Written Testimony of JA Blume,HB Seed,S Smith & Others in Response to ALAB-598.Vols 1 & 2.Contains Tera Corp Aug 1980 Fi

UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION

I ~

'TOMIC SAFETY AND LICENSING APPEAL BOARD Richard S. Salzrnan, Chairman Dr. John H. Buck I'n Dr.W. Reed Johnson the matter of PACIFIC GAS AND ELECTRIC COIVIPANY (Diablo Canyon Nuclear Power Plant, Units 1 and 2) Docket Nos. 50-275~ 50-323

APPLICANT PACIFIC GAS AND ELECTRIC COMPANY'S WRITTEN TESTIMONY IN RESPONSE TO APPEAL BOARD QUESTIONS; SET FORTH IN ALAB-598 ggO~

VOLUME 1

August 8, 1980

PA.C IF IC GA.B A.KD ELECTR,XC C G

MPA.NT'0081

8 0 4P~

0 I. The October 15, 1979, Imperial Valley Earth- quake (IV-79, ML=6.4-6.9) provided an extensive set of strong motion records in the near field of a rather severe

earthquake. ~33 The parties should compare the horizontal peak acceleration values recorded for various instrument positions with earlier predictions and compilations of such

motion, ~e. .. those contained in the Final Safety Analysis Report (FSAR) on the Diablo Canyon Nuclear Power Plant, Amendment 50, Appendix D LL llB, Figures 2, 3 and 4; and the United States Geological Survey (USGS) Circular 795, Figures 4, 24, 47, and 48. Those comparisons should (if possible) address whether there is magnitude independence or a saturation effect for ground motion intensity in the near

field of earthquakes. ~34

~33 Preliminary Summary of the U.S. Geological Survey Strong-Motion Records from the October 15, 1979 Imperial Valley Earthquake by R.L. Porcella and R.B. Mathiesen (October 1979), included in Board Notifi- cation, December 17, 1979. ~34, See, for instance, TR. 8597; 10,105; 5889-90.

Written Testimony

DR. JOHN A. BLUME

Q. Dr. Blume, the parties to this proceeding have been asked to answer certain questions regarding the Imperial Valley Earthquake of October 15, 1979. One questions deals with a comparisonof'hose of peak horizontal accelerations recorded in the free field with predictions and compilations of such motions as presented at the ASLB seismic hearings of 1978-79. Would you please do that at 10 this time? The Imperial Valley earthquake of October 15, 1979 '12 (IV-79) yielded an extensive amount of near-field seismic data. As to be expected, there is considerable scatter in the data. While the information obtained from the

15 recordings of the earthquake will prove extremely valuable,

16 care must be taken in its interpretation and possible

17 extrapolation to other geologic and seismological 18 conditions. 19 In this testimony, I have plotted the IV-79 data 20 in the form of peak instrumental acceleration versus 21 distance from the Imperial Fault. These data points are 22 compared to the predictions obtained with the same equations

23 used in the Diablo Canyon hearings and as in the record. In addition, the IV-79 data have been plotted on each of the

25 figures referenced in the question. 26

Blume I-1 , I IV-79 Peak Acceleration Versus Distance And Prediction E ations Figure I-1 is a plot of peak horizontal accelerations recorded at ground level during the October 15, 1979, Imperial Valley earthquake. The distances are those normal to the Imperial Fault and its assumed projection. This procedure may be conservative for stations 2, 3, 4, and 5 whose peak accelerations may have been caused by the closer Brawley Fault which also ruptured. All points plotted are "instrumental" accelerations obtained from 10 Reconaissance ~Re ort, ~lm erial ~Count, California, ~hk, hhl'hhykh khh k 12 h,, Engineering Research Institute, February 1980. 13 Also shown in Figure I-1 are prediction curves for 14 the same earthquake obtained with the use of the Blume SAM V 15 equation (Amendment 50, Appendix D-LLll B) and appropriate 16 constants for Imperial Valley. ~ This equation was also used 17 in the probabilistic studies of shaking at the Diablo Canyon 18 site as in references D-LLll, D-LL41, and D-LL45. 19 Appropriate values for the variables in the equation were 20 * selected from published data

22 23 The magnitude of the IV-79 event has been reported from 6.4 to 6.9. It is believed that 6.6 is becoming the more accepted value for ML, and that value for M has 25 been used. However, dash lines also indicate where the 26 [footnote continued next page]

Blume I-2

Figure I-1 shows that the SAM-V equation accurately fits the recorded IV-79 accelerations. The actual data points fall as follows relative to the prediction curve:

Distance Total Points exceedin rediction for range, no. of Km points The Mean Plus One Std. Dev.

O-ll incl. 37 46/ 13/ 4"8 incl. 20 55K 10% 0-26 incl. 63 38/ 10/ 10 Theoretical value: 50/ 14'/

12 13 The agreement is excellent for all three bands considered. It is to be noted that the Diablo Canyon distance is 5.8 km, 15 about the midpoint of the first two distance bands listed. 16

17 18 [continued from previous page] 19 prediction would be for M< = 6.8. The parameter R in 20 the SAM equations is the slant distance to the point of assumed energy release. The Imperial Fault has been 21 considered as a vertical plane of energy release and 22 the focal depth has been taken as 8 km to calculate the 23 R values; this is the average of the 6 to 10 km reported. Duke reported 520 fps for the shear velocity in the upper layer and a soil weight of 128 p.c.f. 25 These values were used = to obtain pVs 1070fps and = 26 b 2.05, the Blume site factor.

Blume I-3

Figure I-2 is a plot of the IV-79 recorded peak accelerations with the Trifunac equation superimposed. f', 11 ' th ~1'~ f February, 1976.) All of the Trifunac constants have been used along with appropriate site conditions. The Trifunac prediction curves, even for the mean, fall far above all the data points for IV-79. While Figures I-1 and I-2 are valid comparisons because the curves are site specific, the remaining 10 comparisons should be used very cautiously as the curves are "set" and not site specific. For example, the heavy-line SAM Curves in Figure I-3 are for pV = 2000 fps, whereas the

13 Imperial Valley region has a pVs value in the range of 1070 to 1400 fps for which the SAM Curves would plot higher than 15 shown. Figure I-3 is a reproduction of Figure 2 of D-LLll-B 16 with the IV-79 points superimposed.

17 Figure I-4 is a reproduction of Figure 3 of 18 D-LLll-Bwith the IV-79 data plotted as dots. Figure I-5 is the same treatment of Figure 4 from D-LL11-B. 20 Figures I-6, I-7, I-8, and I-9 are reproductions 21 of Figures 4, 24, 47, and 48, respectively, from USGS 22 Circular 795 with all IV-79 recorded peak horizontal accelerations (for both components) superimposed thereon. Figures I-6 and I-7 with the superimposed data show clearly, as recommended by the authors of the USGS report, that the 26 curves should not be extrapolated for close-in distances.

Blume I-4

Figures I-8 and I-9 show that the IV-79 data points fall below the curves, and well below TO and T2 (Trifunac, 1976) curves at short distances. In summary, there is nothing in the Imperial Valley data to be considered surprising or anomalous for the local conditions. The predictions by the Blume SAM equations fit the data excellently. The Trifunac equations do not. Q. Dr. Blume, do these comparisons give additional evidence on the question of 10 ma itude or distance saturate.on? It is my long-held opinion that the near-field 12 magnitude becomes much less important relative to shaking

13 and local damage for values above 6-1/2 M or possibly 7 M. A great earthquake releases energy along a long rupture 15 length and a wide depth as is shown by isoseismal or 16 intensity maps, actual ruptures, and damage patterns. The 17 damage is spread over much greater areas and the shaking 18 lasts longer for a great earthquake like San Francisco in 19 1906, but there is much evidence to support the fact that 20 the local shaking, such as at San Francisco in 1906, is not 21 much, if any, greater for a 6-1/2 to 7 magnitude event than 22 for an 8 M or larger. (Blume ASLB seismic hearing 23 testimony, photos of 1906 buildings)

Figure I-6, which presents data from 5.0 to 7.6 M 25 and IV-79 data, suggests magnitude saturation effects at 26 close in distances. The majority of the IV-79 data from a

Blume I-5

6.6 M earthquake is within the 70/ confidence limits for earthquakes in the magnitude 5.0 to 5.7 group. Figure I-l, which is purposely not plotted on a log scale, shows a relatively small difference between

M = 6.6 and M = 6.8 for the SAM Curves, especially for the mean values. Also, the distance saturation effect is shown by the data points for IV-79 in spite of the extreme recorded value for Bonds Corner at 3 km. The curves reverse 9 close in, and yet they closely fit the actual data on a 10 statistical basis. The opposite effect is shown in Figure I-2 where the Trifunac curves do not reverse and do not fit the data at all. 13 Hanks and Johnson (B.S.S.A., 1976) plotted log peak acceleration versus magnitude and showed that magnitude 15 is a very weak parameter relative to peak accelexation. Figure I-10 is a duplication of their figure with the close 17 in (< 11 km) IV-79 data range superimposed at M = 6. 6. It 18 agrees very well with the other data and clearly indicates 19 magnitude saturation. Note that their "Imperial Valley" 20 symbol is for pre-1979 Imperial Valley earthquakes. Also 21 note that the single high point is for Pacoima Dam which, as 22 noted in the record, is not true ground motion but the 23 amplified response of a ridge. The Blume testimony at the ASLB seismic hearing 25 and other work (2nd U.S.A. National Earthquake Conference, 26 August 1979) covered magnitude saturation and also peak

Blume I-6

acceleration saturation relative to actual damage and the 2 lack of same. The IV-79 data are not surprising in view of prior knowledge and experience. A comparison of the data with the USGS curves as well as the SAM equation and the Hanks and Johnson plot provides additional evidence of magnitude saturation.

12 13

18 19 20 21 22

25 26

Blume Z«7

1.5

1.4 Ex lanation of S mbols x Mexican Station 1.3 ~ "United States Station

1.2

1.0 0 4J 0 9 0 Predict ion Curves by SAM V for: 0 = o 0 Vz 520 fps p = 2.05 Depth = 8 km C 0.7 0.6

Ig 0

~ ~~ 8 0.4 66 ~ 0 'p)„,, X stan Xo lg 0.3 ev)a ~ ~ ) ~ ~ ~ ~ X ~ 0.2 ~ ~ PJ x x mean) X F 00 ~ ~ 0.1 X CO Lg

0 0 10 15 20 25

D i stance from Imper i a I Fau I t (km)

FI GURE I-1. COMPARISON OF OCTOBER 15, 1979, IMPERIAL VALLEy PEAK HORIZONTAL ACCELERATIONS AND SAM V PREDICTION

Blume Fig. I-1

1.5

Ex lanation of S mbols x Mexican Station ~ United States Station 1.3

1.2 Prediction Curves from Trifunac Equations

Ch 1.0 0C e 09

S o 08 g6 ~ c 0.7 Q E (y ~hf ~ 0.6 L

e 0.5

0.4 ~ 0 Xe 0.3 X ~ ~ ~ ~ ~ ~ XX ~ ~ ~ X ~ 0.2 ~ ~ x X ~ OI ~ ~ 0.1 X OO 4%

10 15 20 25

Distance from Imperial Fault (km)

FIGURE I-2. COMPARISON OF OCTOBER 15, 1979, IMPERIAL VALLEY PEAK HORIZONTAL ACCELERATIONS AND PREDICTIONS FROM TRIFUNAC EQUATIONS (BuZ'Letin of the Seismological Society of Ame2ica, ' February 1976)

Blume Fig. I-2

1,000

(12)

(15) SAM (1) ~ ~Q e for pV> = 2,000 fps, W ~ average i+ ~,~~~+ g (8) ~ e (lq) (l3) '~ V~ ~ ~ 4 e

0 f() l ~ 'e ~ ~ ()) 100 O ~ X O San Fernando 1971: Soil + x ++ 0 Rocky (15) Delete ~ + g ee N -6.5

( )+~ Ex lanation

+ Mexican Station ~ United States Station

10 100 Distance to Energy Center (km)

FIGURE I-3. FIGURE 2 OF D-LL11B WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

Blume Fig. I-3

Average of Peak Accelerations For "0" and "I" Site Classifications and for/ = 6.5 (9) SAM IV, y 2 Schnabel 6 Seed"- (15) M = 6.6 '0% 0.25 ~ ««« ~ ~ « ~ ~ ~ ~ «« ~ 0'5 ~ Kana+ = 3 2C (12) M 6.5 «0 ~ 1.0 Mi I ne 8 Davenport: > - (13) ~ Pvs 3,ooo M = 6.5 ~ ----0 Housner"- (11) M = 6.5 pV> ~ 3,000 ~ y ~ 0 = 1U Esteva'4) M 6.5 « W ~ E Cloud F Perezt (14) O M=6.5 C I Donovan"- 0 Blume«(1) «1 = 6.5 (8) 4J M= 6.5 lo B=10km 1U Gutenberg 6 Richter 1U = V (6) M 6.5 O MajoritY of Available Data « Points for M = 6.4 to 6.6 « e

« Blume SAM = 0 IV, y 0, 6'5'S 3

Blume SAM IV, y =1, M = 6.5'S 3'000 +r«~ A Ex lanation + Mexican Station ~ United States Station

miles 2 4 6 10 20 40 60 100 200 400 600 kilometers 5 10 20 40 100 200 400 1,000

Log (Distance, km) NOTES: *Uses Olstance to Causative Fault 'Uses Hypocentral Distance tUses Epicentral Distance

FIGURE I-4. FIGURE 3 OF D-LL11B WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

Blume Fig. I-4

Ex lanation 1.0 + Mexican Station ~ United States Station

5 + ~ ~ ~ oh ~ 0+ ~ ~ ~ 5 0 0) 0 ~ 1

0 4J Ig 0)

~ 0) IU Curves are O SAN V ,for ll = 7, pVS '00

0. 01

Magnitude ~ ~ D 5.0 - 5.9 ~ 6.0 — 6.9

C3 7.0-

0.001 10 100 Distance (km)

FIGURE I-5. FI GURE 4 OF D"LL11B WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

81ume Fig. I-5

1.0; 5.0 - 5.7.

I 0 ~ 6.0 - 6.4 7.1 - 7.6 0 0 '0 I Oq Op 0 lp 0 ~ ' ~ yp' dp I ~ 0 CA 0 0.1 00 0

tD I 0) 0 fp I O U I 0

(D I ape 4J C I 0 ~ N I 0 I'

0. 01

Recorded Data Imperial VaIley 1979

~ Higher Horizontal Component

o Lower Horizontal Component

0. 001 10 100 Distance (km)

FIGURE I-6. FIGURE 4 OF USGS CIRCULAR 795 WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

8lume Fig. I-6

MAGNITUDE 6 ' - 6.4 ALL STRUCTURES

1.0

0 ~ ~ ~ ~ 0 0 ~ 0 p00 0 g 0~ ~ 80 .0 0 ~ C 0 0 0 0.1

O O

tD 4J e 0 N l 0

0. 01

Recorded Data, Imperial Valley 1979

~ Higher Horizontal Component

o Lower Horizontal Component

0. 001,. 1 10 100 Distance (km)

FIGURE I"7. FIGURE 24 OF USGS CIRCULAR 795 WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

Blume Fig. I-7

MAGNITUDE 6.6

TO~ ~ T2~. os~ L ~ N ~ b~ 0 '.. 0 o g ~ O. y o

oS o oo 0

Recorded Data Imperial Val ley 1979

~ Higher Horizontal Component

o Lower Horizontal Component

10 100 Distance (km)

FIGURE I-8. FIGURE 47 OF USGS CIRCULAR 795 WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

Blume Fig. I"8 0 I MAGNITUDE 7.6 TO~ 1.0 T2~

D 0 w ~ '8 -~ . 0

p () ~ ~ ~ so 0 ~ 0 ~ l ~ ~ e 8 a 0.1 005 0 4J (g

Ol Q O O tg C 0 N 10

0.01

Recorded Data 9 Imperial Valley 1979

~ Higher Horizontal Component

o Lower Horizontal Component

0.001 10 100 Distance (km)

FIGURE I-9. FIGURE 48 OF USGS CIRCULAR 795 WITH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPERIMPOSED

Blume Fig. I-9

6= 5 kbar C40 0) 3 8=2 kbar oE E C A 0 ~ —O 4l v fg tl .. ~ ~ ~ Q (2) ' ~ ~ ~ Imperial Valley, 0 tg 1979

lg 0 v Ancona CL L Oroville ~ Imperial Valley, prior to 1979

Magnitude

F I GURE I-10. F I GURE 1 OF HANKS AND JOHNSON, 1976, W I TH IMPERIAL VALLEY OCTOBER 15, 1979 EARTHQUAKE DATA SUPER IMPOSED

Blume Fig. I-10

Written Testimony

DR. H. BOLTON SEED

Q. Dr. Seed would you please comment. on comparisons of recorded free field peak horizontal ground motions from IV-79 . with previous predictions and comment on whether a saturation effect can be seen? The Imperial Valley earthquake of 1979 provided an extensive set, of strong motion records in the near field of 10 a strong earthquake which can be used to evaluate previous concepts of the amplitude of free field motions likely to 12 develop in such events. 13 Figure I-1 shows a plot taken from USGS Circular 795 — "Estimation of Ground Notion Parameters" (Fig. 24 of 15 that report) with the peak horizontal accelerations from 16 records in the Imperial Valley Earthquake superimposed. 17 While the authors of Circular 795 were careful to stop the 18 curves at a distance of 10 km from the fault, it has been hypothesized by some that, statistical analyses of motions 20 recorded at distances greater than 10 km might be 21 extrapolated almost linearly to determine acceleration 22 values in the near-field. The new data from the Imperial 23 Valley earthquake clearly show that (1) statistical results for motions recorded at. 25 distances greater than 10 km should not be 26 extrapolated linearly to directly determine the

Seed I-1

peak accelerations likely to develop in the near-field. This conclusion is in accord with the statements of PGandE witnesses at the Licensing Board hearings in December, 1978. and (2) there is a clear indication of a saturation effect of peak ground motion in the near field with values very near the fault being no greater than those at a distance of about 6 or 7 km from the causative fault. 10 That such results would be anticipated is evidenced by the attached Figure I-2 published in 1976 by Seed, Murarka, Idriss and Lysmer (Ref. 1). These authors also plotted and analyzed statistically the available far-field data for the type of geologic conditions existing 15 in the Imperial Valley (shown within the solid rectangular 16 lines) and also showed the effects anticipated in the 17 near-field by these authors on the basis of physical consid- 18 erations of the problem involved (shown by the dashed lines 19 on the plot). Also superimposed on this plot are the 20 acceleration values recorded in the Imperial Valley earth- 21 quake. It may be seen that all records are in general 22 accord with the predicted range of values for conditions of 23 this type, indicating that both considerations of the basic physics of the problem and the new records of near field motions lead to the conclusion that attenuation curves 26

Seed I-2

flatten out markedly at distances close to the causative fault. Since motions must pass through rock before being modified by local site conditions, the Imperial Valley data provides good evidence that the general form of the acceleration attenuation curve shown in Figure I-2 should also apply for the attenuation curve for any type of local geologic conditions (soil or rock).

16 17 18

19 20 21 22

Reference 1 Seed, H. B., Muraka, R., Lysmer, J., and Idriss, I. M. (1976) "Relationships between Maximum Acceleration, Maximum 25 Velocity, Distance from Source and Local Site Conditions for Moderately Strong Earthquakes," Bulletin of the Seis. Soc. 26 of America, Vol. 66, No. 4, pp. 1323-1342, August 1976.

Seed I-3

MAGNITUDE 6.0 - 6.4 ALL STRUCTURES

1.0

4 Ie ~ Z ohio~ 0

LY. 0.1 U U z 0 0.01 0 \

e imperial Volley Eq. Record

0.001 10 100 DlSTANCE, lN KILOMETERS Figure 24. Peak horizontal acceleration versus distance to slipped fault for magnitude range 6.0-6.4 inc luding data from both l arge and sma l l s tructures. Symbo l s and curves same as in Figure 23. (Taken from USGS Circular 795) Seed Figure I-1

I.O

0 ole I ~ oC I e a OI

«f 4l I M C

CQ 6 ~ GOI„ E 4p 5„

Eorthquoke Mognitude = 6.5 o~

Deep cohesionless soil conditions 'O~ Q Imperial Valiey Data O Data Prior to 1965

I O.OOI I IO IOO IOOO Distonce from Zone of Energy Releose-km

Fig. 7 RELATIONSHIP BETWEEN MAXIMUM GROUND ACCELERATION AND DISTANCE FROM ZONE OF ENERGY RELEASE FOR DEEP COHESIONI ESS SOIL CONDITIONS (Taken from Seed et al., 1976) Solid lines indicate results of statistical analysis of data available to 1975. Dashed lines indicate position of attenuation curve estimated by authors in 1975. Open circles are Imperial Valley Earthquake data. Seed Figure I-2

Written Testimony

DR. STEWART SMITH

Q. Dr. Smith, could you address the question of whether horizontal peak ground acceleration saturates in the near-field of lar e earth akes?

The October 15, 1979 Imperial Valley earthquake (IV-79) has added data which supports the view that 10 horizontal peak ground acceleration (PGA) saturates in the near field of large earthquakes. The saturation effects of 12 interest are twofold; saturation with decreasing distance to the fault rupture surface, and saturation with increasing magnitude. Even by itself, the 1979. Imperial Valley 16 earthquake provides data which supports a saturation of 17 ground motion with decreasing distance. Figure 5-4 of E 18 Exhibit 1 to this testimony (TERA report entitled, "Evaluation of Peak Horizontal Ground Acceleration 20 Associated with the Hosgri Fault at Diablo Canyon Nuclear

21 Power Plant" ) presents the recorded PGA for IV-79 versus

22 distance. Note that eight data points are plotted at 1.2 Km 23 for presentation, even though they were recorded at stations 24 about 0.2 Km from the closest distance to the fault rupture 25 plane. All these data argue strongly for saturation of PGA 26 with decreasing gisrance.

Smith I-1

The curve presented on that figure is based on statistical analysis of IV-79 and 26 other earthquakes. While the 26 other earthquakes had little data recorded at distances less than 1 Km, the total data set suports the saturaton of PGA with decreasing distance. Exhibit 1, which incorporates the IV-79 data, has 7 extended the analysis of horizontal PGA in the near source 8 region. This analysis supports the conclusion that PGA does saturate with increasing magnitude in the near source 10 region. Whether "complete" saturation is achieved for very

large earthquakes (greater than Ms 7.5) is not conclusively 12 established by empirical analysis, due to the paucity of 13 very high magnitude - near source data, and the potential 14 biases of non-North American earthquake data in these 15 analyses. As Exhibit 1 shows, if non-North American 16 earthquake records are excluded, from the empirical analyses 17 or, if they are equally weighted with all data, PGA is 18 predicted to essentially saturate at these large magnitudes. 19 The doubtful quality and lack of information about some 20 non-North American data raises a question as to whether that 21 data should be included in a statistical analysis applicable 22 to California. In addition, regional differences in 23 tectonics and attenuation might obscure the magnitude 24 saturation effect. Sensitivity analyses were done to 25 address these problems, and the results do indeed indicate 26 that PGA approaches saturation with increasing magnitude.

Smith I-2

In a conservative analysis, i.e., when both North American and non-North American earthquake data are included in a multiple regression analysis of near source PGA recordings (Exhibit 1), the predicted median value for a magnitude 7.5 earthquake at a distance of 5.8 Km is 0.48g for instrumental ground motion. If a regression model is chosen which assumes complete saturation of PGA with increasing magnitude (supported on theoretical grounds and by North American data), the predicted median value is .4lg. 10 Thus the resulting prediction of ground motion for DCNPP is little influenced by whether the regression analysis is constrained to require "complete" magnitude saturation. 13

15 16

19 20 21 22 23

25 26

Smith I-3

II. Response spectra have been developed from the near-field (1 to 11 km) ground motion records produced by IV-79. The records contain horizontal peak acceleration values in the range of 0.81g to about 0.2g. The applicant calculated a mean peak acceleration of 0.36g for IV-79 at the 5.8 km site-to-fault distance that characterizes the Diablo Canyon site (Applicant's Brief). Despite the fact that the IV-79 peak acceleration values are generally lower than the 1.15g peak acceleration or 0.75g zero-period acceleration used as the design basis for the Diablo Canyon plant, (resulting from a postulated 7.5M event on the Hosgri fault,),. there're instances (although only those from the El Centro Arrays are significant) for which .the IV-79 horizontal responses exceed the Newmark Design Response Spectrum for Diablo Canyon. (See, staff brief at page 9; Brune Affidavit, Attachments A and B.) In view of this, the parties should discuss whether the Newmark Spectrum is an appropriate and sufficiently conservative representaton of the 7.5M event, at Hosgri. ~35

~35 In other words, if the various IV-79 near-field response spectra were used to generate a smoothed, average response spectrum for a zero-period acceleration appropriate to that event (in accordance with techniques explained in Blume's testimony fol. Tr. 6099 at page 6 and pages 39 and 40), and if this spectrum were scaled to a 0.75g zero-period acceleration, would the resulting response, spectrum be bounded by the Newmark Spectrum for Diablo Canyon?

Written Testimony

DR. JOHN A. BLUNE

Q. Dr. Blume, would you please discuss whether the Newmark and Blume spectra for the Hosgri reanalysis remain conservative in li ht of IV-79'P

In order to study the characteristics of the response spectra shapes obtained from the Imperial Valley 10 earthquake records, the 24 available horizontal records within 11 km of the Imperial fault were collected and their 12 respective response spectra computed and plotted as shown on

13 Figure II-l. All the spectra were plotted for 7% damping, consistent with the value used in the Diablo Canyon Hosgri 15 reanalysis for structures. The specific station records

16 used for the figure are listed in Table II-1. Figure II-1 provides an excellent example of why 18 extreme values are not used to represent data sets. To use 19 the extremely high spectrum shown in the figure (Bonds 20 Corner) as the single Imperial Valley spectrum would be in 21 effect considering that extreme spectrum to be more 22 representative of the Imperial Valley earthquake than any of 23 the other 23 spectra shown. On the other hand, by including the extreme spectrum in the data set and calculating the 25 resultant mean spectrum, each of the 24 spectra receives 26 equal consideration in the analysis. The averaging of

BlumeII-1

spectra to arrive at a design spectrum is an accepted method and consistent with the procedures used in the derivation of the Diablo Canyon Hosgri design response spectra. The calculated mean of the 24 horizontal spectra is shown in Figure II-2, along with the Blume and Newmark Hosgri horizontal design response spectra, both of which were used in the Hosgri reevaluation. The figure clearly demonstrates the substantial margin between the Hosgri design spectra and the mean Imperial Valley spectrum. 10 If the peak ground acceleration of the mean Imperial Valley spectrum is normalized to 0.75g, corresponding to the peak ground acceleration specified in 13 the Diablo Canyon reanalysis, the Imperial Valley spectrum is still within the Hosgri design spectra in the periods of 15 interest. (Figure II-3) 16 In conclusion, both the Newmark and Blume design response spectra used in the Diablo Canyon Hosgri reanalysis 18 remain appropriate and conservative in view of the data 19 obtained from the October 15, 1979, earthquake in the 20 Imperial Valley. 21 22

25 26

BlumeII-2

TABLE II-1

24 FREE-FIELD HORIZONTAL RECORDS WITHIN

11 KM OF IMPERIAL FAULT From October 15 1979 Earth ake

STATION FAULT DISTANCE KM NO. OF RECORDS ' Array 4

Array 5

10 Array 6 Array 7

12 Array 8

13 Array 10 Bonds Corner 3.

15 Differential Array

16 Brawley Airport 2 Calexico Fire Station

18 Holtville Post Office Imperial County Center

20 24 21. 22 23

25 26

BlumeII-3

3.00

74 Damping

2,25 Bonds Corner

C 0 4l tg 0 e 1.50 O M M I 4J 0 0.75

0.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0. 70 Period (sec)

FIGURE ZI-]. HORIZONTAL RESPONSE SPECTRA FOR 24 RECORDS WITHIN 11 KILOHETERS OF THE IHPERIAL FAULT, OCTOBER 15, 1979 EARTHQUAKE

3.00

74 Damping

2.25 Blume Hosgri C Design Spectrum Q7 0 C Newmark Hosgri 3 (D / ~~ > Design Spectrum tD 1 ~ <9 / 1.50 /

M M /r I l 4J 0 // Mean Imperial Cl Valley Spectrum cn 0 75

0. 00 0. 00 0.10 0. 20 0.30 0.40 0. 50 0. 70 Period (sec)

FIGURE II"2. COMPARISON OF MEAN IMPERIAL VALLEY HORIZONTAL SPECTRUM ~ FROM OCTOBER 15, 1979 EARTHQUAKE AND BLUME AND NEWMARK HOSGRI HORIZONTAL DESIGN SPECTRA

.00

74 Damping, Normalized to 0.75g

2.25 ~ Blume Hosgri C Design Spectrum 0 ~ 4l Newmark Hosgr< lg Design~ Spectrum 0

/ Mean Imperial 4J Valley Spectrum, / Normalized to 0.75g 0.75

0. 00 0. 00 0.10 0. 20 0. 30 0.40 0. 50 0.60 0.70

Period (sec)

FIGURE II-3. COMPARISON OF MEAN IMPERIAL VALLEY HORIZONTAL SPECTRUM FROM OCTOBER 15, 1979 EARTHQUAKE, NORMALIZED TO 0.75g, AND BLUME AND NEWMARK HOSGRI HORIZONTAL DESIGN SPECTRA

III. We are told that IV-79 data are not relevant to the Diablo Canyon seismic analysis because that plant is a "rock" site, whereas the Imperial Valley data were obtained on soil sites. * (Rothman - Kuo Affidavit at page 3; Blume Affidavit, Paragraph 8.) What is the significance of this difference in view of the conclusion of the authors of USGS Circular 795 (based on an analysis of data provided in that document) that, for comparable earthquake magnitude and distance, there are no significant differences between peak horizontal accelerations measured on soil or rock? (USGS Circular 795 at pages 1, 17, and 26.) This question should be considered in light of statements by applicant's witness Blume to the effect that acceleration, rather than velocity or displacement, is the critical parameter in the design= of Diablo Canyon (Blume Affidavit, Paragraph 9; Testimony fol. Tr. 6099, p. 33).

Written Testimony

DR. JOHN A. BI'UME

Q. Dr. Blume, you have read question number 3 of ALAB-598. Please describe for the Board how acceleration, velocity, and displacement differ between rock and soil sites and the criticality of these arameters at, Diablo Can on.

The shape of the response spectrum for a soil site 10 differs from that of a rock site for a given magnitude earthquake, the degree of variation depending on the 12 frequency range considered. In the high frequency portion 13 of the spectrum, there is relatively little difference in the spectrum shape between soil and rock sites. In the 15 medium and low frequency portions of the spectrum, the 16 difference in spectrum shapes between rock and soil sites, is 17 more significant. The high frequency portion of the 18 response spectrum is predominately influenced by the level 19 of peak acceleration. USGS Circular 795 stated that "In the 20 distance range used in the regression analysis (15-100 km) 21 the values of peak horizontal acceleration recorded at soil 22 sites in the San Fernando earthquake are not significantly 23 different from the values at rock sites . . ." This conclusion is supported by the results of almost three 25 thousand records from point-source, large underground 26 nuclear explosiops in Nevada and elsewhere, where at short

BlumeIII-1 I distances from the source the average accelerations on rock and on alluvium were found to be about the same (Amendment 50, D-LL11B). The medium and low frequency portions of the response spectrum is primarily influenced by peak velocities and displacements, respectively. It is generally recognized that velocity and displacement are greater in soft alluvium than on rock. Thus, with regard to the Imperial Valley, it would be expected that its deep, soft (low shear velocity) 10 alluvium and high water table would have a greater velocity and displacement for comparable earthquake magnitudes than for a rock site such as Diablo Canyon. Thus, while the high frequency portion of the response spectrum (dominated by peak accelerations) for a 15 rock and a soil site for a given magnitude earthquake will 16 not differ significantly, there will be significant differences in the medium to low frequency portions of the 18 spectrum. 19 The matter of relative importance of acceleration, 20 velocity, and displacement in the Hosgri reanalysis of 21 Diablo Canyon has been addressed in D-LL 42 of Amendment 53. 22 The significant natural frequencies of the'iablo Canyon structures are all higher than 2 Hz. As shown on the 4-way 4 log plot spectral diagrams o f Figure 42-A (D-LL 42 o f 25 Amendment 53), +e velocity and/or displacement could be 26 arbitrarily increased to a considerable degree without

Blume III-2

affecting the design motion for these rigid, short-period structures and facilities. In other words, in the high frequency range, as for Diablo Canyon, acceleration, and not 4 velocity or displacement, is the critical parameter in design. In conclusion, the response of rigid structures, such as those at Diablo Canyon, are predominantly governed by peak acceleration, but, since peak displacement and 9 velocity also influence the shape of the response spectrum, 10 it is inappropriate to use data from soil sites to derive response spectra for rock sites.

17 18

19 20 21 22 23

25 26

Blume III-3

IV. The magnitudes of vertical and horizontal acceleration values measured at IV-79 are generally comparable. (Nean values calculated at a distance of 5.8 km from the fault are virtually identical.) ~36 The response spectra developed for vertical motion within 11 km of the Imperial Fault during IV-79 appear to show generally equivalent values of vertical and horizontal response for periods less than about 0.2 seconds (i.e., frequencies in excess of 5 cps). ~37 Finally, in some instances the higher frequency portions of the IV-79 response spectra for vertical motion exceed comparable portions of the Diablo

Canyon Design Response Spectrum. ~38 Observations made of the IV-79 data and response spectra appear to be consistent with the criteria set forth in NRC Regulatory Guide 1;60. These require that vertical accelerations in the higher frequency range be equal to horizontal accelerations. As the guide states: It should be noted that. the vertical Design Response Spectra are 2/3 those of the horizontal Design Response Spectra for Frequencies less than 0.25; for frequencies higher than 3.5 they are the same, while the ratio varies between 2/3 and 1 for frequencies between 0.25 and 3. 5. +39

The references to vertical motion made in the Diablo Canyon record, however, indicate that a 2/3 ratio between vertical and horizontal motion was apparently utilized at all frequencies. 4~0 The parties should address this apparent inconsistency and explain it, if possible. Should there be substantive and relevant analyses suggesting that vertical motion records do not reflect the true vertical motion, these should be provided. ~41

~36 Blume Affidavit, Table 1, Figures 1 and 2.

~37 Rothman - Kuo Affidavit, Figures. 38/ 'bid.

~39 We note .that elsewhere in the Regulatory Guide freguencies are presented with accompanying units of cycles per second (cps), and assume that these units are inadvertently omitted in the portion we have quoted.

~40 SER Supplement 7, pages 3-18; Knight Testimony, page 13, fol. Tr. 8697, Ghio Test., page 1, fol. Tr. 6993. Blume Testimony, page 41, fol. Tr. 6099.

4~1 See, for example, Newmark Testimony, fol. Tr. 8552, Reference B at, pages 4, 5; Tr. 9349.

WRITTEN TESTIMONY

DR. JOHN A."BLUME

Q. Dr. Blume, would you please tell the Board what earthquake codes call for and what the design practice is as respects vertical accelerations?

Codes for earthquake resistance are constantly being re-evaluated as new information is obtained and more research is conducted. The widely used Uniform Buildincu Code, 10 for example, is revised every three years. The NRC is continually considering new information, and from time to time issues new guidelines or revises old ones. In most 12 cases the changes made are not, extreme. 13 Most codes in use for all but so-called high risk have no 15 facilities provision for vertical earthquake forces.* The have 16 structures considerable resistance to vertical earthquake forces because are 17 they designed for dead and 18 vertical live loads under low allowable stresses. The members, have a 19 therefore reserve capacity for when 20 additional vertical forces considering infrequent and 21 earthquake .forces higher allowable stresses. Diablo Canyon has the 22 benefit of this reserve capacity in all of its vertical load bearing members. 23 4

* The only pr9vision for vertical forces in most codes is that the gravity loads must be reduced somewhat (for 25 earthquakes} where such loads otherwise tend to make 26 members strpnger for lateral forces.

Blume IU-1

At the time Diablo Canyon was designed the 2/3 rule was.in nearly universal effect for nuclear plants. At 'he same time skyscrapers and all other buildings were generally not at all designed for vertical seismic forces. This was primarily because of the inherent values noted above. In other words, the ratio was zero. The Diablo Canyon structures, however, are in fact capable of accommodating far more than a 2/3 ratio. Q. Please explain to the Board what vertical to horizontal ratios were actually used for the Diablo Canyon 10 Hos ri reanal sis.

The 2/3 rule applied to Diablo Canyon was used in the original design. However, the reanalysis the 13 in for Hosgri fault exposure more than 2/3 was actually used in most cases. The Blume the ASLB 15 written testimony in seismic hearings presented in Figures 8 through 17 the Hosgri 16 gives spectra by Blume and Newmark anchored to 0.50g at zero 17 period without any tau reduction. The 0.50g was simply 2/3 of the 0.75g value where tau was zero. A 19 tau reduction could have been taken as the due 20 well for vertical spectra to the lack of wave coherence, embedment and other reasons, but was not. Using the Newmark spectrum as an 22 it horizontal example, the the 23 ratios of peak vertical effective acceleration to Qe peak horizontal are as follows: l// ///

Blume IV-2

Miscellaneous Small Structures: 0.50/0.75 = 0.67 Containment and Intake: 0.50/0.60 = 0.83 Auxiliary: 0.50/0.55 = 0.91 Turbine: 0.50/0.50 = 1.00 Thus, for the most important structures, the effective ratio of peak vertical to peak horizontal is much more than 0.67. When the ratio of horizontal to vertical is studied on a spectral basis, we see that the ratio varies depending on the period of vibration. Figures IV-1 and IV-2

10 show the ratio of vertical design spectrum to horizontal design spectrum for the Blume and Newmark Hosgri spectra,

12 respectively. The figures demonstrate that within the period range of interest. (less than 0.5 sec) only the ratios for the miscellaneous structures approach two-thirds. The

15 primary Category I structures are designed to ratios greater

16 than two-thirds, and in the case of the Turbine Building the ratio approaches 1.4 at about 0.10 seconds for the Blume

18 spectrum. The ratio specified by NRC Reg. Guide 1.60 is also shown on the figures for reference. Thus, while the

20 Hosgri reevaluation criteria specified a two-thirds ratio of 21 vertical to horizontal, in fact the primary structures and 22 components have been analyzed for ratios greater than 23 two-thirds.

25 26

Blume IV-3 h' Q. Dr. Blume, are there considerations leading to reductions of free-field accelerations for desi u oses? There is much in the record about the differences between recorded, or "instrumental" peak accelerations, and "effective," or anchor point, accelerations. The ratio of

N 0.75g to 1.15g for effective to instrumental horizontal motion used at Diablo Canyon was conservative. That ratio is even more conservative for vertical motion, i.e., the difference between effective vertical motion and instrumental vertical motion would be larger. I know of no case of pure vertical failure (not to be confused with moment or shear failure at the base of a structure due to horizontal 13 forces). A plot of all vertical instrumental accelerations versus normal distance to the Imperial Fault for IV-79 leads to a curve which would about, at, 16 produce 0.34g instrumental 5.8km. were 17 If this figure factored by 0.75/1.15, it would become 0.29g which is far below the 0.50g actually used. It could be down 19 reasonably factored even more, say to 0.20g 20 for design for the IV-79 earthquake. On this basis, Diablo. 21 Canyon is 0.50/0.20 or 2.5 times better off vertically than IV-79 22 would justify. Vertical motion recorded on a pad of concrete at the surface, or on top of a basement slab, is not the same motion that "drives" a building vertically at the bottom of

26 its foundations. There is a definite restraint or

Blume IV-4

confinement at the foundation base that is entirely lacking at the instrument locations. This is related to, and actually an extension of, the "tombstone" effect where the instrument and pad together with the soil constitute in effect a dynamic system. Most, if not all, vertical recordings are too high. The instruments, or transducers, should be placed under the instrument foundations to record true verticals. Q- Dr. Blume, is there a difference in arrival time between peak horizontal and peak vertical accelerations, and, if so, 10 what is the effect of

this'2

A study of earthquake time histories reveals not only that the horizontal peaks are usually well separated in the time domain, but that the vertical are separated from the horizontal by even more time, up to several seconds. IV-79 was no exception. The 16 Hosgri reanalysis, however, required that three components be 17 all considered simultaneously, with combined on a 18 responses square-root- of-the-sum-of-the-squares 19 process. In addition, the Hosgri reanalysis two 20 required equal horizontal components; in actuality, one horizontal component always exceeds the other, sometimes The 22 substantially. conclusion is that the combining procedure 23 for the three earthquake components, and the assumption of equal horizontal components is a conservative 25 procedure. 26

Blume IV-5

Q. Dr. Blume,-please tell us what is meant by "peak clipping" and what effect that mi ht have on res onse s ectra. We have reported (Amendment 50, DII30) that the spikes or peaks of horizontal time histories of ground motion have very low correlation with stress or damage in structures or in response. All of the peak accelerations (both negative and positive) in a time'istory can be

reduced approximately 30% with only a few percent, say 5%, decrease in the spectral response. This procedure has been

10 repeated for two IV-79 vertical records. Although the peaks of the vertical records have high amplitudes, they do not

12 contain large amounts of energy and are not significant to

13 structural response. As an example, the peak vertical acceleration for station 3 can be reduced from 0.15g to

15 0.10g with only a 5.46% reduction in the peak value of the acceleration response spectrum.

