Professional Documents
Culture Documents
1.4.12. 2005 Scaling of Spectral Displacement Ordinates With Damping Ratios
1.4.12. 2005 Scaling of Spectral Displacement Ordinates With Damping Ratios
SUMMARY
The next generation of seismic design codes, especially those adopting the framework of performance-
based design, will include the option of design based on displacements rather than forces. For direct
displacement-based design using the substitute structure approach, the spectral ordinates of displacement
need to be specied for a wide range of response periods and for several levels of damping. The code
displacement spectra for damping values higher than the nominal value of 5% of critical will generally
be obtained, as is the case in Eurocode 8 and other design codes, by applying scaling factors to the
5% damped ordinates. These scaling factors are dened as functions of the damping ratio and, in some
cases, the response period, but are independent of the nature of the expected ground shaking. Using
both predictive equations for spectral ordinates at several damping levels and stochastic simulations,
it is shown that the scaling factors for dierent damping levels vary with magnitude and distance,
reecting a dependence of the scaling on the duration of shaking that increases with the damping ratio.
The options for incorporating the inuence of this factor into design code specications of displacement
response spectra are discussed. Copyright ? 2004 John Wiley & Sons, Ltd.
KEY WORDS: displacement response spectra; displacement-based design; damping ratios; seismic design
codes
1. INTRODUCTION
Recognition of the poor correlation between transient inertial forces induced by earthquake
shaking and damage to structures has led to the development of displacement-based approaches
for seismic design and assessment. Amongst the dierent methods that have been proposed
for estimating inelastic displacements in structures, many are based on the response of an
equivalent elastic system (see review by Miranda and Ruiz-Garca [1]). These methods are
based on equivalent linearization, using the substitute structure concept [2, 3] in which the
inelastic deformation is modelled by a reduced stiness and the hysteretic dissipation of
energy is modelled by an increased level of viscous damping.
The adoption of direct displacement-based design (DBD) approaches places the onus on en-
gineering seismologists to provide suitable input in terms of long-period spectral displacements
∗ Correspondence to: Julian J. Bommer, Department of Civil and Environmental Engineering, Imperial College
London, London SW7 2AZ, U.K.
† E-mail: j.bommer@imperial.ac.uk
for a range of damping ratios. Transformation of the acceleration spectra in current seismic
codes to displacement spectra will generally not produce reliable displacement ordinates at the
longer periods that become relevant to DBD [4, 5]. In order for the acceleration spectrum not
to result in spectral displacements that increase monotonically with increasing response period,
there must be a portion over which the acceleration decays in inverse proportion to the square
of the period. The 1990 French seismic code [6] was the rst to introduce such a decaying
branch in the acceleration spectrum for periods beyond 3:0 seconds. Eurocode 8 [7] includes
specications for displacement spectra in an informative annex, amongst the rst codes to
explicitly make provisions for displacement-based design of buildings. The EC8 displacement
spectrum (Figure 1) is designed to be compatible with the acceleration spectrum, following
the proposal of Bommer et al. [8], with the control periods TD , TE and TF based on the work
of Tolis and Faccioli [9].
Although the original motivation for this work is related to dening input for direct DBD,
the issue of over-damped spectral displacements at intermediate and long response periods is
also of clear relevance to the design of buildings and bridges incorporating base isolation or
supplementary damping devices.
