ML19340C482
| ML19340C482 | |
| Person / Time | |
|---|---|
| Site: | San Onofre  |
| Issue date: | 09/29/1980 |
| From: | Luco J AFFILIATION NOT ASSIGNED |
| To: | Reiter L Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 8011170473 | |
| Download: ML19340C482 (11) | |
Text
\\*
4 o
29 September 1980 h
Dr. Leon Reiter, Leader dh)[h Seismology Section, Geosciences Branch Division of Engineering U.S. Nuclear Regulatory Commission Washington, D.C.
20555
Dear Leon:
I have reviewed the reports that you describe in your letter of August 29, 1980.
With respect to the report by Systems, Science and Sof tware (December 1979) I can say that I found it to be of excellent quality.
It confirms my previous opinions that the rise times used in Delta's models are too long, that the frequencies of interest are controlled by stress drop and not necessarily by strength, that Delta's results may be acceptable for earthquakes with stress drops of the or-der of 30 bars, and, that some of the sources of randomness introduced by Delta may distort the results.
My reaction to Delta's Supplement III is covered in the enclosed review where I also discuss t'he implications on the predicted SONGS one ground motion.
Supplement III should be completed by indicating the values for the source parameters used in the comparison with the Imperial Valley 1979 data (" initial" slip velocity, rise times, static offset).
Sincerely yours, r Tr2 E=ay /
4, J. Enrique Luco JEL:bs Enc.
M33170 h 1
1 1
Review of the Report " Simulation of Earthquake Ground Motions for San Onofre Nuclear Generating Station, Unit I.
Supplement III, August 1980".
A Report prepared for the Nuclear Regulatory Commission by J. Enrique Luco 28 September 1980 The report by~ TERA / DELTA compared the calculated and observed strong. ground motion at several stations for the 1979 Imperial Valley earthquake.
The report also describes some additional sources of ran-domness introduced in the numerical model as well as the effects of these new sources of randomness on the predicted motion at San Onofre.
1.
Comparison with 1979 Imperial Valley data.,
1.1 Peak Horizontal Accelerations.
With exception of station 6, where the calculated values exceed the observed response by factors larger than 2, and station Agrarias where 'there is a reasonable match, the calculated horizontal peak accelerations are systematically lower than t.he observed values as indicated in Table 1.
In this table the ratios of the observed to the calculated peak horizontal accelerations are listed for five stations at different distances to the fault.
Table 1.
Comparison of peak horizontal accelerations ratio obs./ calc.
Station Distance (km)
S40E 550W Bonds Corner 3
1.69 2.27*
Station 8 4
1.56*
1.10 Station 4 7
1.72*
1,29 Station 11 13 2.49_
2.21*
Station 1
,22 1.89*
1.34
- Component with la_rgest observed peak acceleration.
^
1.
J
Table 1 reveals that the largest observed peak ho:oizontal accel-
'erations at the different stations can be 1.56 to 2.27 times larger than the calculated values.
For the component with the lowest peak ac-celeration, the' observed values are 1.1 to 2.49 times larger than the calculated peak accelerations.
This systematic underestimatiun of the observed values occurs over the distance range from 3 to 22 km.
The match at station Agrarias is questionable since the mathematical model did not include rupture to the south of the epicenter.
1.2 Peak Horizontal Velocities.
The calculated. peak horizontal velocities are systematically lower than the recorded values as indi-cated in Table 2.
Table 2.
Comparison of peak horizontal velocities
~
ratio obs./ calc.
Station Distance (km)
S40E S50W Bonds Corner 3
1.67 1.35*
Station 8.
4 1.84*
1.35 Station 4 7
1.62 3.35*
Station 11 13 2.43 2
Station 1 22 1.85*
,1.32
- Component with largest-observed peak velocity.
The observed peak horizontal velocities are larger than the cal-culated values by a factor ranging from 1.32 to 3.35.
This deficien-cy in the calculated peak acceleration can also be abserved in Figure S.5 of the' TERA / DELTA report.
1.3 Peak Horizontal Displacements.
The calculated peak horizon-tal displacements _ seem to be, on the average, somewhat larger than the observed values as indicated in Table 3 and in Figure 5.7 of the report.
2 i'
-Table 3.
Comparison of peak horiz6ntal.-displacements.-
ratio obs./ calc.
Station Distance (km)
S40E S50W Bonds Corner 3
0.52 0.38*
Station 8 4
1.47 0.66*
Station 4
~7 0.7 1.50*
Station 11 13 1.07*
0.92 Station 1 22 0.58*
0.77
- Component with largest peak horizontal displacement.
1.4 Horizontal Response Spectra.
The calculated horizontal re-sponse spectra typically underestimates the observed spectra for periods shorter than 0.3 seconds and for periods in the range from 1 to 10 sec-onds.
Some comparisons illustrating this point are presented in Figures 1, 2, 3 and 4.
The deficiency in the range from 1 to 10 seconds is particularly confusing since the calculated peak displacements are, on the average, larger than the observed values.
At high. frequencies, the observed spectral amplitudes are about twice as large as the calculated values.
1.5 Vertical Components of Motion.
In an attet?t at' matching the vertical peak accelerations and velocities the method' of integra-tion over a fault element has been modified by introducing an addition-al. source of " micro-randomness".
I consider this modification a step-backwards since the differential time delays associated with P and.S a
waves are not considered.
The effect of this modification is to ar-tificially increase the high-frequency; ver'tical components of motion.
It should be-noted.' that the calculated peak ~ vertical dis' placement con-siderably? exceeds the observed values--(Figure.5.8).