17 18

19 20 21 22

25 26

Blume IV-6

2.00

Q1 Turbine Building

Q2 Auxiliary Building

Q3 Containment, Intake Structures

Q4 Miscelianeous Smali Structures

gRRC Regulatory Guide 1.60

= g~ 0.67 +

74 Damping

0.00 0. 00 0. 10 0. 20 Oe30 0.40 0.50 0.60 0. 70 0.80 Period (sec)

FIGURE IV-1. RATIO OF VERTICAL DESIGN SPECTRUM TO HORIZONTAL DESIGN SPECTRUM FOR BLUME HOSGRI CRITERIA

2.00

1 Turbine Building 2 Auxiliary Building 3 Containment, Intake Structures c 1.50 E 4 Miscellaneous Small Structures 1 S 0 0 0) CL CL c Q) Vl 1.00 Q NRC Regulatory Guide 1.60 C)

IlJ lg 4J 3 0 C 4' 0 N Q V = 0.67 ) g 0.50 8

7C Damping

0.00 0. 00 0.10 0. 20 0.30'.40 0. 50 0.60 0. 70 0. 80 Period (sec)

FIGURE IV-2. RATIO OF VERTICAL DESIGN SPECTRUM TO HORIZONTAL DESIGN SPECTRUM FOR NEWMARK HOSGRI CRITERIA

Written Testimony

DR. GERALD FRAZIER

Q. Dr. Frazier, IV-79 produced apparently high vertical accelerations. Could you please discuss the recorded vertical accelerations from IV-79 and the impact, if an , this data has on" Diablo Can on? Unusually high vertical accelerations were recorded for the 1979 Imperial Valley (IV-79) earthquake. IV-1 the strong motion 10 These data are presented in Table for stations positioned as shown in Figure IV-1. At close were as on 12 distance, the peak horizontal accelerations high, as accelerations. A few of 13 the average, the peak horizontal the stations recorded vertical peaks that were substantially

15 higher than the horizontal peaks. While vertical exceed horizontal 16 accelerations that rival or the peak accelerations are unusual they are not without precedence. 1.74 18 However, the peak vertical acceleration of g (later corrected to 1.52 g) recorded for Station 6 is without

20 precedence. 21 In interpreting the unusually high vertical examine 22 accelerations recorded at IV-79, it is important to the different characteristics of seismic waves. Acoustic waves in the earth are referred to as compressional waves or "P waves the 25 waves". The particle motion for these is along 26 direction of, propagation. Waves with shearing motions, Frazier IV«1

particle motion perpendicular to the direction of propagation are referred to as shear waves or "S waves." The velocity of P waves is always larger than that of S waves. Hence, the first motions to occur during an earthquake are due to P waves. Because these waves emerge steeply, and because the particle motion is along the direction of propagation, P waves appear principally on vertical

recordings. S waves arrive later and typically provide the / preponderance of seismic energy on both horizontal and recordings of earthquakes. is these waves that 10 vertical It account for most earthquake damage to structures.

For ~t ical Southern California earth structures, vertically polarized shear waves (SV waves), and not 13 P waves, account for the larcaest vertical accelerations. Furthermore, the vertical-to-horizontal amplitude ratio for

the SV waves is less than one. This ratio is 16 typically equal to the tangent of the angle of the emerging SV waves 17 with respect to vertical. The angle is small since material 18 velocities, which increase with depth, cause the high frequency waves to emerge steeply. Consequently, vertical 20 I'1 accelerations are typically lower than horizontal accelerations. 22 In the Imperial Valley, however, the situation is ~at ical. The deep sedimentary layering there causes unusually P-wave amplitudes. Examination of the 25 large IV-79 indicate that the large vertical 26 recordings for Frazier IV-2

accelerations resulted from high frequency P waves, not SV waves. The unusually large P waves that 'appear in the vertical recordings from IV-79 are due to the uncommon wave properties of the deep sedimentary basin found in the

Valley, illustrated in Figure IV-2. Material 'mperial velocities increase nearly linearly with depth from low values within the 300 meters or so of surface alluvium to 7 values typical of bedrock 'at a depth of 6 kilometers. The properties of this deep sedimentary basin are markedly different from other regions of California where rock is 10 encountered within 1 or 2 km of the earth's surface as illustrated in Figure IV-2. 12 The substantial velocity gradient with depth at 13 Imperial Valley causes waves to bend (refract) toward the earth's surface along circular paths in accordance with 15 Snell's law. The sedimentary basin acts somewhat like an echo chamber due to the extreme bending of waves toward the 17 earth's surface. For example, at Imperial Valley only about 18 10% of the seismic energy (P and S waves) emanating from 19 earthquake rupture at a depth of 3 km escapes into the 20 underlying bedrock (e.g. Ray "D" on Figure IV-3). The 21

remaining 90% emerges at the earth's surface within an > 22 epicentral distance of 20 km. In a more typical situation, 23 such as at Diablo Canyon, the energy emerging within 20 km about 60 percent. Therefore, the unusual echo chamber 25 is 26 Frazier IV-3

effect at Imperial Valley which concentrates seismic energy would not be expected at Diablo Canyon. While the echo chamber effect at Imperial Valley

tend to increase both P and S wave amplitudes, the 'ould unique velocity profile there (Figure IV-2) preferentially amplifies P-wave amplitudes. Due to the conservation of energy, the amplitude'f a seismic wave increases as it propagates'nto a medium of decreasing stiffness. Consequently, the amplitudes of the emerging waves increase rather dramatically as they approach the earth's surface.

However, high frequency S waves are severely attenuated in the shallow sediments, thereby compensating for the sedimentary basin. P waves, on the 13 amplification of the other hand, can travel efficiently in the soft surface materials. In the shallow sediments at Imperial Valley,

S waves above 10 Hz can be attenuated by a factor of ten within one kilometer, while P waves above 10 Hz are attenuated only about 20% over the same distance. The and uniform gradient with depth 19 large velocity in the Imperial Valley preferentially amplifies the P waves and thus yields higher vertical acceleration peaks than due to the unique earth 22 horizontal peaks.= This effect is and since the earth structure 23 structure at Imperial Valley at Diablo Canyon is substantially different, such high verticals are not expected. /// Frazier IV-4

Comparative calculations have been performed to examine how differences in earth structure between Imperial Valley and Diablo Canyon lead to differences in surface motions. Figure IV-2 illustrates the differences in two earth structures with lower material velocities present in the Imperial Valley to a depth of about 6 km. Surface motions have been calculated for small (point) earthquake ruptures positioned at varying hypocentral depths and epicentral distances for the two earth structures. The results, termed Green's functions, include all wave types 10 (including P, S, and surface waves) over the frequency band, from 0 to 20 Hz. 12 The most notable difference was the ratio of peak vertical to peak horizontal ,accelerations. These vertical-to-horizontal ratios were computed and averaged 15 over epicentral distances less than 10 km. These ratios, 16 averaged over the near field, are presented in Table IV-2. 17 The vertical-to-horizontal ratio of peak acceleration is 18 about ten times higher for Imperial Valley than for Diablo 19 Canyon for both strike-slip and dip-slip rupture over the 20 range of hypocentral depths that were tested. These results 21 indicate the trend to be expected in actual earthquake 22 motions, namely, considerably reduced vertical accelerations 23 at Diablo Canyon as compared to vertical accelerations recorded the Imperial Valley earthquake. 25 for 26 Frazier IV-5

Further discussion of computed ground motions at Diablo Canyon due to a major hypothesized earthquake along the Hosgri fault may be found in my response to question VII. The results of earthquake modeling yield vertical accelerations less than half the horizontal accelerations. Even accounting for the effects of the unique earth structure at Imperial Valley, the large vertical acceleration recorded at Station 6 (1.74 g later corrected to 1.52 g) during the 1979 Imperial Valley earthquake poses an enigma. Station 7, which is about 1 km SW of Station 6, 10 recorded a peak vertical acceleration of .65 g (corrected to .51g), and Station 5, which is approximately 3 km NE of, Station 6, recorded a peak'ertical of .71 g (corrected to .44g). It is clear from field observations that the Imperial Fault passes between Station 6 and Station 7 and the Brawley Fault passes between Station 6 and Station 5. 16 important to note that the motion of the wedge-shaped 17 It is block, upon which Station 6 is located, is downward relative 18 to the adjacent blocks. 19 Boore and Fletcher (1980)* note that recordings of 20 an aftershock which occurred to the south of Station 6 and 21 7 indicate that the spectral amplitudes of both P 22 Station 23

* Boore and Fletcher (1980). A preliminary study of selected aftershocks of the 1979 Imperial Valley 25 earthquake from digital accelerations and velocity U.S.G.S. 26 recordings. reprint. Frazier IV-6

and S waves were considerably higher at Station 6 than at Station 7, for a wide range of frequencies. Porchella (,1978)* observes that a swarm of earthquakes in 1977 north of the stat'ions show the peak accelerations recorded near Station 6 to be consistently larger than those recorded near Station 7. Thus, the motion at Station 6 is amplified

to Station 7, basically independent of the 'elative direction of incoming events. Of significance is the observation that the

S waves at Station 6 were delayed by .5 seconds relative to . 10 Station 7, (Boore and Fletcher, 1980**). Station 6 is underlain by material of lower velocity than is Station 7. The lower velocities may be due to the down-dropping of the 13 wedge-shaped block between the Imperial and Brawley Faults which Station 6 located. This would result in lower 15 upon is velocity sediments beneath Station 6 that extend deeper'han beneath either Station 7 or 5. However, due to rapid depositional processes, the surface velocities would likely 18 be beneath of these stations. A laterally 19 similar all heterogeneous earth structure, which exhibits a column of 20 21

22 Porchella, R. Z. (1978). Seismic Engineering Program Report, September-December, 1977, U.S.G.S. Circular 23 762-C, 27 p. Boore and Fletcher (1980). A preliminary study of selected aftershocks of the 1979 Imperial Valley 25 earthquake from digital accelerations and velocity U.S.G.S. 26 recordings. reprint. Frazier IV-7

low velocity material beneath the vicinity of Station 6, would trap obliquely emerging waves by refracting the wave toward the region of lower velocities (i.e., into the wedge). This is believed to be the cause of 'the significant, amplification of vertical accelerations recorded at Station 6. Material attenuation, as cited above, prevented severe amplification of the high frequency S waves, which represent the major contributor to the horizontal ground acceleration. Extensive geologic investigations at the Diablo Canyon site indicate that similar local amplifications of ground motion should not occur at that site. Therefore, such unusual recordings as that for Station 6 do not provide suitable analogs for use in evaluating the seismic criteria at Diablo Canyon. While the peak vertical accelerations recorded at IV-79 were unusually high, little damage occurred. This can partially be explained by two facts. First, the few large vertical pulses are isolated and, while large in amplitude, they are relatively low in energy. Second, the P waves with the high peak vertical accelerations arrive well before the

onset of the S waves. To demonstrate the difference in arrival times, acceleration recordings for Stations 7, 5 and Differential Array (DA) are presented in Figures IV-4, IV-5 and IV-6, respectively, as typical examples of the near field recordings. As noted by the arrows, the peak vertical

Frazier IV-8

acceleration precedes the horizontal peaks by several seconds. In fact, the vertical peaks generally occur well before the onset of strong horizontal shaking. The time difference in PGA between the horizontal and vertical peak accelerations for Stations within 25 km are plotted against distance in Figure IV-7. Positive time differences indicate that the vertical PGA precedes the horizontal PGA. For 7 stations within 10 km, the vertical PGA precedes the horizontal PGA by an average of 1.85 seconds. As was the case in the Hosgri reanalysis for 10 Diablo Canyon, typical seismic analyses assume simultaneous horizontal and vertical shaking. This procedure gives one additional margin indeed the high vertical motions occur 13 if in the absence of large horizontal motions. One method for estimating this margin to compare recorded combinations 15 is of horizontal and ground accelerations. 16 vertical For IV-79, vectorial amplitude of the three components of ground motion has been calculated using only 18 the peak value for each component of motion. This is 19 referred to as "Modulus Peaks" Figures IV-4, IV-5, and 20 of in IV-6. For purposes comparison, the actual modulus of 21 of recorded acceleration calculated at each time along 22 is point the digitized recording. This modulus is referred to as "Time Modulus," the same Figures. For 24 illustrated in stations 10 km the actual Time Modulus 24% lower 25 within is than the Modulus Peaks on the average due phase 26 of to Frazier IV-9 differences in the occurrence of vertical and horizontal peaks.,

This - average margin of 24% between actual recordings and the more conservatively assumed motions with concurrent components is indicative of, but probably underestimates, the margin resulting from procedures used in the reanalysis at Diablo Canyon. More precise appraisals-of the margin of conservatism with respect to actual recordings is difficult to appraise, because motions and stresses at various points in the structures depend on the actual time histories used for input to the analysis. The recordings obtained for the Imperial Valley earthquake are inappropriate for this purpose due to differences in earth properties at the two sites. The point is that observed recordings of the Imperial Valley earthquake do not pose as severe a threat to structures as their peak accelerations might, suggest due to phase differences in the individual components. In summary, the vertical accelerations recorded during the 1979 Imperial Valley earthquake are unusually large due to the wave properties of the deep sedimentary basin. Such large vertical accelerations are not likely .at

ll Diablo Canyon. In addition, the required simultaneous combination of the three components of ground motion for Diablo Canyon conservatively ignores the actual recorded large phase differences between the vertical and horizontal components.

Frazier IV-10

TABLE IV-1

CORRECTED IMPERIAL VALLEY STRONG MOTION RECORDINGS

Distance* Accelerate.on Station Coordinates km Azimuth

32.85 N 230 .463 115.50 W 140 .333 Up .514 32 84o 230 .433 115.49 140 .348 Up 1.52 Aeropuerto 32.65 45 .316 10 115.33 315 .240 Up .179 Bond's Corner 230 32.69'15 .786 34o 140 .587 Up .355 13 Agrarius 32 62o 3 .280 115 30 273 .227 15 Up 32.81 230 .467 16 115 53o 140 .61 17 32.86o 230 .375 115 47o 140 .53 18 Up .441 19 20 21 22 Approximate distance to nearest point on the 1940 Imperial Fault trace. Data from strong motion data recorded in Mexico for the October 15, 1979 Imperial Valley earthquake by 25 J. Brune, J. Prince, F. Vernon III, E. Mena, and R. Simons, 26 preprint. Frazier IV-11

TABLE IV-1 (continued)

Distance* Acce eratzon Station Coordinates km Azimuth Diff. Array 32.80 360 .487 115.54 270 .352 Up .659 Brawley 32.99 315 .221 Airport 115.51 225 .165 Up .153 32 86o 230 .357 10 115 43o 140 .494 Up

.203'3 Holtville 32.81 8'15 .217 115.38 225 .251 Up .228 010 32.78 50 .172 115 57o 320 .226 Up .105 16 Calexico 32 83o '15 .201 225 .275 17 115.49 Up .183 18 19 20 21 22 * Approximate distance to nearest point on the 1940 23 Imperial Fault trace. 1 Data from strong motion data recorded xn Mexico for the October 15, 1979 Imperial Valley earthguake by 25 J. Brune, J. Prince, F. Vernon III, E. Mena,, and 26 R. Simons, preprint. Frazier IV-12

TABLE IV-1 (continued)

Distance* Accelerate.on Station Coordinates km Azimuth

Mexicali 32. 62 13 0 .311 SA Hop 115.42 90 .459 Up .332 32.75 13 230 .382 115 59o 140 .363 Up .140 32.89 13 230 .223 10 115.38 140 .267 Up .132 32 92o 16 230 .414 12 115 37o 140 .316 Up .110 Cucapah 32 55o 85 .310 115 23o 355 Up .115 15 Parachute 32 93o 315 .204 16 115.70 225 .109 Up .156 17 e12 32-72 18 230 .116 18 115.64 140 .142 Up .064

20 21 22 * Approximate distance to nearest point on the 1940 23 Imperial Fault trace. 1 Data from strong motion data recorded an Mexz.co for the October 15, 1979 Imperial Valley earthquake by 25 J. Brune, J. Prince, F. Vernon III, E..Mena, and R. Simons, 26 preprint. Frazier IV-13

TABLE IV-1 {continued)

Distance* - Acceleration Station Coordinates km Azimuth el3 32.71 230 .139 115.68 140 .117 Up .043

32.96 22 230 . 139 115.32 140 .142 Up .044

15 16

17 18

19 20 21 22 * Approximate distance to nearest point on the 1940 Imperial Fault trace. 1 Data from strong motion data recorded in Mexico for the 25 October 15, 1979 Imperial Valley earthquake by J. Brune, J. Prince, F. Vernon III, E. Mena, and 26 R. Simons, preprint. Frazier IV-14

TABLE IV-2

Ratio of peak vertical to peak horizontal accelerations averaged over epicentral distances less than 10 km.

IMPERIAL VALLEY RESULTS

Source Mean Ratio For Mean Ratio For h STRIKE-SLIP Ru ture DIP-SLIP Ru ture 2.5 3.1 3.5 4.5 3.0 3.2 10 8.25 2.1 3.9 11.8 2.0 3.7

13 DIABLO CANYON RESULTS

15 Source Mean Ratio For Mean Ratio For STRIKE-SLIP Ru ture DIP-SLIP Ru ture 16 ~h

3.3 ~ 0.3 0.3 5.0 0.2 0.3 7.0 0.2 0.4 18 9.0 0.2 0.4 19 20 21 22 23

26 Frazier IV-15

Calipatria

~ Brawley 33'00'uperstition Airport gountain ¹1

Brawl ey ~ Parachute Faul t ~ Imperial ¹3 Faul ¹6 t -¹5 ¹4

~ ~¹B¹ Hol tvi 1 le ~ Plaster City DA EPICENTER 1940

811 32'45'Bonds4 e ¹12 Corner

¹13 exi co ~ ,'al

U.S.A- ~ Aeropuerto 'EPICENTER 1979 mexico exicall .~grarius Compuertas 30'ucapahS A Hop 32 Chihuahua

Cerro Prieto 0< 10 20 '0 (km) Delta

115~45'15~30'15~15'. Victoria

— Fault;-.trace motion stations the Figure Frazier IV-:1.. .and strong ' .in Imperial Valley;- (U.S.,G;S. -Open-File Report 79.-1654) .: „- -.:;;.. =.:*:..

2 ;

3 1

I ,E 5

CL 1 C)

10 l

0 1 2 3 4 5 6 7 P-wave velocity (km/sec} Imperial. Valley,, P —————'Di'ablo Canyon

Figure Frazjer .lV-. 2...A, comparison .of .two,Southern California -eaIpth structures. The, solid line„is.-representative,oi.".the 'top 11 km.at... imperial Yalley., The'dashed: 1:ice-i's an'.earth.stI.ucture', with ~ less pronounced velocity gradient, as- characteristic of,D'diablo Canyon,

Velocity

+J CL QJ Ch

Figure Frazier IV-3. .Three rays emerging from a source within a vel'ocity gradient. Rays A and B are direct rays. Ray C is a PP surface multiple- trapped in the velocity gradient zone.-- Ray D was a take-off angle steep enough to allow it to escape into the -underlying bedrock.

STATION 7

Horizontal ,46 g

.75

o ~ g Vertical +3 n5 S- tD

QJ O O

-.75 Horizontal .33 g

Modulus of Peaks for 3 acceleragrams = .77g .75

O ~g = Maximum Time.'Modulus .52 g A5 S- QJ

QJ O O

8 12 20 Time {sec)

Figure Frazier IV-4. The top three traces show the horizontal and vertical components of acceleration recorded at Station 7. The small arrows indicate the peak accelerat'ions for each recording. The. bottom trace is the Time Modulus, or vectorial combination of the top three recordings. The maximum Time Modulus '.52 is g. This compares with a modulus of .77 g for the three peak accelerations, represented .by the solid line.

STATION 5

Horizontal .37 g

.75

o I Vertical .44 g 5- ( 4!ld~L~'phd~ I

CJ tJ

-.75

Horizontal .53 g

Modulus of Peaks for 3 accelerograms = .78 g. .75 t le = S Maximum Time Modulus ..54 g CU Cl CJ

12 16 20 Time (sec)

Figure Frazier IV-5. The top three traces show the horizontal and vertical components of acceleration recorded at Station 5; The small arrows indicate the peak accelerations for each recording. The bottom trace is the Time Modulus, or vectorial combination of the top three recordings. The maximum Time Modulus is .54 g. This compares with a modulus of .78 g for the three peak accelerations, represented by the solid line. ~ ' STATION D.A. Nfl~~Horizontal .49 g .75

O I Vertical -66 g AS S- Cl

CJ O /

-. 75

Horizontal .35 g

Modulus of peaks for 3 accelerograms = .89 g

.75

O Maximum Time Modulus =.,66 .g

12 16 20- Time (sec)

Figure Frazier IV-6, The top three traces show the horizontal and vertical components of acceleration recorded at Station D.A. The small'rrows indicate the peak accelerations for each recording. The bottom trace is the Time Modulus, or vectorial combination of the top. three records. The maximum Time Modulus is .66 g. This compares with a'modulus of .89 g for the three peak acce'terations, represented by the solid line.

0 ~

~ ~ QJ O. ~ ~ 0 ~P4 UJ 0

QJC3

10. D I STANCE Kl'1

Figure Frazier IV-7.. Time difference, TH - TV, between the peak horizontal and peak vertical accelerations for the 1979 Imperial Valley earthquake, plotted as a function of fault distance. For all stations within 20 km, the vertical peaks precede the horizontal'peak accelerations..

FINAL'EPORT EVALUATIONOF PEAK HORIZONTAL GROUND ACCELERATION ASSOCIATED WITH THE HOSGRI FAULT AT DIABLOCANYON NUCLEAR POWER PLANT

Submitted to: Mr. F. W. Brady Pacific Gas & Electric Company 77 Scale Street San Francisco, California 94I06

August l980

TERA CORPORATION

2150 Shattuck Avenue Berkeley, California 94704. 415 845 5200

Berkeley, California Dallas. Texas Bethesda. Maryland Baton Rouge, Louisiana Del Mar, California New York. New York San Antonio. Texas Denver, Colorado Los Angeles. California

TABLE OF CONTENTS

Section ~Pa e

I.O INTRODUCTION.

I;I Summary of Results . ~ ~ I-3

l.2 Summary Conclusions . ~ ~ I-7

2.0 DATA AND ANALYSIS TECHNIQUES 2- I

2. I Near-Source Data Base ~ ~ 2-I

2.2 Selection Criteria for DCNPP ~ ~ 2-3

2.3 Description of Data Base. ~ ~ 2-7

2.4 Regression Analysis . ~ ~ 2-l2

2.5 Physical Considerations ~ ~ 2-l4 3.0 ANALYSISAND RESULTS 3-I

3.I DCNPP Data Base Regression ~ ~ 3- I

3.2 Site Geology Effects ~ ~ 3-4

3.3 Fault Type . ~ ~ 3-6

3.4 Building Effects ~ ~ 3-8

4.0 SENSITIVITY RESULTS .

4. I Model Variations ~ ~ 4- I

4.2 Far-Field Decay Rate . ~ ~ 4-7

4.3 Focusing Potential ~ ~ 4-9

4.4 Well-Recorded Earthquakes ~ ~ ~ ~ ~ ~ 4-l4

4.5 Alternative Data Partitioning - Small Building Analysls ~ ~ 4-l5

5.0 COMPARISONS AND CONCLUSIONS. 5-I

5.l Physical Characteristics . ~ ~ 5-I

5.2 Comparison with Other Studies . ~ ~ 5-8

5.3 Statistical Characteristics . ~ ~ 5-l2

5.4 Conclusions ~ ~ 5- I 8

6.0 REFERENCES . 6-I

APPENDICES

A WEIGHTED HISTOGRAMS SHOWING DISTRIBUTION OF DATA FOR THE DCNPP DATA BASE . A-I

THE DCNPP DATA BASE . B-I

TERA CORPORATION

LIST OF TABLES

Table No. Pocae

2-I Geological Site Types Included in the DCNPP Acceleration Data Set 2-6

3-I Distribution of Earthquake Recordings Within Appropriate Dtstance0 t Ranges...... 3-3 3-2 Classification of Horizontal PGA According to Earthquake Fault Type...... 3-7 3-3 Reduction in Free-Field Horizontal Accelerations Due > to the Presence of Embedded Structures ...... 3-I I 3-3A Recording Station Descriptions ...... 3- I 2 3-4 Reduction in Free-Field Horizontal Accelerations Due to the Presence of Non-Embedde'd Structures...... 3- I 3 3-4A Recording Station Descriptions 3-l4

3-5 Reduction in Horizontal Accelerations Due to the Embedment of Structures . 3-l5

3-5A Recording Station Descriptions ...... '- I 6 4-I Sensitivity Results for Model Variations...... 4-4 4-2 Sensitivity Results for Variations in Far-Field Decay Rate...... 4-8 4-3 List of Earthquakes for which Near-Field Focusing at One or More Recording Stations Has Been Reported ~ ~ ~ ~ ~ ~ 4 I I

TERA CORPORATION

LIST OF FIGURES

Predicted Peak Accelerations for M=7.5 -. Physical Model ~ ~ ~ ~ I 5

I-2 Predicted Peak Accelerations for M=7.5- Statistical Model I-6

2-I Weighted Histogram - Showing Distribution of Magnitude 2-9

2-2 Weighted Histogram — Showing Distribution of Significant

Distance ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 2-IO

2-3 DCNPP Data Base-Magnitude vs. Distance 2-II

3- I Comparison of High Frequency Spectral Amplitudes- Imperial Valley I 979 Free-Field Stations 3-I 7

4-I Effect of Data Base Selection on Magnitude Scaling at S.8 km-

Physical Model o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 4 S

4-2 Effect of Data Base Selection on Magnitude Scaling at S.8 km- Stat istical Model. 4-6

4-3 Test for Focusing - Imperial Valley I 979 Data . 4-l2

4 4 Test for Focusing —San Fernando Data 4-l3

5-I Predicted Peak Accelerations for M=7.5 Showing Distribution of Magnitudes . ~ ~ ~ ~ 5 2

5-I A Predicted Peak Accelerations for M=6.5 Showing Distribution of lagnitudes . ~ ~ ~ ~ 5 3

5- I B Predicted Peak Accelerations for M=7.0 Showing Distribution of Magnitudes . ~ ~ ~ ~ S 4

5-I C Predicted Peak Accelerations for M=7.5 Showing Distribution of Magnitudes . 5-5

5-2 Predicted Peak Accelerations for Distance = 5.8 km 5-6

5-3 Predicted Peak Accelerations for Distance = 5.8 km- Data Less Than IO km ~ ~ ~ ~ 5 7

Imperial Valley I 979 Data Compared to Statistical Model at Ms=6.9 ~ 5-9

IV TERA CORPORATION I LIST OF FIGURES

(CONT.)

ure No. ~Fi ~Pa e

5-5 Compar ison with SAM-V 5-IO

5-6 Comparison with USGS Circular 795 Statistical Model at Ms=6.9 (ML=6.6)...... 5-I I 5-7 Normalized Weighted Residuals vs. Distance (Statistical Model).... 5- l3 5-8 Normalized Weighted Residuals vs. Magnitude (Statistical Model).... 5- I 4 5-9 Normalized Weighted Residuals vs. Predicted Value (Statistical

o ~ ~ ~ ~ ~ ~ Model) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 5 I 5 5-IO Distribution of Residuals (Statistical Model)...... 5-l6 5-I I Normal Probability Plot...... 5-l7

v TERA CORPORATION ' I.O INTRODUCTION

The objective of this study was to estimate the horizontal. peak -ground acceleration (PGA) for large earthquakes which could be produced in the near- source region for the range of conditions appropriate at the Diablo Canyon Nuclear Power Plant (DCNPP). The technical approach consisted of the following steps:

ao The development of an extensive near-source horizontal peak ground acceleration data base representing a wide range of parameters addressing properties of the earthquake source, the travel path, and the recording station. b. Determine the set of parameters most relevant to estimates of peak'orizontal acceleration at DCNPP through a detailed statistical and interpretive analysis of these data.

co Determine the PGA for DCNPP by performing linear and nonlinear regression analyses on these data using the above-selected parameters as independent variables. Per- form sensitivity analyses on important parameters and certain physical models.

A near-source data base was assembled from worldwide earthquake recordings using selection criteria which assured high quality data in the range of magnitudes and distances of interest at DCNPP. The data base consists of 229 horizontal peak ground acceleration recordings from 27 earthquakes, defined in terms of source, travel path and recording site characteristics such as magnitude, closest distance to the fault'rupture surface, site geology, instrument type and location, and size of structure. The data were weighted, by earthquake; through several distance intervals to eliminate potential bias from well-recorded events such as the l979 Imperial Valley and the l97I San Fernando earthquakes. These data were used in conjunction with both physically constrained and purely empirical distance attenuation models whose parameters were determined using both linear and nonlinear regression techniques.

TERA CORPORATION

Sensitivity studies were performed, both to test key assumptions in the analyses and to identify important parameters in the data base. Calculations and important conclusions also were made on partitioned sets of the data base to check for biases. For example, a systematic conservative bias was observed in. non-North American earthquake data. This is important for two reasons: first, limited North American data are available for very large magnitude earthquakes in the .distance range of interest (less. than IO kilometers), and therefore, prediction of PGA for DCNPP would have to be based on extrapolation if non- North American data were not used; and'second,'onclusions regarding the saturation of PGA with increasing magnitude are modified by the exclusion of non-North American data.

Because of the lack of substantial data within 3 kilometers of earthquake sources and since, for simplicity, we did not analyze data further than 50 kilometers from earthquake sources, physical boundary conditions appropriate to these two regions were developed and tested. While the near-source data are adequate for" predicting horizontal PGA at DCNPP without these boundary conditions, the inclusion of informati'on outside the range of these data provides added confidence to the conclusions of this study. The two boundary conditions were first, that the far-field slope be constrained to l.75 based upon the studies of other investigators, and second that the PGA at zero distance be constrained to a constant value independent of magnitude, based on widely accepted models of earthquake dynamics.

I-2 TERA CORPORATION

I. I SUMMARYOF RESULTS

We present in Figures I-I and I-2 the results of our analysis for magnitude 7.5. Figure I-I displays the results using the above boundary conditi'ons (termed, henceforth, the Physical Model) in terms of the median, and median plus and minus one-standard-deviation predictions. Figure I-I also shows the data used for this analysis, normalized by the physical model to M 7.5. The identification of non-North American data in this plot suggests a conservative bias to these accelerations. Note that the Physical Model yields a median peak horizontal acceleration for M 7.5 at 5.8 kilometers (the appropriate magnitude and distance of concern for DCNPP) of 0.4I g. Because the regression analysis established that the median plus one-standard-deviation value is l.52 times the median, the median plus one-standard-deviation acceleration is 0.62 g.

Figure I-2 presents similar results for our Statistical Model--a model identical to the aforementioned but without the two physically modeled boundary conditions. Due to the greater influence of the systematically biased non-North American data, this model yields slightly higher. accelerations at the same distance. For M 7.5 at 5.8 kilometers, this model a median accelera- s predicts tion of 0.48 g and a median plus one-standard-deviation of 0.72 g. We will show in a later section that purely statistical treatments of the data yield results very consistent with the Physical Model. We provide, however, in this summary, results based on all the data to establish the technical breadth for our conclusions. In order to provide specific insight into the breadth of the conclusions, we note that the analysis includes some of the largest accelerations ever recorded. These accelerations include Tabas, Iran, 0.80g; Gazli, USSR,

0.8I g and 0.65 g; Koyna, India, 0.63 g and 0.49 gg and l979 Imperial Valley, 0.8I g and 0.66 g. For completeness, we have included an acceleration of 0.73 g inferred for the S25 E component from a seismoscope record at USGS Station IOI3 during the I966 Parkfield earthquake (Trifunac and Hudson, l970), which aguments the 0.5I g peak acceleration recorded for the N65 E component. Many of these data are suspected to be biased on the high side, such as the non-North American data (see Section 4.0). All these data were included in the analysis, in

I-3

TERA CORPORATION 1 t spite of their upward bias, to provide completeness and an added measure of conservatism.

I-4

TERA CORPORATION e

0 E3

NQTE: DATA NQRHALIZEO TQ Hs ~ 7. 5 5 NON NQRTH AHERICA MEOIAM - PHYSICAL MOOEL + I 5 I GMA O 5 I GMA O 10. OISTRNCE tKH)

F I GURE 1 -1

PREDICTED PERK RCCELEHRTIQNS FUR 8=7. S

E3 +o CC CC hJ

LLj

(Z tv'(AT C: OATA tlORMALI ZEO TQ M 5 = 7. S l5 NUN NORTH AMERICA HFOIAN - STATISTI'AL HOOc,l +I SIGMA --l SIGMA o o '0. OESTRNCE tKN)

FIGURE 1 -2

PREOICTEO PERK ACCELERRTIQNS FQR:8=7. 5 0 I.2 SUMMARYCONCLUSIONS

Based upon the results of our studies, we have reached the following conclusions related to peak horizontal ground accelerations in the near-source region of large earthquakes:

The predicted median peak instrumental accelerations of 0.4I g (Physical Model) and 0.48 g (Statistical Model) for a magnitude 7.5 earthquake at 5.8 kilometers firmly establish the conservatism of the DCNPP seismic design.

The results of this study are consistent with the data recorded from the I 979 Imperial Valley earthquake, although the earthquake was only one of 27 used in this analysis. Our results are essentially unaffected by removing the earthquake from the analysis.

Accelerations recorded on sedimentary rock (our Soft Rock classification)~consistent with the DCNPP site, are lower than those recorded on either soil or Hard Rock, confirming the conservatism of incorporating both soil and Hard Rock in predictions of PGA for DCNPP.

The results of this study establish that accelerations tend to saturate with increasing magnitude. This conclusion is more firmly established when certain conservatively biased data are removed from the analysis. Both the l979 Imperial Valley data and the results from this analysis show that accelerations saturate with decreasing distance. This firmly establishes that linear extrapolation from far-field data is inappropriate.

In the next section of the report, we first present the overall acceleration data base assembled for this study. We then describe the selection criteria that resulted in the DCNPP data set. The section concludes with a presentation of the regression models employed in the analysis. Section 3.0 presents the basic results developed from the analysis and provides the basis for the parameter selection. Section 4.0 of the report summarizes the sensitivity analyses we have performed. Finally, Section 5.0 presents a comparison with other studies and our conclusions. The appendices to the report provide additional detail regarding the acceleration data base.

I-7

TERA CORPORATION

2.0 DATA AND ANALYSISTECHNIQUES

In this section are described the organization, selection criteria and analysis of an extensive acceleration data base compiled for this study. This overall data base, termed the Near-Source Data Base in this study, consists of acceleration data recorded within 30 and 50 kilometers of a worldwide set of earthquakes with shallow. rupture. On the basis of various physical and statistical arguments, components of this data base were selected for analysis of peak acceleration at the DCNPP. This data base is termed the DCNPP Data Base. The selection criteria (Section 2.2), the data base (Section 2.3) and the analysis techniques (Section 2.4) are described below.

: 2. I NEAR-SOURCE DATA BASE

The Near-Source Do{a Base, is an extensive collection of over l,000 acceleration- distance-magnitude data points from l66 earthquakes. The data base contains earthquake information concerning the event, such as event name, date, time, latitude, longitude, quality of the location, local Richter magnitude, body-wove magnitude, surface-wave magnitude, depth, and focal mechanism. The station information in the data base includes the site of the strong-motion instrument, such as USGS station number, latitude, longitude, structure type, size classifications, instrument type and location, owner, geology and station name. The strong-motion portion of the data base contains the largest and smallest horizontal acceleration components, earthquake identification, date, USGS station number, epicentral distance, and significant distance.

This 'data base represents, to the best of our knowledge, all available published peak acceleration data, recorded in the United States through March I 979, that meet the following criteria: 'I

Errors in earthquake location less than 5 kilometers.

Distances within 20, 30, and 50 kilometers for magnitudes less than 4.75, between 4.75 and 6.25, and greater than 6.25, respectively.

2-I

TERA CORPORATION

~ Earthquakes with rupture surfaces within 25 kilometers of the surface.

~ Accelerograms having a minimum PGA of 0.02 g, which triggered early enough in the record to capture the strong phase of shaking.

~ Accelerograms recorded on instruments either in the free field, in the basement of buildings, or on the ground level of structures without basements. Recordings from upper levels of buildings were specifically excluded, as were ~ records from instruments in the upper basements of buildings.

In addition, several significant earthquakes which occurred either outside the United States or since March l979, and which also met the selection criteria were included: the August 6, l979 Coyote Lake (ML 5.9) and October l5, l 979 (M 6.9) earthqaukes in California; the December I 0, I 967 Imperial Valley s India earthquake (M 6.5); the July I 967 Fairbanks, Alaska earthquake Koyna, s 2l, l (mb 5.6); the December 23, l972 Managua, Nicaragua earthquake (M 6.2); the Alaska earthquake (M 7.6); the October 3, l974 Lima, Peru July 30, l972 Sitka, s earthquake (M 7.6); the May l7, I 976 Gazli, USSR earthquake (M 7.0); the I 978 St. Elias, Alaska earthquake (M 7.2); and the September l6, l978 Tabas, Iran earthquake (M 7.7).

The Near-Source Data Base was developed without any restriction on either the age of the record, the recording instrument, the recording site geology, the tectonic province of the earthquake, the earthquake fault type, or the earthquake magnitude.

The primary sources for the strong-motion data were the Department of Commerce's annual publication, United States Earthquakes, and the U.S. Geological Survey's quarterly Seismic Engineering Program Reports. Information on earthquake locations and magnitudes was obtained from local seismic networks whenever possible, such as the seismological centers at the California Institute of Technology (Southern California) and University of California at Berkeley (Northern California). Other major sources included NEIS's Preliminary Determination of Epicenters and the Bulletin of the International Seismological Centre. In addition, many special reports and papers in journals were used either to verify original sources or to develop additional related data.

2-2 TERA CORPORATION

2.2 SELECTION CRITERIA FOR DCNPP

The criteria used in defining the DCNPP Data Base considered appropriate for the analysis of peak acceleration for the site are described below.

Peak Acceleration

Except for a few older recordings, peak accelerations scaled from the actual recorded accelerograms were used in the analysis, wherein each horizontal component was treated as a separate observation. The data used included undigitized and unprocessed accelerograms for two reasons: first, the limited amount of digitized data precluded their exclusive use in near-source studies of strong ground motion; and second, peak accelerations scaled from the processed accelerograms are generally smaller than those scaled from the original accelerograms due to the 0.02 second decimation of the records. For, some older recordings, we chose to account for the questionable quality of the reported unprocessed peaks by using'the digitized and corrected records. In every case, this resulted in the use of a higher peak acceleration compared to the unprocessed data, thereby addressing this uncertainty in a conservative manner.

Macenitude

A magnitude (M) consistent with the moment-magnitude scale (Hanks and

Kanamori, I 979) was used as a uniform basis for characterizing earthquake size. Surface-wove (M was used when both local magnitude and magnitude s ) (ML) surface wave magnitude were greater than or equal to 6.0. Local magnitude was were below this value. Where M or used for earthquakes where both magnitudes s ML was not available, an appropriate value was estimated based upon empirical relationships among magnitude scales. The analysis was restricted to earthquakes of magnitude (M) 5.0 or greater. Since accurate significant distances could not be determined for earthquakes of smaller magnitudes, we restricted the analysis to the range of magnitude most appropriate for predictions of peak accelerations at DCNPP, magnitude (M) 5.0 or greater.

2-3

TERA CORPORATION . ~ Source-To-Site Distance

Peak acceleration data were restricted to recording stations for which an accurate estimate of significant distance (the shortest distance between the station and a fault rupture surface) was available or could be determined. These distances were computed from surface fault expressions and areal-depth distri- butions of aftershock sequences. Data were selected if distances were within 30 kilometers for M ( 6 I/4 and within 50 kilometers for M ) 6 I/4. The significant distance of 5.8 kilometers for DCNPP was determined using the closest distance to the surface trace of the Hosgri fault.

Site Geolo

Consistent with the overall approach described in Section l.0, we have included accelerations from a range of site conditions in the DCNPP data set. Although we believe that the site geology at DCNPP is best represented by sedimentary rock (our Soft Rock classification), we included several other site types in our DCNPP data set so that statistical trends between the site types could be examined. This examination would then provide a basis for the final selection of site types relevant to estimating peak accelerations at DCNPP. Stations known to be situated at sites underlain by shallow soil deposits or extremely soft soils are not consistent with site conditions at DCNPP and were not included in the data base. Statistical analysis has shown that the accelerations recorded at these sites are significantly different from those recorded on the other site types. The site types included in the analysis are summarized in Table 2- I. The Pacoima Dam record of the San Fernando earthquake was specifically excluded from the analysis for two reasons. First, the site experienced extreme, topographic amplification (Boore and Zoback, l974). Second, the large gradation in wave propagation velocities near the surface combined with low damping (Duke et al., l97 I) creates a condition of high frequency resonance, thus placing the site in a very suspect category.

. 2-4

TERA CORPORATION

Instrument Location

Peak acceleration data, recorded on instruments located in basements of buildings and at ground level were selected for analysis. Ground-level instruments included those located on the ground level of buildings without basements, free-field stations housed within small shelters, and a few located near the abutments of dams. The Koyna Dam record was actually located in the lower gallery within the dam. This recording was used in the analysis, since it was believed to be representative of the motion at the foundation of the dam (Krishna et al., 1969).

ln order to mimimize any possible bias associated with the effect of large buildings during the 1971 San Fernando earthquake, we have used the San Fernando data reported by Boore, et al. (1978) in Circular 795. They applied criteria that selected only a few stations from densely instrumented locations, such as down'town Los Angeles, so as to obtain a reasonable distribution of site types, distan'ces, and instrument locations.

~De th

Consistent with regional seismicity and the crustal thickness of the Diablo Canyon site, only shallow-focus earthquakes —those which ruptured within 25 kilometers of the surface--were used in the analysis.

Earth uake Location

ln accordance with the overall approach presented in Section 1.0, we have included a worldwide set of earthquakes in the data base. Subsequent sections in this report will address the sensitivity of our results to the inclusion of these earthquakes, especially the non-North American events.

2-5

TERA CORPORATION

TABLE 2-I

GEOLOGICAL SITE TYPES INCLUDED IN THE DCNPP ACCELERATIONDATASET

Site Geolo Descri tion Classif ication

Recent Alluvium Holocene Age soil deposits with rock > IO m deep

Pleistocene Deposits Pleistocene Age soil deposits with rock > IO m deep

Soft Rock Sedimentary rock, soft volcanics and soft metasedimentary rock

Hard Rock Crystalline rock, hard volcanics and D hard metasedimentary rock

2-6

TERA CORPORATION

2.3 DESCRIPTION OF DATA BASE

Application of these criteria resulted in the selection of 229 horizontal components from 27 earthquakes considered appropriate for the analysis of peak acceleration at DCNPP. The distribution of these data, with respect to magnitude and distance, is displayed in Figures 2-I and 2-2. These figures represent "weighted" histograms, in which the relative frequency is measured with respect to the weight each point carried in the regression analysis. (The weighting scheme will be discussed in detail in Section 3.I.)

The distribution of magnitude with respect to distance is presented in the scattergram of Figure 2-3. This plot demonstrates a good variation of earthquake magnitudes over the distance of approximately 3 to 30 kilometers. This is important to our results since multiple regression analyses of horizontal PGA are made with magnitude and distance as independent variables.