There is currently a signicant level of disagreement regarding appropriate values for the
control periods of the displacement spectrum. In EC8, the Type 1 spectrum (applicable in
high seismicity areas) control period TD has a value of only 2 seconds; TE is set at 6 seconds
and the spectral ordinates are expected to converge to the peak ground displacement (PGD) at
a period (TF ) of 10 seconds. Guan et al. [10] propose displacement spectra based on Japanese
and Californian accelerograms recorded between 1989 and 2001; the control periods in their
proposal are very similar to those for the EC8 Type 1 spectrum. Other proposals assign much
larger values to the control periods: as Faccioli et al. [5] report, the procedure of Newmark and
Hall [11] yields values of 10 and 30 seconds for TE and TF respectively. The 2003 proposed
revision of the NEHRP guidelines [12] also includes a constant displacement branch in the
denition of the response spectrum; maps are provided that show the variation of the period
TL , equivalent to the period TD in the EC8 formulation (Figure 1). Values of this control
period specied for the contiguous United States vary between 4 and 16 seconds, increasing
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 147
Figure 2 compares the spectral ratios, in this case normalized to the 10% damped spectral
ordinates, from the studies of Newmark and Hall [11] and Wu and Hanson [13], which for
the 20% damped spectral ordinates are in good agreement. The ratios from both studies tend
towards unity at very short and very long periods, as would be expected. Both studies were
based on pseudo-velocity spectra and hence are applicable to displacement spectra. Lin and
Chang [14] found that whilst the ratios obtained from spectra of relative displacement show
relatively little variation with response period (and hence the same is true for pseudo-velocity
and pseudo-acceleration spectra), the ratios obtained from spectra of relative velocity and
absolute acceleration display a much more pronounced variation with period. Lin and Chang
[14] also point out that the spectral ratios for dierent damping values have generally been
determined from studies of displacement spectra and then applied to acceleration spectra for
the calculation of seismic design forces. This criticism is not relevant to the present study
Figure 2. Spectral ratios, with respect to the 10% damped ordinates, from two studies; the ratios for
the Newmark and Hall [11] study are obtained using median values.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
148 J. J. BOMMER AND R. MENDIS
since we are expressly concerned with the scaling of intermediate- and long-period ordinates
of spectral displacement, SD.
In EC8, the 5% damped spectral ordinates can be scaled for higher damping values, (%),
using a simple expression derived by Bommer et al. [8]:
SD() 10
= (1)
SD(5%) 5+
The scaling factor applies to the ordinates at all response periods between TB (the start
of the acceleration plateau) and TE (Figure 1); at shorter and longer periods the factor
increases linearly to reach a value of 1.0 at T = 0 and T = TF . Figure 3(a) compares the
resulting spectral ratios, normalized to the 5% damped spectral displacements, from Equation
(1) with values obtained by Lin and Chang [14] from their study of the median spectra from
more than 1000 strong-motion accelerograms. Although there is considerable divergence be-
tween the two sets of curves at long periods, this reects as much as anything the uncertainty
associated with obtaining long-period spectral displacements and the sensitivity of the long-
period spectral shape to record processing. Certainly over the period range that corresponds
to the constant displacement plateau in the EC8 spectrum between 2 and 6 seconds, there is
reasonable agreement with the results of Lin and Chang [14].
Equation (1) replaced a similar expression included in the original draft of EC8 [15], the
origin and derivation of which have not been documented:
SD() 7
= (2)
SD(5%) 2+
Figure 3(b) compares the ratios from Equation (2) with the values obtained by Lin and
Chang [14] using the 84-percentile spectral ordinates. For periods up to 6 seconds, the agree-
ment is excellent, which may be fortuitous but might also give a clue as to the derivation of
Equation (2). We believe that the scaling factors adopted in seismic codes for adjusting the
5% damped spectrum to other damping ratios should be based on median estimates, since the
Figure 3. Spectral ratios, with respect to the 5% damped ordinates: (a) Equation (1) compared with the
values of Lin and Chang [14] based on median ordinates; and (b) Equation (2) compared with the values
of Lin and Chang [14] based on 84-percentile ordinates. In both plots, the ratios for the EC8 spectra
have been obtained using the control periods for the Type 1 spectrum and site class C.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 149
inherent variability in the ground-motion predictions is already accounted for in the proba-
bilistic derivation of the 5% damped spectrum. As can be appreciated from Figure 3, using the
ratios based on 84-percentile values will produce lower spectral ordinates for higher damping
values and it is not clear to us how such a choice could be rationalized for safe design.
A third variation of the scaling factor was proposed by Tolis and Faccioli [9], whose study
was based primarily on the recordings of the 1995 Hyogo-ken Nanbu (Kobe) earthquake:
SD() 15
= (3)
SD(5%) 10 +
In a few other seismic design codes, alternative scaling factors are encountered, although
it would appear that most code drafting committees have not envisaged the use of damping
ratios other than the nominal value of 5% assumed for reinforced concrete. The 1990 French
code [6] and the 1994 Spanish code [16] both included the following scaling factor for spectral
accelerations, SA:
0:4
SA() 5
= (4)
SA(5%)
The 1983 Portuguese seismic code [6] and the 1984 Indian seismic code [6] both include
graphical representations of acceleration spectra at more than one damping level, but do not
provide scaling factors. Figure 4 compares the ratios obtained from these graphs, averaged
over the period range covered (in both cases the variation with period is small), and the ratios
obtained from Equations (1)–(4).
