Professional Documents
Culture Documents
Abstract Italian strong-motion data were used to study the attenuation of response
spectra and to simulate artificial accelerograms as a function of magnitude, distance,
and site geology. The database has already been utilized for the study of the atten
uation of peak ground acceleration (PGA) and velocity and consists of 95 accelero
grams from 17 earthquakes of magnitudes ranging from 4.6 to 6.8. Using multiple
regressions, we developed empirical predictive equations for the vertical and hori
zontal components ofresponse spectracorresponding to 14 frequencies ranging from
0.25 to 25 Hz. Predictive equations, aimed at the ground-motion simulation, were
also estimated for time-dependent frequency parameters, strong ground motion du
ration, and Arias intensity.
The shape of the predicted spectra is strongly dependent on magnitude and nearly
independent of distance. Alluvium sites show an amplification effect, with respect
to stiff sites, in different frequency ranges according to the thickness of the soil
deposit. The vertical/horizontal spectral ratio in far field varies, with magnitude and
frequency, from 0.35 to 0.85. The resulting response spectra are compared with the
predictions of sorne recent attenuation relationships and with those proposed by the
Eurocode EC8.
The simulation of nonstationary strong ground motions is achieved through an
empirical method where time and frequency features of the motion are represented
through the physical spectrum, extending the spectral moments theory to the nonsta
tionary case. The simulated time histories fit the recorded accelerograms in terms of
several ground-motion amplitude measures, such as peak acceleration, peak velocity,
Fourier spectra, and response spectra. The principal advantage of the proposed
method consists in correlating the simulation parameters with earthquake magnitude,
source distance, and soil conditions.
Introduction
In the field of seismic risk assessment, the prediction of fore, the engineering practice of using peak acceleration to
shaking intensity in future earthquakes plays a key role. Re scale constant spectral shapes is likely to lead to erroneous
liable estimates of the ground motions that structures must estimation of the frequency content of ground motion. Be
be designed to withstand are obtained mainly through em sides, sorne engineering methodologies require artificial ac
pirical predictive relationships developed after statistical celerograms to be used as input in nonlinear dynamic anal
analysis of data recorded during strong earthquakes. yses.
Many predictive equations are by now available in dif In recent years, many seismological and engineering
ferent regions of the world (Campbell, 1985, 1993) for the methods have been proposed for simulating ground motions.
prediction of peak ground acceleration (PGA); many fewer While both seismologists and engineers recognize the sto
are the relationships developed for the prediction of response chastic nature of high-frequency ground motions, they esti
spectral values (Idriss, 1991; Boore et al., 1993), and very mate the ground motion in fundamentally different ways.
few of these also take into consideration the vertical com The stochastic and deterministic models used in the seis
ponent of the ground motion (Campbell, 1990; Sadigh, mological approach (Boore, 1983; Boit, 1987; Atkinson and
1993). Somerville, 1994) rely on the estimation of the motion from
The ground-motion attenuation relations show that the the rupture process of a fauIt of specified dimensions and
shape of the response spectra depends strongly on earth orientation or from a point source; these models usually re
quake location, magnitude, and local site conditions. There- quire the knowledge of many parameters, characterizing the
337
©200S INIST CNRS Tous droits d 'été' t Il . II . '"
. e propn ln e ectue e réservés. ReproductIOn, représentatIOn et ddfuslon interdites. Loi du 01/07/92. Articles 5 ,6 el 7 des CGV
)+~;"V'JO INIST CNRS. Tous droits de propriété intellectuelle réservés. Reproduction, représentation ct difTusion intcrdites. Loi du 01/07/92. Articles 5,6 et 7 des CGV
Estimation of Response Spectra and Simulation of NOllstalionary Earthquake Ground Molions 339
8111I
--e- deop a1hNium
--&- shailoW a11l.NilJ11
10.0 10.0
1.0 1.0
Figure 2. Dependence of the predicted horizontal Figure 4. Dependence of the predicted horizontal
spectra on magnitude and comparison between spec spectra on site geology.
tra calculatcd with raw and smooth coefficients (see
Table 2).
correlated with h and was therefore constrained to the value
of - 1. This assumption, besides reducing the numbers of
PREDICTED HORIZONTAL RESPONSE SPECTRA 5% DAMPING coefficients to be estimated, agrees with the basic principles
M=6 Epîcentral distance Site=stiff
100.0 '---rI 1 i Iii 1 -- '-~-·TI, l "TIniITI- - - - - , - - - "- ' - ,TITITilrn~
of wave propagation in elastic media.
Tables 1 and 2 summarize the results obtained for the
coefficients in equation (4) in case of the larger of the two
horizontal components of the motion (N-S and W-E), re
spectively, for fault distance and epicentral distance. Tables
3 and 4 provide the same results in case of the vertical com
10.0
ponent of the motion. The coefficients previously obtained
(Sabetta and Pugliese, 1987) for the horizontal PGA and PGV
are reported in Tables 1 and 2, and the new vertical PGA and
[ R=O,,",
PGV coefficients are reported in Tables 3 and 4.
