ML19310A249

Safety Evaluation Re Geologic & Seismic Design Bases for Facility Modifying NRC 790927 Rept Stating That Surface Offset of 2.5 Meters Could Occur Beneath Getr.Surface Offset of One Meter Beneath Facility Is Appropriate Design Basis

ML19310A249
Person / Time
Site:	Vallecitos File:GEH Hitachi icon.png
Issue date:	05/23/1980
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML19310A248	List: ML19310A247 ML19310A249 ML19310A250 ML19310A251 ML19310A252 ML19310A253 ML19310A254 ML19310A255 ML19310A256
References
NUDOCS 8006060452
Download: ML19310A249 (38)

Text

.. _ _

o O

a arc UNITED STATES ft

,og y

,#(

g NUCLEAR REGULATORY COMMISSION c

WASHINGTON, D. C. 20555 a

SAFETY EVALUATICN BY THE OFFICE OF NUCLEAR REACT')R REGULATION FOR THE GENERAL ELECTRIC TEST REACTOR GENERAL ELECTRIC C0i1PANY DOCKET NO. 50-70 Introduction On October 24, 1977 the NRC issued an Order to Show Cause requiring that the General Electric Test Reactor (GETR or the facility) be placed in cold shutdown pending further Order of the Comnission. The basis for this action was new geologic evidence which placed in question the adequacy of the GETR's seismic design. The issues of the Order are as follows:

(1) What the proper seismic and geologic design bases for the GETR facility should be; (2) Whether the design of GETR structures, systens and components important to safety can be modified so as to remain functional, considering the seismic design bases determined in issue (1) above; and (3) Whether activities under Operating License No. TR-1 should be suspended nending evaluation of the foregoing.

By letter dated November 11, 1977, the General Electric Company (GE or the licensee) responded to the Order. Based on geologic, seismic, structural and analysis infor-mation accompanying their' response, GE requested authorization to resume operation s

following conpletion of proposed modifications. Since that time extensive site and facility reviews have been in progress. The staff, in this safety evaluation, presents its partial conclusions regarding the proper seismic and geologic design bases for the GETR (Issue 1 of the Order). The evaluation is presented in three sections as follows:

A.

NRC Geosciences Branch Review B.

Review of Probabilistic Analyses C.

Engineering Seismic Design Parameters

( 't 800606o 45 A p

..w.-

,;-..a-4,_mm 4

m A

.y

.a a.

m

+e.a.m4_mah r.

4-

+=

_r a

+- -->- - - +em a

--.w

,-aa-

~

,b

># 8 tog g_

jo UNITED STATES y

g NUCLEAR REGULATORY COMMISSION j

WASHINGTON, De c.20555 3

5 r

%.,...../

t I

i GETR 4

l, f

1 SECTION A i

i k

i i

NRC GEOSCIENCES BRANCH REVIEW 4

I 4

i i

t i

i i

I 1

---m-*-,em--.

,,w..._.g v.

-,w--

- -,~- - - -. -.,

m 4

w-.

i-,--

,y-c.

,. ~ -

..n

-wyy,-

v n--m---

n i

SECTION A NRC GEOSCIENCES BRANCH REVIEW I.

Background

In July 1977, the Geosciences Branch was requested to perform a review of the geology,and seismology aspects of the General Electric Company's (GE) application to renew the Operating License of the General Electric Test Reactor (GETR) at Pleasanton, California. As a result of that initial review anc a trench investigation which revealed the existence of a low angle thrust fault near the reactor, a Show Cause Order was issued on October 24, 1977.

Since that time, GE has continued to acquire data to better establish the mode of origin of the faults, their locations, degree of recurrent activity, and potential for reactivated movements. On September 27, 1979, the staff issued a report entitled, "Geosciences Branch Safety Evaluation Report Input". That report documented the Geosciences Branch review up to September 6,1979 and defined the NRC staff position regarding'the proper seismic and geologic design bases for GETR.

In that review, the staff concluded that a conservative description of surface displacement on the Verona fault zone during a single earthquake event was that a surface offset of two and c,ne half meters could occur beneath the GETR.

On November 14, 1979 both GE and the staff presented their conclusions to a subcommittee of the Advisory Committee on Reactor Safeguards (ACRS).

(Transcript of USNRC ACRS Subcommittee meeting on GETR November 14, 1979 held at Burlingame, California, 331 pp.).

As a res0lt of that meeting and the questions raised by the Subcommittee and its consultants, further review of the seismological parameters and a probabilistic assessment of the surface fault potential was

I e undertaken. As a result of the staff's review of GE's April 12, 1979 submittal of a Jack R. Benjamin and Associates probability study and our generally negative conclusions regarding it, GE undertook a new probability study.

In addition, we have received a number of reports from GE relating to the probability study, supporting bases for geolologic assumptions in the study, a fault evaluation of GETR excavation photographs, dip of faults, discussions of the Livermore Valley regional seismicity, and the significance of observations of the 1979 Imperial Valley earthquake. These reports are:

(1) General Electric Company, " Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building, General Electric Test Reactor," March 12, 1980, by Jack R. Benjamin and U soc., Inc.

(2) General Electric Company, Additional Analysis, supplement to

" Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building, General Electric Test Reactor," March 17, 1980.

(3) General Electric Company, " Dip Angle for General Electric Test Reactor (GETR) Site Shears," March 31, 1980, memo from R. C. Harding to D. L. Gilliland dated March 27, 1980,3 pp.

(4) General Electric Company, " Response to Questions Raised by the GETR Subcommittee of the Advisory Committee on Reactor Safeguards Consultants-License-TR-1 Docket 50-70," April 14, 1980; includes B. A. Bolt and R. A. Hansen, " Seismicity of the Livermore Valley in Relation to the General Electric Vallecitos Plant," dated March, 1980.

(5) General Electric Company, " Proposed GETR Landslide Investigation, License TR-1, Docket 50-70," April 17, 1980, 3 pp., 2 figs.