The 2001 Caltrans Seismic Design Criteria [17] allow damping to be increased from
5% to 10% if the bridge is heavily inuenced by energy dissipation at the abutments and is
expected to respond like a single-degree-of-freedom system. The spectral displacements can
then be scaled using the following equation, originally derived by Kawashima and Aizawa
[18] for spectral ordinates of absolute acceleration:
SD() 1:5
= + 0:5 (5)
SD(5%) 0:4 + 1
Figure 4. Spectral reduction factors for dierent damping values from a number of seismic design codes
and from the study of Tolis and Faccioli [9].
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
150 J. J. BOMMER AND R. MENDIS
The 1997 edition of the Uniform Building Code (UBC) [19] provides scaling factors at
a number of discrete damping values for the design of seismically isolated structures. These
factors, and those from Equation (5), are also shown in Figure 4.
The curves and individual values shown in Figure 4 reect a signicant disagreement
amongst models for the scaling of 5% damped spectral displacements for higher damping
values. Since the studies are presumably all based on the analysis of elastic single-degree-
of-freedom (SDOF) oscillators, one may assume that the dierences have not arisen from
variations in the structural models. The logical way to explore the possible cause of the
divergence is to investigate the extent to which the scaling factors are dependent on the
characteristics of the seismic demand.
3.1. Magnitude
Using the four predictive equations listed in Table I, median response spectra for dierent
damping levels were constructed for a rock site at 10 km from earthquakes of dierent mag-
nitude. For each magnitude, the spectral ordinates were then divided by the ordinates of the
5% damped spectrum and the ratios plotted (Figure 5). A consistent pattern was observed in
all cases: for periods greater than about 0:3 seconds—which are the periods of relevance to
direct DBD (based on equivalent linearization)—the reduction of the spectral displacements
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 151
Notes: 1. Distance measures as dened by Abrahamson and Shedlock [24]; 2. PSV = pseudo-velocity response
spectrum, PSA = pseudo-acceleration response spectrum, SD = relative displacement response spectrum; 3. As
with some of the design codes reviewed previously, values of damping lower than 5% are not considered
in this study since they are unlikely to be relevant to displacement-based design; 4. Modied hypocentral
distance that accounts for source dimensions.
Figure 5. Variation of the ratios of 20% damped spectral ordinates to the 5% damped
ordinates with earthquake magnitude from the equations of (upper) Trifunac and Lee
[21] and (lower) Bommer et al. [23].
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
152 J. J. BOMMER AND R. MENDIS
for higher damping ratios increases with the earthquake magnitude. Figure 6 shows the vari-
ation of the scaling factors for a response period of 2:0 s—the longest period covered by the
equations of Boore et al. [22]—with magnitude for dierent damping ratios from all four
equations.
Stochastic simulations were also generated for the same scenario of a rock site at 10 km
from earthquakes of dierent magnitude, using the program SMSIM [26]. The same trend
of decreasing spectral ratios with increasing magnitude was observed, as shown in Figure 7.
The default source, path and site parameters for coastal California were employed for the
simulations.
The variation of the spectral ratios with magnitude shown in Figure 6 clearly varies from
one equation to another, and may in some cases appear to be not very pronounced; the
curvature in the plots corresponding to Boore et al. [22] is the result of the quadratic term in
magnitude in their equations. However, the ratio of 20% to 5% spectral ordinates are similar
for all four equations and it is reasonable to assume that the pronounced gradient of the curve
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 153
Figure 7. Variation of the spectral ratios (relative to the 5% damped ordinates) at T = 2:0 s with
earthquake magnitude inferred from stochastic simulations.
for 30% obtained from the equations of Bommer et al. [23] would also be observed for these
other equations had they provided coecients for this damping level.
