You are on page 1of 12

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

net/publication/281212369

Normalized Response Spectrum of Ground Motion

Article · March 2015

CITATIONS READS

3 3,097

1 author:

Praveen K. Malhotra
StrongMotions Inc.
74 PUBLICATIONS   1,521 CITATIONS   

SEE PROFILE

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

Performance-based seismic design of a liquid nitrogen tank in California. View project

Ground Motion Analysis and Interpretation View project

All content following this page was uploaded by Praveen K. Malhotra on 25 August 2015.

The user has requested enhancement of the downloaded file.


Normalized Response Spectrum of Ground Motion

Praveen K. MALHOTRA
StrongMotions Inc.
Praveen.Malhotra@StrongMotions.com
www.StrongMotions.com

Praveen K. Malhotra, Ph.D., P.E. is a Principal at StrongMotions Inc. in the Boston Area. He has more
than 25 years of experience in practice and research of structural and geotechnical earthquake engineering,
including teaching multiple short-courses throughout the United States and abroad. He specializes in
transparent assessment and cost-effective mitigation of risk. He provides consulting services related to
hazard analysis, risk analysis and performance-based engineering. He is often consulted to peer-review
major projects.

Summary Keywords: strong-motion, response spectrum,


The peak ground acceleration PGA, peak ground normalized response spectrum, ground motion
velocity PGV and peak ground displacement PGD are prediction equations, damping. 
the fundamental strong-motion parameters (intensity
measures). The response spectrum of ground motion Introduction
relative to PGA, PGV and PGD is known as the The response spectrum provides valuable information
normalized response spectrum (NRS). In the past, regarding the ground motion. It allows an engineer
the number of records has not been sufficient to to estimate forces and deformations in structures
conclusively establish the shape of the NRS and to due to ground shaking. The spectral values at short-,
study its sensitivity to various factors. In this study, the intermediate- and long-periods correlate well with
NRS is derived from 13,192 strong-motion records. It PGA, PGV and PGD, respectively. Numerous
is shown that the shape of the NRS is sensitive only to studies1-6 have shown that a smooth response
the normalized velocity PGVn = PGV/(PGA•PGD)1/2. spectrum of ground motion can be constructed from
For the same PGVn, earthquake magnitude, distance PGA, PGV and PGD. The response spectrum relative
and local soil conditions have insignificant effect on to PGA, PGV and PGD is known as the normalized
the NRS. For the same PGVn, the direction of motion response spectrum (NRS). The attractiveness of
(horizontal or vertical) has insignificant effect on the the NRS is that it reduces the number of intensity
NRS. The shape of the NRS is rooted in structural measures to just three: PGA, PGV and PGD. In
dynamics; thus it should be preserved in predicting the past, the ground motion records have not been
ground motions for future earthquakes. It is found sufficient to conclusively establish the shape of the
that the latest empirical ground motion prediction NRS and to study its sensitivity to various factors. In
equations do not always preserve the shape of the this study 13,192 strong-motion records are used to
NRS. Therefore, it is recommended that ground establish the NRS and to examine its sensitivity to
motion prediction models should only be developed earthquake magnitude, distance, local soil conditions
for PGA, PGV and PGD and that the response spectra and the direction of motion. This study also examines
for various damping ratios should be generated from the shape of the response spectra generated by latest
PGA, PGV and PGD by using the NRS. ground motion prediction equations (GMPE).