The distribution of other earthquake and recording parameters may be found in Appendix A. These histograms show that the strong-motion data selected for analysis at DCNPP are predominantly recorded by modern strong-motion instruments at the ground level of buildings I to 2 stories high. The predominant focal depths of these events fell within the range of S to IO kilometers.

The selected set of earthquake and strong-motion data used in the analysis of peak acceleration at DCNPP is tabulated in Appendix B. In order to maintain a high standard of quality and consistency in both the data and the analyses, the selection process (described above) resulted in the elimination of many of the older earthquake recordings obtained since l932 at El Centro, Long Beach, Hollywood, Hollister, Ferndale, Eureka and other long-term recording stations.

Reasons for not including many of these older recordings included imprecise magnitude determinations, late instrument triggering, and inaccurate locations. Due to limited distribution of local seismometer networks in southern and northern California prior to two decades ago, magnitude determinations generally were reported to the nearest one-half magnitude unit, which by our analyses

2-7

TERA CORPORATION 0 would represent an unacceptable variation in predicted peak acceleration. The triggering mechanism of the older USGS strong-motion instruments caused relatively large trigger delays, with the result that many of the older :.near-source recordings begin well within the strong phase of shaking. This, of course, results in unreliable estimates of peak acceleration from these recordings. The lack of an adequate distribution of seismometers resulted in errors in epicentral locations of I5 kilometers or greater for certain older earthquakes. Most often, focal depths could not be sufficiently determined from these data and were therefore constrained to l6 kilometers in order to locate the epicenter. Such errors are unacceptably large for meaningful analyses of peak accelerations within 30 to 50 kilometers of the source, and therefore, were not used in our near-source studies. Furthermore, the unavailability of aftershock data of sufficient quality and completeness precluded the determination of significant distances for many of these older recordings.

We emphasize, however, the magnitude range of about 5-I/2 to 6-I/2 associated with these older earthquakes that have been excluded is well represented in the selected near-source data base. Therefore, their exclusion is not expected to significantly affect the predictions of peak acceleration at DCNPP offered by our analyses.

We believe that the application of the above criteria has resulted in the selection of a set of consistent, high-quality peak accelerations appropriate for the analysis of ground motion at DCNPP. As we will show later in this report, the abundance of near-source data meeting the high standards established by these criteria result in statistically robust estimates of peak acceleration at the magnitudes and distances associated with the Hosgri Fault.

2-8 TERA CORPORATION

DISTRIBUTION OF MAGNITUDE

40 Uz

30 29%

22% 22% 20

IO I I%

4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0

MAGNITUDE

WEIGHTED HISTOC.RAM

FIGURE 2- I

2-9 TERA CORPORATION

DISTRIBUTION OF SIGNIFICANT DISTANCE 30

27 28%

2l zU IS Cf ) I5 I4%

9% 9% 9%

0 5 IO I5 20 25 30 35 40 45 50

SIGNIFICANT DISTANCE (km)

WEIGHTED HISTOGRAhh

FIGURE 2-2

2- I 0 TERA CORPORATION 'e {9 NON NORTH RHERTCR

LtJ Za e (A M P C3

CI C) 4. 00 s. oo s. aa v. oa NRGNITUOE

F IGURE 2-3

DC'NPP ORTR BASE NRGNITUDE VS. DISTRNCE

2.4 REGRESSION ANALYSIS

The general attenuation function used for modeling near-source peak accelerations at DCNPP is expressed by the following expression:

PGA = a e R + C(M) (2- I) where PGA is peak ground acceleration in gs, R is significant distance in kilometers, and M is magnitude. This functional form was selected because, when used with nonlinear regression analyses, it is capable of modeling possible magnitude and distance saturation effects that may be suggested by the data. A form with properties similar to this was originally proposed by Esteva (l970). Our investigation of this form is unique in that we employ nonlinear regression techniques to quantitatively evaluate all the coefficients in the equation. While the coefficients in this expression can be determined directly from the data, we chose to investigate both a Physical Model, which would allow predictions outside the range of data used in this analysis, and a Statistical Model, which would allow us to statistically test whether PGA saturates with decreasing distance and increasing magnitude. After discussing several important characteristics of these models, we show (in the next section) that the predictions at DCNPP are rather insensitive to the assumptions used in these models.

The coefficient a in Equation 2-I scales the amplitude of the peak acceleration at magnitudes and distances equal to zero. The coefficient b controls magnitude scaling of PGA at large distances (R77C). The coefficient d controls the rate of decay of PGA with distance at large distances. The C-term 'provides an added degree of freedom by allowing the magnitude scaling of PGA to be a function of distance in the near-source region.

Again, following Esteva (l970), we characterize the C-term as an exponential function of magnitude:

c2M C(M) = cl e (2-2)

2-12

TERA CORPORATION

C modulates the rate of decay of PGA with distance (for distances close to the fault, where little attenuation is expected based on geometric considerations). The distance at which this effect takes place is a function of the size of the fault rupture surface —especially fault length for larger shallow-focus events. Since fault rupture dimensions scale exponentially with magnitude, one would expect C also to scale exponentially with magnitude and, thus, take the form of Equation 2-2. The sensitivity of the predictions to the form of this function are discussed in Section 3.0.

There are several reasons why the expression given in Equation 2-1 was chosen to model near-source attenuation of peak acceleration. Its strongest asset is its ability.to model both near-source and far-source motions by independently scaling both magnitude and distance decay-rate in both of these regions. The C-term allows (but does not receuire) possible saturation of PGA at very small distances. This natural limit of PGA at small distances has been proposed by seismologists and geophysicists, based on the physics of the earthquake generating process (Hanks and Johnson (l 976), Boore (l 974), Kanamori (l 977), Ambraseys (l 978), Aki and Richards (l980), Jennings and Guzman (l 975)). From a.purely statistical viewpoint, this model incorporates all the features of other empirical models using a minimum number of parameters. A comparison of results obtained by Equation 2- I and another model similar to that recently proposed by Donovan and Bornstein (l 978) is presented in Section 4.0.

2-l3

TERA CORPORATION

2.5 PHYSICAL CONSIDERATIONS

As indicated earlier, we analyzed the DCNPP acceleration data in two ways. The first, the Statistical Model, involved straight regression analysis to statistically determine the coefficients in Equation 2-I. In a second analysis, the Physical Model, we used regression'echniq'ues together with two widely- accepted ground motion characteristics to test hypotheses on magnitude and distance saturation of PGA at near-source distances.

Because this study was not directly concerned with predict'ing far-field ground motions, peak accelerations recorded farther away than 30 to 50 kilometers from the source were not included in the data base. However, in order that our results may be compared with others and give more realistic predictions at larger distances, the far-field slope d was constrained to a value of l.75 in the Physical Model. This value was selected from a survey of published attenuation relationships as being representative of the far-field decay rate of PGA. As will be discussed in Section 4.0, the predictions appropriate at DCNPP for a distance of 5.8 kilometers are insensitive to the value of d for values ranging from about I.O to 2.0.

The second constraint involves the prediction of PGA at significant distances closer than 3 kilometers, where strong-motion data are extremely limited. As indicated earlier, it is the understanding of most seismologists and geophysicists that, at or very near the rupture surface, peak accelerations are essentially independent of earthquake magnitude. The physics of the rupture process would suggest that PGA is controlled by dynamic stress-drop, which is related to rock strength, and consequently independent of magnitude. Additionally, for sites near the rupture surface, PGA essentially would become independent of the size of the ruptured surface and thus independent of magnitude. Based on these physical arguments, another boundary condition was included in the Physical Model requiring a constant peak acceleration, independent of magnitude, at the

2-l4 TERA CORPORATION fault rupture surface (i.e., a significant distance of R=O). This condition requires that the parameter c2 in Equation 2-2 be given by the expression

(2-3)

2-l5

TERA CORPORATION

3.0 ANALYSISAND RESULTS

As indicated in Section I.O, the approach developed for this study was to establish a DCNPP acceleration data set representing a range of earthquake and site parameters. A final assessment of the accelerations for DCNPP follows from a detailed analysis of these parameters with regard to their relevance to predictions of PGA. In the following sections we assess the significance of these remaining parameters to show that their inclusion in the analysis is either conservative or null. ll

3.I DCNPP DATA BASE REGRESSION

In this section we present the results of a regression analysis on the DCNPP Data Base (Section 2.0) for both the Physical and the Statistical Models.

Consistent with the assumption that PGA is distributed lognormally, the regression analysis was performed on the logarithmic form of Equations 2-I and 2-2, as given by

In PGA = In a + bM - d In (R + c e 2 ) I

Due to the nonlinear form of Equation 3-I, the coefficients were determined from a nonlinear, multiple regression technique in which both M and R were treated as independent variables.

Weights were assigned to each recording in order to reduce the bias of the few well-recorded earthquakes in the data base. It was decided that these weights should depend on distance so as to account for the added information on attenuation represented by data that are well-distributed with respect to distance.

In order that each earthquake be given equal representation in the analysis, a relative weighting factor of I/n- was used, where n- is the total number of acceleration components for the i earthquake within the j distance range.

3-I

TERA CORPORATION

The weights were then normalized so that their sum totalled the number of components ysed in the analyses. This assured that the statistics of the analyses would represent the correct number of degrees-of-freedom. The definition of the distance ranges and the distribution of earthquake recordings w'ithin each range are presented in Table 3- I.

The results of the regression analysis for the median value of peak horizontal acceleration yield the following'expression for the Physical Model:

- ' l.75 PGA = 0.0I I6e 'R+ O.l I5e (3-2) where d was constrained to a value of I.75 and c2 was defined by Equation 2-3 as discussed in the previous section. The median plus one-standard-deviation value of PGA may be obtained by multiplying the median value by a factor of l.52.

The results of the regression analysis when all coefficients in Equation 2-4 are determined by the analysis yield the following expression for the Statistical Model:

l.09 PGA p p I58e0.868M (R + p p6p5e0.700M)- (3-3)

The median plus one-standard-deviation value of PGA may be obtained by multiplying the median value by a factor of I.50.

These results for M 7.5 at 5.8 kilometers yield values of 0.4I g and 0.62 g for the median and median'plus one-standard-deviation values of PGA for the Physical Model, and 0.48 g and 0.72 g for the median and median plus one-standard- deviation values of PGA for the Statistical Model. Plots of Equation 3-3 as a function of magnitude and distance showing the data, as well as various plots of the residuals, are displayed and discussed in Section 4.0, Comparisons and Conclusions.

3-2

TERA CORPORATION C LE 3-1

DISTRIBUTION OF EARTHQUAKE RECORDINGS WITHINAPPI

Distance Range No. of Distance Range No. Earthquake Ear of (Km) Components (Km) thquake Components

0 - 2.4 Parkfield 1966 2 14.1 - 19.9 Parkfield 1966 2 Imperial Valley 1979 12 Fairbanks, Alaska 1967 2 Lytle Creek 1970 2 2.5 — 4.9 Tabas; Iran 1978 Santa Barbara 1978 2 Koyna, India, 1967 Malibu 1979 2 Gazli, USSR, 1976 Coyote Lake 1979 4 Coyote Lake 1979 Imperial Valley 1979 4 Imperial Valley 1979 San Fernando 1971 14 20.0 — 28.2 Santa Barbara 1978 5.0 - 7.4 Long Beach 1933 2 Daly City 1957 Park field 1966 2 Lytle Creek 1970 Managua 1972 (m 5.6) 2 Point Mugu 1973 Coyote Lake 197k 2 Bishop 1978 Imperial Valley 1979 10 Coyote Lake 1979 Long Beach 1933 7.5 - 9.9 Helena, Montana Malibu 1979 Daly City 1957 Imperial Valley 1979 Parkfield 1966 San Fernando 1971 20 Holiister 1974 28.3 — 40 .0 Bear Valley 1972 2 OroviI le 1975 Peru 1974 2 Bishop 1978 Lima, St. Elias, Alaska 1978 2 San Fernando 1971 Creek 1970 4 Coyote Lake 1979 Lytle Bishop 1978. 4 Imperial Valley 1979 Imperial Valley 1979 4 Santa Barbara 1 978 Oroville 1975 . 6 San Fernando 1971 10 10.0 — 14.0 Imperial Valley 1940 2 Santa Barbara 1941 2 40.1 - 56.6 Kern County 1952 2 Santa Barbara 1978 2 Borrego Mountain 1968 2 Holi ister 1974 4 Sitka, Alaska 1972 2 Daly City 1957 6 Lima, Peru 1974 2 Imperial Valley 1979 10 Imperial Valley 1979 2

3.2 SITE GEOLOGY EFFECTS

As presented elsewhere, our approach to site geology has been to include several types of site geology in the analysis, and to test for systematic differences between them. Establishing that the differences are small would provide a basis for using site types other than Soft Rock in the DCNPP analysis. As has been

noted by other investigators (e.g., Boore, et al., I 978), it is difficult to quantify the effect of site geology because of structural effects. Most of the recordings in the DCNPP data base were obtained in buildings and these buildings are usually sited on soil. Furthermore, the larger the building, the more likely that the instrument is located in a basement. Thus, the effects of site geology, building size and instrument location are extensively interrelated.

'e attempted to extract the effects of site geology from the other effects by selecting data recorded at ground level in small structures (one- and two-story buildings consistent with Boore et al., l978) or in the free field. The results of an analysis on these data suggested that Soft Rock sites had lower recorded PGAs than either soil or Hard Rock sites. The limited number of Hard Rock recordings precluded definitive conclusions regarding their behavior; however, PGAs recorded on these sites were quite similar to those recorded on soil. Similar results, although more difficult to interpret, are obtained from analysis of the whole DCNPP Data Base. These results would appear to be contrary to some past investigations and current opinion among some earthquake engineers, which suggest that in the near-source region rock would be expected to record higher accelerations than. soil. In order to resolve this contradiction, field investigations by a geologist were conducted for each site initially considered to be rock in the DCNPP Data Base. This study determined that many of these "rock" sites were actually underlain by shallow soil or fill—shallow being defined- as depths less than about la meters. The inclusion of these recordings as rock sites resulted in "rock" predicting significantly higher accelerations than soil, consistent with past studies.

We emphasize that these conclusions are restricted only to peak accelerations in the near-source, and cannot be extended without further study either to other

3-4

TERA CORPORATION 0 ground motion parameters (peak velocity, displacement, or spectral ordinates) or to other distances (less than 3 to 5 or beyond 30 to 50 kilometers).

Since'e judge that the Diablo Canyon site is best represented by sedimentary rock (our Soft Rock classification), including data that are slightly higher than the most appropriate data for the site is conservative and allows for more robust statistical predictions of PGA.

It should be emphasized that this partitioning of the data base was performed simply to test the importance of site geology. As will be shown in Section 4.5, the entire data base is sufficiently complete to permit robust estimates of PGA for DCNPP.

3-5

TERA CORPORATION

3.3,FAULT TYPE

Of the 27 earthquakes originally used in this study 14 are associated with strike- slip faulting mechanism, IO are characterized by reverse or thrust, 2 with normal and the remaining 2 by a combination of strike-slip and dip-slip faulting. The inference is based on geological field reports, seismological source studies and tectonic environments.

The distribution of the DCNPP acceleration data according to the earthquake fault type is given in Table 3-2. As seen in this table, while all types of sources are represented in the data base, the majority of the data are strike-slip (about 60 percent), consistent with the Hosgri and other nearby faults of the San ( Andreas system. Furthermore, note that most of the non-North American data are from reverse faulting.

We have statistically analyzed the worldwide data with respect to their fault type and have determined that reverse faults are systematically higher than any other fault type. The analysis further indicates that this conclusion passes a 90 percent confidence test. We have tested for, but not found, such systematic differences between other fault types.

When we similarly analyze the North American data, our results change appreciably. For this restricted, but more relevant, data set we are unable to determine, at a 90 percent confidence level, any systematic bias between the various fault types. The comparison of the two tests is indicative of the strong conservative bias introduced by the l974 Lima, 1976 Gazli and l978 Tabas earthquakes, all having reverse source mechanisms.

3-6

TERA CORPORATION

TABLE 3-2

CLASSIFICATION OF HORIZONTALPGA ACCORDING TO EARTHQUAKE FAULT TYPE

Wor Idwide North America

Str ike-slip l37 60% l35 6 I%

Reverse 78 34% 7I 32%

Normal IO 4% IO 5%

Mixed 2% 2%

Total 229 I 00% 220 I 00%

3-7

TERA CORPORATION

3.4 BUILDINGEFFECTS

In Section 3.2 w'e pointed out the difficulty in separating the effects of site geology, building size and instrument location. These difficulties motivated partitioning the DCNPP data base in order to isolate important trends. We follow the same approach in this section. For this analysis we partition the data base into two subgroups, both situated on soil, represented by embedded and ground-level recordings as follows:

e Small (1-2 story) buildings or free field stations

~ Large (3-20 story) buildings

The effects of both embedment and building size were studies by regression analysis of the above selected data. Due to limitations in these data, valid comparisons could only be made between small building/free field recordings at ground level and those obtained in basements of large buildings. This comparison indicated that PGA recorded in basements of large buildings were on the average 20 percent to 25 percent lower than those recorded at ground level.

Although the entire DCNPP Data Set tends to support these results, as will be shown in Section 4.0; the inclusion of large buildings in the analyses have not significantly affected the predictiop of free field PGA for DCNPP.

We feel that this effect is sufficiently important that we have performed a separate case study into this topic. In order to minimize as many of the above- mentioned biases as possible (site geology, distance, magnitude), we restricted the investigation to paired recordings from nearby stations.

Three types of comparisons of peak accelerations from strong motion accelerograms are presented in this report. First,'peak accelerations measured in the free field (accelerograph located on the ground surface in an instrument shelter 'or small building) were compared with those measured by an accelerograph located in the basement of an embedded structure (Tables 3-3 and 3-3A).

3-8 TERA CORPORATION . ~ Second, peak accelerations recorded in the free field were compared with those recorded at the ground level of a nearby building without a basement (Tables 3-4 and 3-4A). Third, peak accelerations recorded at ground level sites in structures are compared with peak accelerations measured in the basement of closeby embedded structures in order to study the effect of embedments (Tables 3-5 and 3-5A). These effects are summarized in terms of the mean reduction in percent in the mean horizontal free field or ground-level recorded values as compared to nearby building or embedded recordings. A positive reduction means that free field/ground-level recordings were higher, and a negative reduction means they were lower than those recorded in nearby structures.

The Imperial County Services Building is the only structure to record higher peak accelerations than its nearby free-field station. There are several reasons why this might have occurred. First, there is an indication that the free-field recording may be anomalously low compared to other nearby free-field recordings (Figure 3- I). A possible explanation includes the saturated near-surface soil conditions due to intense summer and fall irrigation practices. Second, the recorded motion in the building may have been complicated by the failure of the structure, that occurred before or just after the occurrence of the peak accelerations in the recording.

The compar ison between the El Centro Differential Array and El Centro Station 9 may not be totally justified due to the relatively large (l.3 km) distance between these two stations. However, examination of recordings within the vicinity of these two stations shows no consistent attenuation over this distance range. Furthermore, soil and geologic conditions, including the depth of the water table, are very similar at these.,two stations. This suggests that the comparison offered by these stations is valid.

Omitting the Imperial County Services Building, the mean reductions in

free-'ield horizontal accelerations due to the presence of structures and in ground- level building recordings due to embedment range from 22 percent to 66 percent with a mean of 39 percent. The data suggest a slight tendency to smaller reductions for the larger magnitude earthquakes. When only magnitude 5.0 or

3-9

TERA CORPORATION 0 1 greater events are considered, excluding the Japanese data, the mean reduction reduces to 33 percent. This reduction should be compared against the reduction factor of 20 to 25 percent derived statistically from the DCNPP data base.

3-IO

TERA CORPORATION

TABLE 3-3

REDUCTION IN FREE-FIELD HORIZONTALACCELERATIONS DUE TO THE PRESENCE OF EMBEDDED STRUCTURES

Magnitude Mean istance Station Reduction ~arthcauake Date (ML) (k ) Location (%)

Kern County 07-2 I -52 7.2 I07 Hollywood Storage 22 Bldg.

Southern 08-30-64 4.0 23 Hollywood Storage 66 California Bldg.

Lytle Creek 09-12-70 5.4 80 Hollywood Storage 34 Bldg.

San Fernando 02-09-71 6.4 35 Hollywood Storage 37 Bldg.

Lillis Ranch 08-03-75 4.9 18 Pleasant Valley Pump 38 Plant

Ferndale 06-07-75 5.3 25 Humboldt Bay Power Plant 54

Imperial IO-I5-79 6.6 El Centro Station 9 25 Valley El Centro Diff. Array

3-II

TERA CORPORATION

TABLE 3-3A

RECORDING STATION DESCRIPTIONS

STATION NAME

Humboldt Bay Nuclear Description of Hollywood Storage Site Pleasant Valley Site El Centro Differential Embedded Building Power Plant (Free Field (Building and (Building and Array (Free Field) and Storage Building and Centro Station Free Field) Free Field) Reactor Caisson" El //9

Distance from Free Field Station 47m 400m 23m l,3 I 0m

Site Geology Alluvium, 130m Alluvium Pleistocene Hookton Alluvium )300m Formation (hard soil)

Number of Stories/Height Above Ground l4 stories I story I lm high 2 stories

Plan Dimensions l7m by 72m 37m by l4m 7m by 20m

Depth of Embedment of the Instrument 3m 25m 7m

The Reactor Caisson is an ISm diameter fuel containment structure located completely below the ground surface. Directly above this structure is the 37m long by 14m wide by I lm high refueling building. The storage building is a small I-story non-embedded structure, considered to represent free field conditions.

TABLE 3-4

REDUCTION IN FREE-FIELD HORIZONTAL ACCELERATIONS DUE TO THE PRESENCE OF NON-EMBEDDED STRUCTURES

Mean Magnitude Distance Station Reduction ~arthcauake Date (ML) (km) Location (%)

Imperial Valley IO- l5-79 6.6 Imperial County -2l Services Building and Free Field

Average of 8- I 9-68 (6 50 Hachinohe Technical 50 Four After- to College and Free shocks of the 8-25-68 Field, Japan Tokachioki Earthquake of May l6, I 968 (Mag = 7.8)

Average of I 97 I (6 70 - l4I Apartment House 33 20 Earth- to and Free Field, quakes Near 1975 Inage, Japan lnage, Japan

See text

3-l3

TERA CORPORATION

TABLE 3-4A

RECORDING STATION DESCRIPTIONS

STATION NAME

Description of Building Imperial County Hachinohe Technical Apartment House Services Building College and Free Field, and Free Field and Free Field Inage, Japan

Distance from Free Field Station II4 m IOm to 50 m l9m

Site Geology Alluvium 300 m Silt, sand and clay to 24 m

Number of Stories

Plan Dimensions 46 mby 28 m 3 wings: Each 80 m by 60 m by l5 m IO m separated by Hallways

Depth of Embedment of the Instrument Ground level Ground level Ground level

TABLE 3-5

REDUCTION IN HORIZONTALACCELERATIONS DUE TO THE EMBEDMENT OF STRUCTURES

Magnitude Mean istance Station Reduction Earth uake Date ') (k ) Location (i)

San Fernando 02-09-7 I 6.4 l5 l4724 Ventura Blvd., LA 30 I 5250 Ventura Blvd., LA

l9 l 760 N. Orchid, LA 28 7080 Hollywood, LA

20 6430 Sunset Blvd., LA 30 6464 Sunset Blvd., LA

24 6200 Wilshire, LA 46 5900 Wilshire, LA

39 3407 W. Sixth, LA 6I6 S. Normandie, LA 3470 Wilshire, LA 28 34I I Wilshire, LA 3550 Wilshire, LA

Southern OI-01-76 4.2 9 Diemar Filter Plant 60 California

NOTE: Other station pairs (one station in a building recording at ground level, the other recording at depth) exist in downtown Los Angeles. However, the effect of embedment in fhe central downtown area is complicated by the great density of large structures, the presence of underground transit tunnels, and complex geologic conditions. These complications add additional scatter to the downtown Los Angeles records, which have not been included in these comparisons.

3-I5

TERA CORPORATION

TABLE 3-5A

RECORDING STATION DESCRIPTIONS

Separation Number Area of Depth of Building Embedment Station Name Between Site Geology of Beneath Stations Stories of the Surface Instrument

2 l4724 Ventura Blvd., LA Alluvium l4 89 m Ground level 9I4 m 2 I5250 Ventura Blvd., LA Alluvium . I2 100 m Basement

1760 N. Orchid, LA Alluvium 23 Ground level 450 m 7080 Hollywood, LA Alluvium l2 Basement

6430 Sunset Blvd., LA Alluvium l5 Ground level l00 m 6464 Sunset Blvd., LA Alluvium l2 Basement

6200 Wilshire, LA Alluvium over asphaltic l7 Ground level sands 500 m 5900 Wilshire, LA Alluvium over asphaltic l7 Basement sands

2 3407 W. Sixth, LA Alluvium, Shale at l3 m 7 2,000 m IYz m 2 3470 Wilshire, LA 536 m Alluvium, Shale at l3 m II 2,700 m Sm 2 34 I I Wilshire, LA 396 m Alluvium, Shale at IO m 3I 8,500 m I8m 6I6 S. Normandie, LA 4I6 m Alluvium, Shale at IO m l7 l,200 m 3550 Wilshire, LA 666 m Alluvium, Shale at 33 m 21 2,300 m 7 m

MRXIMUtt HGf) I 20NTRL CQMPOWEHT

EC CEHTAd FAKE flELP EL CEHTAd RAART «6 EL cEHrAa RAART «5 EL CEHTAa RAART «8 EL cEHrAa RAART «4 EL cEHrAa OlFFEAEHrlRL R EL CEHTAa RAART «2

tllGH FRFQUCllCT 5t'ECTHAL AHPL I TUDF5

III'FATAL VFlLLf..r l979 FAFLF 1 t=t 0 STATt(l,<5

I r ~ P' /t I t fg A/ //~

n. f P I/ j~lr

Cl o Q. 01 0 l. PEBICJO — SEC F1GUHL= 3-l 3-l7 l 4.0 SENSITIVITYRESULTS

A sensitivity study was conducted to determine the robustness of the predicted peak accelerations for DCNPP, with respect to the data base and various assumptions incorporated in the analyses. Studies were concentrated in five main areas: (I) the effect of model variations, (2) the effect of the far-field, decay rate, (3) the effect of focusing, (4) the effect of well-recorded earthquakes, and (5) the effect of an alternative partitioning of the DCNPP data base.

4. I MODEL VARIATIONS

In addition to the Physical Model and Statistical Model defined by Equations 3-2 and 3-3, respectively, three other attenuation models were proposed and developed for this study so as to check the sensitivity of the results to the choice of Equations 2-l, 2-2 and 2-3.

Four of these models involved the choice of C in Equation 2-I. In the Statistical Model, the parameters cl and c2 were allowed to be statistically fit by the regression analysis; whereas, in the Physical Model, c2 was determined by Equation 2-3. In the third model, C w'as constrained to be a constant independent of magnitude. In the fourth model, C was set equal to zero and the r'emaining constants fit by the regression.

The four models involving the choice of C have the following physical properties:

C physically determined: PGA must saturate to a constant value at zero distance for all magnitude earthquakes.

C statistically fit: The standard error is minimized, and PGA can saturate or not with distance and/or magnitude.

C equal to a constant: PGA can saturate with distance, but . cannot saturate with magnitude.

C equal to zero: PGA cannot saturate with either distance or magnitude.

4-I

TERA CORPORATION

A fifth model, with properties similar to a form proposed by Donovan and

Bornstein (I 978), was a log-linear relationship of the form ~

2 In PGA = A '+ BM + DlnR + Eln R + FMlnR

To assess the effects of the non-North=American data, these models were tested against three partitions of the data; the worldwide data base (weighted), only North American data (weighted), and the worldwide data base (unweighted). Many investigators have noted that foreign recordings may be inappropriate for inclusion in predictions of ground motion for western U.S. sites, due to tectonic differences and questionable data quality. On the other hand, sufficient high- magnitude recordings do not exist for the western U.S. to allow meaningful analysis without extrapolation at the magnitude 7.5 level. These three data base partitions test the sensitivity of non-North American data to conclusions regarding PGA saturation with magnitude and the prediction for DCNPP. The first data base gives non-North American earthquakes substantial weight in the analysis. The second data base eliminates these recordings altogether. The third data base is analyzed without weighting, thereby treating each data point individually, consequently reducing the effect of non-North American recordings which are represented by only a few recordings.

Mean predictions at 5.8 kilometers for magnitudes 6.5, 7.0 and 7.5 for these various models are presented in Table 4-I. Also included in this table are the ratios of the median plus one-standard-deviation estimates to the median value 2 and the r values of the regression. The r values represent the percentage of the variance in the observations which could be explained by the model. The results of an F-test on the mean square errors from each of these models as compared to the Statistical Model suggested that the C = 0 model had a significantly higher standard deviation at the 90 percent confidence level, and thus, it should be rejected as statistically invalid. Results from the C=O model are, therefore, not presented. Among the other models, little variation is observed for the magnitude 6.5 predictions, and variations between the statistical and physical models are less than l6 percent are observed at the other

TERA CORPORATION 0 magnitudes. This, of course, should be compared with a one-standard-deviation value that is approximately 50 percent higher than the median (significantly higher than variations among the median predictions).

Thus, the statistical analyses of this near-source acceleration data base showed that, statistically, the proposed Physical Model for predicting near-source accelerations consistently fit the different data as well as or better than these other models. The proposed Physical Model is in close agreement with the Statistical Model which also supports saturation of PGA with increasing magnitude.

Figures 4-I and 4-2 plot the Physical and Statistical models for the weighted North American and the weighted and unweighted worldwide data bases. The close agreement between the Physical Model and the Statistically Model for the weighted North American and the unweighted world wide data bases tend to support saturation of PGA at small distances to a constant value for all magnitude earthquakes.

4-3

TERA CORPORATION

TABLE4- I

SENSITIVITY RESULTS FOR MODEL VARIATIONS

WORLDWIDE DATABASE (WEIGHTED)

Peak Acceleration at 5.8 km(g) Model Med;an + I r2

C = Physically Determined .32 .37 .4I I.52 .76

C = Statistical Fit .3I .39 ~ .48 I.50 .78 C = Constant .3l .43 .6l l.5l .77

Log-Linear .3I .39 .48 ~ I.47 .80

NORTH AMERICANDATA(WEIGHTED)

Peak Acceleration at 5.8 km(g) Model Median + 2 65 70 75 Median

C = Physically Determined .29 .32 .36 I '.48 .76

C = Statistical Fit .29 .34 .38 I .48 .76

C = Constant .29 .40 .55 I .52 .73 Log-Linear .29 .35 .43, l.47 .76

WORLD-WIDEDATA BASE (UNWEIGHTED)

Peak Acceleration at 5.8 km(g) Model Median + r2 6.5 7.0 7.5 Median

C = Physically Determined .29 .33 .37 l.50 .75

C = Statistical Fit .29 '33 .36 l.50 - .75 C = Constant .28 .38 .52 l.53 .72 Log-Linear .30 .37 .45 I.48 .76

TERA CORPORATION

CL CL

CE PHYSICAL JIODEL HEIGHTEQ NIRLO itIDE WEIGHTED NORTH RHERICR +++++ UNWEIGHTED WORLD HIDE

4. 00 5. QO 6. 00 7. QQ 8. 00 MAGNITUDE

FIGURE 4-i

EFFECT OF DATA BASE SElECTION ON NAGNITUOE SCAlING AT 5. 8 KN.

4-5

e CL hJ Lu,

STATISTICRL HANDEL HEIGHTfD NdRLO HlOE HEIGHTED NORTH AMERICA +++++ UNHEIGHTED HORLD HIDE

5. 00 8. 00 7. 00 8. 00 MAGNITUDE

fIGURE 4-2

EFFECT QF DRTR 8RSE SELECTION CN MAGNITUDE SCRLING RT 5. 8 KM.

4-6

4.2 FAR-FIELD DECAY RATE

The far-field decay rate d was constrained in the Physical Model to I.75, consistent with other investigators'ar-field studies. The sensitivity of the predictions to this assumption at 5.8kilometers was studied by varying the assumed value of d. Equation 3-3 allowed the parameter to be fit by the regression, which selected a value of 1.09 (reflecting, further, the limits of the near-field decay rate defined by data within 50 km). Two other analyses constrained this parameter to values of I.S and 2.0, respectively. The range of values selected represent a reasonable variation of this parameter, as deter- . mined from a literature survey of available attenuation models.

The results of the analyses are presented in Table 4-2. Variations in the predictions at 5.8 kilometers over the range of magnitudes of interest in this study are less than ten percent among all values chosen for Equation 3-2, demonstrating relative insensitivity to this parameter. An F-test on the mean square errors also confirmed that there is no significant difference among these models at the 90 percent confidence level..

4-7

TERA CORPORATION l' TABLE 4-2

SENSITIVITY RESULTS FOR VARIATIONS IN FAR-FIELD DECAY RATE

Peak Acceleration at 5.8 km(g) Decay Rate Median+ I r2 (d) 6.5 7.0 7.5 l.09 (Statistical Model) .3I .39 .48 l.50 .78 l.50 (Physical Model) .32 .38 .43 l.49 .79 1.75 (Physical Model) .32 .37 .4l 1.52 .77 2.00 (Physical Model) .3I .35 .44 l.5l .77

4-8

TERA CORPORATION

4.3 FOCUS! NG POTENTIAL

While focusing may be a result of either near-source geometrical factors or dynamic properties of the source, or both, most of the questions raised in near- field studies so far are concerned with the dynamic aspect. The focusing of energy in the direction of rupture propagation, at the expense of points in the opposite direction from which the rupture front moves away, has been extensively utilized for the far-field modeling of rupture dynamics. The far-field observations involve, low frequencies capable of representing gross (average) dynamical properties of the source. Table 4-3 lists the earthquakes of our data base for which focusing has been suspected to occur at one or more stations. It is therefore generally accepted that all strong earthquakes with extended source dimensions produce focusing to varying degrees.

in order to test for the effect of focusing an peak acceleration, we have conducted a special study of the accelerations from two well-recorded earth- s quakes--the I 97 I San Fernando and the l 979 Imperial Valley earthquakes.

Our data base contains 46 readings of PGA from the former and 52 readings from the latter. We have, therefore, chosen these two earthquakes to test for the effect of focusing, which we have quantified at each recording station by the focusing potential factor (Ben-Menahem, I 961; Boore and Joyner, I 978)

P cos VR with as the angle subtended by the direction of a ray from the hypocenter to the station, and by the direction of rupture propagation. and VR are the speed of propagation of the shear waves and rupture front, respectively. We have adopted a vertical fault plane striking N37 W for the Imperial Valley earthquake and a I IO striking fault plane dipping 42 to the northeast for the San Fernando earthquake. The direction of rupture is taken to be northwest and nearly updip

(rake angle of 70, from Langston, I 978), respectively, having an average rupture s

4-9

TERA CORPORATION / r velocity of VR — 0.7P. There is no unanimous agreement on any of these parameters, but they represent an overall average suitable for the test. One would not expect changes in V< to dramatically influence the correlation pattern.

Figures 4-3 and 4-4 show how the actual data recorded from these two earthquakes compare to our statistical model result. If the parameter F was a significant parameter in the analysis, these plots would reveal trends between the residuals and F that could be used to incorporate F into the analysis. The apparent random scatter of the data establishes that F is not a significant parameter. On the basis of these observations and the data of Figures 4-3 and 4-4, it may be concluded that the deterministic evaluation of the focusing effect in a specific direction from the source requires more detailed understanding of the focal processes. The best one can hope for at this stage of knowledge is to treat it as a random effect distributed throughout the data base, with equal likelihood.

4-l0

TERA CORPORATION

TABLE 4-3,

LIST OF EARTHQUAKES FOR WHICH NEAR-FIELD FOCUSING AT ONE OR MORE RECORDING STATIONS HAS BEEN REPORTED

Rupture Earthquake Date Direction Source

I) Kern County 52072 I- NE Benioff, l955

2) San Fernando 7I0209 updlp Boore and Zoback, 1974; Allen et al., l975

3) Pt. Mugu 73022I bilateral Boore and Stierman, l976

4) Oroville 75080I updlp Lahr et al., l976

5) Gazli, USSR 7604I7 updlp Hartzel, I 980

6) Santa Barbara 7808I3 NW USGS Circular, l979

7) Tabas, Iran 7809I6 NNW Berberian et al., l979

8) Imperial Valley 79IOI5 NW Archuleta and Sharp, l 980

0. 50 2. 00 3. 00

F IGURE 4-3 IMPERIRL VRLLEY 1S79 DRTR TEST FOR FOCUSING

4-l2 a O

CV I

0. 00 0. 50 I. 00 I. 50 2. 00 2. 50 3. 00 F

F IGUBE 4-4

SRN FERNRNOQ DR,TR TEST FQR FQCUSIMG

4 (3

4.4 WELL-RECORDED EARTHQUAKES

Due to the abundance of data from the, l97l San Fernando (46 components) and the l 979 Imperial Valley (52 components) earthquakes, each of these events was removed from the data base and the analysis repeated.

The results of this analysis establish the insensitivity of our conclusions to data for- a single earthquake. Using our result for the Physical Model of 0.4! g (median) as a base, excluding the Imperial Valley l979 data we obtain median

DCNPP accelerations of 0.4S g, and excluding the I 97 I San Fernando earthquake we obtain 0.42 g. The difference is less than IO percent, and the effect is therefore obviously very small. Furthermore, because there was a smaller standard error in the regression when these earthquakes were excluded, the difference between the one-standard-deviation accelerations are even less.

4-l4

TERA CORPORATION

4.5 ALTERNATIVEDATA PARTITIONING - SMALL BUILDINGANALYSIS

In Section 3.0 during discussion of the effects of site geology, building size, and instrument location, we developed alternative partitions to the data and stated that our DCNPP results were insensitive to this partitioning and indeed reflected free-field conditions. This section provides the basis for those conclusions. ln order to demonstrate this, we extracted from the DCNPP data set all the data recorded either in the free field or in small buildings (I-2 stories). All of the instruments were located at ground level. The regression analysis, based on the Statistical Model as described in Section 2.0, performed on these data yielded a median acceleration of 0.49 g and a median plus one-standard-deviation acceleration'of 0.74 g —results essentially identical to those reported for the Statistical Model using the entire, unpartitioned data base. This confirms the validity of our results for representing free-field predictions of peak horizontal accelerations at DCNPP.

4-I5

TERA CORPORATION

5.0 COMPARISONS AND CONCLUSIONS

In the previous sections, the development of the near-source data base and the regression model were described, and the results of sensitivity studies which tested many of the model characteristics and predictions were presented. In this section we present supporting data, comparisons with other investigators results, and the conclusions drawn regarding the behavior of horizontal PGA in the near- source region.

S. I 'HYSICALCHARACTERISTICS

An important characteristic of an empirical model is a comparison of its prediction with the data. In Figure 5-l, we present .the model's predictions versus distance for magnitude 7.5. The plus and minus standard error bounds are included in the figures to indicate the model's capability to fit the data. The data have been normalized to this magnitude by Equation 3-3 and grouped into three magnitude bands. As expected for the lognormal distribution, one observes about one-third of the data falling outside of the standard error bounds. Note also the distinct tendency of PGA to saturate with decreasing distance.

For closer scrutiny of the data we have included the model's prediction at magnitudes 6.5, 7.0 and 7.5 (Figures 5-IA, 5-IB, and 5-IC) which are plotted against the actual data within a one-half magnitude range of the respective predictions. Non-North American data have been boxed, for reference, demonstrating their bias towards higher accelerations.

A similar plot of the data and model predictions at 5.8 kilometers as a function of magnitude is provided in Figure 5-2. These data have been normalized by Equation 3-3 to 5.8 kilometers. This figure shows a distinct tendency of PGA to saturate with increasing magnitude.

Figure 5-3 displays the model s prediction at 5.8 kilometers plotted against all data with a significant distance of less than IO kilometers. In this figure the data have not been normalized, thus producing a slightly greater scatter.

5- I

TERA CORPORATION

4m+ ~ + g ~4@

+mr pe e~ + ~m+

NOTT:: OATH HORHALIZCO TO N = 7, 5 '3 GRE'ATE'R T))AH Mc ~ 7. Hc = 6 TO H» = 7 + LC5S fttAPJ Hs ~ 6

MEOTAN - STATTOTTCAL MOOr.L ~ l 5 I GHA SIGMA

10. 0 I STANCE (KH )

F I GURE S -?

PREOI'CTEO'EAN ACCELERATI'GNS FGR M=7. 5 SHGNI'NG DISTRlBUT.IGN GF MAGNITUDES

+ + +~ +

+ + + +wg+ ++ ~ -4-'-

NOTE! DATA NOT NORMALIZED + Ms~6.S -7.0 + '+ Ms i 8.0 -S.S @ MOM NORTH RMERlCA

NEDIRN — STRTISTICRL NMEL +1 SIGW -1 SIGMA

10. '1 DESTRNCE (KH)

00.'IGURE

5-1R

PREOICTEO PERK RCCELERRTIQMS FUR 8=6. 5 SHQHIMG DISTRIBUTION QF HRGMITUDES

5-3

* + + + + +

X+ ++ + +

+ + + +

NdTf! DRTR NOT NdRNRLIZf0 + Hs~ 7.0 -7.5 + Hs ~ 6. 5 - 7. 0 8 NdN NdRTH RtffRICR

NEDIRN - STRTISTICRL NODEL +1 SIGNA

~+Q SIGNA C) 1. 10. 100. OESTRNCE (KM)

F IGURE 5-1B

PREDICTED PERK RCCElERRTIQMS FQR M=7. Q SHQNING DISTRIBUTIQN QF HRGNITUDES

5-4

NdTE! CATA HOT HdRifRLIZEO Hs ~ 7. 5 - 7. 7 + H» ~ 7.0 - 7. S ~ HdN HdRTH AMERICA

MEDIRM - STRTISTICRL MdDEL +1 SIGNA »1 SIGNA

10. 100. OESTRNCE (KN)

FIGURE 5-1C

PREDICTED PEAK RCCELERATIVNS FUR M=7. 5 SHINING QISTRIBUTIQN QF MAGNITUDES

5-5

o CZ CC hl

CK NOTE: DATA NORMALIZED TO S. 8 KH 0 NON NORTH AMERICA

AEDIAA - STAT1$TICAL HDDEL +f SlGMA ~e4 C) -i $ IGMA C) 4. 00 5. 00 6. 00 7. 00 MAGNITUDE

FIGURE 5-2

PREOICTEO PERK RCCELERRTIVMS FUR DESTRNCE = 5. 8 KH.