In plotting the ratios for magnitudes up to 8 the equations are extended slightly beyond
their range of applicability. For the 30% damped spectrum, the scaling factors inferred from
the equations of Bommer et al. [23] decrease by 15% as the magnitude is increased from 6
to 7.5.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
154 J. J. BOMMER AND R. MENDIS
Figure 8. Variation of the ratios of 20% damped spectral ordinates to the 5% damped ordinates with
source-to-site distance (measured from the surface projection of the fault rupture) from the equations
of Boore et al. [22] and Bommer et al. [23] for an earthquake of magnitude 7.
based on 30m shear wave velocities, with 360m=s and 760m=s marking the boundaries between
classes. The results from the equations of Bommer et al. [23] do not show a consistent pattern
but it is known that the classication of a large proportion of strong-motion stations in Europe
and the Middle East is highly uncertain. The same qualifying remark applies to the equations
of Berge-Thierry et al. [20]; their results indicate lower scaling factors at ‘rock’ sites than at
‘alluvium’ sites, but for response periods beyond 5 seconds (not shown in the plots in order
for them to be visually comparable) the curves are inverted. Amongst the four studies, the
only one for which most recording station site classications may be considered reliable is
Boore et al. [22] and hence greatest weight should perhaps be given to the pattern indicated
from their results: appreciably lower spectral scaling factors at soil sites than at rock sites.
The number of records from rock sites in their data set, however, was limited, so even here
some caution may be in order in drawing conclusions from the results.
There are some cases where site eects on strong motion have been very pronounced,
foremost amongst these being the recordings from Mexico City of the 1985 Michoacan earth-
quake. Figure 10 shows one of these records, its displacement response spectra for various
damping levels and the variation of the ratios of 30% to 5% spectral displacements with
response period. At the dominant period of this narrow-band signal, the reduction factors are
very small, whereas at other periods the scaling factors are exceptionally high.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 155
Figure 9. Variation of the spectral ratios (20% to 5% damped ordinates) with site classication inferred
from median values obtained from the four predictive equations in Table I.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
156 J. J. BOMMER AND R. MENDIS
Figure 10. Top: N90◦ component of the recording from the soft soil site at SCT
of the 19 September 1985 Michoacan, Mexico, earthquake. Middle: Displacement
response spectra for various damping levels. Bottom: Ratio of 30% to 5% damped
ordinates compared with those obtained from Equations (1) and (2).
The proposal for using the square root of the factors dened in Equation (1) for the special
case of near-source earthquake ground motions was based on the idea that ‘in the near-eld
region, the velocity pulses may reduce the eectiveness of damping’ [30]. Figure 11 shows
a classic near-source accelerogram aected by forward rupture directivity: the Lucerne record
of the 1992 Landers (California) earthquake. The gure shows the time-histories and the
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 157
Figure 11. Fault normal component of the Lucerne accelerogram from the 1992 Landers
earthquake: 5% damped pseudo-velocity response spectrum and time-histories of accelera-
tion, velocity and displacement.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
158 J. J. BOMMER AND R. MENDIS
Figure 12. Displacement spectra of the fault normal component of the Lucerne accelerogram from the
1992 Landers earthquake (upper) and spectral ratios, normalized to the 5% ordinates, compared with
the ratios given by Equations (1) and (6).
Figure 12 shows the displacement response spectra for various damping levels obtained
from the accelerogram in Figure 11, and compares the spectral ratios with those obtained
from Equations (1) and (6). One should, of course, be cautious about drawing conclusions
from a single record, although the time history in question clearly represents a case of almost
all the energy of the ground shaking being concentrated in a single velocity pulse. The
average ratios obtained from the accelerogram (the short-period variations are the result of
high-frequency excitation in the record due to a thin soil layer at the site) at most periods
agree well with those predicted by the equation of Priestley [30]. Only in the region of the
period of the velocity pulse itself (∼ 4 seconds), are the ratios much closer to those predicted
by the existing equation in EC8.
From Figure 12 it is clear that the velocity pulse does manifest on the 5% damped spectrum
as a distinct peak in the displacements, but the eect is smoothed out by higher damping
ratios. This is an area that requires further investigation, which should be undertaken in tandem
with rening models for the displacement spectra due to near-source ground motions. The
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 159
results shown here do suggest that Equation (6) proposed by Priestley [30] for reduced spectral
scaling factors for motions aected by forward directivity is an appropriate modication, but
it is also necessary to ensure that the forward directivity pulse is captured in the denition
of the 5% damped spectrum.