The Bridge and Structural Engineer Volume 45 Number 1 March 2015  1


Strong-Motion Data 18.5 cm/s. Figure 3 shows the processed displacement
histories in two horizontal directions. The PGD in
The strong-motion data from 599 earthquakes in
two horizontal directions are indicated in Figure 3.
seismically active regions around are world7 are used
The rotated-median PGD is 1.39 cm.
in this study. The same data have been used to develop
GMPE for shallow crustal earthquakes in seismically
active regions.8-12 The data consist of 21,335 mostly
tri-axial ground motions. The earthquake magnitude
ranges from M 3 to M 7.9, closest distance R ranges
from 50 m to 1,533 km and the average shear-wave
velocity in top 30 m of the site (VS30) ranges from 94
m/s to 2,100 m/s.7 The response spectra of horizontal Fig. 1: Processed acceleration histories at strong-motion
and vertical ground motions were read from ‘flat- station CSMIP # 68206 during the 2014 M 6 South Napa
files’ released by the Pacific Earthquake Engineering Earthquake in California14
Research (PEER) Center.7 The response spectra in Figure 4 shows a tripartite plot of the 5% damping
PEER flat-files were for natural periods between 0 and rotated-median response spectrum. In this plot, the
20 s. The histories of ground motion in three directions peak pseudo-velocity PPV is shown along the vertical
were not available for this study. Nearly 40% of the axis; the peak pseudo-acceleration PPA is shown
response spectra in PEER flat-files could not be used along the -450 (counter-clockwise) axis; the peak
because they did not show the correct asymptotic deformation PD is shown along the 450 (clockwise)
behavior in the long-period (low-frequency) region. axis; and the natural period T is shown along the
This will become clear later in this paper. horizontal axis. PPA, PPV, PD and period T are
related to each other as follows:5,15
Rotated-Median Response Spectrum
PPA × T/(2π) = PPV = PD×2π/T (1)
The horizontal response spectra in PEER flat-files7
are rotated-median. The definition of rotated-median Note in Figure 4 that the PPA at short-periods
response spectrum is discussed here. approaches PGA and the PD at long-periods
approaches PGD. This response spectrum was
The horizontal ground motion at a site is typically generated by simultaneously using the processed
recorded in two orthogonal directions, say north and acceleration, velocity and displacement histories
east. The responses of a single-degree-of-freedom (Figures 1, 2 and 3).16 Therefore, it shows the correct
(SDOF) system can be computed in two orthogonal asymptotic behavior at both short- and long-periods.15
directions by solving the equation of motion. From
the responses in two orthogonal directions, the
responses in numerous other horizontal directions
can be calculated by using vector-geometry. The
median (50th percentile) of the calculated responses
in numerous horizontal directions gives the rotated-
median response spectrum; it is usually denoted as
RotD50.13 Fig. 2: Processed velocity histories at strong-motion
station CSMIP # 68206 during the 2014 M 6 South Napa
Figure 1 shows the processed acceleration histories in Earthquake in California14
two orthogonal horizontal directions at strong-motion
station CSMIP #68206 during the 2014 magnitude M
6 South Napa Earthquake in California.14 The PGA
in two horizontal directions are indicated in Figure 1.
The rotated-median PGA is 0.693 g, where g = 9.81
m/s2 = acceleration due to gravity. Figure 2 shows
the processed velocity histories in two horizontal Fig. 3: Processed displacement histories at strong-motion
directions. The PGV in two horizontal directions are station CSMIP # 68206 during the 2014 M 6 South Napa
indicated in Figure 2. The rotated-median PGV is Earthquake in California14

