ML13333A761
| ML13333A761 | |
| Person / Time | |
|---|---|
| Site: | San Onofre  |
| Issue date: | 06/30/1982 |
| From: | TERA CORP. |
| To: | |
| Shared Package | |
| ML13317A188 | List: |
| References | |
| B-82-231, NUDOCS 8207010383 | |
| Download: ML13333A761 (15) | |
Text
ESTIMATION OF SELECTED RESPONSE SPECTRAL VALUES AT THE SAN ONOFRE NUCLEAR GENERATING STATION Submitted to:
H. Gene Hawkins Southern California Edison Co.
Rosemead, California B-82-231 June 1982 TERA CORPORATION 2150 Shattuck Avenue Berkeley, California 94704 415-845-5200 Berkeley, California Dallas, Texas Bethesda, Maryland Baton Rouge, Louisiana New York, New York Denver, Colorado Los Angeles, California
@TERA CORPORATION, 1982 Valley Forge, Pennsylvania 6207010383 820629 PDR ADOCK 05000206 P
ESTIMATION OF SELECTED RESPONSE SPECTRAL VALUES AT SAN ONOFRE NUCLEAR GENERATING STATION INTRODUCTION A multiple regression analysis similar to that performed by Campbell (1981) and TERA Corporation (1982) for horizontal peak ground acceleration (PGA) was used to develop near-source ground motion scaling (attenuation) relationships for horizontal pseudo-relative velocity response spectra (Sv) at four selected periods and three values of damping.
The periods selected were 0.10, 0.12, 0.15 and 0.20 sec., representing frequencies of 5.0 to 10 Hz. The selected damping values were 2, 5 and 10 percent of critical.
The data base was essentially the same as that used to develop scaling relationships for peak ground acceleration (Campbell, 1981).
Some revision of the data was necessary as a result of new information which had become available since the previous studies. Furthermore, in order that the predictions represent as closely as possible "free-field" conditions, neither recordings obtained in large buildings (greater than two stories) nor those obtained on either the abutments or toes of dams were included. Otherwise, the selection criteria were the same as those applied to the previously referenced study on peak acceleration.
Because they have not as yet been digitized, some of the recordings in the original data base were not used in this analysis.
These undigitized records amounted to about 20 percent of the total number of original records and were recorded primarily from smaller earthquakes.
A graphical comparison between the original (Campbell, 1981) PGA data base and the data base used in this study is presented in Figures I and 2, which show the magnitude and distance distributions of each data base.
The earthquake data used in this study is summarized in Table I and consists of 62 recordings from 16 worldwide earth quakes of magnitude 5.0 to 7.7.
As will be discussed in a later section, the net effect of these differing data bases on the prediction of Sv is considered negligible.
B-82-231 I
TERA CORPORATION
50 0
0 40
- 0 E
30 -
0 S
20 4.0
.0 5.0 00 6.070
.0b0 0*
0IUR I
.0-
@0*
4.50 5.00 5.50 6.00 6.50 7.00 7.50 b.00 MAGNITUDE FIGURE 1 DISTRIBUTION OF DATA POINTS PGA DATA BASE TERA CORPORATION
50 400 300 30
- 0.
FG 8
0 200-*
0 I
I.
4.50 5.00 5.50 6.00 6.50 7.00 7.50 8.00 MAGNITUDE FIGURE 2 DISTRIBUTION OF DATA POINTS SPECTRAL DATA BASE TERA CORPORATION
TABLE I EARTHQUAKE DATA USED IN THE SPECTRAL REGRESSION Number Earthquake Date Magnitude Fault Type of Records Imperial Valley 05-19-40 7.1 Strike Slip I
Kern County 07-21-52 7.7 Reverse Oblique I
Daly City 03-02-57 5.3 Reverse Oblique I
Parkfield 06-28-66 6.0 Strike Slip 4
Borrego Mountain 04-09-68 6.7 Strike Slip I
Lytle Creek 09-12-70 5.4 Strike Slip 4
San Fernando 02-09-7 I 6.6 Reverse 8
Sitka, Alaska 07-30-72 7.6 Reverse Oblique I
Managua, Nicaragua 12-23-72 6.2 Strike Slip I
Hollister 11-28-74 5.1 Strike Slip 3
Oroville 08-01-75 5.7 Normal I
Gazli, U.S.S.R.
05-17-76 7.0 Reverse I
Santa Barbara 08-13-78 5.7 Reverse 2
Tabas, Iran 09-16-78 7.4 Reverse I
Coyote Lake 08-06-79 5.9 Strike Slip 6
Imperial Valley 10-15-79 6.9 Strike Slip 26 B-82-231 4
TERA CORPORATION
On the basis of our multiple regression analyses, for the individual spectral and damping values investigated we find that the 0.6 7g Housner Reanalysis Spectrum (HRS) for SONGS Unit I falls within our 72nd and 89th percentile "free-field" predictions for SONGS.