The results of the previous section suggest that the divergence amongst proposed scaling fac-
tors for adjusting the 5% damped spectral ordinates for higher levels of damping displayed
in Figure 4, may in large part be due to dierences in the characteristics of the ground
motions employed in their derivation. Figure 3 also indicates that another cause of the diver-
gence may be the choice of using median values or some other percentile. Although already
stated previously, we again make our case against the use of values other than the median for
determining these ratios, the fundamental point being that since they are to be applied to a
5% damped spectrum derived from probabilistic hazard analysis, the use of values other than
the median is to double count the inuence of the scatter in the ground-motion prediction
equations. Furthermore, it is in contradiction with the probabilistic approach to make any
arbitrary selection of percentiles on the basis of perceived ‘conservatism’. This is even more
the case here where the ‘conservative’ choice of the ratio of 84-percentile ordinates, as pro-
posed by Lin and Chang [14], actually results in lower—and hence not conservative at all—
spectral displacements for higher damping values.
Instead of making subjective decisions regarding appropriate levels of aleatory variabil-
ity, our approach has been to identify causes for the variation in the median values and to
transform random variability into epistemic uncertainty by identifying additional explanatory
variables to be incorporated into the prediction of the scaling factors.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
160 J. J. BOMMER AND R. MENDIS
Figure 13. Maximum spectral displacements of damped oscillators (middle) of period T and dierent
damping levels subjected to harmonic excitation with the same period of vibration (top) and variation
of spectral ratio with number of cycles of motion (bottom).
The patterns identied in Section 3 show that the spectral ratios are strongly dependent on
earthquake magnitude and source-to-site distance, and weakly dependent on the site classi-
cation. All of these patterns are entirely consistent with the ratios being dependent on the
strong-motion duration. In order to be meaningful, any discussion of duration must specify
which of the 40 or so denitions encountered in the literature is being employed [31].
Bommer and Martinez-Pereira [32] grouped denitions of strong-motion duration into three
generic categories of bracketed, uniform and signicant, but also identied that the most im-
portant distinction amongst denitions is whether the time interval is determined from absolute
levels of motion or from relative proportions of the maximum amplitude or Arias intensity.
Since in the present context the discussion is focused on the ratios of spectral displacements
for oscillators with dierent levels of equivalent viscous damping, the appropriate denitions
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 161
Figure 14. Predicted median values of signicant duration from the equations of Abrahamson and Silva
[34] for the signicant duration based on 5 –75% of the total Arias intensity.
are those based on relative measures, such as the signicant duration. This is dened as the
interval between the times at which two specied levels, such as 5% and 95% or 5% and 75%,
of the total Arias intensity are reached [33]. Predictive equations for signicant duration show
that the duration of the shaking grows strongly with magnitude (an observation that holds
for all denitions of duration) and also increases with increasing distance from the seismic
source; the increase of duration at soil sites compared to rock sites is small (Figure 14).
Strong-motion duration is generally not found to be strongly dependent on site classication
when the latter is classied only by the nature of the uppermost layers at the site [32]. This
is not to discount the fact that duration of shaking can be appreciably prolonged at certain
soil sites, but this is interpreted to be due to two- and three-dimensional eects of basins
and the eect of ‘trapped’ energy [35]. The lack of clear trends in most of the graphs in
Figure 9 is therefore still consistent with duration of shaking being the underlying cause of
the variation of the spectral ratios. Of the four equations for predicting spectral ordinates
used in this study, only Boore et al. [22] can be considered to be based on consistent and
reliable site classications. The curves in Figure 9 derived from these equations do show a
clear separation of the behaviour at rock and soil sites, from which it might be concluded
that since the sti and soft soil sites will include some that are in basins, the lower ratios
predicted for these sites are due to the prolongation of shaking at such locations.
The eect of near-source rupture directivity, and the corresponding reduction of the spectral
scaling factors proposed by Priestley [30], is also consistent with the controlling parameter
being duration. The model of Somerville et al. [27] predicts a reduction of the signicant
duration (5–75% of Arias intensity) by 60% due to forward rupture directivity experienced
by the Lucerne accelerogram (Figure 11).
Returning to Figure 4, however, there is a case that would appear to contradict this in-
terpretation: the ratios predicted by Tolis and Faccioli [9], based on data from the Mw = 6:9
Kobe earthquake, are higher than those obtained by Bommer et al. [8], based on a Eu-
ropean data set concentrated in the magnitude range from 5.5 to 7.0. However, there is
once again an explanation that is consistent with the interpretation of duration as being the
key parameter: the Kobe earthquake was an almost pure bi-lateral fault rupture, hence the
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
162 J. J. BOMMER AND R. MENDIS
duration of shaking at near-source sites was about one half of the values that would be
expected from a unilateral rupture [32]. The curves in Figure 14 would suggest that the
impact of bi-lateral rupture on the duration of shaking would have been equivalent to re-
ducing the magnitude by one unit to about Mw = 6. Bi-lateral fault ruptures in major earth-
quakes are rare [36] hence such adjustments for the duration do not often need to be made.