2  Volume 45 Number 1 March 2015 The Bridge and Structural Engineer


A normalized plot of the response spectrum is
obtained in two steps:6
1. The natural period along the horizontal axis is
divided by the central period Tc to obtain the
normalized period Tn = T/Tc. This causes the
PGA to become PGA•Tc and the PGD to become
PGD/Tc. The PGV is not affected by normalizing
the period.
2. The peak pseudo-velocity PPV along the vertical
axis is divided by (PGA•PGD)1/2 to obtain the
normalized peak pseudo-velocity PPVn = PPV/
(PGA•PGD)1/2. This causes the PGA to become
PGA•Tc/(PGA•PGD)1/2 = 2π, PGD to become
Fig. 4: Five percent of critical damping rotated-median
response spectrum of horizontal ground motion at strong-
PGD/Tc/(PGA•PGD)1/2 = 1/(2π) and PGV to
motion station
Normalized CSMIP
Response # 68206ofduring
Spectrum 2014 M 6 South
theMotion
Ground become PGVn = PGV/(PGA•PGD)1/2.
Normalized Response Spectrum of Ground Motion
Napa Earthquake
Figure 5 shows the normalized plot of the 5% damping
Normalized
Normalized Response Spectrum
Response Spectrum
Normalized Response Spectrum response spectrum shown in Figure 4. The normalized
From
, PGV and PGD, twoPGA, PGVparameters
additional and PGD,can twobeadditional parameters
determined,which plot emphasizes the shape of the response spectrum
provide further
can be determined,
o the ground motion. These are thewhich
centralprovide
period Tfurther insight relative to PGA, PGV and PGD. Next, the smooth
, PGV and PGD, two additional parameters can be determined,which provide further
into velocity
c, and the normalized
o the ground motion. Thesemotion.
the ground are the central
These period
are theTccentral
, and theperiod
normalized shape of the NRS will be generated by considering
Tc velocity
and the normalized velocity PGV : 6 numerous ground motion records.
n
T� � ���PGD/PGA (2)
T� � ���PGD/PGA (2) Smooth
(2) Normalized Response Spectrum
PGV� � PGV/√PGA · PGD (3) (3)5% damping response spectra in PEER flat-file7
The(3)
PGV� � PGV/√PGA · PGD
A ground motion is composed of many different were normalized with respect to the PGA, PGV and
motion is composed of many different frequencies (or periods). Tc can be thought of as
frequencies (or periods). Tc can be thought
motion is composed of many different frequencies (or periods). T c canof as
be PGD
thought of asvalues in the PEER flat-file. Figure 6 shows the
period of the ground motion.66PGVn is indicative of the band-width of frequencies
period of thethe central
ground period
motion. PGVofn isthe groundof motion.
indicative 6 plots of 21,335 NRS. Note that for many of these
PGVnofisfrequencies
the band-width
a ground motion.66A higher value of PGVn implies a narrower band-width.PGV n
indicative
a ground motion. of the
A higher band-width
value of PGVn of frequencies
implies present
a narrower in response
band-width.PGV n spectra, the peak deformation PD does not
alue of 1 when the ground motion 6consists of a single frequency of period Tc.
alue of 1 whenathe
ground
groundmotion. A higher
motion consists of avalue
single of PGVn of
frequency period aTc. approach PGD at long-periods. In other words, many
implies
narrower band-width. PGVn reaches a value of 1 when of the response spectra in PEER flat-file do not show
the ground motion consists of a single frequency of the correct asymptotic behavior at long-periods. This
period Tc.

Fig. 5: Normalized
Figure 5. Normalized plot of the plot of thespectrum
response responseshown
spectrum shown 4.
in Figure in Fig. 6: Normalized plots of 21,335 five-percent damping
Figure 5. Normalized plot of the response spectrum
Figure 4 shown in Figure 4. response spectra in PEER flat-file7
6
zed plot of the response spectrum is obtained in two steps:6
zed plot of the response spectrum is obtained in two steps:
tural period along the horizontal axis is divided by the central period Tc to obtain the
tural period along the horizontal axis is divided by the central period Tc to obtain the
ized period TThe Bridge
n = T/T c.Thisand Structural
causes the PGA Engineer
to becomePGA·Tc, and the PGD to Volume 45 Number 1 March 2015  3
ized period Tn = T/Tc.This causes the PGA to becomePGA·Tc, and the PGD to
e PGD/Tc. The PGV is not affected by normalizing the period.
e PGD/Tc. The PGV is not affected by normalizing the period.
trongMotions Inc. ●www.StrongMotions.com● (781) 363-3003 ● 4/6/15 ● Page 8 of 29
trongMotions Inc. ●www.StrongMotions.com● (781) 363-3003 ● 4/6/15 ● Page 8 of 29
is because the response spectra were generated from 13,192 ‘clean’ response spectra. These response
acceleration histories with low-frequency ‘noise’. spectra show the correct asymptotic behavior at both
Had they been generated from acceleration, velocity short - periods and long-periods.
and displacement histories simultaneously,14 they
The NRS in Figure 7 have the same normalized PGA
would have shown the correct asymptotic behavior at
= 2π and the same normalized PGD = 1/(2π), but they
both short- and long-periods.
have different normalized velocities PGVn, ranging
The response spectra which do not show the correct from 0.13 to 1.1. The NRS were sorted by their
asymptotic behavior at long-periods were removed normalized velocity PGVn and placed in different bins
from the data set so that the shape of the smooth with specified normalized velocities. Figure 8 shows
NRS is not corrupted by the long-period ‘noise’. The 380 NRS with median normalized velocity of PGVn =
undesirable response spectra were removed with the 0.4. The median (50th percentile) of these 380 NRS is
help of the 30% damping response spectra in PEER shown by thick red lines in Figure 9; this is the smooth
flat-file because 30% damping peak deformation 5% damping NRS for PGVn = 0.4. The uncertainty
PD approaches peak ground displacement PGD at a in normalized values is indicated by thin blue lines
much faster rate than the 5% damping PD. Response representing 16th and 84th percentile values. Figure 9
spectra for which 30% damping PD(20 s) > 1.05 PGD also shows maximum spectral values relative to PGA,
or < 0.95 PGD were removed from the data set. This PGV and PGD. The maximum pseudo-acceleration is
reduced the number of ‘clean’ usable response spectra 2.46 PGA; maximum pseudo-velocity is 1.82 PGV;
to 13,192. Figure 7 shows the normalized plots of and maximum deformation is 1.86 PGD. For PGVn
= 0.4, the maximum dynamic amplifications relative
to PGA, PGV and PGD are 2.46, 1.82 and 1.86,
respectively.