,o~
AlI of the coefficients were found to be statistically sig
R=10,,",
nificant at levels of confidence exceeding 95%, except for
the following: the coefficient el in the frequency range 0.25
to 1 Hz for the horizontal components and 0.25 to 2 Hz for
i~ ~
thc vertical components; the coefficient e2 in the frequency
PREDICTEO RESPONSE SPECTRA 5% DAMPING The values of the coefficients corresponding to the use
M=6 Epicentral distanœ=30 km Sile=sliff
1 i i IfTll of the epicentral distance, in comparison with the fauIt dis
tance, are higher for the magnitude coefficient and lower for
the constant tenn according to the fact that the difference
between the two distance definitions becomes remarkable
only for high magnitudes when the fauIt dimensions are rel
evant.
10.0
Figure 2 shows the dependence on magnitude of the
predicted horizontal spectra: higher magnitudes are associ
ated to a proportionally greater spectral content at low fre
quencies according to the seismological relationship be
tween corner frequency and seismic moment (Brune, 1970).
1.0 The dependence on distance of the predicted horizontal
spectra is given in Figure 3, except for distances less than
10 km, where the influence of the parameter h is greater, the
spectra attenuate with distance maintaining the same shape
consistent with the findings of other authors (Console and
0.1
0.1
_-----...L.i 1 11111
1.0
1 1 1 l,Ill
10.0
1 1 l " 1 ~I
100.0
Rovelli, 1981; Joyner and Boore, 1982; Boore et al., 1993).
Other studiies (Idriss, 1991; Sadigh, 1993), show a distance
Frequency (Hz)
dependent spectral shape, significant for distances greater
Figure 5. Comparison between horizontal and than 100 km. In our case, most of the data (Fig. 1) were
vertical response spectra. recorded at distances less than 100 km, and the distance
independence is consistent with the lack of statistical signif
icance that we obtained for the coefficient taking into ac
1.1 ~>T!ll!l l-----rTT1111 1 ---,--r count the frequency-dependent anelastic attenuation (Pug
liese and Sabetta, 1989). As a matter of fact, the anelastic
1.0 ------- attenuation acts mainly in the surface soil layers and there
,, fore depends more on the local site conditions than on the
1
0.9 1 source distance.
1
1
The effect of the site geology on the spectral shape is
0
ê 0.6 1 depicted in Figure 4. The amplification effects of shallow
1 1.1=7
1§ 0.7
RG1.60
- .• ---
____ 1
1
and deep alluvium sites with respect to stiff sites are of the
same order of magnitude (40 to 60%) but act in different
frequency ranges (0.25 to 2 Hz for deep alluvium and 1 to
M:6 25 Hz for shallow alluvium), illustrating that alluvium sites
~
0.6
should not be classified together in a single category ne
ECB glecting the depth of the soil deposit.
0.5 M=5
The spectra for the larger of the two horizontal com
ponents are compared in Figure 5 with the spectra obtained
0.4 -- -~..:'-..:,
for the vertical component of the motion. In the same figure,
'.'
~ the curves corresponding to 50 and 84 percentile are re
0.3 1 1 1 l ,,1 1 1 1 (1 rll 1 [~
0.1 1.0 100.0
ported. The vertical spectra show lower values with respect
10.0
Frequency (Hz) to the horizontal ones: the difference is of the same order of
magnitude as the standard deviation and decreases with in
Figure 6. Spectral ratio between the vertical and creasing frequency because of the larger high-frequency
the horizontal component of the ground motion (R =
content associated with the vertical component.
30 km, stiff site), as a function offrequency and mag
nitude, compared with the ratios proposed by euro Figure 6 shows, for an epicentral distance of 30 km and
code ECS and R.G. 1.60. a stiff site, the ratio of the verticallhorizontal spectral values
with varying frequency and magnitude. For law-magnitude
values (M = 5), the ratio is fairly constant with frequency
The analysis of the residuals resulting from the regres and fluctuates around the value of 0.4. For high magnitudes
sion did not show systematic variations as a function of the (M = 7), the ratio varies from a minimum of 0.35 at low
variables used in the model, and in particular, a correlation frequencies to a maximum of 0.85 at high frequencies. The
analysis confirmed that the residuals were uncorrelated with comparison, given in the same figure, with the verticallhor
respect to magnitude, distance, and predicted spectral values izontal ratios normally adopted in the engineering practice
at greater than 99% level of confidence. as suggested by the Regulatary Guide 1.60 (U.S. Atomic
©200S INIST CNRS Tous droits de propriété' 1" t " é é R .
. In C ee uc e r SCry s. eproduellon, représentation et diffusion interdites. Loi du 01/07/92. Articles 5 ,6 et 7 des CGV
10.0 10.0
M=6
M=6
M=5
PREDICTED VERTICAL. RESPONSE SPECTRA 5% DAMPING horizontal ratio ranges from 0.4 (M = 5) to 0.6 (M = 7)
M=6 Fautt distance
for PGA and from 0.5 (M = 5) to 0.7 (M = 7) for PGV.
Such values, particularly in case of high magnitudes, are in
basic agreement with the suggestions of the literature (New
mark and Hall, 1978) and regulations (Commission of the
European Communities, 1993).
Figure 7 compares our results with the horizontal spec
tra predicted by sorne recent attenuation relationships de
;
.. --- .. veloped with up-to-date databases of recordings of western
. - North American earthquakes.