5-6

LU hJ

NdTEt DISTANCES LESS THAN 10 KM CK OATA NdT NdRMALIZED 8 NdN NdRTH AMERICA

REDIRR - STRTISTICRL RDDEL +1 SIGMA ~+t Cl -1 SIGMA C) 4. QQ S. QO 6. 00 7. 00 MAGNITUDE

FIGURE 5-3

PREOICTEO PERK RCCELERRTIQNS FUR OISTRNCE = 5. 8 KN.

5-7

5.2 COMPARISON WITH OTHER STUDIES

The results of the Statistical Model described in Section 2.0 have been compared with the data recorded during the l979 Imperial Valley earthquake, with SAM V predictions for that earthquake (Blume, I 977), and with USGS Circular 795. The l 979 Imperial Valley earthquake was chosen for this comparison because it has the most extensive set of near-source recordings for a western U.S. earthquake.

The comparison of the Statistical Model predictions (at a magnitude of 6.9) with the Imperial Valley data in Figure 5-4 shows that the regression model is "conservative" with respect 'to the data. This trend is expected since the mean predictions for this earthquake were lower than the "mean" for this magnitude predicted by the model.

The comparison with SAM V in Figure 5-5 shows remarkably good agreement. The median SAM V prediction is inside the one-standard-deviation bounds.

Figure 5-6 compares USGS Circular 795 70 percent confidence limits (Figures 23-25 from USGS 795) with the l979 Imperial Valley earthquake data and the Statistical Model predictions for that earthquake. Major differences exist between this study and that in USGS Circular 795 that should be considered when making comparisons. Specifically, USGS Circular 795 included only the highest peak horizontal recording. Note also that the 70 percent confidence results for magnitudes 5.3-5.7 and 6.0-6 4 are essentially based on local magnitudes (ML), which for the l979 Imperial Valley earthquake had a value of 6.6 (the surface wave magnitude was 6.9). USGS Circular 795 results for magnitude 6.0-6.4 are consistent with both the largest component of the peak horizontal acceleration recorded during the Imperial Valley earthquake and the results of the Statistical Model over the applicable distance range.

5-8

TERA CORPORATION

'4gee

HfDIAN — STATISTICAL NDDfL +i $ EGHA -i SEGER io. O,I STANCE (KH )

F IGURE 5-4 IMPERIRL VRLLEY f979 ORTR STRTISTICRL MQOEL RT Hs = 6. 9

5-9 e .>AH-V AT HL 6, 6 VF;)IP.nt - STAT J.">; tCAL H)DCL AT e..::6..'3 I "JCVA .I i t tiHA

io. OISTRNCE (KH)

FI'GURE'-5

CCJNt'ARI'SUN Hl TH. SAN-V

5-!0

' ! ~ i). i ~

I I 4 i~ I + I

C) I i-o (Z yC QC Lu I lU I I I

CZ I Ml:OIVH SLOT|". itCAI. i HOX 1ni iH C )HPOt(F r41 H181H! JN COt1l'~7HF.H t ~ l 51GMA o -l 5l GHA o 1 0." DISTRNCE (KM)

F I GURE 5-6

CGMt'RRISQN HITlf USGS 795 STRTISTICRL MGOEL RT Ms=6. 9 (ML=6. 6) 0 5.3 STATISTICALCHARACTERISTICS

In order to observe biases in the predictions given by Equation 2-5 regarding magnitude, distance or predicted acceleration, plots of the normalized weighted residuals are presented in Figures 5-7, 5-8 and 5-9. The data were normalized with respect to the standard deviation of the regression. If there were systematic trends in the data that were not accounted for by our statistical analysis, such trends would be evident from these plots. Note, however, that the residuals are uniformly distributed with respect to magnitude, distance and the predicted accelerations.

Many of the statistical tests used in the sensitivity studies of Section 4.0 required the assumption that the residuals be distributed normally. Since the regression analysis was performed on the logarithm of peak acceleration, this would require PGA to be lognormally distributed. The actual distribution of weighted residuals normalized by the standard deviation of the regression is given in Figure 5- IO. Visual inspection of this histogram would appear to confirm the assumption of normalcy. A more statistical validation may be obtained from the normal probability plot, Figure 5-I I. Here the normal score, defined as the expected probability of observing a given residual (assuming they are normally distributed), is plotted against the normalized, weighted residuals. The linear trend of this plot again suggests that the residuals are normally distributed. The Kolmogorov-Smirnov test confirmed that the normalcy assumption could not be rejected at the 95 percent confidence level.

5-I2

TERA CORPORATION

8 NOH NORTH RHERSCR

44 %y44~gf

CV I aC7

0. 00 i0. 00 20. 00 30. 00 40. 00 DISTANCE (KN)

F IGURE 5-7 MQRHRLIZED NEIGHTED RESIDURLS VS. OISTRNCE (STRTIST I CRL NQDEL)

5-l3 0 8 NON NORTH AHERSCR

~CJ

LU CC,~~ ~ c8 hJ

~ C) Z~~ Cl

C) .C) Ol

D Cl

4. 00 5. 00 8. 00 7. 00 8. 00 HFI GN E TUDE

FIGUFIE 5-8 NQRHRLIZEO HEIGHTEO RESIDURLS VS. HRGNITUOE tSTRTISTICRL HQDEL)

5-l4

8 NON NORlH RHFRICR

Q ~

I (A LU c) tX ~d tel

~ Cl Z~ K)

C) C)

'.o> O. 1 PREOICTED VALUE

F IGURE 5-S NQRHRLIZED HEIGHTED RESIDURLS VS. PREDICTED VRLUE „(STRTISTICRl YODEL)

5-I5

DISTRIBUTION OF RESIDUALS STATISTICAL MODEL

30 zU

-2.5 -2.0 -I.5 -1.0 -'0.5 0.0 0.5 I.O I.5 2.0 2.5 3.0

. NORMALIZEDRESIDUALS

FIGURE 5- I 0

5-I 6 TERA CORPORATION 0 A D Ol

A A

~o K:

(A a ~a K(Z ~D~a ~~4 l

Aa Oi

A A Ol I -S. 00 -2. 00 -i. 00 0. 00 f. 00 3. 00 MQRNRLlZED RE'SIDURL

FIGURE 5-11

MQRMRL PRQBRBILITY PLOT

5-l7 i 5.4 CONCLUSIONS

Based upon the analyses presented in Section 3.0, the sensitivity studies discussed in Section 4.0, and the comparisons presented in this section, we have drawn the following conclusions with regard to horizontal peak ground acceleration (PGA) predictions in the near-source region:

~ The predicted median peak instrumental accelerations of 0.4I g (Physical Model) and 0.48 g (Statistical Model) for a magnitude 7.5 earthquake at 5.8 kilometers firmly establish the conservatism of the I.I5 g peak instrumental acceleration DCNPP used in the seismic design. The I.IS g value is considerably greater than even the 0.62 g and 0.73 g median plus one-standard-deviation values of PGA predicted by these models. The results of this study establish that accelerations tend to saturate with increasing magnitude at small distances. When conservatively biased non-North American data are omitted, results support a constant value of PGA at the fault rupture surface as suggested from earthquake source dynamics.

Both the l979 Imperial Valley earthquake data and the results of this analysis support saturation of acceleration with decreasing distance. This establishes the inappro- priateness of linear extrapolation of far-field data in estimating near-source accelerations. Accelerations recorded on sedimentary rock (our Soft Rock classification) consistent with the DCNPP site, are lower than those recorded on either soil or Hard Rock, confirming the conservatism'of incorporating both soil and hard rock in predictions of PGA for DCNPP.

A 20- to 25-percent reduction in PGA was found to exist for recordings obtained in the basements of large buildings, when compared to ground-level recordings in small (I- and 2-story) buildings or in the free field. This is consistent with a mean reduction of 33 percent observed for paired recordings from a case study of building effects.

An extensive sensitivity analysis has confirmed the robustness of the predictions of PGA at the distances and magnitudes appropriate for DCNPP. Variations in predicted accelerations under a wide range of parametric perturbations were well within the one-standard-deviation estimates given by the Physical and Statistical Models.

5- I 8

TERA CORPORATION

~ The Physical and Statistical models developed in Section 2.0 are not controlled by individual, well-recorded earthquakes. Eliminating the l97l San Fernando earthquake (46 recordings) or the l979 Imperial Valley earthquake (52 recordings) does not significantly change the estimates of PGA nor other conclusions reached.

~ Statistical analysis supports the conclusion that non-North American accelerations, primarily from reverse type faults, are systematically biased high relative to the primarily strike-slip North American data.

~ . Conservatisms in the results of this study include: (a) the use of several significant high acceleration non-North American earthquakes, (b) the use of data recorded on soil as well as rock, and (c) the use of accelerations scaled from unprocessed accelerograms.

The results from the Physical Model are insensitive to the specified far-field acceleration decay rate over the range I.O to 2.0.

Traditional mathematical relationships of increased peak accelerations resulting from focusing are contradicted by a statistical analysis of data from the l97I San Fernando and I 979 Imperial Valley earthquakes.

Statistical assumptions regarding the lognormal distribution of PGA are confirmed by our analyses.

5- I 9

TERA CORPORATION

6.0 REFERENCES

Aki, K., and Richards, P. (I980) "Quantitative Seismology Theory and Methods," Vol. 2, p. 830.

Allen, C.R., T.C. Hanks and J.H. Whitcomb (l 975); "Seismological studies of the San Fernando earthquake and their tectonic implications", CDMG, Bull. I 96 I, pp. 257-262.

Ambraseys, N.N. (I 978) "Preliminary Analysis of European Strong-Motion Data I 965- I 978," Vol. I Proceeding of EAEE.

Archuleta, R.J. and R.V. Sharp (I 980), "Source Parameters of the Oct. I5, I 979 Imperial Valley Earthquake from Nearfield Observations", An Abstract, EDS, Vol. 6l, No.I7, p. 297.

Benioff, H. (I955), "Mechanism and Strain Characteristics of the White Wolf Fault as Indicated by After Shock Sequence," CDMG Bulletin I 7 I, pp. I 99-202.

Ben Menahem, A. (I96I) Radiation of Seismic Surface Save From Finite Moving Sources, BSSA, 5I-pp. 40I-435.

Berberian, M. (I979), "Earthquake Faulting and Bedding Thrust Associated with the Tabas-e-Golshan (Iran) Earthquake of September I 6, I 978, "BSSA:69,6,pp. I 86 I - I 888.

Blume, J.A. (I 977) "The SAM Procedure for Site-Acceleration-Magnitude Relationships", Paper Presented at the Sixth World Conference on Earth- quake Engineering, New Delhi, India, January l 977.

Boore, D.M. and Joyner, W.B. (I 978) "The Influence of Rupture Incoherence on Seismic Directivity,"BSSA, 68-pp. 283-300.

Boore, D.M., W.B. Joyner, A.A. Oliver III, and R.A. Page (l978) "Estimation of Ground Motion Parameters", U.S. Geological Survey Circular 795, Arlington, VA.

Boore, D.M. (I 974) "Empirical and Theoretical Study of Mean-Fault Wave Propagation: World Conference on Earthquake Engineering," 5th Rome, Italy, I 973, Proceeds, V. 2, p. 2397-2408.

Boore, D.M. and D.J. Steirman (l976) "Source Parameters of the Point Mugu, California, Earthquake of February 2I, I 973", BSSA; 662, pp. 385-404.

,Boore, D.M. and M.D. Zoback (I 974) "Two-Dimensional Kinematic Fault Model- ing of the Pacoima Dam Strong-Motion Recordings of the February 9, I 97 I, San Fernando Earthquake," BSSA, 64, pp. 555-570.

6-I TERA CORPORATION

Boore, D. M. (I973) "The Effect of Simple Topography on Seismic Waves: Implications for the Accelerations Recorded at Pacoima Dam, San Fernando Valley, California," BSSA, 63, pp. I603-I609.

Brune, J.N. (I970), "Tectonic Stress and Spectra of Seismic Shear Waves from Earthquakes", J.Geophys, Res., 75,pp. 4997-5009.

Brune, J.N., F. Vernon III, and R. Simons (l979), "Strong Motion Data Recorded in Mexico For the October l5, l979 Imperial Valley Earthquake", Earth- quake Notes. 50:4, p. 49 (Abstract).

Donovan, N.C., and A.E. Bornstein (I 978), "Uncertainties in Seismic Risk Procedures", Journal of the Geotechnical Engineering Division, ASCE, Vol. l04, No.GT7, July,pp. 869-887.

Das, S. and K. Aki (l977), "Fault Plane with Barriers: A Versatile Earthquake Model", J. Geophys; Res., 82, pp. 5658-5670.

Duke, C. M., J. A. Johnson, Y. Kharraz, K. W. Campbell, and N. A. Malpiede (I 97l), "Subsurface Site Conditions and Geology in the San Fernando Earthquake Area," University of California, UCLA-ENG-7206.

Esteva, Luis (I 970), "Seismic Risk and Seismic Design Criteria for Nuclear Power P lants", The MIT Press (Edited by Robert J. Hansen), Cambridge, Massachusetts, pp. I 42- I 82. -7 Hanks, T.C. (I979), "b Values and w Seismic Source Models: Implications for Tectonic Stress Variations Along Active Crustal Fault Zones and the Estimation of High-Frequency Strong Grown Motion", J.Geophys. Res., 84, pp. 2235-2242.

Hanks, T.C., and Kanamori, H. (I979) "A Moment Magnitude Scale," JGR 84:B5, 2348-2350.

Hanks, T.C. and D.A. Johnson (I 976), "Geophysical Assessment of Peak Acceler- ations," BSSA, Vol. 66, pp. 959-968.

I USSR Hartzell, S. ( I 980), "Faulting Process of the May I 7, 976, Gazli, Earthquake", Submitted for publication (BSSA)

Jennings, P.C. and Guzman, R.A. (I975) "Seismic Design Criteria for Nuclear Power Plants," Proceedings of U.S. National Conference on Earthquake Engineering, EERI. Kanamori, H. (I977) "The Energy Release in Great Earthquakes," JGR 82:20, pp. 298I-87. Kanamori, H. and D.L. Anderson (l 975), "Theoretical Basis of Some Emperical Relations in Seismology", BSSA, 65, pp. I 073- I095.

6-2

TERA CORPORATION

Keilis-Borok V. (I959) "On Estimation of the Displacement in an Earthquake Source and of Source Dimensions," Annali de Geofisica, Rome, Vol. I2, pp. 205-2 I 4.

Krishna, J., A.R. Chandrasekaran, and S.S. Sanai (l969) "Analysis of Koyna Accelerogram of December I I, l 967," BSSA, 59, 4 pp. I7I9-I73I. Lahr, K.M., J.C. Lahr, A.G. Lindh, C.G. Bufe, and F.W. Lester (l976), "The August I 975 Oroville Earthquakes", BSSA: 66, 4, pp. I085- I095.

Langston, C.A. (I 978) "The February 9, I 97 I, San Fernando Earthquake: A Study of Source Finiteness in Teleseismic Body Waves," BSSA, 68, pp. I-29.

Lee, W.H.K, C.E. Johnson, T.L. Henyey, and R.L. Yerkes (l 978) "A Preliminary Study of the Santa Barbara, California Earthquake of August l3, l 978 and Its Major Aftershocks," U.S. Geological Survey Circular 797, Arlington, VA.

Mikumo, T. (I 973), "Faulting Process of the San Fernando Earthquake of February 9, l97 I Inferred From Static and Dynamic Near-Field Displace- ments," BSSA: 63, I, pp. 249-270.

Richards, P.G. (I976), "Dynamic, Motions Near An Earthquake Fault: A Three Dimensional Solution," BSSA:66, I, pp. I-3I.

Rudnicki, J.W. and H. Kanamori (I 980) "Effects of Fault Interaction on Moment, Stress Drop and Strain Energy Release," Preprint. Seekins, L.C., and T.C. Hanks, "Strong-Motion Accelerograms of.the Oroville Aftershocks and Peak Acceleration Data," BSSA, Vol. 68, No. 3, p. 667-689.

Trifu'nac, M.D. (I974), "A Three-Dimensional Dislocation Model for the San Fernando Earthquake of Feb. 9, I 97 I," BSSA: 64, I, I49- l72.

Trifunac, M. D., and D. E. Hudson (I 970), "Analysis of the Sta. No. 2 Seismoscope Record-1966, Parkfield, California Earthquake," BSSA, 60:3, pp. 785-795.

6-3 TERA CORPORATION

APPENDIX A

WEIGHTED HISTOGRAhhS SHOWING DISTRIBUTIONOF DATA FOR THE DCNPP DATA BASE

TERA CORPORATION

MAGNITUDE

40 zU

30 ) 29%

22% 22% 20

IO I I%

4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 MAGNITUDE

WEIGHTED HISTOGRAM

FIGURE A- I

TERA CORPORATION

S IGNIFI CANT DISTANCE

27 28%

U Z 2l

I5 I-

I 4% l2

I I%

9% 9% 9%

7% 6% 5%

0 5 IO IS 20 25 30 35 40 45 50 SIGNIFICANT DISTANCE (km)

WEIGHTED I ISTOGRAV.

FIGURE A-2

TERA CORPORATION

DEPTH

66%

0- Uz 40 IJJ

UJ lV U ) 30

LLI 0

l4%

7% 2% 2/0 0%

0 IO I5 20 25

DEPTH (km)

WEIGHTED HISTOGRAM

FIGURE A-3

TERA CORPORATION

GEOLOC Y

569o

A = RECENT ALLUVIUM

B = PLEISTOCENE DEPOSITS

40 C = SOFT ROCK O z D = HARD ROCK Q

9'O I 5'

WEIGHTED HISTOGRAM

FIGURE A-4

TERA CORPORATION = 1

SIZE CLASSIFICATION

39% I INSTRUMENT SHELTERS OR SHEDS 35 2 BUILDINGS ONE TO. NINE STORIES WITHOUT BASEMENTS

3 BUILDINGS ONE TO NINE STORIES 30 WITH BASEMENTS

4 BUILDINGS TALLERTHAN NINE STORIES 25

22% 20

WEIGHTED HISTOGRAM

FIGURE A-5

TERA CORPORATION

85% 80 BUILDINGHEIGHT

0- ZO 5O

EL 30

7'-2

3-9 I 0-20

NO. OF STORIES

WEIGHTED HISTOGRAAA

FIGURE A-6

TERA CORPORATION

STRUCTURE TYPE

70 70% I BUILDING

2 INSTRUMENT SHELTER 60 3 DAM

4 BRIDGE 50 z 5 CONCRETE VAULT

30 K

20 20% IO 8'''to

2 3 4 STRUCTURE TYPE

WEIC HTED HISTOGRAM

FIGURE A-7

TERA CORPORATION 0 LOCATION OF INSTRUMENT

67%

I BASEMENT 50 2 GROUND LEVEL

3 ABUTMENTOF DAM Uz 40 4 GALLERYOF DAM

289o

2'EIGHTED

HISTOGRAM

FIGURE A-8

TERA CORPORATION I INSTRUMENT TYPE

AR-240 25% 24 2 CRA- I 3 DCA-300 4 MO-2 5 RFT-250 2I 6 S-hh 7 SMA-IT 8 'MA-I 209o 9 SSRZ I8

0- I 7' zU I5 G UJ

5~/o

p%%d

I 2 3 4 5 6 7 8 9

WEIGHTED HISTOGRAM

FIGURE A-9

TERA CORPORATION

APPENDIX B

THE DCNPP DATABASE

TERA CORPORATION

TABLE B- I

THE DCNPP DATA BASE STRONG-MOTION DATA

DATE MAG (M) USGS DISTANCE GEO PEAK HORIZONTAL (YR-MO-DAY) NO. (km) ACCELERATIONS (g) HI "z

LOS BEACH 3303}i 6.2 131 6.4 B ~ 200 ,160 LIIBEACH 330311 6.2 136 28.0 C .098 .064 L96 KACH 330311 6.2 288 22.0 A .150 .130 %1EHA NNTANA 351031 5.5 2229 8.0 D . 150 . 150 WERIAL 1940 400519 7.1 1}7 10.0 A .350 .210 SANTA BARBARA 41 410701 5.9 283 10.0 B .240 .180 KERN COI(TY 520721 7.7 1095 42.0 A .197 .177 9ALY CITY 570322 5.3 1049 24.0 B .047 .029 DALY CITY 570322 5.3 1065 14.0 A .055 .050 DALY CITY 570322 5.3 1078 14.0 A .049 .046 DALY CITY 570322 5.3 1080 12.0 A .103 .062 DALY CITY 570322 5.3 1117 8.0 C .126 .105 PARKFIELD 660628 6.0 1013 ',1 A .730 .510 PARKFIELD 660628 6.0 1014 5.5 A .470 .400

PARKFIELD 660628 6.0 1015 9.6 B ~ 280 .270 PARKFIELD 660628 6.0 1016 14.9 A .072 .060 FAIRBANKS ALASKA 670621 5.7 2721 15.0 9 .140 .090 KOYNA INDIA 671210 9000 3.2 D .630 .490 B(IREM HTH 680409 6.7 117 45.0 A .142 .061 LYTLE CREEK 700912 5,4 111 18.0 9 .086 .059 LYTLE CREEK 700912 5.4 113 29.0 A .045 .04} LYTLE CREEK 700912 5.4 274 28.0 A .120 .060 LYTLE CREEK 700912 5.4 278 32.0 C .022 .020 SAH FERNANDO 710209 6.6 104 '27.9 D .240 .}80 SAN FHWeDO 710209 6.6 125 29. 6 A .170 .}20 SAN FERNANDO 710209 6.6 133 21.3 .150 .110

SAN FH5ANDO 710209 6.6 135 20.5 A ~ 220 .190.

SAN FERNANDO 710209 6.6 137 24.1 C ~ 200 . 140 SAH FH5ANDO 710209 6.6 141 16.9 9 . 180 .160 SAN FH5ANDO 710209 6.6 181 25.2 B .140 .140 SAH FERHAt(DO 710209 6.6 190 25.5 C .080 .070 SAH FERHAt(00 710209 6.6 220 15.4 C . 180 .130 SAH FERHANDO 710209 6.6 229 36.1 B .060 .060 SAN FERNANDO 710209 6.6 241 7.5 A .270 .140 SAN FH(tQHDO 710209 6.6 244 36.1 B .040 .040 SAN FERHANDO 710209 6.6 247 36.1 B .030 .030 SAN FERt(ANDO 7}0209 6.6 253 15.4 A .260 .190 SAH FERHAHDO 7}0209 6.6 262 27.6 A .130 .110 SAH FERNANDO 710209 6.6 264 21.8 B .210 .180 SAN FERNANDO 710209 6.6 266 18.4 D .190 ,110 SAN FH5AHDO 710209 6.6 267 14. 8 B ~ 220 .170 SAN FERNANDO 7}0209 6.6 288 30.7 A .110 .090 SAN FHtt(ANDO 710209 6.6 458 9.7 A .120 .110 SAN FH5AteO 710209 6.6 461 14.3 A }50 .}30

SAN FERNANDO 710209 6.6 15.4 ~ 230 .140

SAN FERNAt(DO 710209 6.6 475 22.5 B .110 ~ 100 SAN FH(t(ANDO 7}0209 6.6 482 24.8 B .130 .110

"'.1 BEAR VALLEY 720224 1028 31.0 A .030 .020

TERA CORPORATION

TABLE B- I (CONT.)

EARTHQUAKE DATE MAC (M) USGS DISTANCE GEO . PEAK HORIZONTAL NAME (YR-MO-DAY) Na (km) ACCELERATIONS (g)

HI HZ

SITKA ALA%A 720730 7.6 2714 45.0 A .110 .090 t}AWOVA 721223 6.2 3501 5.0 A .390 .340 POINT NQJ 730221 5.9 272 24.0 A .130 .ON'210 LIN PERU 741003 7.6 4302 38.0 B .240 Lit}APERU 741003 7.6 4304 40.0 .2% .200 KU.ISTER 741128 5.1 1028 10.8 A .170 .100 l%LLISTER 741}28 5.} 1250 10.8 B .140 .}00 lKLLISTER 741128 5,1 1377 8.9 A .120 .050 OROVILLE 75080} 5.7 1051 8.0 B .110 .100 HNVILLE 750801 5.7 1291 30.0 A ,070 .060 NNVILLE 750801 5.7 1292 31.0 A .080 .060 NNILLE 750801 5.7 1293 32.0 C .040 .030 6ALLI USSR 760517 7.0 9110 3.5 C .810 .650

SANTA BAISARA 78 780813 5.7 885 7.7 A ~ 390 ;240 SNTA BARBARA 78 780813 5.7 941 18.1 C .040 .040 SNTA BARBARA 78 780813 5.7 5093 7.7 B .440 .270 SNTA BARBARA 78 780813 5.7 5135 25.4 C .060

SNTA BARBARA 78 780813 5.7 5137 10.1 B ~ 220 .110 SNTA BARBARA 78 7808}3 5.7 9019 9.8 8 .2}0 .100 TABAS IRN 780916 7.7 9124 3.0 A .800 BISHOP 781004 5.8 1325 34.2 A .060 .060 BISHOP 781004 5.8 1444 7.6 C' .260 .170 BIMIP 781004 5.8 1490 29.0 .070 .050 BISHOP 781004 5.8 9030 27.1 A .060 .030 t}ALIBV 790101 5.0 657 20.7 B .OA .030 t}ALIBU 790101 5.0 757 26.2 B .060 .030 t}ALIBU 79010! 5.0 5080 15.6 C .060 .050 ST ELIAS 790228 7.2 2734 38.3 A .160 .110 COYOTE LAKE 790806 5.9 1251 23.3 C .030 COYOTE LAKE 790806 5.9 1377 14.4 A .1}0 .090 COYOTE LAKE 790806 5.9 1408 8.9 C .130 , }00

COYOTE LAKE 790806 5.9 1409 8.0 A .260 ~ 200 COYOTE LAKE 790806 5.9 1410 6.3 A .270 .260 COYOTE LAKE 790806 5.9 1411 4.9 A .260 .240 COYOTE LAKE 790806 5.9 1422 24.8 A .050 .040

COYOTE LA)Z 790806 5.9 1445 3.9 B 0 230 .160 COYOTE LAKE 790806 5.9 1492 16.2 B .120 .110 ItfERIAL 1979 791015 6.9 117 5.8 A .400 .270 ltfERIAL 1979 791015 6.9 286 24.5 II .210 .120 ltfERIAL 1979 791015 6.9 412 8.2 A ~ 230 ~ 200 ltfERIAL 1979 791015 6.9 724 34.0 A . 100 .074 ItfERIAL 1979 791015 6.9 93} 18.0 A .150 .110 ItfERIAL 1979 791015 6.9 942 1.4 A .720 .450 ItfERIAL 1979 791015 6.9 952 1.0 Z .560 .400 ItfERIAL 1979 791015 6.9 955 4,42 A .610 ~0 ItfERIAL 1979 791015 6.9 9M 3.5 A .640 ltfERIAL 1979 791015 6.9 5028 0.2 A .520 .360

ItfERIAL 1979 791015 6,9 505} 13. 1 ~ 200 .110

TERA CORPORATION

TABLE B- I (CONT.)

EARTHQUAKE DATE MAG(M)I USGS DISTANCE GEO PEAK HORIZONTAL NAME (YR-MO-DAY) NO. (km) ACCELERATIONS(g) HI HZ

ItfERIAL 1979 791015 6.9 5052 30.5 ,070 ItfERIAL 1979 791015 6.9 5053 10.1 .280 .220 ltfERIAL 1979 791015 6.9 5054 2.8 .8}0 ,660 I}fERIAL1979 791015 6.9 5055: 7.3 .260 .220 ItfERIAL 1979 791015 6.9 5056 16.4 z .150 .150 ItfERIAL 1979 791015 6.9 5057 9 3 .270 .220 ItfERIAL 1979 791015 6.9 5058 12.2 .380 .380 ItfERIAL 1979 791P}5 6.9 5059 21.5 .150 .120

. ItfERIAL 1979 791015 6.9 5060 7.0 .ZO .170 }IMPERIAL 1979 791015 6.9 5061 22.2 . 130 .090 ItfERIAL 1979 791015 6,9 5062 28.0 .130 .100 ltfERIAL 1979 791015 6.9 5066 47.7 .140 .110

ItfERIAL 1979 791015 6.9 5090 7.0 ~ 319 .29} ItfERIAL 1979 791015 6.9 5115 'Q.pz .430 .330 ltfERIAL 1979 791015 6.9 5154 7.0 .243 .237 ItfERIAL 1979 791015 6.9 5165 4.8 .5}0 .370 ItfERIAL 1979 791015 6.9 9028 12.6 .}06 .081 ItfERIAL 1979 7910}5 6.9 9031 0.2 .326 .279 ltfERIAL 1979 7910}5 6.9 9032 0.2 .408 .264 'ItfERIAL }979 791015, 6.9 9033 0.2 .359 .303

I SEE PACE 2-2

2 DISTANCES WERE USED TO THE BRAWLEYFAULT SINCE IT WAS KNOWN TO HAVE RUPTURED. THIS ALLOWEDFOR A BETTER STATISTICAL EXPLANATIONOF PGA.

, TERA CORPORATION

. UNITED STATES OF AMERICA 'NUCLEAR REGULATORY COMMISSION

ATOMIC SAFETY AND LICENSING APPEAL BOARD Richard S. Salzman, Chairman Dr. John H. Buck Dr.W. Reed Johnson

In the matter of PACIFIC GAS AND ELECTRIC COMPANY (Diablo Canyon Nuclear Power Plant, Units 1 and 2) Docket Nos. 50-275 50-323

APPL CANT PACIFIC GAS AND ELECTRIC COMPANY'S WRITTEN TEST MONY N RESPONSE TO APPE AL BOARD QUESTIONS; SET FORTH IN ALAB-598

VOLUME2

August 8, 1980

~ ~ k» PA.C IF ZC GA.S A.ND ZLE~~TRIC COMM.NT

V. Peak horizontal acceleration values measured at the base of the Imperial Valley Services Building during IV-79 exceed those measured in the free field 103 meters away from the building. The motion records are described similar amplitudes but greater low frequency motionas'howing in the building than in the free field. ~42 No response spectra for the two recording locations have been provided. The acceleration data, however, may be taken to indicate that no reduction in building motion due to the tau effect was realized in this instance. Based on these observations, intervenors question the validity of the tau concept as well as its use to reduce the higher frequency portions of the Diablo Canyon Design Spectrum. The staff and the applicant answer that, because the Imperial County Services Building was supported on piles in a deep soil structure, these observations are irrelevant to the use of a tau effect in the seismic reanalysis of

Diablo Canyon, which is built on a rock site. ~43 Staff

1 witness Newmark, however, used recorded earthquake motions at the Hollywood Storage Building to demonstrate the use of a tau effect analysis. ~44 The Hollywood Storage Building itself is built on piles in soil. Thus, the "built,-on- piles" rationale appears insufficient to explain why no tau effect was evident at, the Imperial Valley Services Building. One feature distinguishing the two buildings that no party commented upon is that the Hollywood Storage I Building has a basement and the Services Building does not. Intervenors'itness, Dr. Luco, used this fact to explain in part why he believes the Hollywood building should have a large tau value. ~45 Rojahn and Ragsdale's discussion implies that to some extent ground level instrumental responses within the Imperial Valley Services Building may have been influenced by the response (and failure) of the building itself. ~46 In any event, given the apparent similarities between the structural foundations of the two buildings, the explanations provided thus far for a seeming lack of a tau effect at the Imperial Valley Services Building are inadequate. The parties should provide additional information on this point and relate their analyses to both geologic and structural conditions prevailing at the Diablo Canyon site.

~42 See "A Preliminary Report on Strong-Motion Records from the Imperial county Services Building by Christopher Rojahn, U. S. Geological Survey and J.D. Ragsdale, California Division of Mines and Geology (undated but issued early January 1980), pages 7 and 8.

~43 Blume Affidavit, Paragraph 10; Rothman - Kuo Affidavit, page 7.

~44 SER Supplement 5, Appendix C. 45/ Tr. 8949.

~46 Rojahn and Ragsdale, pages 7 and 8. That report also reflects information regarding the Services Building asymetric structure (at, pages 2 and 3) which may explain why is was susceptible to damage (see Newmark Testimony fol. Tr. 8552, Attachment B, pages 14 and 15). 0 Written Testimony

DR. JOHN A. BLUME

Q. Dr. Blume, you have read the Board's question on the issue of tau and the ICSB building. Please discuss the ICSB building relative to IV-79, tau, the Hollywood Storage Building and Diablo Can on'?

It is important to note that in most aspects of engineering seismology and structural dynamics there are

10 many factors, often "competing" phenomena, that lead to a net, measured response. In other words, it is a complex

12 subject requiring the consideration of many things. One has to be extremely careful that consideration of only one or two of many factors may well lead to faulty conclusions.

15 The tau concept is only one of several possible factors that 16 might affect building response.

18 Com arison of the Holi ood Stora e Buildin , the Im erial Count Services Buildin and the Diablo Can on Structures In order for the so-called "tau effect" to occur, 20 the building length must be greater than the ground wave 21 length under consideration and, in, addition, the base of the 22 building must be rigid in the horizontal plane. The latter '4 requirement is to prevent the building from simply responding to the waves along its length instead of 25 averaging those waves. 26

Blume V-1

There are ma jor structural and foundation differences between the Hollywood Storage Building ("HSB"), the Imperial County Services Building ("ISCB") and the Diablo Canyon structures. The HSB has a basement with heavy and continuous basement walls and it has a full-area concrete base slab extended to these walls. Thus its base in effect constitutes a rigid horizontal plate girder with a web (slab) and flanges (walls). The base of this structure is very stiff horizontally. Moreover, there is passive

10 resistance of the soil against the basement walls. The ICSB, on the other hand, has no basement; it has only one short segment wall at story 12 of its first perimeter, and its base slab does not extend to either line of foundations in the longitudinal direction. There are small between the caps have 15 ties pile but these little to act as a horizontal frame. The 16 rigidity rigid connections of the slab to the few wall 17 interior first-story segments 3 32 The 18 is minimal: 0 bars at in. centers. moment slab on assumed 19 of inertia of the grade, generously be 5 inches as 20 to thick, is 7,140 ft compared to that of the second and beams 4 a 21 floor spandral at 42,800 ft , ratio 1 6. Moreover, the ICSB slab the 22 of to connection of the to such as make even 23 structure is not to this minor rigidity effective. Therefore, the ICSB does not have adequate "tau 25 lateral stiffness at its base to develop much, if any, The ICSB 26 effect," whereas the HSB has sufficient stiffness.

Blume U-2

is founded on much softer and deeper soil than the HSB. ICSB also has relatively long (45 ft to 60 ft) piles which do not reach to a significantly harder material. The shear velocity of the soil in which these relatively flexible ICSB piles are embedded is only in the range of 400 to 700 fps and the material below the pile tips does not reach 1050 fps velocity until a depth of over 200 feet (from Shannon and Wilson, et al, data, 1975). At the HSB the piles are relatively short (12 ft to 30 ft) and thus stiffer, and they extend a much a V 10 into firmer layer of material with estimated at 1050 fps. Thus the lateral stiffness o f the two buildings The size and shape the 12 differs greatly. of two buildings vary as well, as shown in Figure V-1; thus the 13 buildings are not comparable in size or proportion. The HSB has much more 15 total pile capacity per square foot of building area than the ICSB because is 16 it higher and is designed heavy storage loading. we 17 for If conservatively assume EI value* 18 that (per unit floor area) of the the same the two and the 19 piles is for buildings, that the same, we can then compare the 20 soil is also effect of a beam loaded 21 pile depth alone. For fixed-but-guided one end, the deformation D 22 laterally at lateral is 23 — Pl D 12EI

25 * The product of the modulus of elasticity- and the moment 26 of inertia of the piles.

Blume V-3

But because of the assumptions above, only D and L are as 20 L and left variables. Assign ft, average for S 50 ft average for LIC B, then

D ICSB 3 = = 15 6 D —3 HS 20

In other words the ICSB piles are at least, 16 times more flexible in horizontal translation than the HS piles. With realistic assumptions for the differences in and the 10 soils different density of piles, this value could easily be doubled. In summary as the comparison o the two 12 to f buildings, they are not at all alike in spite of both having piles. The ICSB is much more flexible in soil and/or pile movement horizontally and the horizontal rigidity of the ICSB structure base a 16 at its level is not only fraction of that of the Hollywood Storage considered 17 Building, it is a "tau To 18 inadequate to support effect." state that these two buildings are alike because they both have piles is because have 20 similar to stating that they are alike both and 21 walls roofing. The a 22 Diablo Canyon structures are obviously great deal more HSB 23 rigid laterally than the for the following 24 reasons: 25 26

Blume V-4

They are based on rock They are embedded in the rock They are large They are massive Their base slabs and walls are very thick In short, Diablo Canyon is far better suited for the tau concept than the HSB or any building in Los Angeles or San Francisco. Natural Mode Sha es mode 10 Natural shapes are directly related to the question of relative motion of the free field and the ground at the ICSB. a on alluvium, or on 12 If building is situated any base material that is not "infinitely" rigid, its natural mode shapes will indicate some deformation at the base, whether a 15 at grade or at basement level. This compliance the may be 16 in soil translational or rotational or both (Blume, 1956). For a slender the 17 tall building more 18 rotational effect is usually significant than the translational. This phenomenon will occur whether or not

20 there are vertical pilings because the typical vertical have 21 piles little rigidity in the lateral sense if they extend some 40 more 22 to depth, say or feet, into soft soil. The move 23 piles will with the soil, or in some cases they may move more than the soil if forced into it by the inertia of

25 the superstructure's motion. 26

Blume V-5 0 That such mode shapes exist has been shown by the forced vibration of real buildings, by model studies on shaking tables, by mathematical models, and by theory. For example, the plotted mode shaped and/or resonance curves for the forced vibration of such buildings as the Alexander Building (Blume, 1956), the Hollywood Storage Building (USCGS Preliminary Report, 1938), and the Mt. McKinley Apartment building in Anchorage (Blume, 1967) show ground response. Forced 10 vibration of the 15-story Alexander Building in San Francisco with the vibrator located at the top of the showed was 12 building clearly that there translational motion not only at the street floor but at the basement, level, and even outside the building across the

street! Reference made 9 and 10 from Blume 15 is to Figures (1956) shown herein as V-2. The basement motion 16 Figure in the third mode was 4% on and the 17 of that the 10th story, in second mode was 18 it about 3% of that on the 10th story. This on 100 19 building is situated about feet of layered alluvial The mode was 20 material. corresponding fundamental data not obtained because 21 only the power of the machine at that was 22 frequency too small to induce measureable motion. All were no 23 of these tests conducted late at night. with disturbances from traffic or any other cause except the vibrator.

Blume V-6

Similar results were obtained for the Hollywood Storage Building for which the "Amplitude ratio of the 13th floor to the basement floor is about 15:1, and of 13th floor to ground 1.2 mi. south of the building, it is about 400:1" (USCScGS, 1938). For the Mt. McKinley Apartment Building located on sand and gravel with clay below that layer, not only was basement motion recorded at the natural periods of the building, but the same periods were recorded in the basement

10 of another building approximately 500 feet away! Forced vibration does not provide a different mode shape than 12 ground motion disturbances. A natural mode shape is induced by adequate forces applied anywhere except at the base nodal point. An example of this is shown in Figure V-3 where 15 the periods measured are essentially the same whether the perturbation is wind, earthquake, or forced vibration. 16 i the 17 If periods are the same, the mode shape is the same, for the same 18 structure. Two things are self-evident: Compliance in the soil leads to 19 slightly longer natural periods in the structure-soil system than if 20 the base were infinitely rigid. 21 2. Lateral compliance in the soil (as part of the natural mode shape) 22 leads to a greater absolute motion in the soil at the base level than 23 in the true free field.

Blume V-7

This does not mean that the response of the superstructure is increased with softer soil (it may be, or it may not be), nor does it mean that the "tau effect" is not present also. It does mean that the "tau effect" can be overwhelmed, or cancelled, by the soil-structure-system mode-shape phenomenon. However, the ground per se does move more than if the building were not there.

Measured Motions and Res onse S ectra Figure V-4 is a plan view showing the ground

10 record locations and directions at the ICSB. The peak accelerations are also provided.

12 Figure V 5 shows 7% damped spectra for the ~ north-south direction, namely for records 3, 10 and 11, normalized arbitrarily to 0.75g so as to compare shapes.

15 Figure V-6 shows 7% damped spectra for the 16 east-west. direction, namely for records 1 and 13, normalized

17 arbitrarily to 0.75g so as to compare shapes. 18 It is clear from Figure V-5 that the spectra for 19 north-south records 10 and 11 in the ICSB are similar and 20 show much response in the range of 0.1 to 0.4 second period. 21 It is also clear that the spectra from 10 and ll in the 22 building are higher than the "free field" in the range of 0.2 to 0.4 second period. The opposite condition prevails beyond 0.5 sec. Obviously the ground was responding with the building; i.e., the piles were moving with the soil and/or binding into the soil.

Blume V-8

Figure V-6, for the east-west direction, shows similar but not so pronounced effects, especially at 0.3 sec period. Obviously, as others have noted,* the soil at, the building was responding differently than the free field. It was, in other words, responding as part of the mode shape of the soil-structure system. This response is entirely logical for the conditions noted, namely: The ICSB soil is soft and its piles are long and flexible laterally. 10 2. There is inadequate lateral stiffness at the base level to provide for much, if any, "tau 12 effect." 3. Even if there were some "tau 0 effect," the natural mode shapes of the soil-structure system would tend to, and probably did, more 15 than cancel any "tau effect." Only the net effect was recorded. 16 Ground Motion Is Not Uniform

18 In this analysis it can not be overlooked that 19 peak ground motion is quite variable from place to place, 20 even within short distances. The ICSB free field record is, 21 in fact, lower than several other records at even greater 22 distances from the fault. The variation between the free 23

* Rojahn and Ragsdale (EERI Reconnaissance Report.): "In other words the motion recorded at the base of the building incorporates, to some signifiant extent, building/foundation system response."