(6) General Electric Company, " Analysis of Slip Rate of Shear Surfaces i

i at the General Electric Test Reactor (GETR) Site - License TR-1-Docket l

50-70," April 29, 1980, 5 pp.

l l

i (7) General Electric Company, " Analysis of the General Electric Test Raactor (GETR) Foundation Excavation Photographs - License TR-1-Docket 50-70," April 29, 1980, 3 pp.

(8) General Electric Company, " Responses to NRC Questions on Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor," April 30, 1980, 19 pp.

(9) General Electric Company, " Resumption of Operation of the General Electric Test Reactor (GETR) - License TR-1 Docket 50-70," April 30, 1980, includes (1) "A Seismological Assessment of the Probable Expectation of Strong Ground Motion at the GETR Site," April 28, 1980, by R. L. Kovach to Engineering Decision Analysis Company, 14 pp.;

(2) " Probability Analysis for Combined Surface Rupture Offset and Vibratory Ground Motion, General Electric Test Reactor," April 29, 1980 by Jack R. Benjamin and Assoc., Inc.

We and our consultants and advisors have reviewed the data provided in these reports submitted up to May 1, 1980. The reviews and recommendation of our consultants and advisors are contained in a variety of reports attached as Appendices A, B, C, D, E and F to this review and listed below:

(A)

N. M. Newmark and W. J. Hall, Letter report, " Seismic Evaluation of Vallecitos Site," April 14, 1980 to C. Nelson and W. Burkhardt, 6 pp.

(B)

D. G. Herd and E. E. Brabb, " Faults at the General Electric Test Reactor Site, Vallecitos Nuclear Center, Pleasanton, California, A Summary Review of Their Geometry Age of Last Movement, Recurrence, Origin,andTectonicSettingandtheAgeoftheLivermoreGrahels,"

U. S. Geological Survey Admin. Rept., April 1980, 77 pp.

~

(

e (C)

W. L. Ellsworth and S. M. Marks," Seismicity of the Livermore Valley, California Region, 1969-1979," U. S. Geological Survey Open File Report 80-515, 1980, 42 pp.

(0)

M. G. Bonilla, J. J. Lienkaemper, and J. C. Tinsley, " Surface Faulting Near Livermore, California Associated with the January, 1980 Earthquakes," U. S. Geological Survey Open-File Report 80-523, 1980, 27 pp., 5 figs., map ~

(E)

D. B. Slemmons, Letter report to R. E. Jackson, review of " Probability Analysis of Surface Rupture Beneath Reactor Building, General Electric Test Reactor," April 28, 1980, 19 pp.

(F)

D. L. Bernreuter, Letter and TERA Corp. Report to 0. Eisenhut

" Seismic Rupture Hazard at the General Electric Test Reactor: A Review and Analysis," May 8, 1980 Our conclusions and supporting bases are presented below. The September 27, 1979 Geosciences Branch Safety Eva';ation Report Input is referenced in the following discussion and, excep+ for its conclusions, is considered part of the NRC's safety evaluation.

II. Current Staff Position The information available at the present time leads us to conclude that:

(1) This evaluation represents our finding with respect to the Show Cause Proceeding. The information developed for this site does not completely meet the investigative requirements of Appendix A to 10 CFR Part 100.

o

t./

• (2) Geologic data indicates that the GETR site is located within a zone of faulting (the Verona fault) which is at least 3200 feet wide based on the latest USGS report.

(3) Since the Verona fault displaces Holocene (less than 10,000 years old) soils it is a capable fault within the meaning of Appendix A to 10 CFR Part 100 and, therefore, poses a potential for surface faulting near or beneath the reactor site.

• (4) Future displacements in the GETR site area have a higher likelihood of occurring along existing fault breaks than between them. Although it is our opinion that such estimates cannot be specifically quantified for the Verona fault zone, such a judgement can be made based on general geologic observations and probabilistic considerations (Slemmons, 1980, Appendix E attached; Bonilla, 1979 and probabilistic estimates LLL/ TERA, 1980, Appendix F attached). The possible existence of faulting has been identified in photographs of the GETR excavation.

• (5) One meter of reverse-oblique net slip along a fault plane which could varyindipfromabout10to45degreesprovidesanapphopriatedescription of surface displacement which could occur on a Verona fault zone strand (splay) beneath the reactor during a single event.

(6) Maximum vibratory ground motion at the GETR site would result from a magnitude 7 to 7.5 earthquake centered on the sector of the Calaveras fault nearest the site. Acceleration peaks at the free-field surface could l

be slightly in excess of 1 g.

This acceleration value is based on.seismolo-gical principles and since it is a free-field value does not incorporate factors dependent on soil-structure interaction or the behavior of the structure.

• denotes positions which have been modified since the September 27, 1979 report, Letter from H. R. Denton, USNRC to R. W. Darmitzel, GE.

a

. (7) The horizontal vibratory ground motion at the GETR site resulting from an earthquake of magnitude 6 to 6.5 centered on the Verona fault could contain acceleration peaks as high as 1 g.

However, the 'overall l'evel and duration of shaking would be less than for a magnitude 7 to 7.5 earthquake centered on the Calaveras fault approximately 2 kilometers from the site.

• (8) Combined loads caused by fault offset at the surface and vibratory ground motion must be considered to act simultaneously because there is no reasonable way to conservatively forecast the location of rupture initiation, the mode of rupture propagation and the potential source area for radiated seismic energy or the sequence of possible interaction among the Calaveras, the Verona and the Las Positas faults.