Fig. 7: Normalized plots of 13,192‘clean’ 5% of critical


damping response spectra in PEER flat-file

Fig. 9: Smooth 5% damping median, 16th and 84th percentile


NRS for PGVn = 0.4
Smooth NRS were also generated for PGVn = 0.6 and
PGVn = 0.8. They are shown in Figures 10 and 11,
respectively. For PGVn = 0.6, the maximum dynamic
amplifications relative to PGA, PGV and PGD
are 2.7, 2.26 and 2.1, respectively (Figure 10). For
PGVn = 0.8, the maximum dynamic amplifications
relative to PGA, PGV and PGD are 2.78, 2.67 and
2.39, respectively (Figure 11). With increase
in the normalized velocity PGVn, the dynamic
Fig. 8: Normalized plots of 380 5% damping response amplifications increase but the response spectrum
spectra with median PGVn = 0.4 becomes narrower. This is expected because higher

4  Volume 45 Number 1 March 2015 The Bridge and Structural Engineer


values of PGVn imply narrower band of frequencies
in ground motion leading to resonance-like response
of the single-degree-of-freedom system.

Fig. 12: Smooth 5% damping median NRS for PGVn = 0.4,


0.6 and 0.8

Fig. 10: Smooth 5% damping median, 16th and 84th Parametric Study
percentile NRS for PGVn = 0.6
The smooth NRS in Figure 12 were generated
from ground motions due to different magnitude
earthquakes, at different distances and at different
soil conditions. The effects of magnitude, distance
and local shear-wave velocity on the smooth NRS are
examined in this section.
To determine if the magnitude of the earthquake
has any significant effect on the shape of the NRS,
the data were sorted in terms of the earthquake
magnitude and split into two nearly equal parts.
The NRS were separately generated from each part.
They are shown in Figure 13. Note that the smooth
NRS are not significantly affected by the earthquake
magnitude. It is more likely for a bigger earthquake to
generate ground motions with wider frequency-band
Fig. 11: Smooth 5% damping median, 16th and 84th (or smaller normalized velocity PGVn), but for the
percentile NRS for PGVn = 0.8 same PGVn, the shape of the NRS is not significantly
Notice in Figures 9, 10 and 11 that the uncertainty affected by the earthquake magnitude. The effect of
is highest in the velocity-sensitive region of the magnitude on the NRS is further diminished by the
normalized spectrum. As PGVn increases, the fact that the site-specific response spectrum is not
velocity-sensitive region becomes smaller and the determined by a single magnitude earthquake but by
uncertainty reduces. Figure 12 compares the median earthquakes of many different magnitudes. Therefore,
NRS for three different values of PGVn = 0.4, 0.6 and it is considered appropriate to ignore the effect of
0.8. magnitude on the NRS.