1.0
1 ..••• In Figure 7a, the comparison with the study of Boore et
R= 15 km al. (1993) is made for different magnitudes, a fault distance
of 30 km, and the site category shallow alluvium, corre
1 \
sponding approximately to the site class "B" used by BJF
R=30km (shear-wave velocity averaged over the upper 30 m between
I-~'stfll 360 and 750 mfsec). At internnediate and high frequencies,
~993,roc:I< the results are similar, with higher values predicted by our
model for M = 5 at high frequencies (2 to 10 Hz) and for
0.1 L ! ! ! 11111
M = 7 at intermediate frequencies (1 to 3 Hz). At low fre
0.1 1.0 10.0
Frequency (Hz) quencies (kss than 1 Hz), our relationship gives significantly
lower estimates of spectral velocity than those predicted by
Figure 8. Comparison of the median vertical spec BJF.
tra obtailled by Sadigh (1993) with those from this
study, for a stiff site, M = 6, and different values of In Figure 7b, the comparison with the relationship by
the fault distance, Sadigh (1993) is made using the multiplying coefficient of
1.09, suggested by the author to take into account the am
plification effect of oblique faults with respect to strike-slip
Energy Commission, 1973) or by the Eurocode EC8 (Com faults. The predicted spectral values, for M = 5, 6, and 7
mission of the European Communities, 1993) shows that our and R = 30 km, are in good agreement, except for frequen
ratios are comparable to those proposed by EC8 but much cies lower than 0.5 Hz, where, as previously discussed, our
s
lower than the RG-1.60 ratios. results could be underestimated by the high-pass filtering.
As far as the peak values are concerned, the vertical! The vertical spectra predicted by our relationship, com
,)'m l:r>2I)08 INIST CNRS. Tous droits de propriété intellectuelle réservés. Reproduction, représentation et dilTusion interdites. Loi du 01/07/92. Articles 5 ,6 et 7 des CGV
Estimation of Response Spectra and Simulation of Nonstationury Eurthquake Ground Motions 343
C_--J
HORIZONTAl.. RESPONSE SPECTRA 5% CAMPING HORIZONTAl.. RESPONSE SPECTRA 5% DAMPING
M=5 Epiœntral dstanœ=15 km PGA=O.œ (9) M=7 Epicentral disfanœ=:15 km PGA=O.31 (9)
0.20 ------.-~-----,-~--,- 1.0 -,-- 1 1 -----,-- [
a) b)
;~::::-
- - EC8."'_A - - EC8, _A
-- . EC8. adlsci B EC8 _B
0.16 0.8 •••••••• - - This Iludy. stlIl
\ 1 •••••••L -_ _lll_Is_Iludy
__ •deep
__ IIIhN_.--'
\
\
..........
0.12 \
,, 0.6
".
, §
i1j
"......'\
...
...
... ...
Q.
.....•....
0.08 0.4
. ~ .
O'04 0.2
- .:~
0.00
0.0
l --,--_~I__ ~~~_-lI
0.5
_ _.L.-_ _--'
1.0
Period (s)
1.5 2.0
0.0
0.0
--l-------l_ _-'----_--'-_---''--_-'---_--'-_.---J
0.5 1.0
Period (s)
1.5 2.0
Figure 9. Comparison of the horizontal spectra predicted by our model and the
spectral shapes proposed by eurocode EC8 scaled at zero period to the peak acceler
ation corresponding to (a) magnitude = S, distance = 15 km; (b) magnitude = 7,
distance = 15 km.
------------,
PS(t, f) I w(t - u)x(u)e- i21ifu du (5)
T1 T2 T3
220 r - - - - - I > - - , - - "- - - - - - - , - ,- - - - - - ,
200 'é---R/7 ----»:
1~ 1.5 DV ----», where t and f represent time and frequency, u is a dummy
lBO " ,
;;:- 160
"
"
1
,
variable, and w(t - u) is a rurming time window function
: 140 1 l , whose characteristics have been described in a previous ar
~ 120 :. 1 ~
ticle (Sabetta et al., 1986). Figure 10 shows an example of
'E 100 :
~ 80 ; PS calculated, with a Gaussian moving window of 2.S-sec
0.. 60 , width, for an accelerogram recorded in 1984 at the "Garig
40 :
20 1
liano" station.
o 1
PS is constituted by a series of power spectral densities
o 5 10 15 2~ 25 30 (PSD' s), calculated at different times, which can be fitted
,
1
o
- -- _._~----
5
--------------
10 15 20 25 30
À;(t) = f lpS(t,f) df i = 0, 1,2; (6)
Time (s)
Pa(t) ,.1.0 (t)
Fe(t) À1(t)/Ào(t) (7)
Figure 12. Top: time envelope function used for Fb(t) [)02(t)/À O(t) - Fe2(t)] 112.