Inhiewoftheabohe, there is insufficient evidence to support the proposition that strong ground motion and surface fault displacement will be separated in time. Although we recognize that the entire one meter of displacement noted in (5) above will probably not occur coseismically (during the time of the earthquake), we are unable to quantify what proportion of that displacement will occur during strong shaking. We recomend further consideration of this aspect by the structural engin!ers and their consultants dependent upon the critical significance of this observation to th' structural evaluation.

e

• (9) While it is the staff's position that the evidence strongly supports tectonic origin of the offsets observed in the trench exposures at the site, there is also evidence for a potential landslide hazard at the site. This is based on (1) location of the GETR within a shear zone; (2) j evidence for repetitive displacements on these shears; (3) youngest offset l

l during the Holocene; (4) topographic relief adjacent to the site, and (5) potential for seismic loading. Landslides often occur as a result of seismic events,

• denotes positions which have been modified since the September 27, 1979 report, Letter from H. R. Denton, USNRC to R. W. Darmitzel, GE.

1

and, therefore, the landslide hazard is considered to be part of tne overall geologic and seismic hazard to the GETR site. The proposed GETR landslide investigation program, dated April 17, 1980 provides a generally acceptable investigation program and should provide sufficient data regarding the stability of the hillside deposits for the staff to evaluate the potential hazard. Our review of the results of this program will be provided in'an SER supplement after the information acquired during the investigation is docketed and completely reviewed.

A schedule for implementation of the program has not been provided by GE.

III. Discussion 1.0 Geology 1.1 General - A general discussion is included in the September 27, 1979 staff report.

1.2 Verona Fault and Verona Fault Zone The Verona fault lends its name to the zone of faults recognized in the trenches and boreholes near GETR. Each of the principal faults identified in trenches 8-1/B-3, B-2 and H are referred to as the Verona fault.

Since some new or reorganized information was presented by GE at the ACRS subcommittee meeting the U.S. Geological Survey (USGS), as advisors to the NRC staff, undertook a comprehensive review of arguments and data provided by GE relating to the absence of the Verona fault. Their detailed review, entitled, " Faults at the j

General Electric Test Reactor Site, Vallecitos Nuclear Center, Pleasanton, California, A Sumary Review of Their Geometry, Age of Last Movement, Recurrence, Origin,

s

. andTectonicSettingandtheAgeoftheLihermoreGrahels,"isattachedas

~

Appendix B to this report. Their report provides further supporting bases for the conclusion that the Verona fault should be considered to be a tectonic (earthquake) fauit. Several items from their report which are of particular significance are noted below; however, only a thorough study ofthereportandcomparisonswithahailabledatacanprohideacomplete understanding of their review.

Slickensides in trenches near GETR show essentially dip-slip movement, with a small component of lateral displacement.

Boreholes suggest thrust fault zone steepens from 14 degrees at the surface to 35-40 degrees with depth.

There may be faults that actually surface beneath the reactor.

The evidence gathered by Earth Sciences Associates, GE consultants, is both permissive and supportive of fault displacement in the last 2,000 to 4,000 years because the modern soil is formed in< colluvium. deposited atop the stoneline and the stoneline may bec.as: young as.10,000 years; the last glaciation did hot conclude until around i0,000 years befoe present (BP); and the C-14 ages for the soils are reasonable and there is no need to apply radical age corrections.

Faulting occurred after the modern soil achieved its present degree of maturity; if this faulting.had occurred 8,000 or 10-15,000 BP, a new soil profile should have developed across the faulted soil; therefore faulting is less than 2,000-4,000 BP.

There were 2 to 5 feet of thrust movement on B-1/B-3, T-1 in the last 4,000 years.

Aheragesliprateof.0004ftlyr(0.012cm/yr)fitsacurveofcumulative apparent dip slip separation versus age of displacement.

The Verona fault occurs within one of the most unusual tectonic settings in coastal California; the Greenville fault is the easternmost seismically active right slip fault in the central Coast Ranges; the Las Positas is one of only two known active left lateral strike slip faults in Coast Ranges north of Transverse Ranges.

l i

l

. The occurrence of the January 1980 earthquakes near Lihermore, California and observed surface breakage confirmed the USGS tectonic model of the Livermore Valley and revealed that faulting can occur simultaneously on more than one of Valley's faults.

Assuming that alluvial deposits in 8-1 extend beneath GETR, the reactor rests on beds older than 70,000-130,000 years and younger than 300,000 years.

In addition to this review, the USGS also completed a study of the Livermore Valley region st.lsmicity. This study entitled " Seismicity of the Livermore Valley, California Region 1969-1979, Open-File Report P N 15" was prepared by W. L. Ellsworth and S. M. Marks and is attached as xppendix C to this report. With respect to the Verona fault, this study indicates that the Las Positas, Pleasanton and Verona faults are identified as probably seismically active faults.

In addition, Ellsworth and Marks conclude that earthquake focal mechanism solutionsforehents nearVallecitosValleydemonstratethatthisregionisazoneofactihe thrust faulting and that some of these thrust events are in probable association with the Verona fault. Finally, they suggest that better microearthquake instrumental coverage and knowledge of crustal structure in the Vallecitos Livermore area should allow this relationship to be demonstrated.

A modification to our past position relates to the age of the youngest offset soil. The staff prehicusly indicated that the younger soil (estimated 8,000-17,000 years old) horizons are displaced approximately 3 feet. After carefulrehiew,theUSGS(AppendixB)indicatesthatthemostrecentfault movement-isbeliehedtohaheoccurredsince 2,000-4,000 years before present.

Dr. D. B. Slemons, an NRC staff consultant on fault evaluation, indicates in his letter report of April 28, 1980 (attached as Appendix E to this report) that

l i l he would place an error band for fault displacement in the soil between approximately 1,500-2,000 years to 4,000 years before present for trench B-1.

Based on these recommendations, we conclude that offset of the youngest soilhorizoncouldhaieoccurredwithinaboutthelast2,000 years. This determination is important because it relates to surface slip rate and recurrence e'stimates which a-e discussed in the following section on surface offset. Duringthiscontinuingrehiew,theMs=5.8 January 24, 1980 1

and Ms=5.2 earthquakes of January 26, 1980 occurrednorthofLihermore, I

California and were accompanied by surface faulting in the Greenville fault zone and apparently on the Las Positas fault zone also. An Open-File Report 80-523 entitled"SurfaceFaultingNearLiYermore,CaliforniaAssociated with the January 1980 Earthquake," has been prepared by M. G. Bonilla, J. J. Lienkaemper, and J. C. Tinsley and is attached as Appendix D to this report. TheirreportindicatesthatthemainrupturewithintheGreenhille faultzonewasabout4.2kilometersandmayhaYeextendedto6.2 kilometers.