This can be contrasted against the 0.67 g zero-period value of the HRS which represents our 98th-percentile prediction for peak ground acceleration (PGA).
ANALYSIS The ground motion model is the some as that used previously to develop a scaling relationship for peak acceleration, InSv = a+bM -din
[R+clexp(c2M)]
(1)
Sv was defined as the mean of the response spectral values selected from the two horizontal components in cm/sec.
As in the previous studies, the para meters M and R represent Richter magnitude and closest distance to the fault rupture surface in kilometers, respectively.
Our analysis indicated significant systematic bias associated with reverse/re verse-oblique fault types and embedded instruments.
Therefore, scaling vari ables were added to the above expression to account for this bias based on the expression, In Sv = eF + fE + g(M,R)
(2) where e and f are the scaling variables, F and E are variables representing the classification by fault type and embedment, and g(M,R) is the function of M and R given by Equation (I).
The parameters F and E are equal to 0 or I, depending upon the classification:
F = 0 for strike-slip, normal and normal oblique fault types, F = I for reverse and reverse-oblique fault types, E = 0 for ground-level instruments, and E = I for buried instruments or instruments in the basement of small buildings (1-2 stories).
The scaling variables and the coefficients of Equation (I) were statistically determined from weighted nonlin B-82-231 5
TERA CORPORATION
ear regression analyses based on the method of least squares, without the imposition of further physical considerations or constraints. Accordingly, using the terminology of Campbell (1981), the model is "unconstrained."
Use of the constrained regression model would result in even lower predictions.
In our previous analysis on peak acceleration we found that records obtained in areas of steep topographic relief were systematically high with respect to PGA obtained from other records. To explore this further for spectral ordinates we classified all rock sites into four categories based on the topography of the region:
- 1.
Sites located on the top of hills, ridges or steep slopes, II.
Sites located on the side of such features, Ill.
Sites located at the bottom of such features, and IV.
Sites located in relatively flat terrain.
Our preliminary analyses of both Sv and PGA have indicated that recordings obtained on the top and side of hills or slopes (Classes I and II) were higher than those obtained at the bottom of slopes or on relatively flat ground (Classes Ill and IV).
Although at present there are very little data for the lost two categories, we have conservatively chosen to leave topographically affected sites in the data base for the present analyses even though the SONGS site is not affected by topography.
Considering the low residuals associated with sites considered to be more similar to SONGS, we conclude the final SONGS results reflect a conservative estimate of both the median and 84th-percentile value for both Sv and PGA.
The median and 84th-percentile predictions of Sv and PGA based on the current scaling relationships for "free-field" conditions appropriate for SONGS (E=O, F=O) are tabulated in Table 2.
The 84th-percentile values are plotted for the 0.12-second, five-percent damped spectral value in Figure 3 for magnitudes of 5.0, 6.0, and 7.0.
The 84th-percentile values are compared to the Housner Reanalysis Spectra in Table 3. This comparison is shown graphically in Figure 4.
B-82-231 6
TERA CORPORATION
TABLE 2
SUMMARY
OF EMPIRICAL FREE-FIELD PREDICTIONS OF PSEUDO-RELATIVE VELOCITY FOR SONGS Sv for M = 7.0, Period Damping R = 8 km (cm/sec)
(sec)
(o/o) 84th percentile 0.10 2
11.05 16.42 5
8.57 12.24 10 7.05 9.96 0.12 2
14.96 22.72 5
11.46 16.67 10 9.29 13.27 0.15 2
20.68 31.98 5
15.35 23.16 10 12.24 18.11 0.20 2
29.61 45.40 5
22.13 33.34 10 16.96 25.49 PGA(g) 0.313 0.456 B-82-231 7
TERA CORPORATION
TABLE 3 A COMPARISON OF THE 84TH-PERCENTILE PREDICTIONS OF PSEUDO-RELATIVE VELOCITY FOR SONGS WITH THE SONGS UNIT I 0.67g HOUSNER REANALYSIS SPECTRA Sv for M = 7.0, Period Damping R = 8 km (cm/sec)
Ratio (sec)
(
Housner TERA 84th/HRS (HRS)
(84th) 0.10 2
14.19 16.42 1.16 10 10.77 9.96 0.92 0.12 2
19.08 22.72 1.19 10 13.12 13.27 1.01 0.15 2
27.43 31.98 1.17 10 16.86 18.11 1.07 0.20 2
43.31 45.40 1.05 10 23.86 25.49 1.07 PGA (g) 0.667 0.456 0.68 B-82-231 8
TERA CORPORATION
PSEUDO-RELATIVE VELOCITY (cmlsec) m ao n
m Z
m z
a 7
C) z c
C)a n
C aI 1a aU n
Z iC' af a
- -5za z
-0 c
n a
0 a
a 0
a ZC)
- H
30 20 u
O Q) 0 0
LUJ REANALYSIS SPECTRA TERA 84th PERCENTILE 5
.03
.04 0.1
.2
.3
.4
- 1.