Blume V-9

field peaks and the other peaks at the building are not greater than between other free field stations at similar distances from the fault. The "Net" Ground Motion The E-W field ground motion reported at record 01 was subtracted from the E-W ground motion recorded at the roof center, the 4th floor center, the 2nd floor center and the ground floor east, which was record 013. The ground floor record of "net" motion (the absolute recorded motion

10 minus the free field motion) shows several predamage cycles where all the motion is in phase. The building and the soil

12 are in a modal response which is reflected in the ground 0 floor net motion. The maximum net ground floor response during the

15 entire record occurred about 6.3 seconds after the start of the record, and is in the order of 0.20g. However, this

17 elapsed time period does not coincide with the maximum 18 recorded ground motion at the free field which was at about 19 7.8 seconds. 20 At 7.8 second elapsed time, the net E-W ground 21 motion at the building ground level was about 0.12g and this 22 corresponds phase-wise with the movement of the structure 23 above. It could thus be proposed that the maximum E-W free field motion of 0.24g plus the 0.12g of net E-W building

25 ground motion, the sum of which is 0.36g, is about the same as the absolute ground motion measured in the E-W direction

Blume V-10 %l of the building, namely 0.33g. The agreement is reasonable considering the nature of the problem. The conclusion is that the record of the building at ground level was not true ground motion but a combination of ground motion and the structural response. This has nothing to do with the "tau effect," which is another phenomenon, probably absent, or at very low level in this case because of the conditions discussed earlier.

~aammar

10 Several reasons have been shown above, any one or combinations of which could account for the measured motions at the building site and at the free field station. It is not possible to determine which factors were most significant from the single event. Four things are certain

15 however:

16 The Hollywood Storage Building and the ICSB are not at all similar. 17 2. The ICSB is not rigid enough 18 laterally at ground level (as is the Hollywood Storage Building, and 19 Diablo Canyon to a much greater degree) to develop much, if any, 20 "tau effect." 3. Nothing in the IV-79 data negates 21 or impairs the tau concept. 22 The Diablo Canyon structures and, site are fully capable of 23 developing the "tau effect" used in the Hosgri reanalysis.

Blume V-11 4 ~ ~ ~ ~ ~

~ ~ ~ ~ ~ I ~ ~

~ ~

Motion Parallel to Montgomery Street Tenth Story

0 0.2 0.4 0.6 0.8 1.0 1.2 Period (sec)

Motion Parallel to Montgomery Street ~ Street Floor a Basement 15 ElBush Street

0 0.15 0.20 0.25 0.30 0.35 0.40 Period (sec)

FIGURE V-2. RESONANCE CURVES FOR FORCED VIBRATION OF THE ALEXANDER BUILDING

Motion Parallel to Montgomery Street

0.1

0. 01

0 ~ 001

O. Fifteenth Story - Earthquake b-Tenth Story - Earthquake 0.0001 Q-Basement - Earthquake ~ -Fifteenth Story - Mind 1-Tenth Story - Forced Vlbratlon I -Third Story - Forced Vibration 0.00001 ~ - Basement - Forced Vlbratlon 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Period (sec)

FIGURE V-3. OBSERVED VIBRATION OF THE ALEXANDER BUILDING

Blume Figs. V-2 and V-3

North

Trace Number

136'10"

340'"

I "Free-Field" Station lA OO

Imperial County Services Building Peak Station Acceleration

1 0. 24g 3 0. 21g 10 0. 34g ll 0. 29g 13 0. 33g

FIGURE V-4. GROUND"LEVEL INSTRUMENT LOCATIONS, IMPERIAL COUNTY SERVICES BUILDING, SHOWING PEAK ACCELERATION FROM OCTOBER 15, 1979 EARTHQUAKE

Blume Fig. V-4

3.00

ICSB Free-Field 2.25 Trace 3

C 0 !i / / x O 1. 50 U 1 O I /' ICSB ! Trace 11 ! 4J 0 ICSB Q 0.75 Trace 10 PGA Trace 3 = 0.2lg PGA Trace 10 = 0.34g PGA Trace 11 = 0.29g 74 Damping, Normalized to 0.75g 0.00 0.00 0.10 0. 20 0.30 0.40 0.50 0. 60 0. 70 Period (sec)

FIGURE U-5. COMPARISON OF FREE-FIELD AND GROUND-FLOOR SPECTRA FROM OCTOBER 15, 1979 EARTHQUAKE, IMPERIAL COUNTY SERVI,CES BUILDING, NORTH-SOUTH DIRECTION

3.00

PGA Trace 1 = 0.24g PGA Trace 13 = 0.33g 2.25

ICSB Free-Field Trace 1 1.50 / r I

i 1 / J ICSB ~ I Trace 13 0.75

74 Damping, Normalized to 0.75g

0. 00 0.00 0.10 0. 20 0.30 O.4O 0. 50 O.6O 0.70 Period (sec).

.FIGURE V-6. COMPARISON OF FREE-FIELD AND GROUND-FLOOR SPECTRA FROM OCTOBER 157 1979 EARTHQUAKE, IMPERIAL COUNTY SERVICES BUILDING, EAST-WEST DIRECTION

Written Testimony

DOUGLAS~H. HAMILTON

Q. Mr. Hamilton, could you please describe the geology and seismotectonics of the Im erial Fault? The Imperial fault is the most active element in the transition'one between the spreading center-transform complex of the Gulf of California to the south and the San 10 Andreas transform-plate boundary fault system to the north. The Imperial, and the en echelon Cerro Prieto fault between 12 it and the head of the Gulf, represent the main locus of 13 inter-plate movement at the latitude of the international boundary. Recognized movement along the Imperial is 15 dominantly right lateral strike slip except, at its northerly 16 end, where partial transfer of motion to the Brawley fault results in downdropping of a block between these faults with 18 consequent vertical slip along each. 19 The Imperial fault extends longitudinally down the 20 axis of the deep structural basin of the Imperial Valley. 21 The basin itself is bounded on the northeast by the San 22 Andreas, Sand Hills, and Algodon faults, and on the southwest in part by the Cucapa fault and in part by uplifted bedrock highs without mapped continuous boundary 25 faults (Figure 1). The area where movement transfers 26 between the northern part of the Cerro Prieto and the

Hamilton V-1 '

southern part of the Imperial faults is the site of the Cerro Prieto volcanic and geothermal field and is thought, by some (e.g., Elders, et al., 1972) to be underlain by a segment of a spreading rise between the two transform faults. Although the spatial distribution of faults in the 6. Imperial Valley region is such as to suggest that the Imperial fault might be considered part of a zone extending southward from the San Jacinto fault, the pattern of seismicity clearly defines a subsurface extension from the 10 Imperial and Brawley faults northward to the San Andreas fault, northeast of the Salton Sea (Figure 2).

12 The Imperial Valley has been the site of very

13 rapid and extensive tectonic subsidence since Pliocene time, about 4 m.y.b.p., when the Gulf of California spreading 15 regime, and with it the current tectonic regime of the San 16 Andreas fault, became established. The structural basin

17 beneath the Valley now contains up to 6 km of sediment. 18 Because it accumulated rapidly, in considerable part under 19 continental conditions of delta flood alluviation, this 20 sedimentary fill is poorly consolidated and consequently is 21 characterized by relatively low seismic velocity. (See V-17 22 ~et se .) The overall tectonic regime in the region also is characterized by both horizontal movement along the 25 transform faults and differential vertical movements 26 associated with the pulling apart and downdropping of the

Hamilton V-2

basin. The latter effect is illustrated by scarps along parts of some faults, and also by geodetic changes in elevation (e.g., as shown in Elders, et al., 1972, Figure 6). Geophysical studies of the character of the basement beneath the Imperial Valley indicate that the crust there is thin and unusually hot. Because of this, the depth of brittle fracture and, hence, of earthquake generation is limited to about 8 km, which is shallow even for California. This condition probably contributes to the high frequency of 10 recurrence and the low stress drop of earthquakes originating in the region. Also because of the high geothermal gradient in 13 the crust beneath the Imperial Valley, the sediments accumulated there are subject to low grade metamorphism at 15 depths as shallow as 1 km. Elders et al. (1972) calculate

16 . that temperature and pressure conditions that would lead to melting of rocks of granitic composition may exist at depths 18 as shallow as 6 km beneath the surface in parts of the 19 Imperial Valley. 20 P-wave velocities in the basin fillsection of the

~ shallow crust in the Imperial Valley range from about 5700 22 to 12,500 feet/sec (1.8 to 3.8 km/sec) with increasing

23 depth. The basement floor of the basin has a velocity of . about 18,000 ft/sec (5.5 km/sec). Near surface velocity 25 measurements in the vicinity of the El Centro array yielded 26 P-wave velocities ranging between 880 and about 1250 ft/sec

Hamilton V-3

to depths of about 3 to 19 feet, between 1380 and

2 4940 ft/sec at depths between about 3 and more than 110 feet 3 (locally), and from 4480 to 5310 ft/sec at depths between 11 and more than 110 feet. The geodetic strain accumulation across the

6 Imperial fault has been determined for the period 1972-1978 as 4.7 +2.1 cm/yr (Savage, et al., 1979). The rate of strain resulting from the opening of the Gulf of California since

9 the onset of spreading there in Pliocene time, has been 10 estimated as about 6 cm/yr (Larson, et al., 1968). The Imperial fault is currently the most active 12 fault in California or, in fact, along the boundary between 13 the Pacific and North American plates, both in terms of seismicity and of frequency of surface offsets. Six 15 earthquakes in the M 6-7 size range have occurred along it 16 since 1852, as well as thousands of smaller shocks. Surface 17 o ffset associated with the 1940 earthquake amounted to a 18 maximum of 5.8 m. right slip and as much as 1.2 m. vertical 19 slip. Lesser amounts of surface offset have occurred on six 20 occasions since 1940, three in connection with earthquakes

,21 on the Imperial (including the M 3.,6 one of 1966 — the 22 smallest shock with associated surface rupture on record) and three from sympathetic response to earthquakes on other, 24 nearby faults. 25 /// /// Hamilton V-4

References - Im erial Fault

Biehler, S., R. L. Kovach, and C. R. Allen, 1964. Geophysical framework of northern end of Gulf of California Structural Province: American Association of Petroleum Geologists Memoir 3, p. 126-143. Crowell, J. C., and A. G. Sylvester (eds.), 1979. Tectonics of the juncture between the San Andreas fault system and the Salton Trough, southeastern California — a Guidebook: Annual meeting Geological Soc. of Amer., San Diego. Elders, W. A., R. W. Rex, T. Meidav, P. T. Robinson, and S. Biehler, 1972. Crustal spreading in southern California: Science, v. 178, p. 15-24. 10 Johnson, C., and D. M. Hadley, 1976. Tectonic implications of the Brawley earthquake swarm, Imperial Valley, California, January 1975: Seismological Soc. of 12 Amer. Bulletin, v. 66, p. 1133-1144. 13 Keller, B., 1979. Imperial Valley earthquake swarms in Crowell, J. C., and A. G. Sylvester (eds.), Tectonics of the juncture between the San Andreas fault system and the Salton Trough, southeastern 15 California; Guidebook for Annual Meeting, Geological Soc. of Amer., November 1979. 16 Kovach, R. L., C. R, Allen, and F. Press, 1962. Geophysical investigations in the Colorado Delta region: Journal of Geophysical Research, v. 67, no. 7, 18 p. 2845-2871. 19 Larson, R. L., H. Menard, and S. Smith, 1968. Gulf of California: A result of oceanfloor spreading and 20 transform faulting: Science, v. 161, p. 781-784. 21 Leivas, E., E. W. Hart, R. D. McJunkin, and C. R. Real, 1980. Geological setting, historical seismicity 22 and surface effects of the Imperial Valley Earthquake, October 15, 1979, Imperial County, California; in Earthquake Engineering Research Institute, Reconnaissance Report, Imperial County, California Earthquake, October 15, 1979, p. 5-19. 25 26

Hamilton V-5

Muffler, L. P. J., and D. E. White, 1969. Active metamorphism of Upper Cenozoic sediments of the Salton Sea Geothermal Field and the Salton Trough, southeastern California: Geological Soc. of Amer. Bulletin, v. 80, p. 157-182. Rockwell, T., and A. G. Sylvester, 1979. Neotectonics of the Salton Trough, in Crowell, J. C., and A. G. Sylvester (eds.), Tectonics of the juncture between the San Andreas fault system and the 6 Salton Trough, southeastern California; Guidebook for Annual Meeting, Geological Soc. of Amer., November 1979. Savage, J. C., W. H. Prescott, M. Lisowski, and N. King, 1979. Deformation across the Salton Trough, California, 1973-1977: Journal of Geophysical Research, v. 84, no. B6, p. 3069-3079. 10

Sharp, R. V., and others, 1972. The Borrego Mountain . earthquake of April 9, 1968: U. S. Geological Survey Professional Paper 787, 207 p. 12

15 16

19 20

25 26

Hamilton V-6

Q. Would you now briefly describe the geology and seismotectonics of the Hos ri Fault? The Hosgri fault is the southernmost segment of the system of faults and flexures that form the boundary between the offshore basins and the coastal uplift of central California. The Hosgri has been mapped over a total length of about 90 miles (145 km) between end points where it dies out in folds north of Point Piedras Blancas on the north, and offshore from Purisima Point on the south 10 (Figure 3). The dominant sense of movement along the Hosgri is right obligue, with an accumulated vertical component of 12 offset since mid Miocene time of about 4 to 5 km. The

13 horizontal component may amount to a maximum of two to four times this amount. The amount of strain represented by 15 direct offset along major strands of the Hosgri fault

16 diminishes toward its end points, although the total cummulative offset along the boundary fault system increases 18 again along the San Simeon fault north of the Piedras 19 Blancas uplift. 20 Along much of its length, the Hosgri fault 21

Hamilton V-7

downwarps of the San Luis, Santa Maria Valley, and Santa 2 Ynez Valley synclines east of the fault. The dip of the main, central reach of the Hosgri 4 is vertical to about 80 degrees east, to depths of 2000 feet; below this depth, it dips about 70 degrees east. The "basement" rocks through which the Hosgri fault cuts are principally the deformed metasedimentary and metavolcanic rocks of the Mesozoic Franciscan Complex, with 9 local remnant bodies of oceanic crust ophiolite suite rocks. 10 Gravity and magnetic studies do not suggest significant 11 differences in the character of the basement of opposite 12 sides of the Hosgri along most of its length. By projection from points where measurements exist. farther east (e.g., 14 Page, et al., 1979), the crust, (above the Moho) in the 15 vicinity of the Hosgri is about 20 km thick. 16 The Hosgri fault has experienced local surface 17 offset in several locations since the late Pleistocene 18 Wisconsinan low stand of sea level, as much as 17,000 years 19 b.p. It has no perceptible general surface expression on 20 the sea floor, however, and therefore has not, undergone 21 extensive or recurrent surface displacement during late 22 Quaternary time. The level of seismicity associated with the Hosgri 24 fault is very low, corresponding to the geologic evidence 25 for only minor movement along it over a span of thousands of 26'ears, and for a moderately low (0.05-0.10 cm/yr) long term

Hamilton V-8 0 slip rate. With the exception, of the M 7.3 earthquake of 1927, which is now considered to have not occurred on the Hosgri (see, for example, response to Appeal Board question IX), published epicenter maps (e.g., Real, et al., 1978; Gawthrop, 1975) show only a few earthquakes as large

as the M 4.0-4.9 range in the close vicinity of the Hosgri trace during the last 50 to 80 years, and historical records extending back another hundred years do not change this

picture (Figure 4). . A recent earthquake (M 4.9, May 29, 10 1980) that occurred in the vicinity of the Hosgri near Pt. Sal has been shown to have a focal mechanism that is not 12 compatible with the typical sense of movement on the Hosgri, 13 and to instead apparently be associated with an offshore extension of a fault along the south margin of the Santa 15 Maria Valley. 16 In the vicinity of the Diablo Canyon site, the 17 Hosgri fault lies along the southwest (seaward) flank of the 18 Point San Luis structural high. The site is situated 5.8 km 19 east of the fault, along the structural inflection between 20 the crest of the Point San Luis high and the trough of the 21 San Luis Range syncline. The site is underlain by 22 moderately deformed sedimentary and volcaniclastic rocks of 23 the Miocene age Obispo Formation, which in turn overlie extensively deformed sedimentary and metamorphic rock of the Franciscan Complex basement at relatively shallow depth 26 (about 3000 feet directly beneath the site). The "basement

Hamilton V-9

complex" rocks are exposed at the sea floor over the crest of the Point San Luis high, between the site and the trace of the Hosgri fault.

10 References — Hos ri Fault

Gawthrop, W., 1975. Seismicity of the central California 12 coastal region; U. S. Geological Survey Open-file Report 75-134. 13 Page, B. M., H. C. Wagner, D. S. McCulloch, E. A. Silver, and J. H. Spotts, 1979. Tectonic interpretation of a geologic section of the continental margin 15 off San Luis Obispo, the southern Coast Ranges, and the San Joaquin Valley, California, cross 16 section summary: Geological Soc. of Amer. Bulletin, Part I, v. 90, no. 9, p. 808-812. 17 Real, C. R., T. R. Toppozada, and D. L. Parke, 1978. 18 Earthquake Epicenter Map of California: California Div. of Nines and Geology Map Sheet 39, 19 scale 1:1,000,000. 20 /// 21 /// 22

24 25 26

Hamilton V-10

Q. Mr. Hamilton, would you now make a geological and seismotectonic comparison of the Im erial and Hos ri Faults? The characteristics of the Imperial and Hosgri faults are compared and contrasted in the following tabular summary. Three main elements of fault characterization are given to establish the comparison. These are 1) character of the crust in the region of each fault; 2) character of the fault itself; and 3) character of sites near each fault, as summarized in the attached tables.

10 The information shown in Tables A, B, and C indicates that, with the exceptions that, the Imperial and

12 Hosgri faults have roughly comparable lengths, and each has a component of right-lateral strike slip to its sense of offset, these faults are dissimilar in nearly all respects. Sites located between 5 and 6 km (as well as over a much

16 wider range of distances) from the fault traces also are highly dissimilar.

18 It is thus concluded that the geological 19 conditions which gave rise to the ground response effects 20 recorded during the IV-79 earthguake do not, exist at the 21 Diablo Canyon site.

23 /// /// 25 26

Hamilton V-11

Table A Character of the Crust in the Re ions of the Im erial and Hos ri Faults Characteristic Im erial Fault Ml' Composition Crystalline Sedimentary and volcanic igneous rock trench accumulation of the Franciscan Complex 6 ,(at least in upper approx. 10-12 Km). Thickness 20 km. (approx.) Seismic P-wave 5-6.2 km/sec 3.2-5.7 km/sec for velocity of Franciscan upper part 10 basement of crust; 6.8 km/sec for crust below about 12 km (Coast Ranges). 12 Heat flow "high" (3-20 "average", without C/100 ft) with unusual concentrations young volcanism of hot springs or and areas of related phenomena geothermal indicative of high 15 fluids heat, flow.

17 18

19 20 21 .22 23

25 26 Hamilton V-12

Table B Character of the Im erial and Hos ri Faults Im eri'al Fault Fault type Right-lateral Right-oblique, with strike slip, fold and reverse fault with rhombo- transition to San chasmic transi- Simeon fault to north. tion to Brawley fault to north Fault length 70 km 145 km overall; 95 km for central part of fault. Fault setting Central, active Fault-offset part of 10 strand of boundary flexure boundary between between offshore Pacific and Santa Maria basin North American and Coast Ranges plates; uplift. originates at, 13 south end as transform across the Cerro Prieto spreading center. 15 Fault geometry Single trace or Single trace locally, 16 narrow zone of mostly two or more traces; vertical major traces, over a zone several 17 near surface, up to 'm dips 70'E at in width; dips 18 base of sedimen- vertical to 80'E to tary section. about 2000 feet depth, 19 70'E below 2000 feet depth. 20 Fault wall Fault cuts Fault cuts Tertiary 21 character through alluvial sedimentary rocks and and lacustrine juxtaposes Tertiary 22 deposits to rocks against Francis- depth of about can complex rocks to. 6 km. depth of 3 to 5 km.

25 26

Hamiltori V-13

Table B (Cont'd) Character of the Im erial and Hos ri Faults Im erial Fault Rate of fault At least 2 cm/yr, 0.05-0.10 cm/yr; local slip up to about surface offsets with 5 cm/yr includ- vertical component ing recurrent totalling 1-2.5 m surface offsets since 10,000-17,000 of as much as 6 yrs. b. p. meters horizontal and 1.2 m vertical. About 7+ meters horizontal and 1.5 meters vertical slip directly 10 associated with earthquakes since April 1940. 12 Associated 6 earthquakes 2-5 earthquakes in the seismicity in the 6.0-7.0 4.0-4.9 magnitude range 13 magnitude range since 1900; no large since 1852; shocks at. least back to thousands of about, 1800. smaller shocks 15 during this 16 interval. 17 18

1'9

20 21 22

25 26

Hamilton V-14

Table C

Character of Sites at 5-6 km Distance From the Im erxal and Hos rx Faults Im erial Fault M ' Setting Imperial basin, Moderately deformed with about 6 km Tertiary sedimentary thickness of and volcaniclastic generally flat- rocks, to a depth of lying Pliocene about 1 km, overlying and Quaternary Franciscan Complex alluvial and basement rock. lacustrine deposits, non- indurated at least in upper 10 several km. Seismic velocity Layered near- Rock at surface with characteristics surface section, velocities mostly 12 with P-wave governed by weathering velocities and local fracturing; 13 increasing from P-wave velocities in 880-1250 fps in upper 10 to 30 feet upper 3-20 feet, range between 2350 to 2000-3570 fps and 3200 fps; and 15 down to the water between 5350 and table at around 5700 fps below these 16 20-40 feet depth. depths for another 20-50 feet. The P-wave 17 velocity of the deepest measurements up to 18 about 100 feet range between 6700 and 19 12,500 fps. 20 Ground water Water table in Fractured rock probably conditions alluvial-lacus- saturated below sea 21 trine sediments, level, at depths of generally at 20 50-80 feet beneath the, 22 to 40 feet depth. site. 23

25 26

Hamilton V-15

Q. Mr. Hamilton, it is my understanding that you have conducted an investigation of seismic velocities in the Imperial Valley. Would you please describe the results of that investi ation?

SUMMARY In July 1980, a seismic refraction survey was conducted in Imperial Valley, California to evaluate seismic velocities of subsurface materials in the vicinity of the El Centro accelerograph array. The seven stations in the central part of the array, and the stations at Bonds Corner, 10 were surveyed with two seismic refraction lines* recorded at approximate right angles to and crossing each other. Com- pression wave velocities were recorded in both forward and

13 reverse directions along all lines and shear wave velocities were recorded in both forward and reverse directions along a

15 portion of one of the lines at each station. The locations 16 of the surveyed accelerograph stations, and of the individual seismic lines, are shown on Figures V-5 and V-6.

18 Profile A-A'hrough the central part of the El Centro

19 accelerograph array was prepared showing a summary of 20 subsurface compression wave velocity zones along with the 21 interpreted ground water table, and is presented as Figure 22 V-7.

(*) For the purpose of this discussion, a seismic refrac- 25 tion line consists of twelve geophones placed at equal intervals in a linear array and monitored simultaneous- 26 ly as shots are detonated off both ends of the array.

Hamilton V-16

Seismic refraction compression wave results indicate that the areas surveyed are generally underlain by low velocity materials (880-1250 ft/sec) to depths of 3-19 feet, typically 5-10 feet, beneath the ground surface. These low velocity materials are generally underlain by low to medium velocity materials (1380-3570 ft/sec) which extend to depths of 11-50 feet, typically 20-30 feet, beneath the ground surface. The low to medium velocity materials are, in turn, underlain by medium velocity materials (4480-5310

10 ft/sec) which extend to depths of more than 100 feet beneath the ground surface. In a few cases, higher velocity

12 materials (6870-8250 ft/sec) may underlie the medium

13 velocity materials at depths greater than 100 feet, but less than 125 feet beneath the ground surface.

15 Seismic refraction shear wave results indicate that the areas surveyed are generally underlain by low

17 velocity materials (410-510 ft/sec) to depths of 6-18 feet 18 beneath the ground surface, below which they increased 19 slightly to 540-690 ft/sec. A few areas surveyed indicated 20 homogeneous shear wave velocities of 580 and 740 ft/sec to 21 the full depth of information obtained (approximately 37 22 feet beneath the ground surface). 23 Table V-D summarizes the results of the seismic refraction survey.

25 The following discussion covers the interpreted 26 results of individual stations, the method and ecpxipment

Hamilton V-17

used, and the limitations involved for the seismic refraction survey. INTERPRETED RESULTS Bonds Corner Seismic refraction lines S-1 and S-2 were crossed at right angles to each other as shown on the exploration map on Figure V-6. The point where these lines were crossed is located approximately 140 feet due west of the U.S.G.S. strong motion instrument. Data and results are presented on

10 Figure V-8. Geophones were spaced at equal intervals of 25 feet in both lines. Interpretation of compression wave data indicates

13 that the site is underlain by low velocity material (about 1140 ft/sec) to depths of 3-19 feet beneath the ground

15 surface. Except for beneath the south end of line S-2, the

16 low velocity material is underlain by low or medium velocity material (2130-2500 ft/sec) which extends to depths of 25-34

18 feet beneath the ground surface. The low to medium velocity

19 material (and the low velocity material at the south end of 20 line S-2) is underlain, in turn, by medium velocity material 21 (4910-5000 ft/sec) which extends to a depth of at least 108 22 feet beneath the ground surface, the approximate depth-limit of compression wave information obtained. Evidence was observed in data from line S-1 to indicate the possibility

25 of higher velocity material (about 8250 ft/sec) near the 26

Hamilton V-18

approximate depth-limit of compression wave information obtained (about 108 feet). Interpretation of shear wave data indicates that the site is underlain by low velocity material (about 450 ft/sec) to depths of 13-18 feet beneath the ground surface. The low velocity material is underlain by slightly higher velocity material (about 690 ft/sec) which extends to at least 37 feet beneath the ground surface, the approximate depth-limit of shear wave information obtained.

10 All materials surveyed at this site are composed of alluvial deposits, and the increase in compression wave

12 velocity to 4910-5000 ft/sec is interpreted to represent 13 saturation of these deposits (ground water table). Im erial Count Services Buildin Station

15 Seismic refraction lines S-1 and S-2 were crossed

16 at approximate right angles to each other as shown on the

17 exploration map on Figure V-5. The point where these lines 18 were crossed is located on a 'lawn east of the Imperial

19 County Services Building (ICSB) between the building and the 20 free field accelerograph station across the street. Data 21 and results are presented on Figure V-9. Geophones were 22 spaced at egual intervals of 10 feet in both lines. Interpretation of compression wave data indicates

24 that the site is underlain by low velocity material (1040-1150 ft/sec) to depths of 7-15 feet beneath the ground

26 surface. The low velocity material is underlain by low to

Hamilton V-19

medium velocity material (1380-1850 ft/sec) which extends to depths of at least 40 feet, the approximate depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that the site is underlain by low velocity material (about 410 ft/sec) to depths of 10-11 feet beneath the ground surface. The low velocity material is underlain by slightly higher velocity material (about 570 ft/sec) which extends to at least 30 feet beneath the ground surface, the approximate

10 depth-limit of shear wave information obtained. All materials surveyed at, this site are composed of alluvial deposits (or possibly fillmaterials) that. are 13 apparently unsaturated to a depth-limit of compression wave information obtained (about 40 feet).

15 Differential Arra 16 Seismic refraction lines S-1 and S-3 were crossed 17 at right angles to each other as shown on Figure V-5. The 18 point where these lines were crossed is 210 feet west of a 19 point, midway between the fourth and fifth substations (along 20 a north-trending line) of the differential array. Data and results are presented on Figure V-10. Geophones were spaced

22 at intervals of 50 feet in line S-l and 25 feet in line S-2. 23 Interpretation of compression wave data indicates that the site is underlain by low velocity material

25 (1090-1140 ft/sec) to depths of 3-9 feet beneath the ground 26 surface. The low velocity material is underlain by low to

Hamilton V-20

medium velocity material (1560-2070 ft/sec) which extends to depths of 25-50 feet. The low to medium velocity material underlain by medium velocity material (4880-5040 ft/sec) which extends to depths of 103-123 feet beneath the ground surface. The medium velocity material is, in turn, underlain by medium to high velocity material (about 6870 ft/sec) which extends down to at least 200 feet, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that (about, 490 10 the site is underlain by low velocity material ft/sec) to a depth of about 14 feet. The low velocity material 12 material is underlain by slightly higher velocity at least 37 13 (about 640 ft/sec) which extends to depths of feet, the depth-limit of shear wave information obtained. consist of 15 All materials surveyed at this site The increase in 16 alluvial deposits or fill materials. compression wave velocity to 4880-5040 ft/sec is interpreted water saturation. 18 to represent the presence of ground

19 Station No. 4 S-1 and S-2 were crossed 20 Seismic refraction lines on V-5. The 21 at right angles to each other as shown Figure west the accelero- 22 lines were crossed at a point 197 feet of on Figure V- 23 graph station. Data and results are presented 11. Geophones were spaced at equal intervals of 25 feet in

25 both lines. 26 Hamilton V-21

Interpretation of compression wave data indicates that the site is underlain by low velocity material (910-1140 ft/sec) to depths of 14-19 feet. The low velocity material is underlain by medium velocity material (4710-4940 ft/sec) which extends to at least 108 feet beneath the ground surface, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that the site is underlain by a single low velocity zone (about 10 740 ft/sec) to at least 37 feet, the depth-limit of shear wave information obtained.

12 All materials surveyed at this site are composed 13 of alluvial deposits or fillmaterials. The increase in compression wave velocity to 4710-4940 is interpreted to

15 represent the influence of the ground water table.

16 Station No. 5

17 Seismic refraction lines S-1 and S-2 were crossed 18 at right angles to each other as shown on Figure V-5. The 19 lines were crossed at a point 130 feet east of the accelero- 20 graph station. Data and results are presented in Figure 21 V-12. Geophones were spaced at equal intervals of 25 feet 22 in both lines. 23 Interpretation of compression wave data indicates that the site is underlain by low velocity material (1090-1110 ft/sec) to depths of 7-10 feet beneath the ground 26 surface. The low velocity material is underlain by low to

Hamilton V-22

medium velocity material (2730-3030 ft/sec) which extends to depths of 24-29 feet beneath the ground surface. The low to medium velocity material is, in turn, underlain by medium velocity material (4910-5310 ft/sec) which extends to a depth of at. least 100 feet beneath the ground surface, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that the site is underlain by a single low velocity zone (about 580 ft/sec) to at least 37 feet beneath the ground surface,

10 the depth-limit of shear wave information obtained. All materials surveyed at this site represent

12 alluvial deposits and are interpreted to become saturated where the compression wave velocity increases to 4910-5310 ft/sec.

15 Station No. 6 16 Seismic refraction lines S-1 and S-2 were crossed

17 at approximate right, angles to each other as shown on Figure 18 V-5, at a point 25 feet north of the accelerograph station. 19 Data and results are presented in Figure V-13. Geophones 20 were spaced at equal intervals of 25 feet in both lines. 21 Interpretation of compression wave data indicates 22 that the site is underlain by low velocity material 23 (880-1040 ft/sec) to depths of 9-12 feet beneath the ground surface. The low velocity material is underlain by low to

25 medium velocity material (2130-2500) ft/sec) which extends 26 to depths of 25-34 feet. The low to medium velocity

Hamilton V-23

material is underlain by medium velocity material (4910-5000 ft/sec) which extends to a depth of at. least 108 feet, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that the site is underlain by low velocity material (about 430 ft/sec) to depths of 6-11 feet. beneath the ground surface. The low velocity material is underlain by slightly higher velocity material (about 540 ft/sec) which extends to at least 37 feet beneath the ground surface, the depth-limit of

10 shear"wave information obtained. All materials surveyed at this site are composed of alluvial deposits and are interpreted to become saturated

13 where the compression wave velocity increased to 4910-5000 ft/sec (ground water table).

15 Station No. 7 16 Seismic refraction lines S-1 and S-2 were crossed at right, angles at a point 200 feet north of the accelero-

18 graph station as shown on Figure V-5. Data and results are presented on Figure V-14. Geophones were spaced at equal

20 intervals of 25 feet. in both lines. 21 Interpretation of compression wave data indicates that the site is underlain by low velocity material

23 (1040-1190 ft/sec) to depths of 4-6 feet beneath the ground surface. The low velocity material is underlain by low to medium velocity material (3330-3570 ft/sec) which extends to depths of 11-27 feet. The low to medium velocity material

Hamilton V-24

is underlain, in turn, by medium velocity material (5250-5260 ft/sec) which extends to a depth of at least 108 feet, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that. the site is underlain by low velocity material (about 510 ft/sec) to depths of 9-11 feet, beneath the ground surface. The low velocity material is underlain by slightly higher velocity material (about 670 ft/sec) which extends to at.

10 least 37 feet, the depth-limit of shear wave information obtained. All materials surveyed at this site are composed

13 of alluvial deposits and are interpreted to be saturated- (ground water table) where the compression wave velocity exceeds 5250-5260 ft/sec.

16 Station No. 8 Seismic refraction lines S-1 and S-2 were crossed at right angles to a point 10 feet south of the

19 accelerograph station as shown on Figure V-5. Data and 20 results are presented on Figure V-15. Geophones were spaced 21 at intervals of 25 feet. in line S-1 and 10 feet in line S-2. 22 Interpretation of compression wave data indicates 23 that the site is underlain by low velocity material (1090-1250 ft/sec) to depths of 5-7 feet beneath the ground

. surface. The low velocity material is underlain by low to medium velocity material (2000-2270 ft/sec) which extends to

Hamilton V-25

depths of 24-28 feet. The low to medium velocity material is, in turn, underlain by medium velocity material (5000-5050 ft/sec) which extends to at, least 100 feet beneath the ground surface, the depth-limit of compression wave information obtained. Interpretation of shear wave data indicates that the site is underlain by low velocity material (about 430 ft/sec) to depths of 13-14 feet beneath the ground surface. The low velocity material is underlain by slightly higher 10 velocity material (about 660 ft/sec) which extends to a depth of at, least, 37 feet, the depth-limit of shear wave

12 information obtained.

13 Materials at this site consist of alluvial deposits and fillmaterials and are interpreted to become 15 saturated where the compression wave velocity increases to

16 5000-5050 ft/sec.

17 METHOD AND E UIPMENT 18 Two seismic refraction lines were recorded at 19 approximate right, angles (south to north and west, to east) 20 at each accelerograph station using a 12 channel seismograph, photographic oscillograph, and associated equipment. 22 The seismic refraction survey procedure used for 23 measuring compression wave velocities consisted of placing twelve 4-1/2 or 7-1/2 Hz geophones in a linear array, spaced 25 at equal intervals of 10, 25, or 50 feet. A 33 pound sledge 26 hammer equipped with impact timing and oscillograph drive

Hamilton V-26

switches was impacted on a flat metal plate positioned at the ground surface 10 or 12.5 feet off both ends of each geophone array to provide energy for near surface seismic information. Shot holes were augered to depths of 3 feet, 25 or 50 feet off both ends of each geophone array using an electric auger and subsequently loaded with 1/3 to 1 pound charges of ammonium nitrate mixed with nitro benzene to, provide energy for evaluating deeper seismic information. Charges were detonated with Vibrodet geophysical blasting

10 caps using a 400 volt electric blaster equipped with a detonation timing device. The instant of impact or detonation (time break)'nd

13 the 12 geophones were monitored simultaneously as compression waves produced by each impact or shot were

15 refracted by the subsurface materials and traveled to and 16 excited each geophone in turn. Compression wave signals picked up by the geophones were amplified by the seismograph

18 (the amplification of signals from individual geophones was

19 attenuated to dampen background noise) and recorded 20 permanently on photographic paper. Photographic 21 oscillograph records were developed in a mobile photographic 22 laboratory in the field and checked for accuracy following 23 each shot or hammer impact. The seismic refraction procedure used for measur-

25 ing shear wave velocities consisted of placing five 4-1/2 Hz horizontal geophones in a linear'rray (along a portion 26 A Hamilton V-27

of one of the compression wave seismic refraction lines at each site), spaced at equal intervals of 20-25 feet. A 33 pound sledge hammer equipped with impact timing and oscillograph drive switches was impacted alternately at both ends of a large wooden plank (8 inches high by 12 inches wide by 8 feet long) placed on the ground surface perpendicular to the geophone array alignment at 10 to 12.5 feet off both ends of the geophone array. The front wheels of a station wagon were parked on top of the plank to hold 10 it in place. The instant of impact (time break) and the 5 shear 12 wave geophones were monitored simultaneously as horizontally 13 polarized shear waves produced by striking the plank were refracted by the subsurface materials and traveled to and 15 excited each geophone. Shear wave signals were amplified by 16 the seismograph and recorded on photographic paper. By alternately striking the ends of the plank, the polarity of 18 horizontal shear waves was reversed, thus facilitating a 19 reliable pick of shear wave arrival times on the record. 20 Both compression wave and shear wave records were 21 reduced and analyzed in the office using computer programs 22 and standard seismic refraction techniques. 23 LIMITATIONS The subsurface velocity profiles of Figures V-8 25 through V-15 represent the most reasonable interpretation of 26

Hamilton V-28

seismic refraction survey data based on our knowledge of existing geologic conditions. The model used for interpretation consisted of multiple gently dipping velocity layers of increasing velocity with depth. Velocity zone inversions (higher velocity materials overlying lower velocity materials) were not interpreted for the sites investigated. The reliability of seismic refraction data for this survey was limited by slightly irregular terrain, 10 cultural objects, and background vehicle and airplane noise. Although avoided as much as practical, these factors

12 produced some scatter in the recorded data and limited the

13 accuracy of velocity zone depth determinations. Because of the wide spacing of shear wave

15 geophones and consequent. low resolution, shear wave

16 velocities and depths of velocity zones should be taken as

17 general indicators of conditions rather than absolute 18 values. 19 Since there was very little existing subsurface 20 information available to correlate seismic information with, 21 subsequent exploration may reveal the need for additional 22 refinement of interpretations. 23 The maximum depth of reliable seismic refraction information obtained during this survey can be assumed to be 25 roughly one-third of the length of the individual lines. 26 For example, a shot detonated 300 feet, from the farthest

Hamilton V-29

geophone in the array will yield reliable data on materials within 100 feet of the middle one-third of the line.

17 18

19 20 21

25 26

Hamilton V-30

TABLE V-D Table Representing Results of Seismic Refraction Survey

July 29, 1980

Bonds Corner

Compression wave velocities: 0 to 3-19 depth, V - 1140 ft/sec ft. cl 3-19 to 12-42 ft. depth, V 2980-3120 ft/sec C2 12-42 to >108 ft. depth, V 4480-4870 ft/sec c3 108 ft. depth? V 8250 ft/sec '? c4 Shear wave velocities: 0 to 13-18 ft. depth, V - 450 ft/sec '1 13-18 to >37 ft. depth, V - 690 ft/sec s2

Count Services Buildin (East Lawn)

Compression wave velocities: 0 to 7-15 ft. depth, V = 1040-1150 ft/sec cl 7-15 to >40 ft. depth, V = 1380-1850 ft/sec C2 Shear wave velocities: 0 to 10-11 ft. depth, V = 410 ft/sec s1 10-11 to >30 ft. depth, V = 570 ft/sec s2

Differential Arra

Compression wave velocities: 0 to 3-9 ft. depth, V = 1090-1140 ft/sec c1 3-9 to 25-50 ft. depth, V = 1560-2070 ft/sec c2 25-50 to 103-123 ft. depth, V = 4880-5040 ft/sec c3 103-123 to >200 ft. depth, V = ™ 6870 ft/sec c4 Shear wave velocities: 0 to - 14 ft. depth, V = 490 ft/sec '1 14 to >37 ft. depth, V = - 640 ft/sec s2

Hamilton V-31

Station No. 4

Compression wave velocities: 0 to 14-, 19 ft. depth, V = 910-1140 ft/sec c1 14-19 to >108 ft. depth, V = 4710-4940 ft/sec c2

Shear wave velocity: 0 to >37 ft. depth, V = 740 ft/sec s1

Station No. 5

Compression wave velocities: 0 to 7-10 ft. depth, V = 1090-1110 ft/sec c1 7-10 to 24-29 ft. depth, V = 2730-3030 ft/sec c2 24-29 to >100 ft. depth, V = 4910-5310 ft/sec c3 Shear wave velocity: 0 to >37 ft. depth, V = 580 ft/sec c4

Station No. 6

Compression wave velocities: 0 to 9-12 ft. depth, V 800-1040 ft/sec c1 9-12 to 25-34 ft. depth, V 2130-2500 ft/sec c2 25-34 to >108 ft. depth, V 4910-5000 ft/sec c3 Shear wave velocities: 0 to 6-11 ft. depth, V 430 ft/sec s1 6-11 to >37 ft. depth, V - 540 ft/sec s2

Station No. 7

Compression wave velocities: 0 to 4-6 ft. depth, V = 1040-1190 ft/sec c1 4-6 to 11-27 ft. depth, V = 3330-3570 ft/sec c2 11-27 to >108 ft. depth, V = 5250-5260 ft/sec c3 Shear wave velocities: 0 to 9-11 ft. depth, V = ™ 510 ft/sec s1 9-11 to >37 ft. depth, V = 670 ft/sec s2

Hamilton V-32

Station No. 8

Compression wave velocities: 0 to 5-7 ft. depth, V = 1090-1250 ft/sec c1 5-7 to 24-28 ft. depth, V = 2000-2270 ft/sec c2 24-28 to >100 ft. depth, U = 5000-5050 ft/sec c3 Shear wave velocities: 0 to 13-14 ft. depth, V = 430 ft/sec s1 13-14 ft. to >37 ft. depth, V = - 660 ft/sec s2

Hamilt:on V-33 'IL SAN EXPLANATION Fault

Exposed basement rock.

l Inferred spreading center PALMSPRINGS 0 bounded by transform faults.