Maximum displacements which were measured indicated about 25 m of right slip including afterslip with Yertical components of up to 50 m.

The reportalsonotesthatthemainbreakintheGreenhillefaultzoneishery close to a fault strand mapped by Herd (1977).

In addition the authors note that some minor amounts (0.5 mm-6mm) of left lateral strike slip tectonicdisplacementslieonorclosetoprehiouslymappedstrandsofthe Las Positas fault. Thisreportisappendedinordertoprohidefurther informationrelatihetotheneotectonicsoftheLihermore,Californiaarea.

11 -

Based on the above considerations, in addition to our past review activities, we sustain our conclusion that a fault zene (Verona) passes through the GETR site area.

1.3 Surface Fault Offset 1.3.1 Previous Positions Previously GE concluded that measurements of offset soil stratigraphic markers confirm that the maximum amount of offset that has occurred on a single shear surface within the last 20,000 years is less than 1 meter and, therefore, an assumed offset of one meter is conservative. Based on further investigations (Phase II, including probability analyses), GE indicated that a zero offset for the reactor building and no more than three feet in a plane 15-25 degrees from horizoatal on observed she es at the site should be used for design criteria.

s As a result of the staff's analysis and on the advice of consultants, it was concluded in the staff September 27, 1979 report that 2.1/2 meters of reverse-oblique net slip along a fault plane which could vary in dip from 10 to 60 degrees provides a conservative description of surface slip on the Verona fault zone during a single event. This judgement was based in part on observations and comparisons with the maximum calculated net slip displacementobserhedduringthe1971SanFernando,Californiaearthquake.

The position was based also on comparisons with the ahailable worldwide fault offset information for reverse and reYerse-oblique slip faults as well as the recomendations of the USGS and Dr. Slemmons.

In addition, because of an inability to quantify the likel3 cod of new rupture between the existing shears, the staff concluded that this offset could also occur beneath the reactor.

, 1.3.2 Current Reconinendations of Review Groups Based on the review of all available information, the USGS's recommend-ation to the NRC remains the same as contained in their September 5, 1979 report which is attached to the staff's September 27, 1979 report.

The USGS indicated that insufficient investigations had been accomplished to adequately assess the potential for surface faulting at the GETR site. The USGS continues to maintain that the present deficient information base does not preclude that the net slip that could occur on the Verona fault could greatly exceed the one meter orocosed by GE. In addition, the USGS advised that unknown.taults could exist beneath GETR. As a result of the critical significance of the pre-sence or absence of surface f aulting to the probabilistic. determination, a recent examination of photographs taken during the excavation for the GETR foundation was undertaken. The USGS has notified us verbally that several offset features can be identified on several phbtographs.

A written report will follow upon completion of their review.

Dr. David B. Slemmons in his letter report of April 28, 1980 indicates (Appendix E) that ruptures on the site are most likely to have surface offsets from 2 to 3 feet, but any future displacements could have maximum offsets of 2 - 2.5 meters as noted in the Staff's position of September 27, 1979.

The current data, in his opinion, do not significantly change the position outlined in the September 27, 1979 Staff's Position.

Drs. N. M. Newmark and W. J. Hall in their letter report of April 14, 1980, attached as Appendix A, reconnend a maximum fault motion on the order of about 1 meter. This fault motion should be taken as the resultant gross motion, but it may occur in any arbitrary direction.

The LLL/ TERA report (Appendix F) primarily discusses a review of the JBA/GE probability study and another proposed probabilistic method. The LLL/ TERA review indicates that because of the geologic data, which indicates that the GETR site is located within a zone of active faulting, it is difficult to dismiss the need to consider some surface rupture coupled with a high"gfound acceleration,"g". They further indicate that, given the uncertainty in all parameters, it is possible that surface offset could occur beneath GETR. LLL notes that the probabilistic analysis by TERA tends to show that this rupture is a low likelihood event, thus it seems reasonable to select the offset and "g'.' value near the mean of the data for a magnitude 6-6.5 earthquake. Typical values of about 0.5 to 1.0 meters would be a reasonable range to select for offset.

In its letter, LLL also notes that even if all of the activity on the Verona is assumed to occur on a shear beneath GETR, the probability of occurrence of a one meter displacement (based on the TERA method) would be on the order of 5x10-3 per year. A further recommendation is made that the peak acceleration should be associated with a lesser value of offset and that further analyses should be accomplished on this aspect if necessary.

. The opinion of the California Division of Mines and Geology (CDMG) has not been solicited by the staff during this latest phase of the review and no further reports have been received. In their August 16, 1979 report attached to the September 27, 1979 staff report CDMG concludes that approximately three feet of surface displacement at the reactor site represents a conservative judgement as to the ground rupture which might be associated with either the landslide interpretation or tectonic mddel.

1.3.3 Staff Evaluation

1. 3. 3.1 Probability Analyses In April 1979 GE submitted a probabilistic study in support of an argument that indicated a quantifiable lower likelihood of fault rupture between the existing shears than on them during an earthquake event. Review of this report was accomplished by the staff and discussed with the licensee during a meeting on January 31, 1980. As a result of these discussions the licensee j

submitted a new probability analysis by Jack R. Benjamin and Associates (JBA) on March 12, 1980. Based on the JBA results, which indicate that the probability of offset beneath GETR is 1 - 1.2 x 10-6, GE maintains that a zero-offset design criteria be used for analyses.

As a result of questions by the ACRS subcomittee members, the staff initiated a new review of this information and solicited the review assistance of Lawrence Livermore National Laboratory (LLL) and a subcontractor to LLL, the TERA Corpora-tion (TERA).