the simulation in case of M = 6 and R = 50 km;
middle: corresponding simu1ated acce1erogram; bot
tom: time behavior of the central frequency. Pa(t), instantaneous average power, is the time envelope -fi
(,:((\YL'UVO INIST CNRS. Tous droits de propriété intellectuelle réservés. Reproduction, représentation ct dilTusion interdites. Loi du 01/07/92. Articles 5,6 et 7 des CGV
EstinUltion of ReJponse Spectra and Simulation of Nonstationary Earthquake Ground Motions 345
40 120 250
30 #1 PGA=40.0 cm/52 100 #1 PGA=88.7 cm/52 200 #1 PGA=212.8 cm/52
80 150
20 60
40 100
~ 10
20 50
Ê 0 0 0
"
;( -10 -20 -50
·40 ·100
-20 ·60
·150
-30
·40
-80
-100
-120
·200
·250
l
0 5 10 15 20 25 0 10 15 20 25 0 10 15 20 25
40 120 250
#2 PGA=40.9 cm/52 100 #2 PGA=123.8 cm/52 #2 PGA=237.0 cm/52
30 200
80 150
20 60
40 100
N 10 50
.!!! 20
E 0 0 0
" -10
:< ·20 -50
-40 -100
-20 -60
-80 -150
-30 -200
·100
·40 ·120 -250
0 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
40 120 250
#3 PGA=37.0 cm/52 100 #3 PGA=120.2 cm/52 200 #3 PGA=201.3 cm/52
30
80 150
20 60
40 100
~ 10
20 50
È 0 0 0
:<" ·10
·20
-40
·50
·100
-20 ·60
-150
-30 ·80
-100 -200
-40 -120 -250
0 5 10 1(5) 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
1
l(s) 1(5)
~
function describing the amplitude variation of the ground ln P(t) = ln Fe(t) - 15 2/2 (10)
motion. Its integral in the time domain is equal to the integral
of PS in the time-frequency plane and corresponds to the 15 = jln [1 + Fb 2 (t)!Fe 2 (t)]. (11)
Arias Intensity lA:
As an example, Figure Il depicts the fit of the PSD, corre
lA = f x(tf dt = ff PS(t, f) dt dl = f PaCt) dt. (8)
sponding in Figure lOto a time of 8 sec, with the lognormal
function of equation (9) calculated at the same value of t.
Values of Pa, Fe, and Fb are calculated from the time history
Fe(t), central frequency, and Fb(t), frequency band recorded at Garigliano through equations (5), (6), and (7).
width, represent the nonstationarity of the frequency content
and correspond, respectively, to the centroid of PS and to Estimation of the Time-Dependent Functions
the radius of gyration of PS with respect to Fe on the fre Defining the Physical Spectrum
quency plane.
The functions PaCt), Fe(t), and Fb(t), used for the sim
With the above-defined parameters, il is possible to de ulation of the physical spectrum, were derived from regres
rive a lognormal function approximating the PS: sion analyses of the 190 horizontal components of the above
described database.
PS (t f) = PaCt) e - [ln /-ln p(t)f/20
2
(9) The nonstationary analysis of the Italian strong-motion
approx ' 1ji;:15 recordings (Sabetta et al., 1986) points out that, in the time
domain, PaCt) shows a lognormal shape, Fe(t) shows an ex
where P(t) and 15 are derived from Fe(t) and Fb(t) in the ponential decay, and the ratio FblFe is time independent.
following way: Regression analyses of the last two parameters as a function
©2008 INIST CNRS - Tous droits de propriété intellectuelle réservés. Reproduction, représentation ct diffusion interdites. Loi du 01/07/92. Articles 5 ,6 ct 7 des COY
Tl = RI7 T2 = Tl + 0.5·DV
T3 = Tl + 2.5 - DV Tot. durat. = 1.3. T3, (16)
where Tl corresponds to the time delay in seconds between the different coefficients and for the standard deviation are
Sand P waves and is calculated by dividing the epicentral reported in Table 5.
distance R in kilometers by the factor Vp *\I,I(Vp - Vs) as Both energy and duration are positively correlated with
sumed to be equal to 7 km/sec. lA and DV (respectively Arias magnitude, whereas duration increases with distance, ac
intensity in cm 2/sec 3 and duration in seconds of the ground cording to seismic waves dispersion, and energy attenuates
motion strong phase) have already been defined. The choices with distance nearly as l1R2 , according to geometrical
of T2, T3, and (J were derived from several tests with real spreading on a spherical surface. Shallow soils amplify the
accelerograms in order to have a time envelope function energy and, acting as a high-pass filter, produce a reduction
Pa(t) with the following characteristics: in the significant duration. Deep soils cause an increase in
both the energy and the duration. lA and DV completely de
a modal value, at time t = n, correlated to the epicentral fine the envelope function Pa(t) characterizing the amplitude
distance; nonstationarity of the ground motion.
a standard deviation proportional to the strong-motion du Figure 12 illustrates, for M = 6 and R = 50 km, the
ration DV; function PaU) derived from equation (14), the corresponding
an area equal to the Arias intensity lA; and simulated time history, and the trend of the central frequency
a total duration 30% greater than the value of T3, corre derived from equation (12) and shifted in time so that the
sponding to the modal value plus 2DV (see Fig. ) 2). exponential decay of Fe begins when Pa(t) starts to rise.
Regression analyses were then carried out to obtain em
Time Histories Simulation
pirical prediction equations for the significant duration DV
and the total energy lA of a given strong-ground motion. After the evaluation of Pa(t 1 M, R, S), Fe(t 1 M, S), and
Equation (4) was fit to the data and the results obtained for FblFe (M, S), it is possible, for a given magnitude M, dis
< ©200S IN IST CNRS. Tous droits de propriété intellectuelle réservés. Reproduction, rcpréscntatiorn et dilTusion interdites. Loi du 01/07/92. Articles 5 ,6 ct 7 des COV
Estil1Ultion of Response Spectra and Simulation of NonstaIionary Earthquake Ground Motions 347
.. 50 50 50
.!!!