The Bridge and Structural Engineer Volume 45 Number 1 March 2015  5


Fig. 13: Effect of magnitude M on the smooth NRS for 5% Fig. 15: Effect of VS30 on the smooth NRS for 5% damping
damping
Next, the data were sorted in terms of the distance R Next, the data were sorted in terms of the average
and split into two nearly equal parts. The NRS were shear-wave velocity in top 30 m VS30 and split into
generated from each part. They are shown in Figure two nearly equal parts. The normalized response
spectra were generated from each part. They are
14. Note that the smooth NRS are not significantly
shown in Figure 15. Note that the smooth NRS are
affected by the distance from the source of the
not significantly affected by the average shear-wave
earthquake. Although, distance from the source can
velocity VS30. Although, local soil conditions can
affect the frequency-band (or PGVn), for the same
affect the frequency-band (or PGVn), for the same
PGVn, distance has insignificant effect on the shape of
PGVn, local soils have insignificant effect on the
the NRS. The effect of distance on the NRS is further
shape of the NRS.
diminished by the fact that the site-specific response
spectrum is determined by earthquakes that could In this section the median NRS are shown to depend
occur at various distances from the site. Therefore, only on the normalized velocity PGVn; they do not
it is considered appropriate to ignore the effect of additionally depend on the earthquake magnitude,
distance on the NRS. distance and local shear-wave velocity. Due to space
limitation, the results for 16th and 84th percentile NRS
could not be shown in this section. But those NRS
also depend only on the normalized velocity PGVn;
they do not additionally depend on the earthquake
magnitude, distance and local shear-wave velocity.

Smooth NRS of Vertical Ground Motion


The smooth NRS were also generated for vertical
ground motions. In Figure 16, the NRS of horizontal
and vertical motions are compared with each other.
Note that there is no significant difference between
the shapes of NRS of horizontal and vertical motions.
This should not be misunderstood. Only the NRS of
horizontal and vertical motions are nearly identical.
Since PGA, PGV and PGD of vertical motion are
Fig. 14: Effect of distance R on the smooth NRS for 5% significantly different from those of horizontal motion,
damping the actual response spectrum of vertical motion is

6  Volume 45 Number 1 March 2015 The Bridge and Structural Engineer


generally quite different from that of horizontal Figure 20 shows a plot between damping and PPVn
motion. Response spectra of vertical motions tend to for three selected values of normalized period T/Tc
have wider frequency band (smaller PGVn), but for = 0.3, 1 and 3. The damping along the horizontal
the same PGVn, the NRS of vertical and horizontal axis is shown on a logarithmic scale and PPVn along
motions are almost identical. the vertical axis is shown on a linear scale. The
open circles in Figure 20 are the actual values of PPVn
for different values of damping. The straight-lines
in Figure 20 pass through the PPVn values for 0.5,
5 and 20% damping. Figure 20 shows that the PPVn
values for damping other than 0.5, 5 and 20% can be
obtained by assuming a piece-wise linear relationship
between log(ζ) and PPVn.

Fig. 16: Smooth NRS of horizontal and vertical ground


motions for 5% damping
Smooth NRS for Different Values of Damping
Smooth NRS were generated for different values of
damping. Figures 17, 18 and 19 show the effect of
damping on the smooth NRS of horizontal ground
motion for PGVn = 0.4, 0.6 and 0.8, respectively.
Table 1 lists the smooth NRS for PGVn = 0.4, 0.6 and
Fig. 18: Effect of damping on NRS of horizontal ground
0.8 and damping ζ = 0.5, 5 and 20% of critical. NRS motion for PGVn = 0.6
for any other value of PGVn between 0.4 and 0.8 can
be generated through interpolation. NRS for PGVn =
0.4 can also be used for PGVn< 0.4. NRS for PGVn =
0.8 can also be used for PGVn> 0.8.