In addition to its review of the GE methodology and model, LLL/ TERA developed its own model and methodology. The validity of the geologic parameters input assump-tions was assessed and a judgment regarding overall applicability of this proposed method was made with the assistance of Dr. David B. Slemons.

Dr. Slemmons' letter report is attached as Appendix E to this evaluation. The review of GE's " Additional Probability Analysis" by LLL and TERA is attached as Appendix F.

The staff's ovaluation of the proposed JBA/GE probabilistic model and the LLL/ TERA report is contained in Section B of this evaluation.

i

. Dr. Slennons has specifically reviewed the geologic parameter input assumptions used in the JBA/GE study (Appendix E).

Highlights of his review are listed below.

The time indicated by JBA since last offset, t, does not indicate thepossibilitythatthetimeoflastsoildisSlacementcouldbeas recent as 1,000 to 4,000 years BP.

The total accumulated offset on the B-1/B-3 shear may be greater than the upper bound value of 420 ft, as indicated by the hypsometric values for the Vallecitos Hills. There could have been from 990 to 2047 ft slip across the zone of 3 faults; this suggests an upper bound that is 2 to 4 times the upper bound values of 420 ft used in the JBA analysis.

The JBA/GE analysis does not consider the third dimension of the GETR foundation. The effect of foundation depth significantly increases the potential for fault rupture at the foundation.

The dates used for the analysis reflect ages to three significant figures in.some of the tabulations, but are actually assumed values that could have errors in the first significant figure.

Many of the assumptions presented in the JBA/GE probabilistic analysis are reasonable and conservative, but the overall effect of the GETR foundation geometry and possible errors in inferred soil ages appear to make the overall probability assessment a non-conservative evaluation.

In general, the probability of offset is about 1/ number of years since last offset beneath the reactor. The application of both the JBA method anc !he TERA method is critically dependent upon the absence of faulting beneath the GETR. This might not be the case as is discussed later. The TERA method, in general, estimates that the probability of any size offset cccuring under GETR ranges from 7.6 x 10-5 to 1.4 x 10-6 per year, assuming no offset observed under GETR in 40.000 years, and 3.5 x 10-5 to 4.5 x 10-7 per year assuming no offset under GETR in 128,000 years.

l To check the validity of the model, a comparison is made to the geologic observations on the shears at the site. This comparison indicates that

)

geologists should observe 1 meter of offset per 20,000 years. The staff curintlybeliehesthatgeologicobserYationsbetterestimate1meterper 1

. 2,000-8,000 years.

If this latter estimate is used to calibrate Fig. 3-2 (TERA Report) the probability estimates change about one order of magnitude.

After making these parameter changes, the probability of 1 meter of displacement on the recognized faults would be on the order of about lx10-3 to 1x10-4 per year compared to lx10-7 to 3.3x10-8 per year between shears assuming no faults under the reactor (with no offsets in 40,000 and 128,000 years, respectively, and a relative error of 50%); a difference of about three orders of magnitude.

Other input assumptions such as fault length, effect of upper magnitude cutoff, slip rate, limited larger magnitude data sets, and use of a worldwide data set for fault observations rather than thrust fault specific infomation must be considered and sensitivities assessed. Of great importance is a primary assumption in the model which implies that an earthquake occurrence at each source forms a Poisson process. This implies that the occurrence or absence of a seismic event at one site does not affect the occurrence or absence of another seismic event at some other site or the same site. This assumption must be carefully tested because geologic observation seems to indicate, at least in the shorter term, that rel2ase of stress in one area of a fault would concentrate stress at another point on the fault or on another part of the system.

Deciding the proper surface offset design basis for a facility within a fault zone by using the proposed probabilistic methods is not favored by any of the geological personnel involved in the review of this site. Several specific areas of concern were outlined above. Far more important, however, is the judgment that such methods are highly dependent on very uncertain input parameters and the critical effects of localized site specific conditions, that such methods have yet to be critically tested against sensitivity to a variety of parameters, and finally, that such methods suffer from a lack of testing against observations of fault behavior in well-known geological areas.

(

- The probabilistic calculations do however provide a frame of reference for making a judgment on geological offset parameters that are not at the upper bound for the dispersion of the available data. Furthennore, they help provide a perspective of the type of data which is needed and which is most critical to making a conservative estimate of the surface offset displacement.

1.3.3.2 Location of Of fset Based on the importance of the presence or absence of faults in the GETR foundation area, we initiated a reexamination of the available photographs of this area. We have also requested that the USGS provide such a review. Although we have not received the U53S report, we have been notified that possible fault-like features can be observed in the photographs. The seven photographs (GE April 29,1980 and January 5,1978 submittals)of the GETR foundation excavation provide the only known direct visual evidence of the geological and pedological characteristics and structure under and immediately adjacent to the reactor. Each photo showc that the foundation material consists of horizontal or nearly horizontal, layered, alternating light-and dark-col'. red sedimentary strata or soil units.

Several photos,.as noted below, suggest the presence of at least one surface of stratigraphic or structural discontinuity. Photos, A, C and F show an east-dipping light-colored unit which apparently is in angular discordance with the flat-lying strata or soil units to the north and west (left side of photos).

Photos B and E show an angular discordance of layers on the face diametrically opposite the discordance in photos A, C and F.

These zones of apparent discordance mayormaynothahebeencontinuous;thatis,theymayrepresentseheralseparate features or features of differing origin rather than being a single planar or curhiplanarfeaturethatwastransectedintwoplacesintheexcavation. The apparent dips in circular sections and physical separation of about 80 feet of these zones of apparent discordance along with rough-cut walls of the excavation, distant camera-hiewsofsuspiciousareas,amongotherdehiations l

from the ideal, allow for a variety of interpretations of the features shown.

l l

s We also considered:

(1) that the apparent absence of faults in the nearby trench B-1 might be due to the shallowness of the trench where the GETR foundation can be projected onto the trench logs; (2) that there is an absence of topographic or air photo evidence for surface offset at the foundation site; (3) that stratigraphic and structural discontinuities may be difficult N recognize in trenches in Livermore Gravel and deposits derived from it, even under the best conditions; (4) that the zones of apparent disc)rdance may be faults as well as unconformities or sedimentary structures such as channel fill; and (5) that several geologists who have observed faults in similar materials in nearby trenches dug in 1979 recognize features shown in the

~

photos as possible faults.