E
u
0 o o
<" ·50 ·50 -50
o 5 10 15 20 o 5 10 15 20 o 5 10 15 20
10 10 10
8 PGV=9.28 cm/s PGV=6.81 cm/5 PGV=7.47 cmls
8 8
5 6 5
4 4 4
~u ~
2 2
o o
:> ·2 ·2 -2
·4 ·4 -4
-5 -5 -5
·8 ·8 -8
.10 L....~~~_~~~~~~~~_ ·10 '--~~~~-~~~~-~- -10 .......~~~~~-~~~~~~
o 5 10 15 20 o 5 10 15 20 o 5 10 15 20
2 2 2
1.5 PGD=O.96 cm PGD=1.12cm PG0=2.02em
1.5 1.5
1
0.5 0.5 0.5
Ê
~ 0 o o
o
-0.5 -0.5 -0.5
-1 -1 -1
-1.5 ·1.5 -1.5
-2 L..~~~~_~~.........~~~~ ·2 ......~~~~~~~~~~~~ -2 .......~~~ .......~~~~~~~~
o 5 10 I(e) 15 20 o 5 10 tes) 15 20 o 5 10
tes)
15 20
. -L --'--'- -------'
Figure 15. Simulated time histories of acceleration (top), velocity (middle), and
displacement (bottom) for M = 6, R = 20 km, and three different site conditions.
,-----~----------------
tance R, and site condition S, to calculate an approximate PS Epie. Distance =20 Km Site =stlff
from equations (9), (l0), and (lI). The simulation of artifi 100.00
1.00
~~'_L~~.~
CIl(t) = j2nfoPS(j", t), (18)
,----------------------------------,
51MULATED 1 vs, PREDICTED RESPONSE SPEClRA (5% DAMPING) quency content of the signal. The PGA of the simulated sig
100 --,----------,-~-------- naIs, reported on each time history, is close to the value
- sîmulated
_ _ predlcled predicted by the attenuation law, reported on the top of the
_____ 51. dey.
figure, and converge to the expected value when a mean of
10 several simulations is performed.
The attenuation of the ground motion with epicenfral
distance is represented in Figure 14. As expected, larger dis
tances correspond to smaller accelerations and higher dura
tions.
Figure 15 displays the effect of the different site con
l 0,1
ditions on the simulated time histories of acceleration, ve
0,1 10 100 locity, and displacement. Going from a stiff site to a shallow
Frequency (Hz) alluvium site causes a slight decrease in the duration of the
------
time histories and an increase in the high-frequency content
Figure 17. Response spectra of four simulated corresponding to a greater value of PGA. On the contrary, a
accelerograms compared with the spectrum predicted deep alluvium site with respect to a stiff site shows higher
by the attenuation relationship in case of M = 6, duration, greater content of low frequencies, and higher peak
R = 20 km, for a stiff site. displacement.
Figure 16 shows a comparison between the Fourier
spectra obtained from our model and the spectra derived
from the seismo!ogical omega-squared mode! with a high
Results frequency cutoff and a constant stress parameter (Boore,
Figure 13 shows an example of simulated accelero 1983; Rovelli et al., 1988). The values of the parameters of
grams for a stiff site at an epicentral distance of 20 km and the omega-squared mode! <p = 3.2 km/sec, p = 2.7 g/cm 3 ,
for different values of magnitude. The increase of magnitude L1() = 100 bars, Q = 300, lm = 15 Hz) are the same as
causes an increase of amplitude, duration, and low-fre those used by Boore (1983). The similarity of the results
I l
00 .--_E_P_1C_.D
__I_st_B_nc,e_"_20_K_m
_ _S_It,e_=_s_tl_" , 1 NPlc.Dlstance= 20 Km
100
Slte~ shall. alluv.
1 10 ~::""'-----c""""+----"'l:l.c-"\1;.------------I
I~
1
~
E
o 0
M=7
:>
en :>
~
1
0.. 1 1 1 ~-~'r---+---------t~~M~6
M~5
0,1
~
0.1
0,1
____1_F_req_Ue_nCy(HZ_)_1_0
L 0,1 1 Frequency(Hz) 1 0
100
-----------------,
Eplc.Dlstance" 90 Km Slte= stl" Eplc.Dlstance" 20 Km Site: deep. alluv.
_.,-, ;
Prad. Spectrum 1
100 . - - - - - - " T " - - - - - - - - , - - -
~ 10 h-~-_="~~~.......~~---
'"
E
o
:>
1
~ M=7
M=6
M~5
1 0,1 0.1 1
L
1
1 10 100 0,1 1 10 100
0,1 Frequency(Hz) Frequency(Hz)
Figure 18. Comparison of the mean response spectra derived from six simulated
accelerograms, with the spectra predicted by the attenuation relationship for different
magnitudes, distances. and site conditions.
ST CNRS. Tous droits de propriété intellectuelle réservés. Reproduction, représentation et dilTusion interdites. Loi du 01/07/92. Articles 5 ,6 et 7 des CGV
ation of Response Spectra and Simulation of Nonstationary Earthquake Ground Motions 349
Table 1
Regression coefficients of equation (4) for psv, PGA, and PGv, for the larger horizontal component and fault distance.
Freq. Const. Term Mag. Coeff. Dist. Coeff. Site Coeff. Sile Coeff.