However, the 1989 Loma Prieta (California) earthquake (Mw = 6:9) was another case of
almost pure bi-lateral rupture; recordings from this event constitute 44% of the dataset of
Boore et al. [22] hence the magnitude-dependence inferred from their equations may, in ef-
fect, be even higher.
The ratios in the lower graph on Figure 13 are much lower than any of those depicted in
Figure 4, conrming that although general trends may be inferred from consideration of har-
monic signals, they are not comparable with non-stationary earthquake signals. Nonetheless,
Figure 13 does serve to support the interpretation of duration as the controlling parame-
ter. Denitions for counting the eective number of cycles of motion in accelerograms vary
almost as much as those for measuring the duration [37], but use is made here of the equation
of Liu et al. [38] for the equivalent number of cycles of motion at 65% of the maximum
amplitude. For a site at 10 km from the source, the equation predicts that the number of full
cycles will be 3 for Mw = 5:5, 5 for Mw = 6:5 and 10 for Mw = 7:5. According to the graph at
the bottom of Figure 13, increasing the number of cycles from 3 to 10 results in a reduction
of the ratio of the 30% to the 5% damped spectral displacement of about 40%, which is
about twice the reduction inferred from the equations of Bommer et al. [23] for the same
increase of magnitude (Figure 6). This dierence emphasizes the fact that transient earthquake
ground motions are not harmonic signals and hence the spectral ratios from accelerograms will
generally be much higher than those shown in Figure 13. However, the accelerogram from
Mexico City shown in Figure 10 came very close to reaching the steady state as a result of
the response of the lacustrine deposits underlying the city; it can be seen that at the dominant
period of about 2 seconds, the spectral ratio for 30% to 5% damped ordinates was about 0.25,
tending towards the value of 0.167 for steady-state response to a harmonic excitation.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 163
this is not recommended because disaggregation of the hazard would often show that the
duration at a given location is dominated by a dierent earthquake scenario than that driving
the spectral displacement at the period of interest. This would then lead to the 5% damped
spectral ordinates and the scaling factors for higher damping levels being incompatible. Since
in current code formats the magnitude and distance of the controlling events are generally not
visible to the user, use can be made of the parameters used to dene the spectral ordinates.
The denition of long-period spectral displacements will require at least three parameters to
be mapped, the rst two being spectral ordinates at short and intermediate response periods,
the third being either a long-period ordinate or, as in the NEHRP guidelines [12] discussed
in the Introduction, the control period for the constant displacement plateau. Bommer et al.
[8] proposed the alternative approach of mapping PGA, PGD and the peak ground velocity
(PGV), and then dening the corner periods of the spectrum from their ratios. Since, like
the NEHRP period TL , the ratio between PGD and PGV at rock sites is a function primarily
of the earthquake magnitude—and only weakly of distance [39, 40]—the adjustment to the
spectral scaling factor could be dened from these values.
5. CONCLUSION
Direct displacement-based seismic design and assessment require input in the form of displace-
ment response spectra over long period ranges (up to the product of the yield period and the
square root of the ductility demand factor) and for a number of damping levels (up to about
30% of critical). Spectral displacements for long periods and high damping levels are also
directly relevant to the design of bridges and buildings with base isolation and supplementary
damping devices. In seismic design codes, the spectra for damping levels higher than 5%
are obtained by applying scaling factors to the ordinates of the 5% damped spectrum. These
factors have been shown to be weakly dependent on response period other than in those re-
gions where the spectral displacements converge to zero or to PGD. At intermediate response
periods, the spectral scaling factors are currently dened only in terms of the damping ratio
and there is signicant disagreement amongst the proposed factors.
This study has shown that the spectral scaling factors vary with seismological features: the
factors decrease with increasing magnitude, decrease with increasing distance, and to a lesser
extent, increase for softer site conditions. These variations all reect a consistent trend of
the scaling factors decreasing with increasing duration of the ground motion. We have not
presented duration-dependent scaling factors because these should be obtained concurrently
with the derivation of displacement spectra for wide period ranges, including the eect of
near-source rupture directivity. The purpose of this paper is to demonstrate the inuence
that duration has on the spectral scaling factors and to explore the extent and nature of
this inuence. These ndings can be incorporated into ongoing work to dene displacement
response spectra for design, thus producing more reliable models and improved estimates of
the design motions.