El Centro array station Other stations: x DA Oifferential array X CS' County Services Building BC Bonds Corner

~>c ~ 8p

BORREGO SPRINGS 4 +c~ Jgp

L BRAWLEY

lEL CENTRO~e 13 ~

'i'co'g Fg

SCALE 0 20 40 kilometers GULF OF CALIFORNIA

FAULTS IN THE IMPERIALVALLEYREGION (Modified from Biehler, Kovach, and Allen, 1964) Hamilton, Figure V-1

Seismic stations Towns ~ Epicenters < I.O km

t 7 I k {l940) O M t 6.5 km ( l942)

S ALTON C+ KILOMETERS ~ SEA r~g ~ 0 IO 20 ~ OQ ~ 0 IO MILES 0".:.

r ~ ~

~ BRAWLEY>,~I ~ ~ ro ~ ~ ~ ~ ~ ~ ~ gg

~ ~ ' r

~ EL CENTRO + 32'45'

CAL~FORNtA MEXlcO MEXICALi

I5 I)6'I5jrc 79

Seismicity of Imperial Valley, 1973-78, showing earthquake epicenters located with a precision of 1 km or less. Redrawn by permission from map compiled by Livermore Berkeley Laboratory (Yonder Haare, personal communication, 1979) from Johnson (1979).(from Keller, 1979)

Hamilton, Figure V-2 0 p SCALE )p miles rr

A 1

«t lltlto

~ at

. ~I 1 oslsro r a

t l) .) l Xa OCtattO

OCeanOy ala

Sat/IA +II

/ IS'.r' \

r rara A

Itn/r

rr a rr( Iot//OC

( rr„, rrr, t r1 a / ~

""X:,.FAULTS~BATHYMETRYANO'L'OCATIONOP lMPORTANT =WTR4TIGRAPHIC'FEATURES~COASTAL REGLON.BETWEEN. ~;. -— -"/,—---

- — , /~WE.. POINT,'CONCEPTION'AND CAPE SAN MAR715,' '. .':.: ~ - ', Hamilton,'Pigura St/'~ l..'-A/i't'jI /.) I g-', M g .'", V3>- e XX + + x' (PX X Pj + X~ P + + lt S i ~ X X 'x%<+, x x A X X ++9~ 'x X+ /6 X Xq + X ~ -+ ~ WX X ~ X X XX x + X 0 o X xxt ~ X X +W X X X x ~X + I P4 + x X„ g ,'X g X X X OY'

xx X

..X xt ''. '", X +ps

x X «X-

a ~ x 0 X Pj~i 00 ytj

1 N x/ I Iiixx ~ xxxxxx~ xxNxxx ~ ~ . x JVES '% ~x + 0 S x X lA/Y<

X 0 ==//9.-m~ r ~- Xy X bx +''(6.3, JV/r4, 9. /9P~

+ ~4% @- X X X X

//I. 0 /rib 6 /xb630b/bb CO@AT/0// /4 SI xt 4J xo xxxxxI b 60. Z/. /bx/Z Zllr//00'lIG brlG. 7.0 l 0.$

EPICENTER MAP OF WEST-CENTRAL CALIFORNIA

Hamilton, Figure V4

EXPLANATIONFOR FIGURE V%A EPICENTER MAP OF WEST-CENTRAL CALIFORNIA

ikfrRUMEHrACCY Cd CArfb Akb CARGER //IS Tdl/CACC V Rf'PORTfb flRruQulKE EP/C EKrfRf V/irklk 75 h/ICEJ 0F r/IE,DIAbcb clNYDN po/I/ER pclNr flrE, lddd-/972

f VM80/ NA GA//AUD E 7.9 > M 7.0 P 1 h/A GNI 7 Ubf AA/b 6.9 > M > G.O S DATE Clfrfb'

5.9 > hf > 5.0 0 49 >M y 40 X 3.9 )) M g 8.0 + 2.9 gM > 2.0 (FIGUR E 0 IKDICATEJ N//hid ER OF EPICfk TERS RECORDED AT JlME COCATION.)

JOURCEJ'OR FlUC7 AAID fAR TN QUAKE fPICENTER DATA ARE lf Fd'CCO/Vf) I. Fl UC T DATA FROM JENNINGS) C. Y)f.) /972, 6'ROC'0 6/C h/AP 0F cAC IFDRHIA; foUT/I NAcF. (PRE elhi/NARY) 2. FoR EARTHQUAKES dF 2 40 MAGNITUDE OCCURRIAIG buRING TNE TIME INTER,vlc /954 TNA'U J UAIE 80, /97/: cAc/FdRHIA bivlfldAI dF h/IN'EJ AND 5E 0 Cd GV PR d Vlfid A/A C El RTNQUAKE EPICEAITEP. AIAP, fCA C E I: I, Odd, ddd, l972

Fd R EAR TffQUAKES OF ~ Z d hIAGN'/TUBE OCCURRIAIG bURIAIG Tfff rlh/E IN TE R VA C /952 THRII l9 7/) BUT NOT IAIC C UD/f6 El RTN QulKEJ., GIVEN ONCY AN /NrfkflrV RATING/ ffifMOGRAPffiC fTA TIOH BfRKE CE Y (UNIVERSITY OF CACIFdRAIIA, d ERKEC E V).

4 FOR C ARGE El RTI/ QUAKEJ dCC VRRING b URIHG TIIE TIME IAITER vAc looo TNRdUGH /98/, Td Ivlllcff E J Tlh/ATES MAGNiruDE'lTiNGJ I/AVE SEEK Aff/GHE'b: CACiFoRA/IA DIY/fid A/ dF htIHEJ A AID G E'0C 0 6Y, PRO vlf id Hl C EARrffQul Kf fPICEK'TER h/AP., JCA C f I: I,ddd,ddd, /972. 5. FdR E.ARTNI/UAKEJ OF 2 $.0 MAGAIITUDE OCCURRIHG DURIAIG TNE TiME /N7EA,vAC JUNE'0 rffRu DEC. Zl,/972/ ff/fh/OCOGICAC Cl BURA TORY dF PAfADEAIA, (CACIFORHIA IHJ Tl TUTC dF TECNNOC dGY).

Horfs Fog REvlsf D FAUCr AAID Ed/ fAIT'R blrl

A. FlVCT blTl Rfvlffb IN ACCORblllCE /VITN Kdrf 6, FIGVRf 5 (d/ABCO FSAR)

d. E PIC f// TERS OF /9 Sf C ECrf D El RTNQUAKE'S', RECdhlPUTED BV S. W. SMITH (/974). fPICfNTER- h/AGNITVDf l'VMSOC OF TNEJ'E EVENTS IS /lib/CA TED SY //OR IZOHTA C blJllf S gK-, -oR Nuh/SER ~ J'Ud SCRIPT I//b/Cl TES E'V fHr HVMSE R

o-fPI(ENTER/ (hllGNIrdbf d 4-1o) Rf CPRbfb Affl Co(lrfb bv WICCIlh/ GAWTN'ROP, (IOTA). A/ bffCRlbfb IN 'PRE CI/4ARY RfPORT off l Jffdfr rffh/ I'flfhllC Jl'Vbv OF rfff JAN CUIJ ld/JPP REEION IN Alfv, /97P." b. OL APPRPKIA/lrf EPICENTER /PR /IISlol/ClCCY REPdRrfb EARTROURKf OF 2 hlhf Yll INTENSITY /9/5 Hamilton, Figure V4A - ~ LArc urae4c -r WHIT'COMB DRAIN , BONDS CORNER — I

0 L0, ~l'r I BONDS CORNER DRAIN ~ ~ s O PQll$ I r4

II , 0 r44

~ II ~ ~

HEMCOCK s DRAIN ,P ~: Bonds r'. 'P S1 I "Corner Pr'll S-2 z P 0 0 P rIL R

OCIC

Gaging Sta t /JI DRAIN \ 'IV MEXy.CCO I, ~+pi

SCALE 1:24 000 Earth Sciences Associates 1 0 mile ~ Palo Afro. California

0 1000 2000 3000 4000 feet IMPERIALVALLEYSEISMIC VELOCITY INVESTIGATION BONDS COB NE R, LOCATION MAP

Note: Base map provided USGS,7N minute ~ by Checked by O Project No. Figure No. quadrangle, Bonds Corner, California. Date~ Date ~o 1310 V-6 t „I; ~ I~- !!!i!:!I', i)! !!I' :jt .:!';: :: i';lli s.'i ~ I I !!i~, ::II ':.:Il";:i li 00 ~ t 110 of1 scale at 101 '." In: i:,I: .l:I !I:::)li :::I '. I;: iii! iii; ,I:I i!'!II:I! !!Ii I !!l: i!i! !I!: : ~ A:I :! I jilt ;I:! I :.."'I:: lift! il !II! ::I!. "i.'Ni'i:: itii :II i!i:: !!I':. ,:II! :i i!::::::: !!I 'i/! li)! !!' i'!i!'"':Ij }III) ';! I. I! 1'1 iii ;:i";. ihj ! I I .,".: ! I:! ttn ! ! I ili'j ',i::! "I il!i N!: 'll 'Ir ..:.'.. !)iI II I: ii!! ;I:; Iili ':i!! t I: I 'Ill: ::i: !;i. 'It! :!i! .'is( !'-' I.: lt,': !ii: l).'I Ill: 'i fj „':.L !!f::,:, jj:: j':l I: 'ill li'I xiii:;fbi !II! li'I tn: I:!: !'iii '.I!: !".! II . I:,!:! I; lh Liljl )! )Iij I !i 10 'l[! ! I :,'~j::i '!i!I !j:i r!jt t'jtj ii I;i; I I: I !!!i I!!i Il!I I: i!I':i I! II!!: jl .I!! :I:,i '::.::f i::: t i'Ilj: : ~: A:I !I!I iiii I! N ! I:I! I I;I; !.I,::I! IjriI III,! I;i Ii f ,." I I ! !I 'j!l: O 1!I) 'I'; lj!I 'll: i„'iii! ':f':i! i; ill! !::: '"I ii': ! ii!f(!lILLL 'I: 50 Iji ,—,: rh f, ::II :I: V ii,'i ::iii iii::!:i:: !II'i I I!II !iii if::': ~t !iif ''l' j!i! ji !!ii ii .'!!i , f, ))'t ~ ;i!i '.! '.ii!I ! f I I'l:l li! 'll! I". ii:. 'il! ii!!. II V ~s: ji';I i! ti!i :I j ;its 50 !! :I!i "" 'LLL I I lt., i IT! ~ I I "." I II in Tli.'I;: 1 I'! Ilii i!::.:: :!iI': I !III '.j !:iij!Ijiji!! I;r ll:;:: ! if l',! ll!I !Ih Il! 'jii !!" I!I i'!i::i I};: ij: i):i II:: :)Ii ! !Ii! i.jj'i !iij jjIIIIII eo li ii: !i r.'! fl!i jli fjlI 1 Ii'i!! 'It'f ir',li I7lt II"I i:II t i I 'll I t i!! !Ill i'.I'i liI! '. '. Ilji ;It. 'jjl ilL',:! Il I:!j ~ ; ~ I:. !Il I;! ij t.-. ji!I Ii : I.!! I i!ti ji :: . III. I'i ill! f i!:! I!i IL ill'!':i it I!!I )ii ih) ! !I 'I j ;l. 'I'.:, js'j ii': pl ~~!. it jill tl: II .' 1 ill! '.:i!i:ili lii! ii!! ii!i j i1 ) I!If I !I j.:I it'. I:.!i !:;I .RI !,I'. '.'. I TTi !I .: Ir:f il lh:::I ;';ti j;I: "i I I. ~ in jt ttl: ;I: :. , !It, ::II ii!',: !! iii:"i .!I! !!ji !iii 'I-::I: I !! !i !iii ~!!fIIII 20 I." !IL': I".!: :: I:.: ,' I IV U;I::ji t! I",. '".I ':i:lj'lin I ~: li: !',: .'ll 'iijj .:! I! :'.:':I i I! '::l,l I'' ;:!I ! i!I! ::I: I.ll f iff) I! )Iii II '. I! !)!!:: !t; :II: I I I il:i :tj! li:.; I! tlt; '. In !!j'I II'jl Al !! I: 'I:I 10 ".; U!I j:I! '. '.: i-,! "!ri li:~ ll!! I ii I'i :I 'll! "!' 4 : II; :I !Ij ! Ii jl !! i !:I! !'ltj I!'ll! !it! I I if': 'I:Ii i!i:. jiff '.i'! I'l Ii !)!I !I:i !f iii! ! ji Ih ; I IL i isf Ih: 'E 5 S 't 01$ 'TANCE IN FEE'T S.2

Ground surface1 — 1140 ft/sec S-2 cranes = = = = Vcf =— = Ground surface 8 1 crosses !Vel 1140 Ijfsactc = Vs 1 450 1 thee = =Vc 2980 h/s ae —!VC2 3120hlsec= =. V 800 hhec s Vca 4810 fth~4 ~=f=—Lower Ilmh of stt ~ar w~ se Infermatl VC3 44801thac =

w ic sf ~To 0 iLowtr limitof comprwtlon ware ln =Lower limlI ol corn preulon wase Informs I TV 8250 fthe» 7 M

~ Ilo REDUCED COPY NOTES ,.HOTC5 Eatth Sciences Associates TIME5CALE IN MILLtSECOND55HOVLO . t. TtME~STANccoanettsAT Totot ttouac actaesettf t. OC smsutc act AAcwottsunvcv oATAtDoss ott nor sort tele inn Ettetut DovoLED ton SHEAR WAVEDATAt(o)AHol!i!Il. Ltnc ot ORAetts astaESEHT oeowtoNE LDCATtons, x'5 ReeaE5ENf SHOTeOINf LOCATIONL : LSEISMIC RCtaACTIOHUNC55 I ANO5 t ton IMttrVALVALLEYSEISMIC VElOCITYINVE51ICATIQH ttoattoNTALscALct 1 seteET DDNDs corutER attcATED oN etovac v+. vcnftcAL5CALCt I teMILUSECOHOS DATAAND INTEAPRETEO SUBSUA FACE VELOCI'1Y PAOFILE = Sunsuatncc vecoctf v taot tees AT ooffoMot tsovac t. rlstneSEHf tHTEnetlETA'ftoNSot'etSMtc RetnAC'stoH BONDS CORNEA LINES AND DATA ANO ARE ttrTENOEO fOst OESION tuaeoses OttLY, Dttt cltltQ Haters No rtswt No VEATICALANDHORIZONTALSCALEt I se tECT Atttaetcbf~tLtlt~otttd ErrrD 1310 vtt

' !!f Ii:: I:: !.!t 'I:: !IN II ji!j: I:-: jii!! 90 ii!! !!i! 'I;I I:I 'Il! : IN:! I ! I:I liii !jlj il'!.: !i!i jlU I:I''I @ji: ijiI }i! Illl 90 's', ':r.'}.'. I::i :II! :lit r IU :j'i ', I! iih} iji} I !:II I!II il il I:I;IIIU! Ii! "i t. I }!!i I I !1 i'!!,'!.".: , . ! I 'jjt: !i} till I :ii! 70 j!j! Ij!I .II! I! ! I!! I I't i!!! :II !I I il! :!Ii ji !NI se jii! } ii,:i ji) lil! I: !ii! 0 I! I!'}.': !Ijl li iji I N!I ')Ij! it 'ii! eo L ili j!j} II!. ji!i !ji Il jj: L'! U hii Il I lj I ]I Il:I I!l. w Ilr I! I ijjj ;ii Ij H i''. :lt! i. ! :jii i Iij:: so LL I! iji. jd ;Ui !IU I! Il! I!ii Z !I ~ I ~ 'l !jl iT! ij I }Il'. },::Ij II 'I lllj :II! ! . I ~ o jg!! jjlj II Ui :Iii S Ii ":I Ir !:I: li s ! Ijj I:I ::i -::ij Tti il ! t.; I jl' U!:Ii !U! jj '. I:! I I I I !I! !Ui J}ii I! lr: I ;i ::I.':I II:! :I.!! :jir:jl: ! jl: U!! jl ij i,!:. \ ~ t ii!i rL 'llj ,ii 'I!I } I ! jj!! i!! !jan !:II "'I I! I I I:! ijl: ii!! 20 ~It I il lj ii} ;:I! I:I: :j i:jll !:i, .'i'!.I jij I!I! iil; : it: :I:! ijit I ji,'i! 10 :I I I I I !UI ji .: :fili !LL !II: : ~ I! jjil C I''ii!I : I!! jj jl: I Ui :;Ii ,: ii i 'j ilflf!I' si! I E IS'T ANCE IN FE KT S.2

Groond swfece 6.2 crosses =- Ground ssnfsce S.l crones:

cl -11 10 ft/see=. = =-Vct 1040 ftfsee—

= VI2 670 ft/sec'. =

= VC2 1380 ft/see = s . Lower Ilmftof sheer weve fnfonnetion „=—

Z Lower ltm!I of compression wave Informa don Z = ower limitof comprmston ware Informs 0 r oo

REDUCED COPY NOTES :NOTES TIME OISTANCKORAPH5ATTOPOC fsOUrrc rsEPRESKNT Earth Sciences Associates I. TiME5CALE sN MILUSECOrtOS 5HOULO BE 50$ MIC REP RACT ION SURvEY DASAa DOTS ON BOTTOM DOUBLKDPots SI

~ J ' aJ ~ J ~ I I I! !!I ii I: ! Ii \ ~ ~ i!(I ! jt I ~ :1» ',ll. ii: iji nl '::I I:lj .:I ii Ii ':ii! »j ! I'. ' Ill I I I! :. I!~ '.: I'I nl'ji: I fjf6i ii ~ ic :tft i!'.:. J»! ff ji:i ~ i' '»: »L:::: I Ii ~ : I I I ti;I i! ~ II! :I.'! n;I ji 'lil ti 1 : .I I I \ ~ J lj jil! i'h its il II J !:i: lji,' :i» ':: }LLLjj » iili !j!l "L I ti! II'I I :!IJ I! .'» ! ii ljje !Ijj !I : j I ah !ii; i! i! fili 1 Ill :I» !li li iijj II!! I 1 it Jl! "» I Jl': II :ji: 20 I ii !!II ii iih ! ! II! iii! » i!'. 'l !I Ii(l II I! i'il :». I: T7 i! ji iji! ~ I j1 I I I!; !jj II :I:: I » :,iiif i !Ii jl Ca II :I;i :lii ltj: e I '.I jl i'! lij IJ I',I: :.I: i'i: i»I ;I j(J 0 40 l! lj I' II )i(! ji! I: cs I I J! i'ii ea ii !I I !» 1 l(l Je jjji nl ill( (I» lij i'j i! ji,'i ~ iij! I i 'I 'I i;il ! i I'! Tll 4 I, I', 'iljI ~\ ~ Ii 40 ji (I ii ljj Ii!1 ~ji! I! 2 jj fi! Ij » fi I ~ Ji(: C tj III! ~l;.! I il I~ I! I ~ . I !ljj ! !!I; I I tl aa i!! I.I i'i ji Ijij II 40 » ~ ~ i! jill -'I J jf S liii:".'i!i ji I I! li I T~l(1 lj i » liil JJjf I i I '1 ':i j!li Ti!.'. i:!i I jl,'i ilij »I I 'n I! I!I! Il;j I ~ jji jif! Lii jlLi i IJI! Jljj II !i!i '»I : I(I » I'I ji J Ij:i ! I! Ijj J. Ii!! J ~ iili'! !!jj ll!! ji! Ii h» I jtj: 1(i. '. I, ji IL! !I ! ~ ~ I I. ~ ~ J ~ ~ : I!, I I J ;I!1 I 11, iTTT:I I I if:.i I I :III i!il 'ti j II'.1 ,I j !I ::r:.i F11 Il» ji j .:I! :i I llj: !Ii :I: L t!.: I ii'! ! I ;!I 20 :I:I "! ,'I;i »! I II !!ii ~ 11 'll: J:I I I!! !' I; ~ I :(JI !I: . j! Ij!! !ii:. I! Jj ijj! I'I: I I Jii! J !!i'!Ii ! 'ill! I:I. i:I! ! I li !1!J I:I 'i I"''il !i!! .I.'I ji ''': : I'! I! II "i !!.I l!Ii :»'I: Inj j :I I' It ! II. i :til ij Ii Iji ::li i(J! Ii I» ,i I j l,t IIJ ijit if:: :ji: Aj ':i I I ".!L! J OISTANCE IN I'ECT S2

=Ground = .-. — surface = 9 2 uossas S 1 crosses= =.Ground surface = —Vat 740 fthsc = « '1140 Vcf ft/sac Vct 9(0 ft/sec:

Lower limitof ala su wsye Informsdon =VC2«4110h/aac = «4840 ft/aac =

a 0 —Lower limitot prels ion ware I r 1imitof pres alon ware Informadon

. REDUCED COPY NOTC5 NOTES I, TIMC5CALE IH M!LLISCCO(IOS5HOULO bC TaMC~(STANcc oRAtaas ATTop op FaoURC RCPRE5ENT Earth Sciences Associates OOUbLEO FOIE SHCAR wAVCOATA 5CaaMJC REFRACTJON SURVEv OATAJ OOISOaa eOTTOaa (re) AN0/c!I(. LINEOf ORAPJJS REPRESENT O(otaaoHE LOCATsoN5 X'S tJle Nw, CJJA«tu 5 RCPIIESCNT 5HOT IJOINT LOCA IONS SE!5M(C REFRACT!ONLsNC5 5 I ANO5-5 FOR lalFERIALVALLEYSEISaiIC VELOCITY 5TATIOH4 LOCATEO Ors FIOU!lLV.S. HORIZOHTAL5CALC» I SOFEET INVESTIGATION VERTICALSCALCJ 1 ~ 10 MSLLISECONOS DATAAND INTERPAETED SUSSUA FACE VELOCI1'YPAOF(LE —IL 5Ub5UR FACC VELOCITY PROF!LES AT bOTTOM Os''aoUAC REPRE5ENT!NTS'RPRETATIONS Ot 5E(SMIC REFAACTION A NE ~ AND OATA ANO ARE INTENOEO FOR OC5(ON WRtoaES ONLY Ctaeraad by Oaae d / /0 tao(act Ho Rawe No VERT(CALANOHORIEONTAL5CALCJ I $0 PECT Attreawtby Et/if~Oared.L80 1210 V 11

!Is ji !:I ii ii,:I :I:: le i)}i :II .!!I: is II) ll!I :I: 50 ii }ii ii ! ~ I, 3 n li ,off eccl ~ at fji I: :In ', I' ':i uit i:I :II! I :I:I t: I ji'Ii I Is! :I!: ilj II) !I:I ;;1 'iti I I' I !I iil! if;! I ln I 50 ili i 'j.i) ii!! Iis jig !I!: !)!i I I I I pf t... ill I ttl l)t! I I;! ! '! I '.:: ii s!I '!i)i U if!I . I!. UI: lji ~ji ii:I I' I! I'i'i'!:L'IU eJ ",I'. il: jf;; ~ s ij',i Iln 'i ij ii lj;I I:I'l ij i! !1!i jl lj:I lo i li ji -.;I III: LI!! il!i )I!i jijj I'! I! ij !III I! !'I!, I II ij !Iii TiiI ,':I Iit; illi M ji I !U! L!I :li'i .It jTj) I !)i ipn! O ij I I I I I'! s Irjl ji 'I j i;}i ji.) I! I'.:: !Il! ji ! Jl jl !i!i ii 50 ~~!! Iji! ! ! .I s I! ss I lii liri ii ifi!: I !I:I )ljl I ',I ! if!i ffi! I i''s;.! ''. j;ji i ii I ijj .!4 i !UI i i'i I! !! 'I I'f II I;i, I: II .:';:ji I! !:jl 'IU I!I; 50 ji ;i LL e'e :; I I I! ',I I tl 2 s jell II I il ji ilII I IIlj fil'! li !Il; Tlj: iii! Z li «I I I I'. i!jl li!i !i ;I!I !i! f)I jiii!iiu I '„". ! I I I I jii',I!LI "! I,I I! li ii s '.:'.I fe I I!I! ao LL)'t! ii ijII! I i'I I ~ ' 'jj,'Iit ji "I'li: fj '! !I:I .". II ;I I 'sls i!':, I fiji I! jsj: WITj I!. !tij ini If I I", I.'I! ts 'jl II, I s fl ili ln: I: ~ . '!I: jl I!if !!I I :Ii i ".:1 iii) Is) .),'I ii I:I II!! 'I)'! I ii!! f :I :I;t '.ll 20 :I. ii.. if'li i!!i I'! I!: iffjs lj I II ili: I li, if;i fi:; )I: I:! s' I !III ifi! !!I i!h i', in I I ! li it ji;I II:! ,i '.:t !:I! ji! !I'.: ij ilj iii! :: ."!I j I is iifi i tel 20 i',ii'i ilii ii ilii .'Ill II!! '.j!i 'll: UI jl! !i', )I !'.i',I ':ll I )s jlii i; i'"! jil ls!: ill: I:! il;i ! I! lh !li! il!i i, ! ",:. ',:.';. I!'III I 'I: I I! jl': !II: I! I ~ s !.I:I !!:. !i !'! '.Ill i,: lj, 10 lt I ! II I I i„i)I !ii I I I il i) )!i; !s'!I T~)i ss !Ii !!I'': III! tl I'."„i II! I I ,:')i }:! !I.I )s II 'j;) ".I i!i I! I Ii jl II!: Ill e'! II : '. )I jjiijfi! ft) I ii I ii, JI !i!i If )iI! ':I ! ih: J;) )LL fil II Is i!! l'IETANC S.2 C IN SECT 5

S.l crosses Ground swface = Growed surface- = 3.2 crosses = V R 1110 li/Hc V 1090 tt — : VS2 2730 If!/sec VC2 3030 It/sec V)1 600 I)/sec— I L Lower l(mf of s))ear waee )ntormauon VC3 6310 ft/sac VC3 f010 It/sec~=

s ee R X r limitof ass)On war ~ lnf ~r lim)10 I corn ss)OA ware In ~ ISO

lsc AEOUCEO COPY NOTCS NOTE5

). TSM C SCALE Itr SS ILU5C CORDS SHOULD OE T)ME~STANCE Ot)APHS ATTOP Ot SICURC REPRC5ENT Earth Sciences Associates DDUoLED son 5HEAA $ fesMsc slsSRACreONSURYfY OATA)DOISONOOTTOM tete eaA CsMsee« wAvc DATAI(0AND+. LINEot'teASHs R fpna5Etet CEotHDHc LocATsores x'5 AftAE5CNT SHOT POINT LOCATIDISL L Sf)5M)C Rfts)ACT)OHLINf$5 ) AND 5 t FOR IMPERIALVALLEYSEISMIC VfLOCITYIHVESTSOATSDH STATION 5 LOCATED OH I'IOUAC 5CALEe SO SCCT V.S. HomtottTAL i OATAANO INTEAPA ETE0 FACE VEATICAL5CALEe toMSLU5ECOHO5 SUSSUA l VELOCITYPAOF)LE $etosUAPAcc vfLoc)TYtAossLfs AT sioTToee os stcunc ArtnssfNT INTErspRE'TATIDNSot'f)5eeec tetr'RAc'rsoN STATION 6 LINES S.l ANO S 2 DATAANDAAC INTENOCO t'OR OC$ ION tUAto5C$ Ot)LY CSereedbr l Dne S Sd 'O )caste HO VERTSCALAHOHORSEONTALSCALCs I SO SECT Apppwed br ~~ Dace )ggSO 1310 V 12

I: ll s H ~ ss t ls sl !!ii "I ts !st ij !!ii ! I st 1 i :I I: In ",:l 's s lljs !!::I:!;" I j:I 's LLl. li!; Il off at 117! olf ~ at 113 !Iti It'.! I!" Ii'!i ::II 'In I ~ in I; ill h!1 h :::: !! il::i tin lit: ,:ii :I jii'Ij! Hj! Ill : ; In ili: ."'! n:1: !1 ji j,.l: I ls nit ii4j 'iiiI'I :!!j !si! !in I; 1; ~fir'l I! I I HI :I s ll ! jjjl i} i j!ilj !Ill Itil ji ljii i!!! is: Il!I !I ili! i!,'i lt [j I!Ii jl jl.'; TO i! !jism jl!! I 1111 i lilj jli II II jl II s I s tijj i'd t.i! H jilt Ii !I I !I,'i i!!i j I ij ,~t 0 I Z I I I 60 t liijj I.!i I iil! j !.'i U ij" jt ll !ji I ':i I li :I: I I j !1 ili! ! ii!! hi tlj ~ I jjii il}1 I I. fl :I;! ':: l..'. s 60 j! j~~ ii! i'l!i iilj I ! li ili! iii JJJ ij 2 I tij) jl ll)l PPi) ilII ii 'I: l! !It! Ili) t.. I lijj I jji ijiijj!j !I I 1 oo jjj! IL li In. I' I !I,"! I li; liii fr! II I I lj lj i!i! I I :11 .Ih : I; I: Ii li jl I ~ 'i s I II! !1 jl!i lli: I!!i ::I: Ii "!I Ii i!Ii ,! I:! ij)I I j I!," II ! jl':I! ii I I! I!'I!. I I: ! I:I !Ill ll ihi 'I! l! LI .ii,! j!1 ;"I I. n I;:! !sr Ii, U::I: I 11! 10 '.il! jjj iij! I! j I! liii :it Ir! r: i! .ili .'It :II "iijl~ r '!i! ii )j!I ij 'il '!I s !!! Ii:I !:I: .'i', :I:; ::I: !.Ii tlj! iin s I!:! ji 10 1 !Ill h!! 1 .! IU !I:I s 1',1 I!s !'i! iiii is}i: i,,lj I II Is I I :jil i s ' I!'le lln: I! li!: I: ,':I!: lj I ~ . Il I!Ii I I! i'Ii rji ::II I I 'iL: I Ill' iI isll I !I N OISTANC E IN SCCT

= —8.1 ccosses- Grounsf swhce 8.2 crosws = Crounsf swfaco— 430 ft/sec =- 1040 It sec= V Tn880fl/MC =. = V52 2500 It/sec = - Vc2- 2130 it/sac 12 Ste ft/sac = = Lower lim ~lshoo wt ~ 4 s s nn = VC3 4910 ft/sac = V 6000 ft/sec =

4 -SOO of comtsc eMton wave infor 0 r limit of cosntsre wave in

REDUCED COPY NOTES NOTC5 I. TIME SCALC IN MILLI5ECONOS 5HOULO AC . sMC~5TANCC OAAtsss AT Tot 0t SIOURE rsttnt5tNT Earth Sciences Associates OOUOLE0 S0 4 SHEAR WAVC OATA Es5MsC AEFAACTsON 5U4VEY OATAs OOT5 ON OOTTOM I(et ANOQ. LINEOt OAAtscs Artsss5ENT CcotHONE LOCATIONS ICS rue kan Cstmsus t. 5615MsC RCFAACTIOHssNE55 I ANO5T ton rsttAESENT SssOZ tOINT LOCATIONS SOll STAT1011 5 LOCATCOON SICUAE VAL H04stON'TAs.scALcs I sosccT I SC SERIAL VALLEYSt iSMICVELOCITY INVESTIGATION VCATICALSCALts I soMsLLsSECON05 DATAAND INTERPAETED SUBSURFACE 5Unsss4SACE Vttests TY tRot ILES AT OOTTOso Oc'sGUAE STATION 6 LINA~R1%ND A Con E5tNT SNT acr SAC'rATSON5 Ot 5CS5MSC 4ES AACTSON S.2 OAT*ANOAACsHTCNOEO ='~L = SOA OCssCN tURSOSCSONLY Coarsen by oscejf r Ss 'a rlteseNA VERTSCAL ANO HOAITON'1AL5CALCs 1 SO SCC'T Atossnec bv ~d(kl Oese.%ltd/O 1310 V.'13

i!i'i :i »I »ii jiil fli iii'i,! i I ljj ,',i:I I ;! j ;:',:,' »L .:..:';I :I !i! I jl f: I I I e jjj II :i! I I:: ':i. I I! 40 j jii'-,::i!I I» jii i":!'!i! I »! I. te !!ii jij !Iji I I I 1 i !TTi,'i jj j jill i'!I I» ll:: li j !i iji! I!Ij ~ I».; ',j jl! jji ijjI jil I III IIii ! :»r j, co JLL! ji ;jij J j>l! ,'rle I «I !i »j: Ii !j li llj I}i! I il i i ': jj- I! li !j i/i I-'. Pji 'll i;I'lies I I ~ I el » Iltl Ilj Ij': jj!j 10 il! lI I! , I :»I Jl I! ji ee » » I I: I Pji': „'ii !iii ti: I'i ji \ j: i}L ij,i iiji je it i!f:; , J~lj Ii D¹ ji j,l i. !!l] il't " I 1 l li I jiij I! i!!I ~ij' Ilii r« e 50 lt 'I'ij llllj!j; 'I!! rs I ii !I I,' iT! I i!ii I iji' I li 77~7! I !I Jj !I! ji li Il !.j. ! ~ ! Ii » I»: ilii !I Ii '. II ! jis I; I ~ ! llj: 1 I!: i,'jj I i!'i:::i,'t I.'.! So ti! e il I! i'.:: » I Ijlj t,i! e ' ll'. !i!i I !i I III il! ! I !I ili' u ! I'„'i I! Il !!!i :tij I:I', I!II I:I I j iil! '.. l I Ije II ij Ii jt II ao jl I l))II j!!j :It l j Ill 2 I'I: I1 ~ ~jj!l'!I: it. rr,'! f e I! I I II !I li s li li Jls:, jlie fij III ! I":; e I il! jJI'll!! i"lji » !ii-: ! II ':::I I !ifi I .!!il !it! lt; ili e s'll Il.! III ';;::I i1 jij I 'h!. I IJ: ,'I « I jj» 'i ~ I ':iii fili 'I IL !I!I !'t: I». !iii !I ii!I .:Il :iii j. II 20 jf« !I lj liLL I ! I! ill ! I' I!j ilij :! I!' ;II:! i e »I '.'. eI I fl" I s I II: li! it! llj: I'il '.ll I,', ji J!1: j!i: j}I:: I:» jii,'I',: !If! le :I :Iij i!, fl !I jl !iii Ie ~ I. !:» :: 'i :t! :i'.I jli! !I» I I I 'I! il iij» ! .I e !I i', 20 j li'! ~l: !LI !I'j li: .»! .:I I' II ii!I::il! »! I: r. I I: ijf.: if!i » li l»r !I» j»! '.:; I .!» I!I: »» I !« li 'I» J!!"I I'I » 'iij I'. :I-. !»: :~,I lj I" il» I iijl ;» ii!:i » iij !I ji j it: il!i I.L !i.: jill jj el!! :I i:"i'i! ~ I' 'i.'j 10 i'i'ji «t! » »» lil 1! I,l: ! I I.'.'fI'l'll ri,. ~ »! 'I ~ ! e I: I:,I fjll II:I 11 'I j:I »ir i! e !ii: '.:il !!~ :-jlf I I 'I'i I 'II; !I~ . j!.!II I i'i ';t': 'I » Ii j ll! jil! ej ;, II rl;I !»! !I'! »I !I I I i » JI!I i!ji iL!j 'll »: lj 'iis,: N 1ST ANCE IN FE ET S.2

S 2uouas Vcl 1010 h/sec —„ =:Ground swface— Ground surface- =Vs! 1190 ft/Mc . Vsf 610 h/sec . S 1 uo»a1

V52 3330 It/sae = = V52 «3510 It/sec -Vi2"670 ft/sec V53- 6260 h/sac— = Lower tof ware Information =

VC3 «5250 roc

4 Z r Iunl10 praulon war~ Inf ~ 110 ower Iiudtof pression ware O

REDUCED COPY rroTcS "NOT CS L T su C 5CALC INassLUSSCONO5 SeeOULO nt Teuc~5TA¹cc oRAMs ATTop or'rouoc rettAt5CNT Earth Sciences Associates DOU4LCD FOR 5teauet 4 sf A Acr en¹ 5UAvtY DATAenots DH 4OTTOM 5H CAR WAVCOATh I(o) ANDfh. Leeet of c4Ap¹5 AttntSCN'r otot¹onc LDCATIDNS x'5 fen ctree Ceereee¹ retPRCSSHT SHOT torNT LOCATrONL t. SCISMSC RtfRACTsON UNC$ 5 1 Are05 t FOR Isrt RIALVALL 5 I MICV L I INV STI A -210 STATION 7 LOCATCDON I'IOURC V.S, HOAetONTAL5CALCe 1 softCT DATAAND INTERPRETED VCRTSCALSCALCe 1 SesarLUSCCOHO5 SUBSURFACE VELOCITYPROFILE 5uetu44 Acc YCLDCITYPAofeLcS ATooTToea ot 4 sou nc STATION Att4tst¹r sNTteeteetTATsous of Steseaec 4tfAAcTsoer 2 LINES S.1 ANDS 2 DATAAHD ARC INTSNOCD FOR 055eCN furltOSC'5 ONLY, Cteeraae uf DHO EP/50 reosHSNA 1st aNA VCRTrCALAHO eeOASZONTAL 5CALCe I SO fCCT eeateoeen ur ~CCK Duo CWr CO 1310 V lC

1!)i !I), iil :I: «I: «I i?i ii" }If j if :I« I"if,-'jt» II.! !i :!11 lp! 00 ~ ~ I 107 iii !i!i I''l: )If ',ii li «I: I }I: ~ I :;I! I Ill II I iiji !III «t CO Ij if I« Ilif I!Ii !« J, i!i I! i! jji! I} «I: jj:j 1::! ljlI ij 111! U" 20 i)F rr !« jlj I fiji! rr .I ll O j I 50 l 1 'll ll I »r I !I !I iil Ij! l!li 'II 50 trf "'t li I j)ti I ~WW~F I 2 Iji. } I fjjf I}j 1!lj jfti 1)l pi! ) ? f "i IU )I: ~ 1','ll ,ji !II1 jj ji" fi'1 Ip! ! I !1! : I I i!ii !i!ijli, I ii" 1 to I I iz! "'» I'.: f :I :".:1 i" I ! t! Iih :I» I !i!: jjf: fi fjrj I ':I: 'I", :iil !!Iti Fi I! r ~ 1 !IL ff I'I f)i I I I« iii P!1 I IJ}! I $ 0 1",.: I j if it!I :il I« I i I I:.:: ..: I :t: :;I'!1) :: p'ir 1 i!- I ~ :: I:I) I lt :':.ll: jij lij r !It i i! 20 p.'1! ".1 !I "„':: «I: '.!Ii I I «i. .!I:: I.L !!i I 1:!' '.:.'. !I,", IC I'; I ~ ;Ii: ~ i t}1 !t ir I I!~ ! '. j iltt I!L Ir» I:1;! i)if !I~ if! i» II'.' Il 10 : !I;i I f I! ':i: . II! I'!''I li!l I ~ jrI j !I! ft; if ll! I'!I «I l ill'i Pii iif it" 1:.i! !}i': f ~ 'il '! ! 'II }'i jl ji:!i Ij[l Lt' Ii: )If ! I'! X jfli Od ~ . fo. I)!I C W D ICTANCE IN FE ET S.2

i Ground Huhco 1090 fthrc— Ground ruff Vct 1250 fthre

~ floe 1 - 2000fthre 2270 = : V52 = V,2» 660 fthre = = VC2 fibre C3» ft HC wrr timifo wrro inror i ~ r I I mtt 0f corn rruon wrrr infornu VC3» 6050 ftlrsc:

? ~114 r Iimito wsr~ Inf O

REOUCEO COPY Irorcs No'rc5 Earth Sciences Associates 1. TIMC5CALC IN frlLLISCCOHOS 5HOVLO RE TIME~STANCE CAAtrr5 ATTOP OS SICUAC REPRESENT 5CISMfC Att RAC1IOH SulfVCV DATAtDO15 ON ROTTOM ~m ria, Cru»ru ooueLED son srf EAR wAvc DATAI(r)AND+. LINCOt GRAPHS nftAtSENT OCOPHOHC LOCATIONS X'$ REPRESENT 5HOT POINT LOCATION5. 1,5CISMIC RCSrfACTIOrl LIHC$ 5 I Arfo S 1 SOR Ilrt RIALVALLEY EISMIC VELOCITYINVESTIGATION sort STATfoN 5 LOCATcDDH F f0uRE Vd HORIZONTALSCALCe I ~ SOSCCT DATAANO INTERPRETEO SUSSUR FACE VCRTICAL5CALCr I Fe MILUSCCONDS VELOCITYPROFILE Suruunpncc vtLocf'tvPAotlL55 AT ooTTorr ot ~ rouRc STATION 0 LINES $.1 ANO S-2 ACPR55CNT INTCRPRCTATIOr@5 Ot 5tr5MIC REFRACTION DATAAfroAifE r HTCNOCO FOR DC5fCN PVRPO5C5 OrrLV Clrrtod lr 804 Oafr 1 dtt Antra No fltrrrHo VCrtTrCALAHDHORICOHTAL5CALC»I So SECT APPR»w lr Mtkac DRr ~Ett 1310 v.15

Written Testimony

DR. H. BOLTON SEED

Q. Dr. Seed, you have read the Appeal Board's concerns regarding the question '6 of a tau effect at the ICSB. Please address the estion.

0.21 g 100m 12 0.'23g 13 0.33g t'.33g 15

18 19 20 The recordings of horizontal accelerations in the 21 vicinity of the Imperial County Services Building (ICSB) are 22 shown above. The acceleration records in the building show 23 higher values than the records of free-field motions and thus could lead to the conclusion that tau-effect is not 25 realized in thiq instance. Such a conclusion is not 26 warranted, however, on the basis of this data.

Seed V-1

At least two good explanations exist, for the apparent absence of a tau-effect in this case: 1. Since there is invariably some scatter in observed levels of peak acceleration at nearby locations in the free-field, the record of free-field peak acceleration values is not necessarily representative of the general free-field conditions in the region surrounding the ICSB. Thus if additional free-field records had been 10 obtained, at other locations nearer to or on other sides of the building, they are likely to have 12 shown peak acceleration values different from 13 those of the single free-field station installed about 300 ft from the building. The occurrence of 15 scatter within a limited area may easily result in 16 motions that vary by a factor of 1.5 or more; e.g., in the San Fernando earthguake, Hudson noted 18 that the motions at the Athenaeum Building and '19 Millikan Library Building at the California I 20 Xnstitute of Technology differed by almost 100 21 percent over a distance of 400 m. Furthermore in 22 the recent Mammoth Lakes earthquakes of May 25 and 27, an acceleration in rock of 0.9 was recorded on the abutment rock of the Long Valley Dam while the 25 acceleration in rock at the base of the dam, only 26 200 m pway and some 75 m lower in elevation was

Seed V-2

only 0.25 g, a difference of over 300 percent. Further light on such variations may well be shown when the USGS releases the data from the Differential Array records obtained in the Imperial Valley earthquake.