(Hz) b e, e, cr
5% damped psv (cm/sec) 0.25 raw -2.320 0.653 -1 O. 0.134 2.3 0.275
smooth - 2.400 0.685 -1 O. 0.130 2.1 0.297
0.33 raw -2.017 0.637 -1 O. 0.149 2.3 0.317
smooth -2.170 0.675 -1 O. 0.151 2.5 0.297
0.50 raw -\.935 0.667 -1 O. 0.194 3.1 0.300
smooth -\.800 0.650 -1 O. 0.184 3.1 0.297
0.67 raw -\.498 0.614 -1 O. 0.200 4.0 0.242
smooth - \.510 0.620 -1 0.010 0.210 3.5 0.295
\.00 raw - 1.112 0.560 -1 O. 0.242 4.2 0.246
smooth -1.120 0.570 -1 0.050 0.242 4.0 0.290
1.33 raw -0.753 0.503 -1 0.120 0.232 4.9 0.261
smooth -0.850 0.530 -1 0.120 0.232 4.4 0.283
2.00 raw -0.496 0.458 -1 0.240 0.156 4.6 0.279
smooth -0.400 0.455 -1 0.220 0.156 5.0 0.271
2.50 raw -0.077 0.397 -1 0.196 O. 4.9 0.268
smooth -0.077 0.400 -1 0.210 0.115 5.3 0.264
3.33 raw 0.467 0.301 -1 0.122 O. 6.1 0.245
smooth 0.400 0.315 -1 0.165 0.065 5.7 0.245
5.00 raw 0.489 0.283 -1 0.103 O. 5.7 0.215
smooth 0.550 0.273 -1 0.130 O. 6.3 0.215
6.67 raw 0.499 0.255 -1 0.135 O. 6.5 0.207
smooth 0.500 0.255 -1 0.130 O. 6.7 0.200
10.00 raw 0.290 0.244 -1 0.168 O. 7.3 0.200
smooth 0.290 0.245 -1 0.130 O. 7.3 0.190
15.00 raw -0.033 0.249 -1 0.112 O. 7.2 0.181
smooth -0.035 0.255 -1 0.130 O. 7.2 0.182
25.00 raw -0.505 0.275 -1 0.147 O. 5.8 0.182
smooth -0.505 0.273 -1 0.130 O. 5.8 0.182
PGA (g) -\.562 0.306 -1 0.169 O. 5.8 0.173
PGV (cm/sec) -0.710 0.455 -1 0.133 0.133 3.6 0.215
Table 2
Regression coefficients of equation (4) for PSV, PGA, and PGV, for the larger horizontal component and epicentral distance.
Freq. Const. Term Mag. Coeff. Dist. Coeff. Site Coeff. Site Coeff.
(Hz) b e, e, cr
5% damped PSV (cm/sec) 0.25 raw -2.409 0.684 -1 O. 0.1\ 1 2.7 0.303
smooth -2.500 0.725 -1 O. 0.100 2.6 0.319
0.33 raw -2.146 0.675 -1 O. 0.109 2.7 0.344
smooth -2.250 0.715 -1 O. 0.108 3.0 0.319
0.50 raw -2.082 0.706 -1 O. 0.157 3.4 0.320
smooth -1.900 0.687 -1 O. 0.150 3.6 0.319
0.67 raw -1.647 0.654 -1 O. 0.165 4.4 0.270
smooth -1.647 0.660 -1 0.010 0.175 4.0 0.315
1.00 raw -1.297 0.605 -1 O. 0.209 4.2 0.271
smooth -1.280 0.612 -1 0.050 0.208 4.4 0.308
1.33 raw -1.006 0.557 -1 0.120 0.208 4.4 0.278
smooth -1.000 0.570 -1 0.120 0.190 4.7 0.303
2.00 raw -0.650 0.498 -1 0.249 0.124 5.1 0.301
smooth -0.595 0.500 -1 0.230 0.124 5.0 0.290
2.50 raw -0.281 0.442 -1 0.219 O. 4.8 0.285
smooth -0.281 0.445 -1 0.222 0.078 5.2 0.280
3.33 raw 0.202 0.354 -1 0.147 O. 5.5 0.260
smooth 0.100 0.377 -1 0.185 0.020 5.4 0.260
5.00 raw 0.296 0.326 -1 0.126 O. 5.8 0.238
smooth 0.296 0.323 -1 0.161 O. 5.7 0.234
6.67 raw 0.222 0.311 -1 0.161 O. 5.8 0.220
smooth 0.222 0.310 -1 0.161 O. 5.9 0.220
10.00 raw -0.019 0.304 -1 0.194 O. 6.3 0.214
smooth -0.019 0.304 -1 0.161 O. 6.2 0.208
15.00 raw -0.312 0.304 -1 0.137 O. 6.5 0.194
smooth -0.312 0.304 -1 0.161 O. 6.3 0.200
25.00 raw -0.817 0.336 -1 0.174 O. 4.7 0.195
smooth -0.817 0.330 -1 0.161 O. 4.7 0.195
PGA (g) -1.845 0.363 -1 0.195 O. 5.0 0.190
PGV (cm/sec) -0.828 0.489 -1 0.116 0.116 3.9 0.249
Coefficients are to be used for 4.6 ;;:; M ;;:; 6.8 and R ;;:; 100 km.
Table 3
Regression coefficients of equation (4) for PSV, PGA, and PGV, for the vertical component and fault distàùce.
Freq. Const. Term Mag. Coelf. Disl. Coeff. Sile Coelf. Sile Coelf.