Fig. 17: Effect of damping on NRS of horizontal ground Fig. 19: Effect of damping on NRS of horizontal ground
motion for PGVn = 0.4 motion for PGVn = 0.8

The Bridge and Structural Engineer Volume 45 Number 1 March 2015  7


Fig. 20: Normalized peak pseudo-velocity PPVn versus
damping for three different values of normalized period T/Tc
and normalized velocity PGVn = 0.6

Table 1. Normalized response spectra of horizontal motion for different damping ζ and normalized velocities.
Damping values are percentage of critical
PPVn = PPV/(PGD•PGA)1/2
Tn = T/Tc

PGVn = 0.4 PGVn = 0.6 PGVn = 0.8

0.5% ζ = 5% 20% 0.5% ζ = 5% 20% 0.5% ζ = 5% 20%

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
0.0147 0.0149 0.0149 0.0149 0.0149 0.0149 0.0149 0.0148 0.0148 0.0148
0.0215 0.0222 0.0222 0.0221 0.0218 0.0218 0.0218 0.0218 0.0218 0.0218
0.0316 0.0341 0.0332 0.0329 0.0323 0.0323 0.0322 0.0322 0.0322 0.0321
0.0464 0.0600 0.0523 0.0500 0.0478 0.0478 0.0476 0.0473 0.0473 0.0472
0.0681 0.134 0.0915 0.0792 0.0731 0.0718 0.0711 0.0703 0.0701 0.0697
0.1 0.331 0.183 0.130 0.131 0.115 0.110 0.111 0.106 0.105
0.147 0.642 0.329 0.208 0.324 0.217 0.180 0.219 0.175 0.162
0.215 1.026 0.529 0.310 0.811 0.454 0.302 0.550 0.338 0.267
0.316 1.323 0.688 0.385 1.56 0.828 0.48 1.25 0.68 0.45
0.464 1.357 0.727 0.401 2.30 1.20 0.65 2.29 1.22 0.72
0.681 1.242 0.673 0.392 2.55 1.35 0.74 3.52 1.87 1.02
1 1.148 0.632 0.365 2.45 1.31 0.73 3.96 2.12 1.13
1.47 1.068 0.599 0.347 2.09 1.17 0.67 2.90 1.63 0.95
2.15 0.952 0.562 0.333 1.60 0.940 0.55 1.78 1.07 0.68
3.16 0.794 0.509 0.313 1.05 0.661 0.41 0.94 0.62 0.44
4.64 0.567 0.395 0.259 0.573 0.391 0.280 0.462 0.337 0.272
6.81 0.351 0.264 0.187 0.282 0.221 0.178 0.228 0.190 0.170
10 0.189 0.158 0.121 0.137 0.125 0.114 0.119 0.113 0.109
14.7 0.0987 0.0918 0.0777 0.0778 0.0760 0.0732 0.0728 0.0721 0.0712
21.5 0.0573 0.0553 0.0504 0.0492 0.0490 0.0483 0.0478 0.0478 0.0476
31.6 0.0361 0.0354 0.0335 0.0327 0.0326 0.0325 0.0322 0.0322 0.0322

8  Volume 45 Number 1 March 2015 The Bridge and Structural Engineer


PPVn = PPV/(PGD•PGA)1/2
Tn = T/Tc
PGVn = 0.4 PGVn = 0.6 PGVn = 0.8

0.5% ζ = 5% 20% 0.5% ζ = 5% 20% 0.5% ζ = 5% 20%

46.4 0.0235 0.0233 0.0225 0.0220 0.0220 0.0220 0.0218 0.0218 0.0218
68.1 0.0156 0.0156 0.0152 0.0149 0.0149 0.0149 0.0148 0.0148 0.0148
100 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
Response Spectra of Predicted Ground The shape of the response spectrum in Figure 21 is
Motions not consistent with the shape of the NRS derived
The response spectra of predicted ground motions from recorded ground motions. Specifically, the
should be consistent with the normalized response amplification of pseudo-velocities relative to peak
spectrum (NRS) derived from recorded ground ground velocity is too low. Next, a 5% damping
motions. response spectrum is generated from PGA, PGV and
PGD by using the following steps:
Figure 21 shows a plot of the predicted 5% damping
median response spectrum of ground motion at a 1. The central period of the predicted ground
‘firm-rock’ site due to a magnitude M 6.8 earthquake motion is calculated from Equation 2 to be Tc =
at distance R = 10 km. This response spectrum is 2π•(0.0783/0.2025)1/2 = 1.125 s.
an average of the response spectra predicted by 5
GMPE.8-12 Figure 21 also shows the predicted median
values of PGA, PGV and PGD. PGA is average of 5
GMPE. PGV is average of 4 GMPE because one of
the equations20 does not predict PGV. PGD is inferred
from the spectral values at long periods since none
of the GMPE8-12 predicts PGD directly due to lack
of confidence in low-frequency content of recorded
ground motions. PGD was assumed equal to the
maximum spectral deformation divided by 2, on the
basis of Figures 9 and 10.