Thus the average free-field motions in the vicinity of the Imperial County Services Building may easily have been as high as 0.33 g (0.22 x 10 1.5) at other ,surrounding points in the free-field. In fact they may have been even 12 higher since the motions shown by the single 13 recording in the free-field near the ICSB were unusually low as compared to other recordings at comparable distances. Based on the entire set of 16 data available the average acceleration at a distance of 7-1/2 km from the fault was about 18 0.34 g, (see attached Fig. V-1) also suggesting 19 that higher values might have, been developed in 20 some locations in this general area. More 21 importantly it suggests that the motions within 22 the building were apparently about the same as the 23 average for the free-field. 24 25 This still shows no apparent tau-effect for this 26 building —and this is to be expected because

Seed V-3

other factors which might cause differential base motions at other sites would not be expected to be significant at the Imperial Valley Services Building.

e.g., (a) Because of the deep soil layer overlying the entire area around the building, incoming waves are likely to be essentially vertical. This fact, combined with the limited dimen- 10 sion of the ICSB would cause differential times of arrival due to inclined wave effects 12 to be negligible in this case. (b) Because of the relative homogeneity of the deep soil layer, significant differential 15 times of arrival due to non-homogeneity would 16 not be expected. (c) Because of the light weight of the building, 18 the influence of soil-structure interaction 19" on the building would be expected to be very 20 low, provided it were firmly embedded on the 21 ground surface. 22 23 Thus a tau-e ffeet (or a base slab averaging effect) producing lower motions inside the 25 building than representative free field values 26 would Pe expected to be very low in this case.

Seed V-4

2. There is also a good possibility that the configuration of the base of the ICSB and the instrument locations on the base may have caused some application of free-field motions for this particular building.

10 100m

FF 12

15 16 17 18 19 20 21 22 The foundation conditions for the ICSB are shown 23 schematically in the above sketch. If the soil had settled away only slightly from the base 'of 25 the pile caps, as sometimes occurs, the response 26 *of the pile/pile cap/ and overlying 6 ft of fill

Seed V-5 ,I could be very complex leading to some amplification of the ground motions by the time they reached the base slab, 2 ft above the ground surface.

Furthermore, the motions recorded at the ground floor of the building could have been increased by torsional effects since they were located near the edges, where torsion could affect the response of 10 the building. Thus there are valid reasons why the motions recorded in the building could have been amplified so that they were higher than those

13 in the free-field and this would have nothing to do with the presence or absence of a tau-effect.

16 In summary therefore, there are good reasons why tau-effects should not be expected at this building, or why

18 the accelerations recorded on the base slab of the building might. even be amplified over the motions in the underlying

I 20 so3.1 21 This situation does not exist at the Diablo Canyon 22 NPP site, however. At Diablo Canyon motions are likely to 23 have much greater non-coherence because of the non-homogeneity of the underlying rock foundation,

25 non-verticality of waves approaching the foundation, and 26

Seed V-6

soil-structure effects which instead of amplifying adjacent rock motions would cause some attenuation of these motions. Accordingly the data from the Imperial County Services Building showing an absence of tau-effects, should in no way reflect on the appropriateness of incorporating the tau-effect in the design of the Diablo Canyon NPP.

15 16

18 19

20. 21 22

25 26

Seed V-7

I 0

Q Q 8 0.

Q ~0. 34g CP

0 Qi 8 Q 8 Q 0

5 7+5 IO l5 C'.IOSgml- I) Isl ance 4 o~ Fn.ufo' l(~.

Written Testimony

DR. STEWART SMITH

Q. Dr. Smith, could you discuss the ICSB free field peak ground acceleration from IV-79 and whether evidence of the so called "tau" effect was present during that earth ake? The fact that the free field recording near the ICSB recorded a lower horizontal PGA than the ICSB should be viewed in a statistical framework. As seen in the results

10 of Exhibit 1 (e.g., Fig. 5-1), substantial scatter of PGA with distance from an earthquake exists. The degree to

12 which this scatter occurred in IV-79 can be observed from

13 the differing PGAs recorded from closely spaced instruments near the Differential Array (Table V-1). Considering the

15 randomness in recordings, it would be unfair to either prove 16 or disprove a reduction in PGA due to the presence of a 17 large structural foundation on the basis of only one 18 empirical observation. The differences between these 19 recordings are placed in their proper context in Exhibit 1 20 (Section 3.4). This report establishes the anomalous 21 character of the ICSB recordings (ground level ICSB vs. free 22 field) through an extensive examination of other relevant 23 data. Furthepnore, examination of the free field

25 recording near +e ICSB has revealed a potentially anomalous 26 behavior in that recording. As shown in Figure V-l, the

Smith V-1

horizontal recording of the ICSB free field (labeled "El Centro Free Field" and plotted in purple) appears "smoother" than the other free field recordings. The lack of high frequency motion relates directly to the recorded PGA since PGA occurs at higher frequencies. To investigate whether the ICSB free field xecording in fact had less high frequency motion and therefore could have been anomalously filtered in the high frequency region and thus have recorded anomalously low PGA, we displayed the response spectra for

other free instruments 10 Km 10 field within of the fault. This plot (Figure V-2) for the maximum horizontal component demonstrates a systematic lack of frequency the ICSB 12 high in free field recording compared with the other free 13 field recordings. While the reason for this lack of high frequency motion not known, 15 is it helps explain the statistically low PGA at the ICSB 16 free field instrument (approximately 0.10g below the mean). 17 18 Table V«1 19 20 Uncorrected Peak Ground 21 Acceleration (g) 22 COMPONENT STATION: SMA-1 02 43 North-South .51 .29 .28 .24 .42 .45 East-West .37 .38 .37 .38 .43 .39 Up-Down .93 .63 .66 .50 .51 .28 25

Smith V-2

In the case of large foundations, seismic wave motion may vary significantly from one part of the foundation to another. Three separate and complex base averaging effects have been recognized and described collectively as the tau effect. Processes which may contribute to variations in ground motion across a foundation include 6 non-vertically incident waves, random fluctuations in motion due to non-homogeneities along the path of wave propagation, and soil-structure interaction. When a foundation is subjected 10 to accelerations that are not simultaneous, as for example when the wave motion is trying to move the ends of the foundation upward 'and the middle downward, the massive and nature the such 13 rigid of foundation will average motion and thus attenuate it by some factor. The USGS 15 Differential Array in the Imperial Valley provides some new on 16 information the spatial variation in the ground motion over comparable 17 distances with large foundations. Since 18 the instruments are located in a free field situation, the data can only address the questions of angle incidence, and due 20 of fluctuations to inhomogeneities. Although no 21 information on soil structure interaction is some base 22 available, important inferences about the other averaging effects can be made. Both the main shock of IV-79 and a large after-

25 shock are well recorded in this array and provide valuable 26 insight into the significance of base averaging effects.

Smith V-3

Unfortunately, the differential array did not, have radio time signals for synchronization, nor was it triggered from a common source, so the relative time between stations cannot be precisely determined. The waves from the aftershock are remarkably coherent across the array, however, and can be used to provide approximate times (accurate to about 10 milliseconds). Simple ray tracing from the earthquake source shows that the waves approach the array at an oblique angle such 10 that the total delay across the array is only 20-30 milliseconds (2-3 samples, 11 since the sample rate is 100/second). In this 12 situation, which corresponds to nearly II vertical wave incidence, base 13 averaging effects due to the angle of incidence could not be very significant. In the case 15 of non-homogenieties, the lack of precise timing information and a 16 is not critical, good estimate of the upper bound on the reduction factor can be made 18 by aligning the accelerograms such that the peak motion coincides, and then summing across the array at each instant of time. This 20 procedure is in fact the first step in a wave number 21 analysis of the array, and corresponds to the zero frequency component mean 22 or value in a conventional frequency a 23 analysis of time series. More detailed analysis and decomposition of the array data into both wavelength and

25 frequency constituents can be done, but the first elementary 26 step already provides some important information about. base

Smith V-4 V 4 ~ averaging factors. Subjecting'he spatially averaged array data to a response spectrum calculation, and then comparing this spectrum with that, of a typical member of the array leads directly to a spectral reduction factor. This calculation was done, and is displayed in Figure V-3. The spectrum of the array average is essentially identical to an individual member of the array up to frequencies of about 5 Hz. ,Above this frequency, the array sum. has its spectral components reduced, the reduction becoming as large as 40% at a frequency of about 30 Hz. The illustration given here is for vertical components, comparing the array average with

12 station 2 of the array.

13 Thus the conclusion from the preliminary analysis of differential array data is that in the case of IV-79,

15 where the closest approach of the fault to the array is

16 about. 5 km, the base averaging reduction factor for a 214 meter array preferentially attenuates vertical acceleration

18 above 5 Hz, and this attenuation increases to at least 40%

19 at 30 Hz. This experience can be extrapolated to other 20 earthquakes in other areas with certain restrictions. Even 21 though the waves are nearly vertical, base averaging effects 22 are still significant. because they are controlled by the crustal structure and the presence of lateral non-homogeneitieS in the earth. The more non-homogeneous

25 the region, the larger the base averaging reduction factor 26 should be. The relative importance of base averaging

factors in other regions can thus be at least qualitatively determined by comparing conditions with those that exist in the Imperial Valley.

16 17 18 19 20 21 22 23

25 26

Smith V-6

CL CCHThd RhhAT «6 "-«HU5TdH hdAO--- 250 OCGhCC5

fL CCHThd eS "--JRIICS hd40--- i%0 OCGhCC5

CL CCHThd od ---ChUJCKSIIRHK tldRO —- JEO OCGhff5

CL CCHThd RhtlAT si --WHOfhSdH hdRO- - 1<0 OCGhff5

CL CCHThd OJI'I'fhfHTJAL RhhRT 560 OCGRCCS

CI. C

CL CCHThd RhhAY 2 —-KCTSTNC hdRO""" 250 OCGhCfS

0. 00 4. 00 ~ 8. 00 12. 00 16. 00 20. 00 TIME — SECQNOS Smith Pigure V-1 J I HFIXJHUM HQHIPOhlTRt. CGHPQNEHT

EL CGHTAO fAEE flELP--- EL cEHrAa RAART ~ 6 ---HV EL cEHrAO RAART «$ ---JR. EL CEH TAO RAART «8 ---CA EL CEHTAO RAART «i ---RH EL CEHTAO OlffEAEHTlRL R EL cEHrAa RAART «2 --"KE

I P ~ /I

C) 4/„ LLIO /g A./ >o

C) C3 0

G .Oi a. 1 PFR I CJD — SEC Smith Figure V-2 1

'/ ANALYSIS OF THE DIFFERENTIAL ARRAY DATA FROM THE IMPERIAL VALLEY EARTHQUAKE OCTOBER I 5, I 979

l0 9 8 7 6

3 0Z 0 < LU LU U U l.o I~ .9 .8 .7 JJ LA Q 0 LL.

3 4 5 6 78910 20 30 40 50 60 70 8090I00

FREQUENCY (HZ)

ShilTH FIC UHE V-3

VI. Throughout the Licensing Board hearings, parties stressed the role of soil-structure interactions as a mechanism that would reduce the magnitude of structure motion relative to ground motion (e.q., Tr. 8878; 8947-46). Staff and applicant's arguments (in response to

intervenors'uggestion of the apparent lack of tau effect during IV-79) point, to soil structure interactions as the reason for building motion exceeding that, of the ground (Blume Affidavit, Paragraph 10; Rothman - Kuo Affidavit, page 7). (a) Describe and explain the circumstances in which soil- structure interactions produce enhanced or reduced structural response. (b) Discuss the relevance and applicability for such interactions to the seismic response assumed for Diablo Canyon.

Written Testimony

DR. JOHN A. BLUME

Q. Dr. Blume, could you please define soil-structure interaction and describe those situations in which it can enhance and or decrease buildin res onse?

In its broadest sense soil-structure interaction (SSI) accounts for all the many differences between the response of a building on assumed infinitely rigid rock and

10 that of the same building on soil or soft rock with compliance. For example, the natural periods, mode shapes,

12 and damping of a building on soft soil will differ from

13 those of the same building on hard rock. The soil site may "tune" 14 have a characteristic period or periods which tend to 15 with those of the building-soil system and thus amplify the 16 motion over that of rock. In addition, there are phenomena 17 such as wave scattering, radiation, wave filtering or 18 averaging due to non-coherence of wave form and foundation. 19 There are effects due to embedment of the structure below 20 the ground surface, and possible slippage effects. There 21 may also be feedback of energy from the structure to the 22 ground. While some of these factors cause enhanced motion, 23 most. of them cause reduced motion. When applied to Diablo Canyon, these factors result in a significant reduction.

25 In the case of Diablo Canyon, the base material is 26 jointed rock and there is some embedment. The latter was

Blume VI-1

conservatively ignored and, in accordance with NRC P standards, the rock material was considerd rigid; thus many other aspects of SSI were also conservatively ignored. However, the non-coherence of ground motions at the foundation level was considered for the larger structures and was included in the term "tau effect." The wording in my affidavit (submitted with Applicant's response to the Motion to Reopen) on tau apparently was not clear. That discussion was limited to a

soil-structure ~s stem, 10 not soil-structure interaction in general. The ~sstem is that. of the building, the piles and the soil of the ICSB, which ~astern is only one aspect of a great many soil-structure interactions. Diablo Canyon is on rock with no piles, and that system was conservatively treated as on The 15 simply individual structures fixed bases. ICSB and Diablo structures are comparable. 16 not in any way When there is compliance in the soil, such as for the ICSB, 18 the true vibrating system becomes not just the a new 19 building but soil-structure system with different than a 20 properties for vertical cantilever fixed at its base. The lowest 21 nodal point is not at the foundation level but sometimes below schemati- 22 that level. This is illustrated cally in Figure VI-1 in which the building on the left is on hard rock and the one on the right on soft alluvium. The

25 hypothetical fundamental mode shapes are shown by the dash 26 lines and the nodal points by the circles. Superstructures

Blume VI-2

A and B are identical and yet the mode shapes are different. The ground displacement for B, due solely to modal response and not to the earthquake motion, is shown as "x". For A, it is zero. Given the same earthquake shaking for each building, the measured peak ground displacement for B would be greater than for A, by the amount "x". This is a phenomenon that enhances ground motion displacements (for B) as compared to either A or the free field. This was ICSB because 10 apparently experienced for the the conditions were right —deep soft soil, high water table, long flexible piles, etc. Unlike ICSB, the nature of the ground and structures at Diablo Canyon results. in a decrease of response.

15 16 17 18 19 20 21 22 23

25 26

Blume VI-3

I l l I r

r Mode A /Mode Shape r Shape / / / l I X / / r /' / / gj/@II//~%/lire.I/i/All~

Hard Rock Soft Alluvium

FIGURE VI-1. EFFECT OF SOIL COMPLIANCE ON MODE SHAPES

BIume Fig. VI-1 ~

Written Testimony

DR. H. BOZTON SEED

Q. Dr. Seed, would you please discuss "soil-structure interaction" in general and specifically as to whether that phenomenon can increase and/or decrease structural res onse? Soil-structure interaction is a complex phenomenon which is generally understood to mean the effect of physical 10 interaction between the building and the adjacent soil or rock on both the response of the structure and the motions developed in the base of the structure.

13 However, many engineers have considered the subject to have a broader context and to include such

15 additional effects as 16 1. The development of strong structural response 17 because of the similarity between the natural 18 period of a building and the characteristic period 19 of the soil formation on which it is constructed. 20 2. The effects of non-coherence of ground motions at 21 the base of a structure and the homogenizing of 22 these motions by a rigid base slab in a structure. 23 3. The variations of ground motions with depth and the effects of this variation on structure

25 response. 26

In the testimony presented at the ASLB seismic hearing, the term soil-structure interaction was generally used in the more restrictive sense mentioned above. Possible effects of similarities in periods of buildings and site were not given any special terminology, the non-coherence of ground motions at the base of a building was generally referred to as the tau-effect, and variations of motions with depth were not considered because the structures under consideration are relatively near the ground surface.

10 Physical soil-structure interaction of the type discussed above has been investigated extensively over the '20 12 past years and in general it would appear that the.

13 magnitude of the effects depends on (1) the relative stiffness of the structure and the

15 foundation medium on which it rests, and 16 (2) the physical size or mass of the structure 17 involved. 18 Thus for example, soil structure interaction effects 19 increase as the ratio of building stiffness to soil 20 stiffness increases, and the effects are greater for massive 21 structures such as nuclear power plants than for more 22 conventional buildings. 23 Where a conventional building is firmly founded on a soil or rock deposit, it seems to be generally agreed that

25 soil-structure interaction effects tend to reduce the 26 response developed at the base of a building compared with

Seed VI-2

those developed assuming rigid base conditions. The magnitude of this reduction in conventional buildings generally ranges from zero to a few tens of percent. The effects of soil-structure interaction of this type are considered in the "Tentative Provisions for the Development of Seismic Regulations for Buildings" -- the most recent guide to seismic design developed by a group of distinguished earthquake engineers representing all parts of the United States, and the guide allows some reduction,

10 typically of the order of 5 to 20 percent in design forces, depending on the influence of the soil and building stiffnesses as discussed above. For nuclear power plants soil structure .interaction may be significantly larger, ranging from about 10 to 50 percent.

15 Q. Please discuss the effects of soil-structure interaction and its role for the structures at Diablo Can on. The influence of soil-structure interaction at the Diablo Canyon NPP was studied by Professors Seed and Lysmer lg who concluded that the effects would be small (for nuclear 20 power plants). Typical results of the Diablo Canyon study 21 are shown on the attached Figure VI-I (Fig. 13 from their original report) and show that soil-structure interaction

23 would reduce the spectral accelerations for the base slab motions by about 20 percent below those of the free-field

25 motions for frequencies above 4 or 5 Hz. It is in this

Seed VI-3

range of frequencies where the tau-effect has its major influence. The Hosgri reanalyses of the structures at Diablo

Canyon NPP are based on the assumption that the structures are founded on a rigid base. This neglects completely any soil-structure interaction effects of the type discussed above and illustrated in Figure VI-1 and thus in this respect is conservative. It leads to the result that the base motions are the same as those in the free-field and

10 thus to the use of horizontal base spectral accelerations at frequencies abov'e 4 or 5 Hz which are about 20 percent higher (the actual amount varying with frequency as shown in

13 the figure) than those which would be computed with allowance for soil-structure interaction effects due to

15 foundation compliance. Other types of soil-structure 16 interaction -- such as the base slab averaging effect due to

17 the non-coherence of ground motions over the area of the 18 slab —would reduce the design motions still further,

19 perhaps by as much as an additional 20 percent. Thus there 20 are ample grounds to justify the magnitude of the base 21 motions used for the Hosgri reanalysis of the Diablo Canyon 22 NPP structures. While the main justification for using a 23 tau-factor to reduce motions in this case was developed on the basis of non-coherence of ground motions, consideration

25 of all types of soil-structure interaction could have led to 26 even greater reductions than those used.

Seed VI-4

In other cases, soil-structure interaction effects in their broadest connotation have, been shown, both

' analytically and observationally, to reduce base motions by as much as about 55 percent (e.g., for the Humboldt Bay NPP in the 1975 Ferndale earthguake) but I do not know of any case, for a structure seated ~firml on the ground, where soil-structure interaction has caused an increase in peak ground accelerations, though theory would indicate that it might do so by something of the order 'of 5 percent in rare

10 cases. Diablo Canyon is not such a case. However, increases could occur for a structure not

1 12 seated firmly on the ground; for example, a structure 13 supported on piles, especially where the base of the structure is supported on pile caps and the ground settles

15 away from the base of the pile caps. In this case the 16 ,foundation system (piles, pile caps, and slab) could behave

17 almost like an extra story to the building, especially where 18 additional geometric features like raising the ground 19 surface are involved, and amplify the motions developed at 20 the base of the pile caps to higher values on the overlying 21 slab. Relatively little is known about such effects but the 22 possibility clearly exists and could be one reason why the motions developed in the base of the Imperial County Services Building may have been higher than those developed

25 in the free-field. A number of engineers believe this to 26 have been the case but an alternative explanation (that the

Seed VI-5

motions at the free-field recording station were abnormally low) is also possible. On the bases of the uncertainties surrounding the motions recorded at and near the Imperial County Services Building I do not think it, is possible to draw conclusions concerning the effect of soil-structure interaction for this building. This does not in any way invalidate the basis for the inclusion of a tau-effect in the Hosgri reanalysis where

10 both soil-structure interaction effects and incoherence of ground motions give a sound basis for allowing the reductions taken in horizontal accelerations at higher

13 frequencies.

18 19 20 21 22 23

25 26

Seed VI-6

R-Wove Fice Field N.P. 63

Rock Rock

P Wove

o S-Wove

S crd P Wove Analysis . Horizontal Molicns S and P Wove Analysis - Vertical tAolrons - r ) = —Fice iield ——Free. held Base of Stivcture (H P 63) —Bast ol Structure (NP 63) c — I oc 4

u uCl Cl u 3 .J I'1 C 1 c J'. 0 u 2 2 u u C' Q 0 1 in

0 )0 2 4 0 20 40 04 (0 2 4 8 23 40 - - Frequency it z F iecvency Ht

I I Rcyteir)h Wave Anclysis . tioiizontol Votions RoytciCh Vrov» Anolysis - VerliCol Motions - — Iield ~ — Fice- ——Fr e e.! ~eld Bose ol Structure (R P 63) —Bcseol S:rvctuie (:tP 63) — P. 4l. c 4 n V o u C' u u C u u 2 2 ui vi rn I

(> 04 tO 2 4 B 20 40 )0 2 4 B 40 Frequency- ltt Frequency -Hz

Fig. 13 SOIL-STRUCTURE'NTERACTION EFFECTS AT BASE OF STRUCTURE (Taken from Seed and Lysmer report on "Analyses of Soil- Structure Interaction Effects During Earthquakes for the Diablo Canyon Nuclear Power Station" )

S(nell FIG. VI-l

VII. Intervenors (Brune Affidavit, page S) and the applicant (Frazier Affidavit, Paragraph 3) have suggested that the strong motion data obtained from stations along the direction of the Imperial Fault evidence the "focusing" of earthquake motion. Yet, when the acceleration data of two such stations, El Centro Array Numbers 6 and 7, are plotted as a function of distance from the fault (e.q.', Blume Affidavit, Figures 1 and 2), the horizontal acceleration values fall well below the regression line mean for the 1 km distance. The vertical acceleration values are also lower than the mean on such a plot. To the extent possible, the parties should analyze the seismic records for the IV-79 earthquake as they pertain to the focusing phenomenon and relate the results of such analyses to the likelihood that, in the event, of an earthquake anywhere along the Hosgri Fault, focusing might result, in amplified seismic motion at Diablo Canyon.

Written Testimony

ROBERT B. EDWARDS

Q. Mr. Edwards, have you been able to ascertain whether focusing significantly affected horizontal peak ground accelerations in IV-79?

Strong motion data recorded during the Imperial Valley earthquake of 15 October, 1979, have been analyzed for evidence of rupture focusing. The'nalysis yielded the

10 result that the horizontal PGA components tend to "saturate" at normal distances of less than approximately 10

12 kilometers. The PGA values for the transverse (normal to the fault) and longitudinal (parallel to the fault) components are roughly constant and nearly equal to one

15 another at these close-in distances. 16 Hypotheses which account, for changes in the character of the radiation pattern of a point dislocation

18 strike-slip source due to rupture propagation usually

19 conclude that of the three components of motion associated 20 with the source (transverse, longitudinal, and radial), the 21 transverse component is generally considered the one most 22 strongly affected by rupture propagation, followed in 23 importance by the radial and longitudinal components, respectively. These intensifications of amplitude of motion

25 in the direction of propagation may be loosely termed 26 "rupture focusing."

Edwards VII«1 M

4 In his testimony at the ASLB seismic hearing, Dr. James Brune stated that the simplest theoretical representations of the phenomenon of rupture focusing \ suggest that enhanced high frequency accelerations might be

produced in a narrow beam (+5 ) in the direction of rupture propagation under special conditions. Figures VII-1 and VII-2 show uncorrected PGA values for the horizontal components of motion from the 1979 Imperial Valley earthquake vs normal distance from the fault. The close-in

10 stations of the El Centro array (those with normal distances I to the fault of 10 km or less) show relatively uniform

12 values of horizontal PGA. A +5'beam" in the direction of

13 rupture emanating from the epicenter includes stations 6 and 7 of the El Centro array. If the simple predictions I based on Brune's testimony are taken literally, then we must

16 conclude that any rupture propagating on the, fault from the 17 epicenter toward stations 6 and 7 would result in focusing 18 being observed at those stations. However, the records from stations 6 and 7 provide no such evidence, according to the

20 data presented in Figures VII-1 and VII-2. 21 High vertical accelerations also were recorded 22 within 10 km of the Imperial fault. These vertical PGA 23 values show a saturation effect similar to that of the horizontal components, though the saturation appears to

25 occur closer to the Imperial fault for vertical motion 26 (compare Figures VII-1 and VII-2 with Figure VII-3).

Edwards VII«2

Appli'cation of the narrow beam focusing criterion to the vertical PGA val'ues for stations 6 and 7 fails to provide conclusive evidence for rupture focusing of the vertical high frequency motion. Available evidence indicates that the vertical PGA value recorded at station 6 was anomalously high (possibly by as much as a factor'f 2). Under these

circumstances, ~ the PGA values of station 6 would not differ significantly from vertical PGA values for station 7 and stations further out.

10 The fact that the vertical PGA values were comparable to the horizontal PGA values within approximately 5 km the 12 of Imperial fault can also be attributed to the nature the 13 of local tectonics of the'Brawley-Impe'rial fault system.

15 Focusing accompanies all earthquakes to a greater

16 or lesser degree. The rough equivalence of the transverse and 17 longitudinal PGA values within a few kilometers of the 18, Imperial fault is not inconsistent with some focusing 19 effect. However, such focusing, if it occurred, did not 20 result i;n abnormally high PGA values. It seems to us most 21 significant that any focusing which might exist does not 22 appear to alter the broader tendency .toward saturation of PGA 23 values in the near-field. The rupture focusing phenomenon is not relevant to the Diablo Canyon site in

25 light of the facQ that, the portion of the Hosgri fault trace 26 which may be "liped up" with the Diablo Canyon site (+5 ) is

Edwards VII-3

so far from the site (nearest approach is -=27 km) that any amplification by high frequency focusing would be eliminated 3 through material damping of the high frequency radiation. (Brune ASLB seismic hearing testimony.)

12 13

15 16 17 18 19 20 21 22 23

25 26

Edwards VII-4

1.0 Bonds Corner

k k

0.1 1.0 10. 0 Norma1 Distance to Imperial Fauit (km)

FI GURE VII-1. PEAK GROUND ACCELERATION FOR TRANSVERSE COHPONENT VERSUS NOfNAL DISTANCE TO IMPERIAL FAULT, OCTOBER 15, 1979 EARTHQUAKE

Edwa rd s F i g. VII-1

10. 0

1.0 C Corner 0 ~Bonds 4J ~ ~ fD ~ 3 Ol 4) V O

~ ~ e Q 1

0.01 0.1 1.0 10.0

Normal Distance to Imperial Fault {km)

FIGURE VII"2. PEAK GROUND ACCELERATION FOR LONGITUDINAL COMPONENT VERSUS NORIIAL DISTANCE TO IHPERIAL FAULT, OCTOBER 15, 1979 EARTHQUAKE

Edwards Fig. VII-2

10. 0-

1.0

Bonds Corner

v v

0.1 VT

0.01 0.1 1.0 10.0

Normal Distance to Imperial Fault (km)

FIGURE VII-3. PEAK GROUND ACCELERATION FOR VERTICAL COHPONENT VERSUS NORHAL DISTANCE TO IHPERIAL FAULT, OCTOBER 15, 1979 EARTHQUAKE

Edwards Fig. VII-3

PROFESSIONAL QUALIFICATIONS OF ROBERT B. EDWARDS

3 Title or Position: Senior Geophysicist, URS/John A. Blume 8 Associates, Engineers. Degrees: B.S. Physics, Worcester Polytechnic Institute, Massachusetts, 1965 M.A. Geophysics, University of California, Berkeley, 1975.

Professional Experience: URS/John A. Blume S Associates, 10 Engineers, Senior Geophysicist, 1977-present University of California, Berkeley, Research

12 Assistant, 1975-1976, 1971-1973 13 U.S. Geological Survey, Topographic Field Assistant, 1974-1975 U.S. Naval Nuclear Power School, Instructor,

16 1966-1970

17 Mr. Edwards'reas of specialization include 18 seismology, geomagnetism, and geophysical field 19 measurements. He conducted a gravity survey near 20 the Sandia Laboratories in California to appraise 21 previously published reports of active faults 22 located within the immediate vicinity of the site of the laboratories, and analyzed the reduced data in terms of theoretical models. 25 /// 26 '// l Mr. Edwards has also perfoxmed seismic velocity measurements to determine dynamic elastic properties of subsurface materials. He has reviewed the Environmental Impact Statement regarding seismic hazards of the Elk Hills Naval Petroleum Reserve in California for the U.S. Navy and has taught courses involving geophysics, seismology and nuclear physics. Mr. Edwards was involved in developing a 10 computer code to predict earthquake ground motion as a function of depth. He has also conducted research on intensity data associated with the

13 '1927 Lompoc earthquake. Mr. Edwards is working on a project to quantify earthquake intensity in 15 terms of recognized seismic parameters such as

16 acceleration, frequency, and duration. Currently, 17 Mr. Edwards is co-principal investigator in a 18 research project to evaluate techniques in, 19 measuring near-surface attenuation properties of 20 soil and rock. This work involves development of 21 analytical techniques to remove near-source 22 effects from attenuation data. He is also 23 conducting a project to investigate near-field ground motion recorded during three recent

25 California earthquakes. 26

Written Testimony

DR. GERALD FRAZIER

Q. Dr. Frazier, could you discuss the henomenon of focusin in eneral?

Recorded ground motions tend to be higher in the direction of spreading rupture than in other directions due to focusing of seismic energy. Conversely, for the case of 10 unidirectional rupture, recorded ground motions tend to be lower in the direction opposite to rupture growth due to defocusing of seismic energy. The bias of large amplitudes of motion in the direction of spreading rupture (i.e., the phenomenon of focusing) has been theorized, and indeed 15 observed, for intermediate to low frequencies (less than 16 about 3 Hz) for more than a decade. One would expect to see 17 the primary effects of focusing in the lower frequency 18 parameters of peak ground velocity and displacement as 19 opposed to the higher frequency parameter .,of peak ground 20 acceleration. As has been previously discussed by Dr. . 21 Blume, the parameters of velocity and displacement are 22 outside the frequency range of primary importance for the 23 Diablo Canyon structures. The effects of focusing on higher frequency ground motion, including peak acceleration, are 25 discussed below. 26

Frazier VII-1

Focusing can be understood in terms of time compression of signals, i.e., the familiar Doppler effect. Consider a unidirectional earthquake that ruptures due north and emits seismic disturbances for a duration of 10 seconds. Because of the approaching rupture, an observer in the near field and north of the source experiences strong shaking for a duration less than 10 seconds, say 6 seconds. The fact that 10-seconds-worth of seismic energy arrives within 6 seconds tends to increase the amplitudes of ground motion 10 in the direction of rupture focusing. Conversely, an observer in the near field, south of the source, experiences 12 strong ground shaking for a duration longer than 10 seconds 13 which tends to decrease the amplitudes of moti.on in the 14 direction of rupture defocusing. 15 Actual earthquake rupture spreads in a somewhat 16 irregular manner, lurching and altering directions in 17 response to stress aberrations and material asperities. The 18 consequence is that, the bias toward large amplitude motions 19 in the path of focusing is subdued for frequencies greater 20 than about 3 Hz with significant bias present for peak 21 velocity and lower frequencies. The effects of rupture 22 focusing and defocusing on lower frequencies are widely 23 observed at distances ranging from near field to teleseismic 24 (more than 1000 km). 25 Data obtained prior to the occurrence of the 1979 26 Imperial Valley earthquake indicated the subdued nature of

Frazier VII-2

focusing at high frequencies. Relevant data were recorded for the 1966 Parkfield earthquake. An accelerometer (Parkfield Station 2) was positioned directly in the line of about 30 km of approaching rupture. The instrument recorded 0.5 g in the horizontal direction. About 5 km away, 6 perpendicular to the surface break, Parkfield Station 5 recorded 0.45 g, a value nearly as large as that directly in 8 the beam of maximum focusing. In contrast, a peak velocity 9 was reached of 80 cm/sec in the beam of focusing 10 (Station 2), a value nearly three times greater than that 11 recorded at Station 5 (28 cm/sec). Additionally, in the 12 Pacoima Dam recording of the 1971 San Fernando earthquake, 4 the effects of rupture focusing are more apparent in the velocity pulses than in the acceleration peaks. 15 The 1979 Coyote Iake earthquake was an example of 16 focusing affecting all three parameters but with the primary 17 effect on the low frequency parameter of displacement. The 18 peak acceleration was 0.42 g directly in the path of focus- 19 ing (Station 6) and 0.25 g within one kilometer of the 20 fault, but, in the path of defocusing (Coyote Creek). The 21 corresponding values for peak ground velocity are 44 and 22 20 cm/sec at the two stations. While effects of focusing are only slightly more apparent for velocity than accelera- 24 tion, the recorded displacements of 9.3 cm and 2.4 cm, 25 respectively, represent a substantially larger effect due to 26 focusing than that for the higher frequency acceleration peaks.

Frazier VII-3

Q. Dr. Frazier, could you discuss focusing as it may or may not have occurred durin IV-79? Recordings of the 1979 Imperial Valley earthquake 4 provide further evidence on the limited effects that rupture focusing has on increasing peak accelerations. At least five strong motion recordings were obtained essentially on 7 the fault (within 1 km) at various points along the path of 8 rupture. The horizontal accelerations of these stations 9 range between about .25 and .75 g with a mean value well 10 under .5 g. These values fit well within the range of peak horizontal accelerations cited above for the Parkfield and 12 the Coyote Take earthquakes. These values are also 13 consistent with the values obtained at stations out 6 or 14 7 km from the fault. Furthermore, there is no significant 15 difference between peak accelerations recorded south of the 16 epicenter, in the defocused zone, and those recorded north 17 of the epicenter, in the focused zone. There is apparently 18 no significant increase in peak accelerations as a result of 19 focusing in the IV-79 earthquake data. 20 Q. Dr. Frazier, it has been suggested that modeling for rupture of the Hosgri 22 fault, including focusing effects, should be done. It is my understanding that 23 you have undertaken that effort. Could you please describe for the Board what ou accom lished in this res ect? 25 Focusing of high frequency ground motions from a 26 large earthquake on the Hosgri fault is not expected to

Frazier VII-4

result in unusually high accelerations at the Diablo Canyon site. A computer model was developed and tested for simulating earthquake processes to provide additional information on ground accelerations and the effects of focusing at Diablo Canyon. The earthquake model simulates effects due to site specific earth structure, complex rupture sequences, radiation pattern, focusing and stress drop. Stringent tests have been and are currently being performed to simulate near-field recordings from past 10 earthquakes. These tests indicate that the earthquake model is suitable for predicting ground motion close to large 12 magnitude earthquakes, albeit conservatively predicting higher than real peak ground accelerations directly in the maximum beam of focusing. 15 As described below, site specific simulations have 16 been performed at Diablo Canyon to examine effects of rupture along the Hosgri Fault focused toward the sit'e in a 18 manner consistent with geologic data.

19 EARTH UAKE MODELING PROCEDURE

20 Ground motion is modeled as a three-step process: 21 (1) Green's functions are calculated for the 22 particular earth structure. That is, surface 23 motions are computed for several hundred point sources distributed over a closely spaced grid of 25 epicentral distances and focal depths. These 26 earth response calculations include all wave types

Frazier VII-5

present in the vertically stratified characterization of the earth over the frequency

range 0 to 20 Hz. (2) Fault slip is characterized in terms of fault type (strike-slip, dip-slip, etc.), rupture velocity, dynamic stress drop (slip velocity at the onset of rupture), static stress drop (fault offset), and duration of slip at each point. Additionally", random processes are included to approximate 10 perturbations or irregularities in the actual earthquake rupture. A single earthquake

12 simulation is repeated several times to 'determine I the range of effects introduced by the random processes. 15 (3) Ground motions for a distributed rupture are 16 synthesized by convolving in time and space the 17 characterization of fault slip (step (2), above) 18 with the earth's response functions (step (1), above). The spatial relationships are assigned at 20 this stage of the calculations, namely hypocenter 21 location, rupture extent, and site location with 22 respect to the rupture. 23 The earthquake model mathematically simulates the actual physical processes that occur during an earthquake. 25 Consequently, the model provides a rational basis for 26 appraising the likelihood of unusual combinations of fault

Frazier VII-6

rupture, including focusing, that could lead to unusually large amplitudes of shaking. TEST CALCULATIONS The model has been tested against strong motion

recordings from four past earthquakes: 1933 Long Beach (Ms ~ 1966 6.3), 1940 Imperial Valley (Ms 7), Parkfield (Ms ~ the 6.4), and 1971 San Fernando (Ms 6.5). In addition, model is, currently being tested against, recorded ground IV-79 ~ motion motions from (Ms 6.9). Strong recordings 10 from the 1966 Parkfield earthquake were used to assist in the calibration of certain parameters in the earthquake model. Strong motion recordings from the remaining three 13 earthquakes were used principally for testing the predictive capabilities of the earthquake model. 15 Results from these studies indicate that. the 16 earthquake model conservatively predicts ground motions within the distance range of interest for the Diablo Canyon 18 site. Specifically the earthquake model apparently produces 19 more severe effects due to focusing of high frequency waves 20 than is indicated by recorded ground motions. 21 Irregularities in the spreading rupture were simulated using 22 random perturbations in such processes as rupture velocity 23 and rupture direction. However, it appears evident that, in spite of these random perturbations, the model overpredicts 25 high frequency effects due to focusing in some cases. The 26 earthquake model predicts peak accelerations directly in the

Frazier VII-7

path of rupture substantially larger than those recorded in the 1966 Parkfield earthquake. Current studies have similarly predicted high levels of peak acceleration when simulating ground motions within one kilometer of IV-79 and yet those values were not recorded on any instruments in the path of rupture. Results from the earthquake model more closely match strong motion recordings for lower frequencies, where focusing effects are significant in recorded data, and for 10 recordings outside the beam of maximum focusing. ll Furthermore, the earthquake model produces synthetic ground 12 motion records which closely resemble actual recordings. 13 The response spectrum for the results of the earthquake 14 model also closely approximates the response spectrum for 15 the station being simulated with the exception noted above: 16 high frequency amplitudes are conservatively over estimated 17 in the beam of maximum focusing. 18 HOSGRI FAULT REPRESENTATION As illustrated in Figure VII-l, the Diablo Canyon 20 site is positioned 5.8 kilometers from the adjacent reach of 21 the Hosgri fault. The dotted line segments illustrate the 22 positioning of the Hosgri fault as represented in the 23 earthquake model. The modeled representation consists of 24 three straight-line segments hinged at points labeled 2 and 25 3. The northernmost segment, extending from point 1'o 26 point 2, has been aligned to point directly at the site.

Frazier VII-8

The three segments extend a total distance of 87.7 km from point 1 to point 4. Rupture over this entire distance (which is most unreasonable for a single earthquake) would correspond to a surface wave magnitude greater than is reasonably postulated for the Hosgri. EARTH PROPERTIES The properties of the underlying earth are represented in terms of material velocities, density, and material attenuation. To realistically model the site 10 specific ground motions, the material velocities and density are assigned depth-dependent values that are characteristic 12 of subsurface conditions at Diablo Canyon. Material attenuation properties, which also vary with depth, are assigned values based on generic relationships used in all 15 prior and ongoing test calculations for past earthquakes. 16 The subsurface properties are presented in Figure VII-2.

Note that material quality factors for P and S waves 18 characterize the attenuation properties of the material by 19 an inverse relationship -- the larger the quality factor, 20 the smaller the attentuation.

21 HYPOTHESIZED EARTH UAKES 22 Ground motions were calculated for seven 23 configurations of earthquake rupture along the Hosgri fault, illustrated in Figure VII-3. Each configuration is / 25 positioned along the dotted-line representation of the 26 Hosgri fault, illustrated in Figure VII-1. The

Frazier VII-9

configurations, A through G, have varying rupture lengths and epicentral locations so as to test, for the most critical rupture sequence with respect to ground motions at the site. Ground motions were calculated for each'upture configuration assigning both. right-lateral strike-slip rupture and 45'-rake rupture with equal components of strike-slip and dip-slip.

CALCULATED GROUND MOTIONS AND RESPONSE SPECTRA The severity of ground shaking calculated at the 10 site was found to be essentially independent of rupture configuration. The similarity of results from earthquakes A 12 through G are illustrated in Figure VII-4, which compares. 13 response spectra for these configurations of strike-slip rupture. Also, the Hosgri reanalysis'pectrum is presented 15 for comparison with calculated values. The reanalysis 16 spectrum is above all computed free-field spectra, at all 17 periods. 18 Ground motions were also calculated using equal 19 components of strike-slip and dip-slip (45'ake) for a 20 selected set of rupture configurations, namely A, C, F, and 21 G. These configurations include the most critical configu- 22 rations with respect to magnitude and focussing at the site. In addition, ground motions were calculated for an even larger magnitude earthquake having a rupture of 120 km, 25 nearly 50 percent longer than has been postulated for the 26 Hosgri fault. This configuration is denoted earthquake H.

Frazier VII-10

Response spectra for these hypothesized earthquakes with equal components of strike-slip and dip-slip are presented in Figure VII-5. These results indicate that lengthening the zone of rupture results in no significant increase in response spectrum (including zero period accelerations) over the period range of interest for the Diablo Canyon structures and equipment. Also for the periods of interest (periods less than .5 second), the strike-slip and 45'-rake rupture resulted in essentially the 10 same level of horizontal ground motions, while the introduction of dip-slip resulted in 10 percent to. 12 30 percent higher vertical response spectra. Here again, 13 the Hosgri reanalysis spectrum exceeds all computed response spectra for free field levels of motion. 15 Mean values for calculated accelerations, 16 velocities and displacements associated with the rupture sequences used in Figure VII-4 (strike-slip) and 18 Figure VII-5 (equal components of strike-slip and dip-slip -45'ake) are given in Table VII-1 and VII-2, respectively. 20 Calculated peak horizontal accelerations range between .36 and .58 g for both strike-slip and 45'ake. The 22 corresponding range of peak vertical accelerations is .12 to 23 .18 g for strike-slip and .18 to .26 g for 45'ake. Time domain results for a typical random simulation for 25 hypothesized earthquake F are shown in Figure VII-6. The 26 calculated acceleration time histories are shown for

Frazier VII-11

strike-slip motion as well as for equal parts strike-slip and dip-slip motion for this particular random rupture simulation. 4 CONCLUSIONS The following conclusions can be drawn from

6 results of the earthquake modeling studies performed for the

7 Diable Canyon site and from results of investigating strong motion data recorded close to earthquakes: (1) The effects of rupture focusing are much less 10 apparent for frequencies of interest at the Diablo Canyon NPP than for peak velocities and lower frequencies. Furthermore, the Diablo Canyon site 13 is not positioned for significant focusing from rupture along the Hosgri. Therefore, adverse 15 effects due to rupture focusing are highly 16 unlikely at the Diable Canyon site. (2) The largest accelerations produced by the 18 earthquake model originated from hypothetical rupture within 20 km of the site. That portion of 20 rupture along the northern extent of the Hosgri 21 fault, which was beamed directly at the site, did 22 not contribute significantly to the peak levels of 23 acceleration. The conclusion is drawn that focused rupture along the Hosgri fault cannot 25 produce unusually high levels of ground 26 acceleration at the site.

