses (Figure 8b), and the resulting conservatism in the estimates,
indicates the approach is viable for design practice. Additional efforts need to be taken
to evaluate the method over a broader range of structural periods and ground motions,
and explore refinements that might improve its accuracy. For example, some immediate
refinements may be to use the secant period Tsec that corresponds to the inelastic dis-
placement predicted by Equation 3 (along with Te for defining the period interval), or to
evaluate simple weighting functions for integrating the area under the elastic displace-
ment spectra. The former would require an iterative assessment of ⌬mean . Other possible
refinements include adjusting the amount of damping (through the use of weighting
functions) or including empirical adjustment factors.

CONCLUSIONS
Accurate estimation of inelastic displacements is important to the overall evaluation
of the seismic response of all structures with desired ductile performance. In particular,
the extended pile-supported bridge structures considered in this study are susceptible to
ESTIMATING INELASTIC DISPLACEMENTS FOR DESIGN 1093

amplified response under long-period velocity pulses, and hence an evaluation of design
methods for estimating inelastic displacement demands is warranted. In this study, a
force reduction (R⫺␮⌬⫺T) and an alternative mean spectral displacement approach was
evaluated in terms of their ability to predict the inelastic displacements calculated from
a series of nonlinear dynamic FE analyses of extended pile-supported bridge structures.
Earthquake motions with a range of frequency contents, intensities, durations and per-
manent displacements were used as input motions. Results indicate the R⫺␮⌬⫺T rela-
tion appears to be reasonable for near-fault motions provided the elastic period of the
structure is longer than the period of the pulse (if present). However, if the elastic period
of the structure is less than the period of the pulse (in the shorter-period range), the R
⫺␮⌬⫺T relation used in this study significantly underestimated inelastic displacements.
The alternative approach for predicting inelastic displacements uses the mean elastic
spectral displacement between two periods that bracket the range of periods most im-
portant to the structure. When these two periods are taken as the elastic period (Te) and
the secant period at the peak displacement demand (Tsec), results showed a substantial
reduction in the standard error of the estimate. These results support the conceptual mer-
its of using the mean spectral displacement method, indicating that the approach is ca-
pable of reducing the uncertainty in predicting inelastic displacement demands for the
types of structures considered when subjected to near-fault ground motions. This im-
provement in the accuracy of predicting inelastic displacements indicates that this ap-
proach has promise and should be evaluated in greater detail. Specifically, a thorough
study of SDOF systems with a broad range of motions and structural characteristics is
warranted for comparison against other established techniques. The alternative approach
also has the advantage of estimating displacements directly; therefore, it can easily be
integrated into displacement-based design methodologies.

ACKNOWLEDGMENTS
This work was supported in part by the Pacific Earthquake Engineering Research
Center through the Earthquake Engineering Research Centers Program of the National
Science Foundation under Award Number 2081999. T. C. Hutchinson was also sup-
ported on a fellowship provided by the Earthquake Engineering Research Institute
(EERI) for the last year of this research. Any opinions, findings, and conclusions in this
paper are those of the authors and do not necessarily represent a policy of either agency
or endorsement by the state or federal government. Dr. Robert Taylor and Dr. Filip Fil-
ippou generously shared their computer codes and provided valuable interaction
throughout this study. Dr. Norm Abrahamson provided ground motions from the 1999
Taiwan earthquake. The above support and assistance are greatly appreciated.

REFERENCES
American Society of Civil Engineers (ASCE), 2000. Prestandard and Commentary for the Seis-
mic Rehabilitation of Buildings, prepared for the SAC Joint Venture, published by Federal
Emergency Management Agency, FEMA-356, Washington, D.C.
Baez, J. I., and Miranda, E., 2000. Amplification factors to estimate inelastic displacement de-
mands for the design of structures in the near field, Proceedings of 12th World Conference on
Earthquake Engineering, Auckland, New Zealand.
1094 T. C. HUTCHINSON,Y. H. CHAI, R. W. BOULANGER, AND I. M. IDRISS

Bertero, V. V., Mahin, S. A., and Herrera, R. A., 1978. Aseismic design implications of near-
fault San Fernando earthquake records, Earthquake Eng. Struct. Dyn. 6, 31–42.
Hutchinson, T. C., Chai, Y. H., Boulanger, R. W., and Idriss, I. M., 2004. Inelastic seismic re-
sponse of extended pile shaft-supported bridge structures, Earthquake Spectra 20 (4), 1057–
1080 (this issue).
Hutchinson, T. C., Boulanger, R. W., Chai, Y. H., and Idriss, I. M., 2002. Inelastic seismic re-
sponse of extended pile-shaft-supported bridge structures, Pacific Earthquake Engineering
Research Center, Report No. PEER 02/14, 215 pp.
Kennedy, R. P., Short, S. A., Merz, K. L., Tokarz, F. J., Idriss, I. M., Power, M. S., and Sadigh,
K., 1984. Engineering Characterization of Ground Motion—Task I: Effects of Characteris-
tics of Free-Field Motion on Structural Response, Vol. 1, NUREG/CR-3805, U.S. Nuclear
Regulatory Commission, Structural Mechanics Associates, Inc. and Woodward-Clyde Con-
sultants, May.
Mahin, S. A., and Hachem, M., 1998. Response of simple bridge structures to near-fault ground
motions, Proceedings of 5th Caltrans Seismic Research Workshop, California Department of
Transportation Engineering Service Center, Sacramento, CA.
Miranda, E., and Bertero, V. V., 1994. Evaluation of strength reduction factors for earthquake-
resistant design, Earthquake Spectra 10 (2), 357–379.
Newmark, N. M., and Hall, W. J., 1973. Seismic Design Criteria for Nuclear Reactor Facilities,
Report No. 46, Building Practices for Disaster Mitigation, Natural Bureau of Standards, U.S.
Department of Commerce, pp. 209–236.
Priestley, M. J. N., 1993. Myths and fallacies in earthquake engineering—Conflicts between de-
sign and reality, Bull. N.Z. Natl. Soc. Earthquake Eng. 26 (3), 329–341.
Tso, W. K., and Naumoski, N., 1991. Period-dependent seismic force reduction factors for
short-period structures, Can. J. Civ. Eng. 18, 568–574.
Uang, C.-M., 1991. Establishing R (or Rw) and Cd factors for building seismic provisions, J.
Struct. Eng. 117 (1), 19–28.
Veletsos, A. S., 1969. Maximum deformation of certain nonlinear systems, Proceedings of 4th
World Conference on Earthquake Engineering, Vol. 2, pp. 155–170.
Veletsos, A. S., and Newmark, N. M., 1960. Effect of inelastic behavior on the response of
simple systems to earthquake motions, Proceedings of 2nd World Conference on Earthquake
Engineering, Vol. 2, pp. 895–912.
Veletsos, A. S., and Vann, W. P., 1971. Response of ground-excited elastoplastic systems, J.
Struct. Div. ASCE 97 (ST4), 1257–1281.
Vidic, T., Fajfar, P., and Fischinger, M., 1992. A procedure for determining consistent inelastic
design spectra, Proceedings of Workshop on Nonlinear Seismic Analysis of RC Structures,
Bled, Slovenia, July.
Ye, L., and Otani, S., 1999. Maximum seismic displacement of inelastic systems based on en-
ergy concept, Earthquake Eng. Struct. Dyn. 28, 1483–1499.
(Received 13 August 2003; accepted 15 March 2004)