Fig. 22: Comparison between the response spectrum derived


from GMPE (Figure 21) and the smooth response spectrum
derived from PGA, PGV and PGD using the NRS
2. The normalized velocity of the ground motion is
calculated from Equation 3 to be PGVn = 2.443/
(0.2025×0.0783)1/2 = 0.46.
3. From Table 1, the normalized peak pseudo-
velocities PPVn are read corresponding to PGVn
= 0.4 and PGVn = 0.6. The values for PGVn = 0.46
are obtained by linear-interpolation between the
values for PGVn = 0.4 and 0.6. The normalized
Fig. 21: Five-percent damping predicted median response pseudo-velocities PPVn are multiplied by
spectrum of horizontal ground motion at a ‘firm-rock’ site (PGA•PGD)1/2 = 0.437 m/s to obtain pseudo-
due to a magnitude M 6.8 earthquake at 10 km distance velocities PPV.

The Bridge and Structural Engineer Volume 45 Number 1 March 2015  9


4. The normalized periods are read from the 1st 4. For the same PGVn, the damping adjustment
column of Table 1; they are multiplied by Tc = factors depend only on the normalized period.
1.125 s to obtain natural periods T. They do not depend on the earthquake magnitude,
distance and local soil conditions.
In Figure 22 the response spectrum derived from
the NRS (using PGA, PGV and PGD) is compared 5. The shape of the NRS is rooted in structural
with the response spectrum derived from GMPE. The dynamics; hence it should be preserved in
difference between the two response spectra is quite predicting ground motions for future earthquakes.
significant in the velocity- and displacement-sensitive The response spectra derived from the latest
regions. It is apparent that the response spectrum GMPE8-12 are inconsistent with the recorded
derived from the GMPE8-12 is not consistent with ground motions.
the recorded ground motions. Similar results were
6. It is strongly recommended that prediction
also obtained for several other earthquake scenarios
models should only be developed for PGA, PGV
defined by magnitude, distance and average shear-
and PGD. Response spectrum for any damping
wave velocity. There are at least two possible reasons
ratio should be generated from PGA, PGV and
for the incorrect shapes of predicted response spectra:
PGD by using the NRS.
 Many of the response spectra in the database, used
for developing empirical GMPE, do not show the References
correct asymptotic behavior in the long-period
1. Newmark, N. M. and W. J. Hall (1969). Seismic
range (Figure 6). This might have corrupted the
design criteria for nuclear reactor facilities,
shape of the predicted response spectrum.
in Proc. 4th World Conf. on Earthquake Eng.,
 The form of the empirical GMPE is unsuitable to Santiago,Chile, B-4, 37–50.
capture the physics of a SDOF system responding
2. Mohraz, B., W. J. Hall and N. M. Newmark
to seismic ground shaking.
(1972). A study of vertical and horizontal
In view of the unreasonable shape of the predicted earthquake spectra, AEC Report WASH-1255,
response spectrum, it is recommended that the Nathan M. Newmark Consulting Engineering
empirical GMPE should only be developed to predict Services, Urbana, Illinois.
PGA, PGV and PGD and the response spectrum
3. Hall, W. J., B. Mohraz and N. M. Newmark
should be generated from the predicted values of
(1975). Statistical studies of vertical and
PGA, PGV and PGD by using the NRS.
horizontal earthquake spectra, Nathan M.
Newmark Consulting Engineering Services,
Conclusions
Urbana, Illinois.
1. The peak values of ground acceleration, velocity
4. Mohraz, B. (1976). A study of earthquake
and displacement (PGA, PGV and PGD) are
response spectra for different geological
the fundamental ground motion parameters.
conditions, Bull. Seism. Soc. Am. 66(3), 915–
Response spectrum for any damping ratio can be
935.
generated from PGA, PGV and PGD by using the
normalized response spectrum (NRS) discussed 5. Newmark, N. M. and W. J. Hall (1982).
in this paper. Earthquake Spectra and Design, Earthquake
Engineering Research Institute, Oakland,
2. The shape of the NRS is sensitive only
California.
to the normalized velocity PGVn = PGV/
(PGA•PGD)1/2. For the same PGVn, earthquake 6. Malhotra, P. K. (2006). Smooth spectra of
magnitude, distance and local soil conditions horizontal and vertical ground motions. Bull.
have insignificant effect on the NRS. Seism. Soc. Am., 96(2), 506–518.
3. For the same PGVn, the direction of motion 7. Seyhan, E., Stewart, J. P., Ancheta, T. D.,
(horizontal or vertical) has insignificant effect on Darragh, R. B., and Graves, R. W. (2014). NGA-
the NRS. West2 site database, Earthquake Spectra, 30(3),