Based upon the above considerations, the staff concludes that a fault or faults might exist under GETR and it is our judgment that the potential for surface offset displacement beneath the reactor must be considered in the design basis analysis for the GETR. This judgement results from considerations of significant information deficiencies (as noted by the USGS) and includes, for example the need for boreholes to determine the projection of faults, further trenching, and mapping to determine the extent of the Verona fault.

It also reflects the need for extensihe demonstration and testing of probabilistic models for fault displacement estimation, the behahior of faalts and soil materials in close proximity to faults, the weakness of'the data base in this area, and inherent uncertainties in predicting fault mohement behahior.

1.3.3.3 Amount of Offset We have detennined that one meter of reverse-oblique net slip along a fault plane provides a description of maximum expected surface slip on the Verona fault under the GETR during a single event. This parameter should be used with engineering judgment as input into the design basis criteria and it should be recognized that probably not all of this slip will occur coseismically.

. Ourjudgementisbased,inpart,onourunderstandingandehaluationof obserhationsoffaultoffsetsmadefollowingthe1971ScnFernando, California earthquake (Barrows and others, 1973). We recognize that the assumption thattheSanFernandoandVeronafaultzonesarecomparableisconservatihe.Wehavedone thishoweherbecauseofthesparsityofdataonthrustfaultmohement. The following indicate the bases for making such a comparison. The Verona fault, including its northwesterly projection along possible splays of the Pleasanton fault, has an estimated surface length of 12 kilometers (Herd, 1977; Earth Sciences Associates,1978).

This fault is either truncated by or merges with the Calaheras fault to the northwest and joins with or is trunchted by the Las Positas fault in the general area of the La Costa tunnel. We believe that utilization of the San Fernando data is a reasonable basis for postulating the amount of offset that could occur on the Verona fault near the GETR because of general similarities: thelengthofobserhedsurfaceruptureduringthe San Fernando event was about 12-15 kilometers; movement was predominantly in a thrust sense with a substantial horizontal component. Assuming the Verona faultrupturesalongitsestimatedtrace,itwouldhaYearupturelengthof about 12 kilometers. Basedonobservationsofarehersethrustmovementinthe trench excavations near GETR and regional stress considerations which would support crustal compression (Lee and others, 1971), we would anticipate the Verona fault toundergorehersemovementasdidtheSanFernandoareafaults. In addition, due to the orientatien of the regional stress, the Verona fault should be expected tohaheahorizontalcomponentofmohement.

Bonillaandothers(1971)calculatedsliphectorsalonganassumedfault planeintheOrangeGroheAhenueandEighthStreetareasthatsustainedsurface ruptureduringthe1971SanFernandoehent. These calculations indicate that 2.4 meters of net slip displacement took place. It should be noted however that vertical displacement for this location is distributed across a zone of breakage 200 meters

. wide which is complicated by a zone of shearing and thrusting and a zone of extension.

At the present time, we do not hahe a compilation of direct measurements of net slip on individual surface ruptures.

Wehaveusedtwoapproachestoanalyzetheahailableinformationonreherse andreherse-obliquefaultmohement. Weanalyzed179obserhationsofhertical surface offsets that occurred during the San Fernando earthquake as compiled by Barrows and others (1973). This analysis indicates that the mean of the cbserhedherticalthrowonagihenfaultbreak is a' bout 34 centimeters

(.34 meters). Of the 179 obserhations, 97% were less than 1 meter and 5 obserhations egaaled or exceeded 1 meter. Themaximumherticaloffsetnotedwhichexceeds 1 meter is 160 centimeters (i.6 meter). Onemeterofherticaloffsetexceeds themeanplustwostandarddehiationsfortheSanFernandodata. Obserhations of offsets in excess of the abohe halues are due to mohement distributed acrossthefaultzoneandreflectrelehelingorarenetslipcalculations (resultantofthedip-slipandstrike-slipmohement). Themeanhalueof thehorizontalmohementwouldbeabo0t40 centimeters (.4 meters). Six of the40horizontalmohementobserhationsnotedequalledorexceeded1 meter with the maximum being 190 centimeters (1.9 meters). Using one meter of net slipresultsinapproximately0.7metersofherticaloffsetassumingrupture on a fault which dips 45 degrees. A comparison with the observations of Barrows and others (1973) indicates that 0.7 meters exceeds the mean plus one standard dehiation for the obserhed data for all segments of the f ault. This statisticalinterpretationmustb.ehiewedcautiouslybecauseofpossiblebias in the sampling and measurement of offsets in the field.

. In addition to the above broader comparisons with San Fernando, detailed obserhation of the fault displacements in the trench excaYations near GETR indicate that the nature of movement on the Verona fault zone occur as episodes of movement of about.7 to 1 meter per event on each recognized fault about every 2,000-8,000 years.

i Both approaches support the staff position that one meter of reverse net slip along a fault plane provides a description of the maximum expected surface slip on the Verona fault under the GETR during a single event.