Frazier VII-12

(3) The computed levels of ground motion are relatively independent of the length of rupture. Conseguently, peak acceleration at Diablo Canyon should be basically insensitive to increases in magnitude for magnitudes greater than about 6.5. (4) The mean values of peak acceleration were well below design levels. Furthermore, calculated free field response spectra are uniformly below that used in the reanalysis. Thus, the earthquake 10 modeling studies indicate that the Hosgri reanalysis spectrum is substantially conservative 12 even when comparing free field predictions with 13 actual reanalysis criteria. /// /// 16 ///

22 23

25 26

Frazier VII-13

STRIKE-SLIP MOTION (0 RAKE)

Acceleration(g) Velocity(cm/sec) Displacement(cm Rupture Sequence Mean Mean Mean

VERTICAL Component A .18 12.8 6.4 B .18 12.5 4.8 C .17 12.5 6.4 D .17 12.4 5.2 E .18 12.5 6.4 10 F .18 12.8 7.3 G .12 7.7 2.0 N65E Component 12 A .51 30.4 11.2 B .50 27.9 9.7 C .48 27.8 11.7 D .49 26.1 9.9 E .48 27.8 10.5 15 F .51 30.4 12.0 G .36 20.2 4.1 16 S25E Component A .55 31.0 23.0 18 B .58 32.3 22.9 C .54 30.3 22 ' 19 D .45 29.5 19.9 E .54 30.4 22.4 20 F .55 31.1 23.8 G .51 25.4 17.8

22 Table VII-1 Calculated peak accelerations, velocities and 23 displacements associated with the rupture sequences used in Figure VII-4 (rake — 0'). 24 25 26

Frazier VII-14

COMBINED STRIKE-SLIP AND DIP-SLIP MOTION (45 RAKE)

Acceleration(g) Velocity(cm/sec) Displacement(cm Rupture Sequence Mean Mean Mean

VERTICAL Component,

H .25 20.9 18.1 A .26 20.4 16.0 C .18 20.6 15.9 F .26 20.4 12.7 G .19 12.7 2.6 N65E Component 10 H .55 33.6 16.9 A .46 27.7 17.2 C .58 29.5 15.4 F .46 27.6 16.3 G .36 15.6 3.9 13 S25E Component,

H .56 26.5 13.3 15 A .56 23.4 11.0 C .52 23.1 9.5 16 F .56 23.5 13.1 G .43 22.9 25.9 17 18 Table VII-2. Calculated peak accelerations, velocities and 19 displacements associated with the rupture sequences used in Figure VII-5 (rake = 45'). 20 21 22

Frazier VII-15

SURFACE COOROltlATES~ (km)

Length of ~Fult 5 ~ t C7g Ql : (-29.5, -39.85) 20.0 ba Q2 : (-17.5, -23.85) ) 41.7 ba Q3 . (0.0, 14.0) ) 26.0 ba Q4: (10.0, 38.0) Site: (0.0, 0.0) , ~ * (East-South coordinate system Mith origin at site)

0 8 . 16 24 32 km

Figure Frazier VII-1. Piecewise representation of the Hosgri strain system in terms of three fault segments which are delineated by the data and described in the box. 0: 200 250 300 350 1 2 3 4 5 50 100 150

Shear Compressional Density Velocity Velocity (gm/cm~) —(km/sec) (km/sec) l I I

I Shear-Wave I Quality I Factor l 10 l Compressi onal- Wave I Quality I Factor I 12

Figure Frazier VII-2 Material properties as a function of depth for the geologic structure p'resented in Table VII-1 for the Di'ablo Canyon site. e ~ 0 ~ ~ ~ ~ ~ ~ ~

0 ~ ~ ~ ~

~ t P~ 4 ~~ye o~ ~gg'3 $0 BS. ~~go

~ ~t ~ ~ ~ y~

0. 8 16 24 32 km ~ ~

Figure Frazier VII-3. Map of seven earthquake rupture sequences used to study the effects of fault location and extent. The rupture plots A through G are separated for purposes of illustration only —all seven earthquakes are aligned with Earthquake A, 5.87 km from the site. The large X's mark the epicentral locations for each earthquake rupture.

DIABLO CANYON SITE SPECIFIC- RESULTS DIABLO CANYON SITE SPECIFIC RESULTS

HORIZONTAL COMPONENT N65E HORIZONTAL COMPONENT S25E

) 00.0

)0.0 I CJ / CI 4J / el C) Ill4LJ 1.0

O. 1 4 0. 0! 0. 10 o.o) 0. 10 ).Oo 1 o. 00 PERIOD (SEC1 PEPIOD ISECI

DIABLO CANYON SITE SPECIFIC RESULTS —'esign Spectra )000.

VERTICAL COMPONENT Strike-Slip Rupture Sequence A B

100. 0 C D E F / 6 )- 10.0 I / Figure Frazier VII-4. Calculated 2X damped response spectra for the W / seven earthquake rupture sequences CI CD j shown in Figure )II-3 'using L4j Jg strike-slip motion. Mean'alues are ).0 obtained at each period from seven random simulations for each rupture sequence. The results are somewhat smoothed and are compared with the reanalysis design spectra (.75 g NEWARK for horizontal components 0.1 vertical 0.01 O.)O, 10.00 and .50 g NEWARK for PERIOD ISEC1 components

DIABLO CANYON SITE SPECIFIC RESULTS 'DIABLO CANYON SITE SPECIFIC RESULTS

HORIZONTAL COMPONENT N65E HORIZONTAL COMPONENT S25E"

100. 0 r

O IA K / LJ / / )- 10.0 f EJ J / Cl / Ili / J / el l r C3 II ~ f EA 0 1.0 )

0.1 0. OI 0. 10 1. 00 10.00 0.01 0. 10 10.00 PER)OD (SEO) PER)OD )SEC)

DIABLO CANYON SITE SPECIFIC RESULTS

'ERTICAL COMPONENT Design Spectra 45 Rake 100. 0 Rupture Sequence H (120 km long) CJ A laJ EA C LJ F G '-Figure Frazier VII-5. Corresponding . results to Figure VII- except that equal parts of dip-slip motion were used and only four random simulations were used to determine the means for each rupture sequence and are compared with the reanalysis design spectra (.75 g NEWMARK for horizontal components NEWNARK 0.1 and .50 g for vertical 0. 01 0. 10 I.00 10. 00 components. PER)OD [SEC)

Earthquake F with Strike-slip Motion (rake = 0')

Vert(cal

O O N65E

IO I 4I ~I CO 4I IIl 4I ci

525E

I c8 0.0 l0. 0 20. 0 30. 0 VO. 0 50. 0 (seconds) Earthquake F-with, Equal Components of Dip-Slip and Strike-Slip'Motion (rake ='5')

Vert(cal

40 tV Cl

H65E

~I4 4I 4I O

525E

ID 0.0 l0.0 20. 0 30. 0 VQ.O 50.0 ~ (seconds)

Figure Frazier VII-6. Typical synthetic acceleration records obtained ,at the Diabl o Canyon si te:

VIII. We have received preliminary reports of the effect of IV-79 on the El Centro Steam Power Station. (Board Notification December 17, 1979, Levin and Martore Observations; Rothman — Kuo Affidavit, page 12). In many respects, the structures and systems of that facility resemble those of the Diablo Canyon plant. Their response to a severe, well instrumented seismic event can be analysed to help confirm or refute analytical technigues and assumptions used in the Diablo Canyon seismic analysis. The parties should prepare and submit such an analysis.

Written Testimony

DR. JOHN A. BLUNE

. Q. Dr. Blume, the Appeal Board has asked for results of analyses which may have been done for the El Centro Steam Plant as a result of IV-79. It is my understanding that your company has performed certain analyses at that plant. Would you please tell the Board what, analyses were done and the results obtained. In addition, please relate the analysis to the Hosgri reanalysis performed at 10 Diablo Can on. The El Centro Steam Plant is a 4-unit electric

12 generating facility of the Imperial Irrigation District located approximately 5 km from the Imperial fault. Each unit of the facility contains three distinct structures: a 15 steel frame and concrete turbine building, containing

16 mechanical and electrical equipment as well as piping 17 systems; a concrete pedestal supporting the turbine and 18 located within, but structurally separated from, the turbine 19 building; and a boiler structure which is a braced steel 20 frame supporting a hanging boiler and structurally connected 21 to the turbine building.

22 Each unit of the plant is structurally

23 independent, and since Unit 4 is the most recent unit (constructed in 1968) it was selected for detailed study. A 25 USGS strong motion accelerograph is located less than 1 km 26 southeast of the plant (USGS No. 5164 El Centro Differential

Blume VIII-1

Array), and its recorded trace was used to derive time histories and response spectra for the plant analyses. Structural Anal sis The turbine building and boiler structure of Unit 4 were modeled and subjected to the response spectra

for 7% damping derived from the differential array recording. A fixed base analysis, similar to that used for Diablo Canyon, was used. It should be pointed out that the steam plant structures are substantially different from the 10 Diablo Canyon structures in terms of design criteria, structural details, and equipment support design. The steam 12 plant was designed for 0.2g horizontal seismic load, using

13 simple static analysis techniques, wi.th no consideration of vertical acceleration. Diablo Canyon was analyzed for three 15 components of motion, all on a dynamic basis. The steam 16 plant walls are 6" thick, with a single layer of reinforcing, and the floors and roofs are metal deck with 18 3-1/2" of concrete fill. The Diablo Canyon walls are 19 thicker, with two layers of reinforcing, and slabs are of 20 structural concrete, 12" thick. 21 The non-symmetry of the El Centro turbine If 22 building, resulting from the presence of a shear wall on the south face of the structure with no complementary element on the north face, is a design feature generally considered 25 undesirable and not found in the Diablo Canyon structures. 26 The, boiler is suspended from the roof girders of the boiler

Blume VIII-2

structure, with lateral seismic stops at, lower elevations of the structure. Actual structural damage to Unit 4 was minor, confined to three buckled bracing members in the boiler structure and bent seismic stops for the boiler itself. Results of the analysis, however, using procedures comparable to those used for the Diablo Canyon Hosgri reevaluation, predicted much more damage than actually occurred. Of 25 panels of bracing in the boiler structure,

10 23 sustained compressive loads in excess of the computed allowable stress according to the analysis. In fact, the compressive forces in 21 of the 25 sets of bracing exceeded

13 the computed Euler stress for the bracing, the theoretical maximum stress for an axially loaded compression member. In

15 addition, the analysis showed that the columns on the west

16 face of the boiler structure sustained bending moments which

17 were twice the yield capacity of the member. 18 The analysis of the turbine building showed that

19 the forces on the shear walls were at or above the computed 20 ultimate capacity of the walls; hence, cracks in the walls 21 would be expected. Cracks would also be expected at the 22 diaphragm/wall interface as the metal deck system reached 23 its ultimate stress. Yet, an inspection of the building resulted in no observed cracking of concrete elements.

25 Obviously, there are physical characteristics of 26 the structure and site that account for much of the

Blume VIII-3

difference between observed and computed responses. As an example, the assumption of a rigid base ignores the reduction effects due to soil-structure interaction and thus leads to a conservative estimate of forces on the structures. While the Diablo Canyon reanalysis also ignored soil-structure interaction, the degree of conservation would be lower. The estimate of bracing member capacities is based on the assumption of pin-supported ends of in actuality, the ends are restrained to the'embers;some 10 degree and will support greater loads than a pin-supported 11 member. The Diablo Canyon reanalysis utilized a similar assumption of "true" pin-supported braces; thus the capacity

13 of the Diablo Canyon bracing members is greater than that assumed in the reanalysis.

15 The analyses of the E1 Centro Steam Plant compared

16 with the actual performance of the structure, demonstrate

17 that the analytical procedures used in the analysis, similar 18 to the procedures used in the Diablo Canyon reevaluation, 19 are conservative. 20 E ui ment and Pi in Anal sis 21 Units 3 and 4 were operating at the time of the 22 earthquake. Units 1 and 2 were down for maintenance. The 23 earthquake caused both operating units to trip off, line. 24 Following the quake Unit 4 was shut down to inspect for 25 damage but Unit 3 continued operating to supply station 26 power.

Blume VIII-4

The piping and equipment at 'the plant was designed for 0.2g horizontal seismic load, using simple static analysis techniques of assuming the seismic load through the center of gravity. Ground motion measurements near the plant recorded approximately 0.5g horizontal acceleration. For equipment on the upper floors of the plant, base acceleration could have, by conventional analysis for nuclear plants, easily exceeded 1.0g, five times the design-basis seismic load. In spite of this very little

10 damage was reported to equipment in the plant. Some leaks were observed in 3 and 4 inch piping in the component,

12 cooling water system. The leaks occurred at plastic

13 couplings or at rust spots on the piping, neither of which would be present in a nuclear plant.'eedwater heaters in

15 Units 2 and 3 skidded on their rollers a few inches, without

16 damaging the attached piping. Positive anchorage of the

17 newer Unit 4 feedwater heater prevented skidding. A 18 lightning arrestor on a voltage stepup transformer was

19 broken in Unit 1 due to excessive rocking of the transformer 20 on its supporting rails. A short in the transformer due to 21 the broken lightning arrestor was a preliminary cause of the 22 two operating units tripping off line. The seismic 23 vulnerability of ceramic insulators is a recognized weakness in electric power systems; however, in a nuclear plant analysis the seismic qualification of the transformer would 26 detect and correct this failure mode.

Blume'III-5

Probably the only significant damage to equipment in the plant was the fracture of an air-actuated valve operator on a 4 inch line supplying steam to an evaporator. Valve actuators constitute an eccentric mass cantilever'ed off a piping system, a structural arrangement highly susceptible to seismic response amplification. The piping system is located on the third floor of Unit 4, such that building motion amplified the seismic input to the piping system, probably incurring a theoretical response of several 10 g's at the location. of the valve actuator. The actuator fractured at the yoke between the valve and the air

12 operator, probably due to a combination of loads incurred by

13 the pipe motion. In a nuclear plant the seismic response analyses of the piping system during the design phase would 15 have detected the excessive flexibility of the line at the actuator location and additional bracing would have been

17 provided. The single-degree-of-freedom quasi-static 18 calculations used for the design of El Centro Unit 4 would

19 not adequately predict the amplification of seismic load to 20 the actuator by the piping system. 21 Three equipment/piping installations in Unit 4 22 were selected for detailed study: 23 the main control panel for Unit 4 two runs of an 8 inch diameter piping system connecting the gland 25 steam condensor with a feedwater heater 26

Blume VIII-6 ' an air-actuated valve similar to the failed valve.

The purpose of the study of these items is two-fold. First, an accurate estimate can be made of the seismic response, and 'hence the dynamic stresses experienced in these items during the quake, due to the availability of a response spectrum developed from ground motion measurements. Second, the Diablo Canyon Hosgri reanalysis seismic criteria and the analytical techniques used in qualifying equipment according 10 to this criteria can be checked for conservatism and accuracy. Finite element models of the 3 items were

13 developed for response spectrum analyses, using the standard codes SAP and PIPESD. In addition, low level dynamic response tests were made on the equipment installations with instrumented impact hammers and accelerometers. The

17 equipment response motion records were processed on a spectral analyzer to measure modal frequencies and mode shapes of the lowest modes of each installation. These

20 measured eigenparameters provide input as well as 21 verification to the finite element models used in the 22 response spectrum analyses. Main Control Panel The Unit. 4 control panel is a walk-in console consisting of a welding frame of steel angles supporting 1/4 inch plate in which are embedded the electrical components.

Blume VIII-7

The panel is approximately 15 feet wide by 5 feet deep by 8 feet high. Anchorage is to the floor with bolts set into the concrete, and also some minimal bracing against the ceiling. Calculations from the finite element model and the results based on test data demonstrated seismic stresses low compared to allowable limits, a normal occurrence for this certain 7 type o f equipment. Given the high acceleration of components on the panel face, however, the hazard o f In spite of 9 electrical malfunction was indeed significant. main 10 this, no operability problems were reported in the control panel or any other electrical control system in the plant either during or following the quake.

13 Gland Steam Condensor Pi in S stem The piping system selected for study carries the 15 feedwater from the main condensor into the tube side of a feedwater 16 gland steam condensor and from there into 40 flange 17 heater. The line is 8 inch schedule pipe with connections at valves and welded connections at branches and nozzles. Welded connections at branches do not involve The 20 welding tees or any apparent reinforcement. pipe is supported only by junction with larger components or an occasional dead load hanger. Standards of construction and

23 support are much higher for piping in nuclear plants ~ Stresses as high as 36,000 psi were calculated at and 25 branch connections between the gland steam condensor the feedwater heater. It is unlikely, however, that fracture or

Blume VIII-8

any yielding except at a local level would occur as long as the pipe is free of cracks and corrosion. Since no damage was observed in the line, it is probable that the stresses experienced in the piping system during the quake did not exceed, or even match, the calculated values. At the same time it should be noted that a design-phase piping analysis based on the stronger Diablo Canyon design basis spectra would certainly have shown this El Centro piping to be inadequately supported. Design changes would have been made 10 to the supports to reduce the seismic induced -stresses in the piping. It can be concluded, then, that although this particular piping system is grossly underdesigned and undersupported by nuclear plant, standards, it survived without damage ground motion on the same order as the Diablo

15 Canyon operating-basis earthquake. 16 Air-Actuated Valve 17 The 6 inch recirculating water blowdown valve is 18 an air operated valve similar to some of the smaller valves 19 used in nuclear plants. The valve actuator consists of a 20 60 lb eccentric mass cantilevered off a 6 inch diameter 21 piping system located at the ground floor of Unit 4. The 22 support conditions for this piping system are far less elaborate than those used in a nuclear plant. Therefore, the larger flexibility of this piping system will result in 25 greater seismic response amplification to the valve 26 actuator.

Blume VIII-9

Critical stresses were determined to be 4.12 ksi (includes operation load) from the computer analysis, and 5.14 ksi based on field testing data. Both methods result in stresses much less than the maximum allowable of 28.875 ksi. If the valve were located at an upper level with an amplification factor on the order of 5.0 or greater, the stresses would be close to or exceed the allowable for the material. This would explain the broken valve actuator at the third level of the turbine building. The items and were 10 three of piping equipment analyzed using the response spectra corresponding to 2% damping and the The use 12 ground elevation. of amplified spectra corresponding to the upper levels of the structure would undoubtedly result in large increases in stress above those stresses or above the 15 calculated, in fact, at level to cause damage. The damage 16 reguired fact that very little occurred further attests the conservatism the 17 to of procedure used the Diablo Canyon 18 in reanalysis. In conclusion, the analyses done for El Centro damage The were done 20 predicted that did not occur. analyses a manner 21 in consistent with the Hosgri reanalyses for Diablo Canyon. The obvious such 22 result is that analyses are conservative in that they theoretically predict damage where

24 in reality none occurs. 25 26

Blume VIII-10

Written Testimony

WILLMER C. GANGLOFF

Q. Mr. Gangloff, would you please describe what analyses were done at, the El Centro Power Plant to assess damage done there as a result of the October 15, 1979 Im erial Valle Earth uake? Westinghouse personnel visited the El Centro Power Plant to assess the impact on piping and supports of the October 15, 1979, Imperial Valley Earthquake. The main

10 structural members of the El Centro Power Plant buildings are wide flanged steel beams and columns with some

12 reinforced concrete walls and floors. Most of the equipment 13 is rigidly attached to the building except the boiler, which does not, resemble any piece of equipment in a nuclear power plant,. The piping systems closely resemble the piping in a

16 nuclear power plant except, there are fewer horizontal

17 supports used to restrain horizontal seismic motion. 18 Although little damage was done to the El Centro 19 plant, pipe motion in excess of two inches could be verified '20 by the crushing of the piping insulation jacket. when the piping moved into the building steel framing (see Figures 1, 22 2 and 3). Most of these large piping motions occurred high 23 in the building, which is to be expected due to the building amplification of the ground excitation. 25 An examination of several piping systems resulted 26 in the selection of the eight inch nominal diameter turbine

Gangloff VIII-1

gland steam line as the line to be analyzed. This line, originating at the air ejectors on the operating floor, is connected to the gland steam condenser and preheater at the ground floor. From these two pieces of equipment the line runs to the deaerator heater on the operating floor. Near the deaerator heater the piping is supported at its highest point vertically from the roof. Near this support, the pipe moved 2-3/4 inches south, 5/8 inches east, and 1/4 inch vertically upward during the earthquake. This was verified

10 by measuring the mark left in the insulation jacket of the pipe when the pipe hit an adjacent hand rail which in turn was welded to a plate embedded in the floor (see Figures 2

13 and 3).

15 Q. Would ou describe the i in model?

16 Using as -built, dimensions (see Figure 5), a computer model of the line was developed. The same methods

18 used in the generation of computer models for Diablo Canyon 19 and other nuclear projects were followed for the El Centro 20 piping model. All the vertical rigid supports on the gland an 21 steam line are similar to the one in Figure 4 which is 22 angle supported by two tie rods. Two variable springs, 23 manufactured by Bergen-Patterson, are included in the model using the actual spring stiffness of the support. A computer plot of the line is shown in Figures 6, 7 and 8.

Ganglof f VIII-2

Q. Would you now describe how floor response spectra were generated in the anal ses? The determination of floor response spectra depends on two inputs — the ground time history acceleration and the El Centro steam plant building model. The time histories used were provided by URS/John A. Blume and Associates, Engineers and were from the Dogwood Road Station No. 5165. The time histories used are the uncorrected acceleration time histories, the same as used by

10 John A. Blume in his study of the October 15, 1979, Imperial Valley earthquake.

12 The El Centro Power Plant is a structural steel

13 building. A four mass model was constructed having an east-west frequency of 2.04 hertz, a north-south frequency of 1.86 hertz, and a vertical frequency of 4.34 hertz. The

16 masses are placed at the elevations of the four floors above 17 ground. The three orthogonal time history accelerations 18 were applied simultaneously to the building model. The 19 modal transient time history was run for 16 seconds and 20 seven percent of critical damping was used for the 21 structure. Seven percent damping is the value specified for 22 bolted steel structures in Reg. Guide 1.61. 23 The resultant history accelerations at each floor were then converted to floor response spectra for two percent critical damping. The frequencies at which

26 accelerations were calculated are in accordance with US NRC

Ganglo ff VIII-3

Regulatory Guide 1.122. The spectra were then broadened fifteen percent. The horizontal spectra at ground level and at the operating floor (floors at which the piping was horizontally attached) were then enveloped for input to the piping model. Since the piping was supported vertically from all floors, the vertical floor response spectra from all floors were enveloped and used, as input to the piping

~ model. It should be noted that in the Hosgri reevaluation 10 of the Diablo Canyon Nuclear Power Plant, torsional response spectra were generated due to the eccentricity of the floor

12 masses. These spectra were, in turn, used to increase the

13 horizontal spectra, depending on the distance of the piping system from the center of rotation. This effect which adds 15 conservatism to the floor response spectra was not, included

16 in the analysis of the El Centro piping system.

Q. Would you now describe the seismic anal ses erformed? 18

19 The piping model as previously defined was analyzed using the envelope building floor response spectra 20'1 which I just described (see Figures 9 and 10). The analyses 22 were performed using program WESTDYN, the Westinghouse 23 proprietary piping computer code, the same code used for the Diablo Canyon piping. Two seismic analyses were run -- one 25 with the applicable spectra applied in the north-south and 26 vertical directions and a second with the applicable spectra

Gangloff VIII-4

applied in the east-west. and vertical directions. The method of solution was a linear elastic response spectra modal analysis. The two shock directions were added absolutely and the individual modes were combined by the square root of the sum of the squares method. In addition, a seismic analysis was performed using the free field ground response spectra (see Figures 11, 12 and 13). This was done to provide a lower limit for the previous analyses. Any differences between the results

10 of these two analyses would be due to the building amplification of the ground spectra.

12 Due to the large magnitude of the vertical ground

13 spectra, a third analyses was run using only the horizontal spectra. Any difference between this run and the run using

15 ground spectra would be due to the vertical input.

16 For each analysis, a deadweight analysis was

17 performed and a design pressure of 175 psi was used. Q. Please describe the results obtained by 18 Westin house from these anal ses. 19 20 The analyses are labeled as: 21 A. Seismic analysis with building floor response 22 spectra. 23 B. Seismic analysis with ground response spectra.

25 C. Seismic analysis with horizontal ground 26 response spectra only. Gangloff VIII-5

For the calculation of stresses the 1973 Edition of the ANSI B31.1 Code was used, as was used on the Diablo

Canyon piping systems. An allowable stress of 2.4 Sh was used. The piping material was ASTM-AS3, Grade B (2,.4 Sh= 36 ksi). Pressure, deadweight and seismic stresses were added absolutely to arrive at a total stress. From Figures 2 and 3, evidence of pipe motion of 2.8 inches can be seen. This point in the piping corresponds to node point 3230 in the computer model.

10 Displacements of this point, maximum stresses and locations, and maximum support loads and locations are tabulated below.

13'5

17 18 19 20 21

25 26

Ganglo ff VIII-6

MAXIMUM STRESS Seismic Maximum Allowable A~nal sis k Stress ksi Node Point A 131.67 36 3810* B 67.68 36 3810* C 65.08 36 3810*

I MAXIMUM SUPPORT LOADS Seismic Maximum Seismic Dwt. ~1'k'aximum~Su ort Load ki s ~Su ort 10 A 8.34 Vert 12 3.'67 Vert 11 3.22 Vert 8 3.67 Vert 11 3.01 Vert 8 3.67 Vert 12 ll

13 DISPLACEMENTS AT 3230 Displacement Seismic Anal sis at 3230 inches 16 21.11 B 12.60 12.39 18 C

19 20 *Node point 3810 is the last elbow before the deaerator heater. 21 22 /// /// 25 26

Gangloff VIII-7

Q. Would you now describe what conclusions you were able to draw from your analyses as res ects the Diablo Can on NPP? The method used for the Diablo Canyon Nuclear Power Plant and other nuclear projects is the method used for Analyses A -- a response. spectra modal analysis using building floor response spectra. As can be seen from the results, the analyses predict that. the pipe is severely overstressed. Even using the ground input, stresses which are almost twice the allowable are predicted.

10 The methods employed for the Diablo Canyon Nuclear Power Plant significantly overpredicted the actual results for the El Centro plant, as can be seen from the predicted

13 displacements at node point 3230. Nonetheless, the pipe showed no indication of failure.

15 In addition, the capacity for a typical support on this line was calculated to be 3.86 kips. For analysis A,

17 five of the ten rigid supports were predicted to fail; 18 however, none of the supports showed signs of failure. 19 Several reasons can be given for the high results 20 predicted. The response spectra method is conservative in 21 the fact that it assumes all modes reach their maximum 22 response at the same time. The most significant reason for 23 the overprediction of stresses and displacements is that the system is assumed to be linear elastic. For small loadings 25 when the system remains elastic, predicted versus actual 26 results are usually fairly close for realistic damping

Gangloff VIII-8

assumptions. However under moderate to high loadings, the t system actually becomes nonlinear. Vertical supports offer resistance to horizontal motion for large horizontal motion, 4 spring hangers "bottom out" and become rigid supports, and friction forces become important. A final point to note is that the effects due to the vertical input are relatively insignificant. This is 8 due to the fact that. vertical modes have higher frequencies 9 than the horizontal modes and, therefore, a smaller 10 response. Less than ten percent of the total response is 11 due to the vertical input. 12 In summary, it has been demonstrated that the 13 analytical methods and assumptions used by Westinghouse in the evaluation of piping systems at Diablo Canyon contain 15 significant conservatisms. These conservatisms become clear 16 when these methods and assumptions are used to evaluate the 17 actual behavior of a similar piping system which has 18 experienced seismic loadings. 19 Significant; reductions in the input seismic 20 excitation still conservatively predict the results in the 21 piping system due to the additional conservatisms in the 22 piping analysis itself. It can, therefore, be concluded 23 that the evaluations performed on Diablo Canyon are 24 conservative and provide substantial margin to the results 25 that would be expected from an actual seismic event. 26

Gangloff VIII-9 «I Figure 1 F)gore

Figure 3 Figure 4

VERT NORTH EAST

yyt

300

VL toy 51 OVstyI5 Oyt

+4 ~EVCNV

! Cyt OO'N 35 EE f 45 Iyt ~

~ VLIL5'3TE

'O VVE ~

4 ~ T Vlt OEICTTVITLI4 T45 - atLTo ~ 4 44 57 VC VOL Og 34

34'4 Vj E+EC

'4C, 45 Tyt 4t Ty 51

44 CL V5 4II ~ O tl to ITOO JT,

LIO 3 tct5C( Vt C ~ETEL: ELC45 KN'C 4( I. ITLI. IILMCNOIONS IN IIICNKS~ 3 CEtolk EIO'3'IVV'tttV'IWNCTLK NOVCO

Figure 5 f'VI

%) FIGURE 6. TURBII'lE GLAtlD STEAt'I "LIt'lE

STRUCTURAL I'MECHANICS DEPRRTNENT

F ILE0 IMP . DFITE: 7/30/80 F I LE USE COUNT 26 t

PLOT 1

VERT 7

"OO V8~8

ygaT <

Nap> pOINT'wlI jV 3~ 50 0260< QV~b ~ n

8$ 90 ~i VFR7 9 g GW'f I o

r„ ts70 VERT II 4C LS60( ) Pl I rO s} iu I soeo ~o I soeo ggRTD susn I %000 I I i SI20

I Sljo snop II I @No I I L J vier Iz I I I I I I D VETcT 2, I Qi yp Veqrl ~ SPA figure 1;. \ .Z STR~JCTUPiRL HECHRNI CS . DEPRRTNENT.

F I LE: IHP DATE: 7/30/80 F I LE USE CQUN'I 26 NQDE 0 RCCQUNT: HEDR 1011 ~ SUPPQR'IC PLQ't 2

'>0

31>n

ann

qO 000 J u

0 )2)A ~l~ 2350

2300

26I0 22IIO 2e00 Nia 2590

iÃv9 2030 2520

a fov 0

DCCC VERT / I080

IQbo

l 020 O I030 620 gV ' FIGURE 8. TURBINE GLAND .STEAM LINE

\ STRUCTURAL M-CHAN I CS DEPARTMENT

FILE: IMP . OFITE0 7/30/80 F ILE USE CQL)NT 26 NQOE: 0 RCCQUNT: NEBR 3011 5UPPQFlT: PLQT 3

veer'I[

3400 )

3070 ( )

0000('000

S730( )

3310 S700 INES NoDg

Po)NV'8JO

oo 0730( 'I

j"!n -I I;.":,:l; . "'!i!>''~; p(c!~> .i)I . FIGURF.. 9; . IIHRI70ih'fAL.BllILDItlh.Fl.AAP. PF5f?oflsF sPFcTRP.

READY-F,2

5.0000 IHPERIAL VALLEY ENVELOPE SPECTRUM SSE HORIZONTAL .02 OAHPINC MASS PT 3 ANO HASS PT 5.

4.0000

F 0000

~ 2.0000

~ I.0000

0.0

Cl Cl Cl Cl Cl Cl Cl Cl Cl CI ' Cl Cl Cl CV IY) IA

F N Y 0 FIhURF. 10. VFRTICAI. BUII.OIflh FI.AAR RFc'PAgc'F. ciPFI:TRP

20.000 IMPERIAL VALLEY ENVELOPE SPECTRUM SSE VERTICAL .02 OAMPING MASS POINTS 1 2 3 ~ 4 ~ ANO 5 17.500

15.000

12.500

~ 10.000

~ 7.5000

4) ~ 5.0000

2.5000

0.0

CD CI CD CI CD CD C) CI CD CI CD CD CI CI Cl CU CP

RE NCY t( H

'IllUR~ 11. IIORTI(.-SOUTH BROtjtlP.-SPFCTRA ~ ~

2.0000 ACCELERA'110M 1C) RESPOMSE SPECTRUH FOR XOOE 5 1M 01RECTIOM XS CALCULATEO US1MG T1HE STEP .01OOOSECONO

1.7500

1.5000

1.2500

1.0000

~ 0.7500 ED

0.5000 ~ ~ ~ ~ 1 ~ ~ ~

0.2500

0.0

ID C) Cl Cb ED C) C) C) CD Cl C3, CI C) C) C5 C) AJ CD CI'RO

5.0000 ACCELERATIOH (C) RESPONSE SPECTRUH FOR NODE 5 IH DIRECTION VT CALCULATEO USINC TIHE STEP 0IOOOSECOHO

0.0000

3.0000

'a.0000

Cl W o CL U

1.0000

0.0

Cl C5n n a C> C) CI n Al 4F

PER OO {HZ CYCLE)

FItlURF. 13. FRST-IIFST hRAI)~]ll c'PFt'.TPh

5.0000 ACCELERATION (C) RESPONSE SPECTRUM FOR NODE 5 IN DIRECTION EV CALCVLATEO USINC TIME STEP .OIOOOSECOND

0.0000

3.0000

9 2.0000 tk

C7 lal CCl

i.0000

~ ~

0.0

C7 C) C) C7 C) C) Cl CD CD CO

PER IOO H V E) t IX. In addition to answering our questions about information from the Imperial Valley earthquake, we would like the parties to address Paragraph E on page 6 of the Mcmullen affidavit (included with the Staff Response to Joint Intervenors'otion to Reopen). That paragraph states that, "in its geologic and seismologic review of the Point Conception LNG site, the USGS reported that 'Existing evidence favors association of the 4 Nov., 1927 (M 7.3) Lompoc earthquake with an east dipping reverse fault such as the Offshore Lompoc or similar reverse fault 10 km to the south that offsets the seafloor.'" Does this USGS statement reflect either evidence not presented in the Diablo ~Can on hearing or a change in the USGS position based on evidence already. in the record? In any event, discuss that, statement's implications for this case.

Written Testimony

DOUGLAS H. HAMILTON

Q. Mr. Hamilton, could you discuss whether the USGS statement reflected in Mr. McMullen's affidavit, is based on "new" evidence and what that statement's implications are for the Hosgri reanalysis of Diablo Can on?

Mr. McMullen's statement, which is partially 10 quoted in ALAB-598, Appendix question IX, is offered as one of the bases for his affidavit conclusion that "... the near 12 shore high resolution seismic reflection profiling conducted 13 by Mr. Leslie ... does not affect the Staff's conclusion

with respect to the assignment of a 7.5 M design basis Safe 15 Shutdoen Earthquake to the Hosgri fault". 16 It is the understanding of the applicant that the

original assignment of a 7.5 M earthquake capability to the 18 Hosgri fault by the USGS was influenced by a study by 19 Gawthrop (1975) in which it was concluded that "... thus the 20 earthquake is assumed to have originated from a rupture along the Hosgri fault, from near Purisima Point on the 22 south to the general vicinity of Port San Luis" (p. 13). 23 The USGS (1976) expressed this view as "... the magnitude of the design basis earthquake for the Diablo Canyon nuclear 25 reactor site should be about 7.5 and located on the Hosgri 26 fault zone. Thi~ is based principally on the fact that the

Hamilton IX-1

November 4, 1927 earthquake had a magnitude of 7.3 and that, the best estimates of its location indicate that it could have occurred on the Hosgri fault" (p. C-4). At the time this assessment was made, the USGS apparently discounted other studies of the location of the 1927 earthquake, such as the one by consultants to the applicant, e.g., the report "Location and source of the 1927 Lompoc Earthquake" (Appendix 2.5E, Amendment 41 to FSAR, August, 1975) or the article "Seismic moments of the larger earthquakes of the 10 Southern California Region" by Hanks, Hileman, and Thatcher (1975) which discussed evidence that favored association of 12 that earthquake with, in the first instance, the offshore 13 Lompoc fault, or in the second, with an offshore fault of the Transverse Ranges Province. The interpretation by the Applicant that the 1927 16 earthquake did not occur on the Hosgri fault, and that it 17 most probably originated on the offshore Lompoc fault was 18 made part of the direct written testimony of Applicant 19 consultants Hamilton and Jahns (p. 57) and Smith (p. 22) for 20 the ASLB seismic hearing. The evidence on the subject 21 available through mid 1978 was also reviewed by the NRC 22 staff, and it concluded in the direct written testimony of 23 Stepp (p. 31) that "The NRC staff considers the weight of the available evidence to support the conclusion that the 25 1927 earthquake was not centered on the Hosgri fault and O, 26 Hamilton IX-2 L,

I ~g

f most likely occurred on structures in the Transverse Ranges." Although Gawthrop (1978) published an article presenting his further interpretation that (p. 1715) "[t]his active fault. (the Hosgri) should be considered as a likely candidate for the causative fault ... (for the 1927 earthquake)." Other articles published then and subsequently presented evidence not favoring association of that earthquake with the Hosgri, vario'usly on geodetic, 10 seismologic, and geologic grounds. Examples include Savage and Prescott (1978), Smith and Hamilton (1978), Payne, Swanson, and Schell (1978), and Hanks (1979). 13 The report referred to in McMullen's affidavit was prepared by an unidentified review team of USGS scientists 15 and was transmitted by a letter dated October 31, 1979, from 16 Joseph I. Ziony, Regional Geologist, Western Region, to Dr. James F. Davis, State Geologist for California. This 18 view on the part of the USGS was subsequently somewhat amplified in'ts Open-File Report 80-229 "Seismotectonic 20 Setting of Santa Barbara, Channel Area, Southern California", 21 by Yerkes, Greene, Tinsley, and Lajoie (1980) . This report 22 discusses some of the data and interpretations given in the references cited earlier herein (Hanks, 1979; Savage and Prescott, 1978; Gawthrop, 1978) and allowing for the 25 possibility that Byerly's (1930) original location for the 26 1927 epicenter (near the Santa Lucia Bank fault farther

Hamilton IZ-3

offshore) may be valid, concluding that "[e]xisting evidence Lompoc earthquake . 2 suggests association of the 1927 (M 7.3) with an east,-dipping reverse fault such as one near Santa Lucia Bank, the offshore Lompoc, or a similar fault 10 km to the south that offsets the sea floor." The conclusion expressed in the two recent. USGS reports is apparently not based on any new geological or seismological evidence which was not presented at the Diablo Canyon hearing. Rather, it appears to represent an 10 evaluation by the USGS of data corresponding to that which was presented in submittals by the Applicant, as early as

12 1975, and that was testified to by both the Applicant's 13 consultants and the NRC staff during the hearing. The USGS thus appears to have independently arrived at an 15 interpretation that corresponds closely with that of the 16 Applicant and the NRC staff regarding the probable source 17 structure of the 1927 earthquake. 18 This contrasts with and tends to make even more 19 conservative the views regarding this issue given in the 20 1976 review by the USGS of the geologic 'and seismologic 21 data relevant to the Diablo Canyon site which, as noted 22 previously, then favored association of the 1927 earthguake 23 with the Hosgri fault. The significance of this to the seismic evaluation 25 of the Diablo Canyon site may be considered to be simply 26 that the high degree of conservatism of the assumption of a

Hamilton IX-4

7.5 M earthquake capability for the Hosgri fault is further affirmed, since the original USGS assumption of an earthquake of 'that size was influenced by its (then) view 4 that an historic earthquake of that size had probably occurred on the Hosgri. 6 /// /// 8 ///

10 ll P 12

15 16 17 18

19 20 21 22 23

25 26

Hamilton XX-5 I REFERENCES

Byerly, P., 1930, The California earthquake of November 4, 1927: Seismological Society of America Bulletin, v. 20, no. 1, p. 53-66. Gawthrop, W. H., 1978, The 1927 Lompoc, California earthquake: Seismological Society of America Bulletin, v. 68, no. 6, p. 1705-1716. 1975, Seismicity of the central Calxfornxa coastal region: U.S. Geological Survey Open File Report 75-134, 87 p. Hanks, T. C., 1979, The Lompoc, California, earthquake (November 4, 1927; M=7.3) and its aftershocks: Seismological Society of America Bulletin, v. 69, 10 no. 2, p. 451-462. Hanks, T. C., Hileman, J., and Thatcher, W., 1975, Seismic moments of the larger earthquakes of the southern California region: Geological Society of America Bulletin, v. 86, no. 8, p. 1131-1139. 13 Payne, C. M., Swanson, 0. E., and Schell, B. A., 1978, Investigation of the Hosgri fault offshore southern California, Point Sal to Point 15 Conception: Report to U.S. Geological Survey, Menlo Park, California by Fugro, Inc., Long Beach, 16 California, 17 p. 17 Savage, J. C., and Prescott, W. H., 1978, Geodetic control and the 1927 Lompoc, California 18 earthquake: Seismological Society of America Bulletin, v. 68, no. 6, 19 p. 1699-1703. 20 S. W., and Hamilton, D. H., 1978, Sea floor expression of the 1927 Lompoc earthquake, abs.: 22'mith, EOS, American Geophysical Union Transactions, v. 59, no. 12, p. 1128. U.S. Geological Survey, 1979, Review of geologic and 23 seismologic technical reports for proposed liquified natural gas storage facility near Point Conception. California: Report to California Division of Mines and Geology, Sacramento, 25 California, 9 p.

Hamilton IX-6

1976, Review of geologic and sexsmologic data relevant to the Diablo Canyon Nuclear Power Station, Units 1 and 2 (NRC Docket- Nos. 50-275 OL and 50-323 OL): Report to the U.S. Nuclear Regulatory Commission, Washington, D.C., 17 p. Yerkes, R. F., Greene, H. G., Tinsley, J. C., and Lajoie., K.R., 1980, Seismotectonic setting of Santa Barbara Channel area, southern California: U.S. Geological Survey Open-File Report 80-299, 24 p.

15 16

19 20 21 22 23

25 26

Hamilton IX-7 ~ .