- 2.
- 3.
PERIOD (sec)
FIGURE 4 COMPARISON OF SONGS I REANALYSIS SPECTRA WITH TERA 84th PERCENTILE PREDICTIONS 2% AND IO% DAMPING T
C R
TEACRPRTO
The 84th-percentile predictions of Sv are found to exceed the Housner Reanaly sis Spectra by no more than 19 percent for the 2-percent damped spectrum, and by no more than 7 percent for the 10-percent damped spectrum.
Predictions for small embedded structures (E = 1) were found to be approximate ly 20 to 30 percent lower for Sv and 20-percent lower for PGA than those obtained on ground-level instruments.
As with our previous studies, predictions of PGA for reverse faults were found to be approximately 40 percent higher compared to other fault types; the effect on Sv was found to be similar. The net effect of including these scaling variables was to lower the prediction of Sv and PGA for "free-field" conditions appropriate for SONGS by approximately ten percent or less, based on the revised data base.
We conclude that the current results more accurately represent "free-field" ground motion than did our earlier, generally more conservative results (TERA Corporation, 1982), in which scaling variables were not included and a more general data base was used.
As an independent verification of the multiple regression analyses we also used a spectral ratio technique for estimating Sv.
The method, identical to that employed by Woodward-Clyde Consultants (1982),
uses the ratio Sv/PGA, or normalized response spectral value, to estimate Sv through the relationship, In Sv = In Sv - In PGA (3) where Sv is the normalized response spectral value.
An advantage to this technique is that PGA may be estimated from a broader data set than S, while Sv, being relatively insensitive to many of the earthquake and recording parameters for small periods, may be estimated from the smaller digitized data set without significant bias.
In fact, for the distances and magnitudes used in this study, embedment was the only parameter associated with 9v found to be statistically significant.
For the present study, the comparison of the median predictions using Equation (3) and those based on multiple regression analysis is restricted to the 0.10-second, five-percent damped spectral value.
A compar ison of the median values obtained from both techniques found them to be within B-82-231 II TERA CORPORATION
three percent of each other, in remarkably close agreement. The results of this independent approach lends further confidence to the results of the multiple regression analysis and indicates that the smaller data set used in the analysis has apparently not seriously biased the median estimates for SONGS.
SENSITIVITY ANALYSIS Based on the analysis of PGA, we find that the net result of accommodating data revisions since our previous analyses has been to increase predictions by about ten percent.
To assess the impact of further selection from the data base and the introduction of scaling parameters on the results of our analyses, we have sequentially invoked the additional selection criteria, repeated our analyses, and compared the median and 84th-percentile predictions. This comparison for the 0.12-second, five-percent damped spectral value and for PGA appears in Table 4.
This table, which presents the incremental percent changes associated with each step, quantifies the conservatisms in the previous PGA studies by showing the conservative bias associated with reverse faults and large buildings or dams.
CONCLUSIONS The results of our analysis indicate that the 84th-percentile predictions of Sv for SONGS Unit I exceed the 0. 6 7g Housner Reanalysis Spectra for the range of periods and damping values investigated by no more than 19 percent for the 2 percent damped spectrum and by no more than 7 percent for the 10-percent damped spectrum.
Therefore, for individual spectral values, the Reanalysis Spectrum falls within our 72nd-and 89th-percentile "free-field" predictions for SONGS.
To put this result in context, it should be noted that our analyses also indicate that the Reanalysis Spectrum always exceeds our 84th-percentile predictions for basements of one-or two-story buildings for the range of periods and damping values investigated. This latter result represents a more realistic but still conservative comparison of our predictions with SONGS design spectra than the one based on our "free-field" predictions.
B-82-231 12 TERA CORPORATION
TABLE 4
SUMMARY
OF SENSITIVITY STUDIES FOR PGA AND SV (0.12 SEC., 5% DAMPING)
Percent variation*
Description Sv PGA Median 84th Median 84th Base Case (Revised Data Base) 0 0
0 0
With Scaling Variables 3
2
-4
-5 Without Large Buildings
-8
-12
-3
-3 Without Dam Abutments
-4
-7
-3
-6 The variation represents the incremental percent change in the median or 84th-percentile prediction for SONGS after invoking the additional criterion described.
Positive variations represent an incremental increase while negative variations represent an incremental decrease in predicted values.
B-82-231 13 TERA CORPORATION
REFERENCES
- Campbell, K.
W.
(1981).
"Near Source Attenuation of Peak Horizontal Acceleration," Bulletin of the Seismological Society of America, Vol. 71, pp. 2039-2070.
TERA Corporation (1982).
"Estimation of Peak Horizontal Ground Acceleration at San Onofre Nuclear Generating Station," Report to Southern California Edison Company, TERA Corporation, Berkeley, California.
Woodward-Clyde Consultants (1982).
"Development of Instrumental Response Spectra for the San Onofre Site." Report to Southern California Edison Co.
B-82-231 14 TERA CORPORATION