1.3.3.4 Dip Angle of Verona Faults The staff concludes that the surface and near-surface dip angle of a Verona fault strand (splay) is likely to fall in the range of 10 to 45 degrees. This rehisionofour?rehiousestimateof10to60degreesasstatedinour September 27, 1979 report is based on the following geologic and seismologic considerations:

(1) all of the shears exposed in trenches at VallecitosCenterhahedipslessthan45 degrees;sehentypercentofdips measuredarethirtydegreesorless;twomainshearsclosesttoGETRhahedips ranging from 0 to 25 degrees (Harding to Gilliland memo dated March 27,1980);

(2) a ncw fault rupture at GETR would likely dip between 9 and 25 degreet north; itispossibleforthrustfaultsinLihermoreValleyareatoflattenwithdepth, but the steepening of dip with depth of reverse-slip f aults is comon; present trench and borehole data for GETR are inconclusive regarding dip angle changes with depth (Slemons, Appendix E); (3) B-1/B-3 fault zone dips 9 to 31 degrees

\\

NE; B-2 and H subsidiary thrusts dip 25 and 27 degrees flE, respectively; bore-j hole data suggest that the thrust fault at the hillfront steepens from 14 degrees at the surface to 35-40 degrees with depth (Herd and Brabb, Appendix B); (4)

i reverse fault planes dipping 45 degrees NE associated with the Verona at depth can be resolved from focal mechanism plane solutions (Ellsworth and Marks, Appendix C).

2.0 Landsl iding Since the issuance of the September 27, 1979 NRC staff position, GE has not conducted a landslide hazard investigation. On April 17,1980, GE submitted a proposed landslide hazard investigation program. This submittal provides no schedule for implementation or completion of the program.

It has beer. the staff's position that the G.E. Landslide Stability Report, July, 1978, was ir. adequate because all important parameters were assumed.

Furthermore it has been the staff's position that detailed investigations and complete analyses are required to assess the stability of the hillside deposits.

The April 17, 1980 proposed GETR Landslide Investigation has been reviewed by the staff and found to be generally acceptable as an investigation program which should provide the data necessary to evaluate concerns regarding the stability of the hillside deposits. The primary goals of the investigation should be (1) to determine to the extent possible the failure plane, (2) to obtain and test samples from the failure plane and (3) detennine groundwater conditions. Monitoring for possible future movement is not included in the current proposed program and should be added. Our review of the results of the program will be provided in an SER supplement, after the infonnation is docketed and reviewed by the staff and consultants. The staff concludes that these investigations and analyses are required to completely define the appropriate seismic and geologic design gases for the GETR.

. 3.0 Seismology The seismic design bases for the GETR are discussed in the September 27, 1979 staff report and that section remains unchanged. Since that time, Drs. Newmark and Hall haYe submitted their review which is discussed in Section C of 'this evaluation and their r oort is attached as Appendix A.

References Barrows, A. G., Kahle, J.

E., Weber, F. H., Jr., and Saull, R.

B.,

1973, Map of Surface Breaks Resulting from the San Fernando, California, Earthquake of February 9,1971; in San Fernando, California Earthquake of February 9,1971, U. S. Department of Commerce, Washington, D. C.,

pp. 127-135.

Bonilla, M. G., 1979, Historic Surface Faulting -- Map Patterns, Relation to Subsurface Faulting, and Relation to Preexisting Faults, Preprint from: Proceedings of Conference VIII, Analysis of Actual Fault Zones in Bedrock, Nat'1. Earthquake Hazards Reduction Program, 21 pp.

Bonilla, M. G., Buchanan, J. M., Castle, R.

0., Clark, M. M., Frizzell, V. A.,

Gulliver, R. M., Miller, F.

K., Pinkerton, J.

P., Ross, D. C.,

Sharp, R. V., Yerkes, R. F. Ziony, J.

I., 1971, Surface Faulting, in, The San Fernando, California Earthquake of February 9, 1971, U. S.

Geological Survey Professional Paper 733, pp. 55-76.

Earth Sciences Associates, 1978, Geologic Investigation General Electric Test Reactor Vallecitos, California; Prepared for General Electric Co...

Pleasanton, California.

Herd, D. G., 1977, Geologic Map of the Las Positas, Greenville and Verona Faults, Eastern Alameda County, California, USGS Open-File Report 77-689, 25 p.

Lee, W. H. K., M. S. Eaton, and E. E. Brabb, 1971, The Earthquake Sequence Near Danville, California,1970, Bull. Seism. Soc. Am. V. 61, p.1771-U94.

a arco

[ja f,,

UNITED STATES g

NUCLEAR REGULATORY COMMISSION y

j WASHINGTON, D. C. 20656

• s.,

/

f GETR SECTION B i

REVIEW 0F PROBABILISTIC' ANALYSES ~~

f i

1

+

_1 SECTION B REVIEW OF PROBABILISTIC ANALYSES By letter dated April 12,1979 the licensee provided the report " Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor". As a result of discussions regarding this report, at a meeting held in Bethesda, Maryland on January 31, 1980, GE submitted a new probability analysis on March 12, 1980 entitled " Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor, JBA-111-013-01".

We have reviewed this report (Document 1) as well as the additional reports listed below:

2.

Letter from R. W. Darmitzel to D. G. Eisenhut,

Subject:

" Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor", dated March 17, 1980 3.

Letter from R. W. Darmitzel to D. C. Eisenhut,

Subject:

" Responses to NRC Questions on Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor",

dated April 30, 1980.

4.

Dr. Mensing's review of JBA-lll-013-01 contained in the letter from Don Bernreuter and Frank Tokay, LLL to Darrell G. Eisenhut, dated May 8, 1980.

5.

Memorandum from B. J. Davis to L. H. Wight,

Subject:

" Review of the 3/14/80 Benjamin and Associates Report - Additional Probability Analyses of Surface Rupture Offset Beneath Reactor Building - General Electric Test Reactor", dated April 14, 1980.

6.

S?ismic Rupture Hazard at the General Electric Test Reactor: A Review aiid Analyses, TERA Corporation, dated May 1,1980 Documents 1, 2 and 3 are docketed material.

Documents 4, 5 and 6 are contained in Appendix F to this evaluation.

Evaluation We have reviewed these documents for the correctness of the modeling tech-niques, contained in these documents, regarding the probability of a surface offset beneath GETR. We have not performed an independent probability analysis of this occurrence nor-associated the probability results with the size of offset. Docunents 2 and 3 describe further sensitivity analyses

a Section B which were performed to supplement the probability analyses in Document 1 (JBA-lll-013-01 ). Documents 4 and 5 contain reviews of the probability models and analyses given in Document 1.