ACKNOWLEDGEMENTS
The authors are indebted to Professors Nigel Priestley and Jose Ignacio Restrepo, amongst others, for
discussions that contributed to the motivation to undertake this work. Special thanks are also due to
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
164 J. J. BOMMER AND R. MENDIS
Dr Robert W. Graves of URS Corporation for providing the rotated Lucerne accelerogram. Thanks are
also due to Dr David M. Boore for his SMSIM program and to Fleur Strasser for guidance on running
simulations. We are also grateful to Jonathan Hancock and Dr Sarada K. Sarma for useful discussions
related to this topic.
A rst draft of this paper was reviewed by the following people: Sinan Akkar, Juliet Bird, David
Boore, Michele Calvi, Helen Crowley, Damian Grant, Jonathan Hancock, Mervyn Kowalsky, Eduardo
Miranda, Rui Pinho and Nigel Priestley. We are indebted to all of them for their insightful comments
and helpful suggestions, all of which have contributed to signicant improvement of the manuscript.
REFERENCES
1. Miranda E, Ruiz-Garca J. Evaluation of approximate method to estimate maximum inelastic displacement
demands. Earthquake Engineering and Structural Dynamics 2002; 31:539 – 560.
2. Gulkan P, Sozen M. Inelastic response of reinforced concrete structures to earthquake ground motions. ACI
Journal 1974; 71:604 – 610.
3. Shibata A, Sozen M. Substitute-structure method for seismic design in RC. Journal of Structural Division
(ASCE) 1976; 102:1–18.
4. Bommer JJ, Elnashai AS. Displacement spectra for seismic design. Journal of Earthquake Engineering 1999;
3(1):1– 32.
5. Faccioli E, Paolucci R, Rey J. Displacement spectra for long periods. Earthquake Spectra 2004; 20(2):
347– 376.
6. Regulations for seismic design: A world list—1996. International Association for Earthquake Engineering,
Tokyo, Japan, 1996.
7. Eurocode No. 8: Design of structures for earthquake resistance, Part 1: General rules, seismic actions and rules
for buildings. Stage 49 draft, CEN, Brussels, 2003.
8. Bommer JJ, Elnashai AS, Weir AG. Compatible acceleration and displacement spectra for seismic design codes.
Proceedings of the 12th World Conference on Earthquake Engineering, Auckland, 2000; Paper no. 207.
9. Tolis SV, Faccioli E. Displacement design spectra. Journal of Earthquake Engineering 1999; 3(1):107–125.
10. Guan J, Hao H, Lu Y. Generation of probabilistic displacement response spectra for displacement-based design.
Soil Dynamics and Earthquake Engineering 2004; 24:149 –166.
11. Newmark NM, Hall WJ. Earthquake spectra and design. EERI Monograph, Earthquake Engineering Research
Institute, 1982.
12. The 2003 NEHRP recommended provisions for new buildings and other structures. Part 1: Provisions (FEMA
450), Federal Emergency Management Agency, Washington DC, 2003.
13. Wu J, Hanson RD. Study of inelastic spectra with high damping. Journal of Structural Engineering (ASCE)
1989; 115(6):1412 –1431.
14. Lin YY, Chang KC. Study on damping reduction factor for buildings under earthquake ground motions. Journal
of Structural Engineering (ASCE) 2003; 129(2):206 – 214.
15. Eurocode 8: Design of structures for earthquake resistance, Part 1: General rules, seismic actions and rules for
buildings. ENV 1998-1-1, CEN, Brussels, 1994.
16. Norma de construccion sismorresistente (parte general y edicacion). Instituto Geograco Nacional, Ministerio
de Obras Publicas, Transportes y Medio Ambiente, NCS-94, Spain, 1994.
17. Seismic design criteria version 1.2. California Department of Transportation, Sacramento, California, 2001.
18. Kawashima K, Aizawa K. Modication of earthquake response spectra with respect to damping ratio.
Proceedings of the 3rd US National Conference on Earthquake Engineering, Charleston, South Caroline,
vol. II, 1986; 1107–1116.