1.6 General Comments. _ItL is apparent that the calculated ' values underestimate. the. observed' horizontal peak accelerations.. peak. velo -
b
-3
?
cities and response spectra by a factor of the order of 2.
As illus-trated in Figure 5-42 of the report the calculated horizontal compo-nents of motion are significantly deficient for components with fre-quencies higher than 3 Hz.
'It seems that a combination of factors have contributed to this poor comparison:
(i) low value for the ' initial' slip veloc.ity Vo =
800 cm/sec.
(ii) high value for rise time (2.4 seconds) (notice that in supplement I a value of 3.8 seconds was used to model the El Centro 1940 record.)
(iii) low values for Qs.
The results of this comparison indicate that a major revision of the TERA / DELTA modeling procedure is required.- This revision should start by establishing realistic values for Q.
2.
Predictions for San Onofre Site.
2.1 Predicted Spectra.
The mean predicted spectrum for the SE component at San Onofre (earthquake D) is compared in Figure 5 with the S40E spectrum at station 4 (IV79, distance - 7 km) and with the S50W spectrum at station 11 (IV79, distance - 13 km).
It is apparent that the predicted spectrum for San Onofre is similar to those obtained at station 4 and 11, and, consequently, may not represent a conserva-tive estimate of the spectrum for a M - 6.5 earthquake and for the distance range of interest.
In my previous review (October 1979) I noted that the predicted spectrum was also lower, for frequencies high-er than 5 Hz, than the El Centro 1940 NS and Parkfield H85E station 5 spectra.
Since the TERA / DELTA mo' del underestimated t.he response for the 1979 Imperial Valley earthquake by a factor of the order of two, and since the standard deviation for spectral amplitudes. also corresponds to a factor of the order of two, I believe that the TERA / DELTA estim'tes 1
a 4
for the San Onofre spectra chould be multiplied, at least, by a fac-
' tor of about 2.
2.2 Predicted peak amplitudes.
The (refined) peak acceleration and velocities predicted in supplement III are compared in Table 4 with those in supplement I.
Table 4.
Comparison of predicted peak accelerations.
Vertical SE NE III I
III I
III I
a (g) 0.178 0.11 0.372 0.34 0.26 0.39 y
(em/sec) 11.8 8
34.3 29 23.1 31 The introduction of new sources of ranaomness (supplement III) has increase'd the peak vertical acceleration and velocity.
The changes in the horizontal components seem surprising; th'e largest values in supplement I were on the NE component,-on supplement III the largest values occur on the SE component.
The peak velocities 23 and 34 cm/sec.
do not se'em conservative when compared with 39 and 78 cm/sec.
at statio'ns 11 (IV79, distance - 13 km) and 4 (IV79, distance - 7 km).
The peak accelerations 0.379 and 0.26g do not seem conservative when compared with 0.38g (IV79, station 11) and 0.49g (IV79, station'4).
2.3 Local Magnitude.
The calculated local magnitudes for the predicted San Onofre event (ML = 6.2 SE, ML = 5.99 NE) are lower than ML = 6.5.
To reach a value of ML = 6.5 the peak amplitudes (at least the peak velocities) would have to be multiplied by "a factor of about 2.
l l
I
~
9 5
e ~
0 g
- 9 e
'\\
(I }\\
kl 11 Y
Yt IW. -
/'/
)^V,I
/
/
I Al
\\
/
\\
f"\\
^
/g
/
\\
Z
/
/s f
/
y
'.c.
D ON k
g,o -
N
.Bernds Corner $50hl
- O$.CU W.)
- ca Ie vin.ted 0 DI 0.1 I.O l0.
it=2iob(G4).
D**D *)~T Yg' o Ju. 2..k L oo 9;f
/
Y k
e 4
1 e
e.
~
I g
/]),
\\.'y.
\\A V
v Il
\\
I 1
q l
4
/
~
6 to.o -
/tl
\\a
/
x
'G
((
h i
~
c, N
N g,o..
I SA
&.g4oE i
..- -... owe,,,c a calesla7cci
.a h
2 a
g,o t 0l~
I.
10 S
.a e
t
+
4
.Z 4
)
A
/
i p o. -.
/s./s
\\_
g i
A
\\
/
g
/
.o
_\\
M
/\\
q/ \\
S 1
b I}l
~
V 10 A.
f/
8 y
w
/
Sie. A. $4-a.(
o bSer e c) gn..
ca/c4 kJ t
Jr i
O.i 1.0 to.
9,o; P G rsj C b [ 5 '- C Fj. 3
~
r f
~
l.
e e*
g O
g I
100,0 y$t
/\\
\\ l Umm.d\\
'\\
l
\\'
/
kio.o.
/
~
(3 a/
x
"/
l~
/
'6
/
o J
Sh //
- 50 v'
- i. o obs.ec ee d co /co. land
~
de O,\\
u 0,01 O.I t.o io, I
R!?2Iop bec.)
$}. 4 O
4 4
d l
~~~~
l I
1 l
SQ 4 54eE i
I l
LOO.
1' f, t t
.]V
\\j\\
,\\
,~'%
I
/
gi
's/
/
I
[\\-
J 1
..a
.5 Mm. \\\\ SFoW
=
V'
(!
S.O."b+ C 5 p
I N
/
/J
/j t.
e
/
V>-
s
- 1. -
O 9
C.cl O.t
- i. o,
1;.
Pcciso (rec) 2D *
- 3* 2 23f $m m). 5-3'
-,