(Hz) b e, e,
"
5% damped PSV (cm/sec) 0.25 raw -2.395 0.621 -1 O. O. 2.8 0.300
smooth -2.395 0.629 -1 O. O. 2.4 0.300
0.33 raw -2.162 0.605 -1 O. O. 2.9 0.328
smooth -2.292 0.636 -1 O. O. 2.4 0.328
0.50 raw -2.154 0.640 -1 O. O. 2.6 0.351
smooth -2.154 0.639 -1 O. O. 2.5 0.351
0.67 raw -1.848 0.607 -1 O. O. 2.4 0.345
smooth -1.900 0.615 -1 O. O. 2.9 0.345
1.00 raw -1.547 0.574 -1 O. O. 3.9 0.305
smooth -1.547 0.574 -1 0.Q10 O. 3.9 0.305
1.33 raw -1.155 0.512 -1 O. O. 5.2 0.280
smooth -1.155 0.515 -1 0.018 O. 4.9 0.280
2.00 raw -0.512 0.402 -1 O. O. 5.1 0.254
smooth -0.712 0.441 -1 0.040 O. 5.1 0.254
2.50 raw -0.523 0.397 -1 0.076 O. 4.9 0.237
smooth -0.480 0.399 -1 0.080 O. 5.1 0.240
3.33 raw -0.482 0.384 -1 0.174 O. 4.7 0.252
smooth -0.350 0.370 -1 0.137 O. 5.1 0.232
5.00 raw -0.135 0.314 -1 0.171 . O. 5.1 0.252
smooth -0.130 0.320 -1 0.173 O. 5.1 0.225
6.67 raw -0.323 0.333 -1 0.153 O. 5.7 0.221
smooth -0.323 0.337 -1 0.185 O. 4.7 0.221
10.00 raw -0.841 0.384 -1 0.192 O. 2.9 0.215
smooth -0.700 0.370 -1 0.205 O. 3.8 0.215
15.00 raw -1.101 0.378 -1 0.231 O. 2.0 0.206
smooth -1.000 0.380 -1 0.220 O. 3.0 0.210
25.00 raw -1.411 0.375 -1 0.235 O. 2.0 0.212
smooth -1.411 0.385 -1 0.230 O. 2.2 0.212
PGA (g) -2.365 0.387 -1 0.182 O. 4.9 0.181
PGV (cm/sec) -1.359 0.525 -1 O. O. 4.9 0.212
Coefficients are to be used for 4.6 ;:;; M;:;; 6.8 and R ;:;; 100 km.
prediction of near-source ground motion for the Diablo Canyon Power velocity from strong-motion records inc1uding records from the 1979
Plant site, Report to the D.S. Nuc1ear Regulatory Commission. Imperial Valley, Califomia, earthquake, Bull. Seism. Soc. Am. 71,
Campbell, K. W. (1993). Comparison of contemporary strong-motion at 2011-2038.
tenuation relationships, Pme. International Workshop on Stmng Mo Joyner, W. B. and D. M. Boore (1982). Prediction of earthquake response
tion Data II, 49-70. spectra, V.S. Geai. SUTV. Open-File Rept. 82-977.
Campbell, K. W. and Y. Bozorgnia (1994). Near-source attenuation of peak Kim, W. Y., D. W. Simpson, and P. G. Richards (1994). High-frequency
horizontal acceleration from worldwide acce1erograms recorded from spectra of regional phases from earthquakes and chemical explosions,
1957 to 1993, Pme. Fifth V.S. National Conference on Earthquake Bull. Seism. Soc. Am. 84, 1365-1386.
Engineering, Chicago, Illinois. Lai, S. P. (1982). Statistical characterization of strong ground motions using
Commission of the European Communities (1993). Eurocode 8: Earthquake power spectral density function, Bull. Seism. Soc. Am. 72, 259-274.
resistant design of structures, CENrrC250/SC8-Part 1-1: Seismic Mark, W. D. (1970). Spectral analysis of the convolution and filtering of
actions and general requirements for structures. nonstationary stochastic processes, J. Sound Vibration 11, 19-63.
Console, R. and A. Rovelli (1981). Attenuation parameters for Friuli region Nau, R. F., R. M. Oliver, and K. S. Pister (1982). Simulating and analyzing
from strong-motion accelerogram spectra, Bull. Seism. Soc. Am. 71, artificial nonstationary earthquake ground motions, Bull. Seism. Soc.
1981-1991. Am. 72,615-636.
Hanks, T. C. and H. Kanamori (1979). A moment magnitude scale, J. Geo Newmark, N. M. and W. J. Hall (1978). Deve1opmentof criteria for seismic
phys. Res. 84, 2348-2350. review of selected nuc1ear power spectra, NUREG-CR-0098.
Hammoutene, M., B. Tiliouine, and P. Y. Bard (1992). A two dimensional Pugliese, A. and F. Sabetta (1989). Stima di spettri di risposta da registra
nonstationary optimized acce1erogram scaled for magnitude, distance zioni di forti terremoti italiani, Ingegneria Sismica 2, 3-14.
and soil conditions, Pme. lOth World Conference on Earthquake En Pugliese, A. and F. Sabetta (1993). Attenuazione della componente verti
gineering, 817-821. cale dei moto dei terreno, Pme. 6th National Conference on Earth
Idriss, 1. M. (1991). Selection of earthquake ground motions at rock sites, quake Engineering, Perugia, ltaly, 1,75-84.