10  Volume 45 Number 1 March 2015 The Bridge and Structural Engineer
1007–1024, Earthquake Engineering Research 4th edition, Prentice-Hall International Series in
Institute. Civil Engineering and Engineering Mechanics.
8. Abrahamson, N. A., Silva, W. J. and Kamai, R. 16. Malhotra, P. K. (2001). Response spectrum
(2014). Summary of the ASK14 ground motion of incompatible acceleration, velocity and
relation for active crustal regions, Earthquake displacement histories. J. Earthquake Eng.
Spectra, 30(3), 1025–1055, Earthquake Struct. Dyn.,30(2),279-286.
Engineering Research Institute.
Nomenclature
9. Boore, D. M. Stewart, J. P., Seyhan, E., and
ζ = Viscous damping (percent of critical)
Atkinson, G. M. (2014). NGA-West2 equations
for predicting PGA, PGV and 5% damped PSA g = Acceleration due to gravity (9.81 m/s2)
for shallow crustal earthquakes, Earthquake GMPE = Ground motion prediction equation(s)
Spectra, 30(3), 1057–1085, Earthquake
Engineering Research Institute. M = Moment magnitude of earthquake

10. Campbell, K. W., and Bozorgnia, Y. (2014). NRS = Normalized response spectrum
NGA-West 2 ground motion model for the PD = Peak deformation
average horizontal components of PGA, PGV,
PEER = Pacific Earthquake Engineering
and 5% damped linear acceleration response
Research Center
spectra, Earthquake Spectra, 30(3), 1087–1115,
Earthquake Engineering Research Institute. PGA = Peak ground acceleration
11 Chiou, B. S.-J., and Youngs, R. R. (2014). PGD = Peak ground displacement
Update of the Chiou and Youngs NGA model
PGV = Peak ground velocity
for the average horizontal component of peak
ground motion and response spectra, Earthquake PGVn = PGV/(PGA•PGD)1/2 = normalized peak
Spectra, 30(3), 1117–1153, Earthquake ground velocity (Equation 3)
Engineering Research Institute. PPA = Peak pseudo-acceleration
12. Idriss, I. M. (2014). An NGA-West2 empirical PPV = Peak pseudo-velocity
model for estimating the horizontal spectral
values generated by shallow crustal earthquakes, PPVn = PPV/(PGA•PGD)1/2 = normalized peak
Earthquake Spectra, 30(3), 1155–1177, pseudo-velocity
Earthquake Engineering Research Institute. R = Distance from the source of earthquake
13. Boore, D. M., 2010. Orientation-independent, SDOF = Single-degree-of-freedom
non geometric-mean measures of seismic
T = Natural period of vibration
intensity from two horizontal components of
motion, Bull. Seismo. Soc. Am. 100, 1830–1835. Tc = Central period of ground motion
(Equation 2)
14. CESMD (2014). South Napa Earthquake of 24
August 2014, Center for Engineering Strong Tn = T/Tc = normalized period
Motion Data, http://strongmotioncenter.org/,
VS30 = Average shear-wave velocity in top 30 m
December 12, 2014.
(100 ft)
15. Chopra, A. K. (2011). Dynamics of structures,

The Bridge and Structural Engineer Volume 45 Number 1 March 2015  11

View publication stats

You might also like