19. Uniform Building Code. Structural Engineering Design Provisions, vol. 2, International Conference of Building
Ocials, Whittier, CA, 1997.
20. Berge-Thierry C, Cotton F, Scotti O, Griot-Pommera D-A, Fukushima Y. New empirical spectral attenuation
laws for moderate European earthquakes. Journal of Earthquake Engineering 2003; 7(2):193 – 222.
21. Trifunac MD, Lee VW. Empirical models for scaling pseudo relative velocity spectra of strong earthquake
accelerations in terms of magnitude, distance, site intensity and recording site conditions. Soil Dynamics and
Earthquake Engineering 1989; 8(3):126 –144.
22. Boore DM, Joyner WB, Fumal TE. Estimation of response spectra and peak accelerations from western North
American earthquakes: an interim report. US Geological Survey Open-File Report 93-509, 1993.
23. Bommer JJ, Elnashai AS, Chlimintzas GO, Lee D. Review and development of response spectra for
displacement-based design. ESEE Research Report No. 98-3, Imperial College London, 1998.
24. Abrahamson NA, Shedlock KM. Overview. Seismological Research Letters 1997; 68(1):9 – 23.
25. Boore DM. Simulation of ground motion using the stochastic method. Pure and Applied Geophysics 2003;
160:635 – 676.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165
SPECTRAL DISPLACEMENT ORDINATES 165
26. Boore DM. SMSIM—Fortran programs for simulating ground motions from earthquakes: Version 2.0 —A
revision of OFR 96-08-A, A modied version of OFR 00-509, describing the program as of February 09,
2003 (version 2.20), US Geological Survey, 2003.
27. Somerville PG, Smith NF, Graves RW, Abrahamson NA. Modication of empirical strong ground motion
attenuation relations to include the amplitude and duration eects of rupture directivity. Seismological Research
Letters 1997; 68(1):199 – 222.
28. Abrahamson NA. Eects of rupture directivity on probabilistic seismic hazard analysis. Proceedings of 6th
International Conference on Seismic Zonation, Palm Springs, California, 2000.
29. Somerville PG. Magnitude scaling of the near fault rupture directivity pulse. Physics of the Earth and Planetary
Interiors 2003; 137:201– 212.
30. Priestley MJN. Myths and Fallacies in Earthquake Engineering, Revisited. IUSS Press, University of Pavia,
Italy, 2003.
31. Bommer JJ, Martnez-Pereira A. Strong-motion parameters: denition, usefulness and predictability. Proceedings
of the 12th World Conference on Earthquake Engineering, Auckland, 2000; Paper no. 206.
32. Bommer JJ, Martnez-Pereira A. The eective duration of earthquake strong motion. Journal of Earthquake
Engineering 1999; 3(2):127–172.
33. Trifunac MD, Brady AG. A study on the duration of strong earthquake ground motion. Bulletin of the
Seismological Society of America 1975; 65:581– 626.
34. Abrahamson NA, Silva WJ. Empirical ground motion models. Report to Brookhaven Laboratory, 1996.
35. Trifunac MD, Novikova EI. State of the art review on strong motion duration. Proceedings of the 10th European
Conference on Earthquake Engineering, Vienna, vol. 1, 1995; 131–140.
36. McGuire JJ, Zhao L, Jordan TH. Predominance of unilateral rupture for a global catalog of large earthquakes.
Bulletin of the Seismological Society of America 2002; 92(8):3309 – 3317.
37. Hancock J, Bommer JJ. Predicting the number of cycles of ground motion. Proceedings of the 13th World
Conference on Earthquake Engineering, Vancouver, 2004; Paper no. 1989.
38. Liu A, Stewart JP, Abrahamson N, Morikawa Y. Equivalent number of uniform stress cycles for liquefaction
analysis. Journal of Geotechnical and Geoenvironmental Engineering (ASCE) 2001; 127(12):1017–1026.
39. Sadigh RK, Egan JA. Updated relationships for horizontal peak ground velocity and peak ground displacement
for shallow crustal earthquakes. Proceedings of the 6th US National Conference on Earthquake Engineering,
1998.
40. Tromans IJ, Bommer JJ. The attenuation of strong-motion peaks in Europe. Proceedings of the 12th European
Conference on Earthquake Engineering, London, 2002; Paper no. 394.
Copyright ? 2004 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2005; 34:145–165