Report prepared for the Structures Division, Building and Fire Re Pugliese, A. and F. Sabetta (1994). Simulation of nonstationary time his
search Laboratory, Department of Civil Engineering, University of tories scaled for magnitude, distance and soit conditions, Pme. lOth
Califomia, Davis. Eumpean Conference on Earthquake Engineering, Vienna, Austria,
Joyner, W. B. and D. M. Boore (1981). Peak horizontal acceleration and 1,247-252.
g2008 [NIST CNRS . Tous droits de propriété intellectuelle réservés. Reproduction, représentation ct dilTusiun interdites. Loi du 01/07/92. Articles 5 ,6 ct 7 des CGV
352 F. Sabetta and A. Pugliese
Table 4
Regression coefficients of equation (4) for psv, PGA, and PGV, for the vertical component and epicentral dîstance
Freq. Const. Tenn Mag. Coeff. Dist. Coeff. Site Coeff. Sile Coeff.
(Hz) b e, e,
"
5% damped PSV (cm/sec) 0.25 raw -2.507 0.652 -1 O. O. 3.1 0.320
smooth -2.507 0.660 -1 O. O. 2.2 0.320
0.33 raw -2.275 0.637 -1 O. O. 3.2 0.346
smooth -2.400 0.670 -1 O. O. 2.5 0.346
0.50 raw -2.294 0.676 -1 O. O. 2.5 0.362
smooth -2.220 0.670 -1 O. O. 2.8 0.362
0.67 raw - 1.946 0.637 -1 O. O. 2.8 0.364
smooth - 1.960 0.645 -1 O. O. 3.2 0.358
1.00 raw - \.702 0.612 -1 O. O. 3.9 0.313
smooth - 1.630 0.605 -1 0.010 O. 4.1 0.313
1.33 raw -1.336 0.554 -1 O. O. 5.1 0.289
smooth -1.336 0.560 -1 0.030 O. 4.6 0.289
2.00 raw -0.655 0.438 -1 O. O. 5.4 0.267
smooth -0.920 0.485 -1 0.090 O. 5.0 0.267
2.50 raw -0.766 0.447 -1 0.100 O. 4.4 0.251
smooth -0.750 0.453 -1 0.120 O. 5.0 0.258
3.33 raw -0.674 0.427 -1 0.196 O. 4.7 0.263
smooth -0.550 0.413 -1 0.155 O. 5.0 0.245
5.00 raw -0.322 0.356 -1 0.193 O. 5.2 0.267
smooth -0.400 0.373 -1 0.190 O. 5.0 0.235
6.67 raw -0.635 0.394 -1 0.181 O. 4.6 0.220
smooth -0.550 0.380 -1 0.210 O. 4.5 0.230
10.00 raw - 1.091 0.437 -1 0.218 O. 1.7 0.227
smooth -0.950 0.411 -1 0.235 O. 3.5 0.225
15.00 raw -1.298 0.422 -1 0.252 O. 1.0 0.221
smooth -1.250 0.424 -1 0.252 O. 2.5 0.221
25.00 raw - 1.626 0.423 -1 0.258 O. 0.5 0.219
smooth -1.626 0.428 -1 0.258 O. 1.5 0.219
PGA (g) -2.637 0.443 -1 0.209 O. 4.1 0.195
PGV (cm/sec) -1.499 0.561 -1 O. O. 5.2 0.228
Coefficients are to be used for 4.6 ;3 M;3 6.8 and R ;3 100 km.
Table 5
Regression coefficients of equation (4) for Arias intensity and duration for the larger horizontal component and epicentral distance.
Consl. Tenn Mag. Coeff. Dist. Coeff. Site Coeff. Site Coeff.
b e, e,
"
lA (cm2/sec 3) 0.729 0.911 -1.818 0.244 0.139 5.3 0.397
DV (sec) -0.783 0.193 0.208 -0.133 0.138 5.1 0.247
RovelIi, A., O. Bonamassa, M. Cocco, M. Di Bona, and S. Mazza (1988). U.S. Atomic Energy Commission (1973). Regulatory Guide 1.60, design
Scaling laws and spectral parameters of the ground motion in active response spectra of nuclear power plants.
extensional areas in ltaly, Bull. Seism.. Soc. Am. 78,530-559. Vanmarcke, E. H. (1980). Parameters of the spectral density function, their
Sabetta, P., R. Masiani, and A. Giuffrè (1986). Prequency nonstalionarity significance in the lime and frequency domain, MIT Civil Engineering
in ltalian strong motion accelerograms, Proc. 8th European Confer and Design 60, 1.
ence on Earthquake Engineering, Lisbon, Portugal, 3.2, 25-32. Vanmarcke, E. H. and S. P. Lai (1980). Strong-motion duration and RMS
Sabetta, F. and A. Pugliese (1987). Attenuation of peak horizontal accel amplitude of earthquake records, Bull. Seism. Soc. Am. 70, 1293-1307.
eration and vetocity from Italian strong-motion records, Bull. Seism.
Soc. Am. 77, 1491-1513. Servizio Sismico Nazionale
Sadigh, K. R. (1993). A review of attenuation relationships for rock site Via Curtatone 3
conditions from shallow crustal earthquakes in an interplate environ 00185 Rome, Italy
ment, Proc. International Workshop on Strong Motion Data, II, 179 Tel. 39-6-4466985
236. Pax 39-6-4466579
Saragoni, G. R. and G. C. Hart (1974). Simulation of arlificial earthquakes,
Earthquake Eng. Struct. Dyn. 2, 249-267. Manuscript received Il May 1995.