The probability models in Document 1 (Approaches 1 and 2) are denoted as the "JBA" models. Document 6 describes a different probability model which was developed as an alternative to that described in Document 1.

We will first briefly discuss the review documents (Documents 4 and 5) then the alternative model (Document 6) and then, atte:npt to summarize the sensitivity evaluations presented in Documents 1, 2 and 3.

Finally, we draw some conclusions from the probability analyses and reviews that have been performed.

Documents 4 and 5 - External Reviews of JBA Probabilistic Analyses Dr. Mensing in Document 4 points out the sensitivity of the JBA models.to the strain rate value r used; increasing r from 1.34 x 10-4 f t/yr to 5.4 x 10-4 ft/yr increases PBS/0tt (probability from an offset between existing shears B-1/

B-3 and B-2, given an offset on unknown undiscovered shears in the region) from.0104 to.0419. Mensing argues that the high r value is a more accurate estimate since it is based on the best estimate of total accumulated offset.

If in addition to increasing r to 5.4 x 10-4 ft/yr,the confidence level is increased from 90% to 99%, then PBS/Ofl is increased to.0837 which in turn results in a probability of.975 x 10-6/yr, i.e., approximately 1 x 10-6/yr for an offset occurring under the reactor due to an offset on B-1/B-3. Mensing notes that if the probability of an offset occurring in 40 years is estimated instead of in 1 year and if r=5.4 x 10-4 ft/yr and 99% confidence is used, then the probability of an offset occurring under the building is approximately 3.9 x 10-5 per 40 years. Dr. Mensing finally points out the sparse-ness of the data and the untested nature (softness) of the JBA models, o

Section B Mr. Davis, in his review of the JBA models in Document 5, points out the seemingly nonconservativeness in the value assigned to PBS/W in the JBA report (Document 1). The problem Mr. Davis has, we believe, comes from tha lack of precise definitions (and units) given in the JBA report for Pg (probability of an offset on unknown undiscovered shears in the region) is inter-g3fg. The JBA values for P g and PBS/ W are consistent if P g and P preted as the probability that an offset can occur at some time in the future because of undiscovered shears and P

_ is interpreted as the probability BS/0N (frequency) "per year" that an offset will occur in a given year given it can occur at all. As Davis points out, even though JBA uses a non-conservative value for PBS/W (under Davis' more natural definitions for Pg anc' BS/M) the resulting value for the product P0N.

BS/ g, which really matters, is P

P still ccnservative. As Davis' concerns indicate, the terms in the JBA models were not defined as carefully as they should have been.

Mr. Davis points out the simple and accurate approximations which can be made for many of the complicated expressions in the JBA model. He also points out that if to (time of no offset occurrence) is always used to estimate t (mean time between offset occurrence), then P decreases with to, which is opposite to what seems ON reasonable and what the JBA model proports - that P should increase with t. The ON o

as a function trouble comes in Davis' always assuming t =t and then treating PON g

of t. The JBA analysis really does not make this assumption since it only o

estimates t =E oased on the present existing data; as more data would become g

available presumably different estimates of t would be used. Thus, if t is increases with t, as it should. This treated as a p'arameter independent pf t, then POP!

3 potential problem in extrapolating to future sets by always incorrectly using t=t de~ not influence any of the numerical results in the JBA report. Davis o

poir's out that the probabilities produced by JBA's Approach 2 are very insensitive e

l w

Section 0 to changes in t ; as t increases from 2000 to 40,000 years P P

changes o

o ON BS/0N by less than 101 As Mr. Davis' concerns point out, JBA did not describe their analyses as carefully as they should have, but the concerns of Mr. Davir regarding extrapolation of the model do not invalidate the JBA model or results.

Mr. Davis finally indicates various sensitivities in JBA's Approach 1; P

is shown to increase by roughly a factor of 4 for different choices of N ON (the number of offsets occurred), t, ts (time period for offsets) and E.

This o

factor of 4 would cause less than a factor of 4 change in the overall probability P of an offset occurring under the reactor building.

Davis does not draw any final conclusions on the reasonableness of the numerical values presented in the JBA report.

Document 6 - Alternative Probabilistic Model In Document 6, the TERA Corporation presents an alternative medel to estimate the probability of a surface rupture occurring under the GETR Building. One purpose of the model was to detemine the validity of the probabilities estimated in the JBA report by using what TERA believes is a superior model approach.

The model is a more traditional seismic model where probabilities of earthquake occurrences, locitions and rupture sizes are estimated. The TERA model bases its data on the Verona Fault data and subjective estimates.

In the TERA model, times between earthquake occurrences are treated as having an exponential probability distribution (giving the Poison distribution for number of occurrences).

For the exponential distribution there is no memory of past occurrence times, i.e., the probability of an earthquake occurrence in the next year does not depend on the past time period of no occurrence. The constant earthquake occurrence rates in the TERA model are localized to given regions. As contrasted to the TERA model, in the JBA models, times between offset occurrences on the shears are assumed to have a nomal or Weibull probability distribution-which have a r..emory.

Section B In the TEPA model, given an earthquake has occurred, the probability that it is of a given magnitude is modeled by the binomial probability distribution (the multinomial distribution when the entire spectrum of magnitudes is con-sidered). A finite number of magnitudes is used in the TERA model and the probability versus maritude function is exponential in shape, to be consistent with the Gutenberg-Richter relationship. Given an earthquake of a given magnitude, the fault rupture parameters in the TERA model (fault length, fault displacement, etc.) are related to the magnitude by linear log regressions (the log of the parameter is a linear function of the magnitude).

In the JBA models, as contrasted to the TERA model, a constant strain rate is assumed for shears B-1/B-3 and B-2 which is used to estimate the number of past offset occurrences on the shears from the total recorded displacement (by forming a simple ratio).

Fault rupture parameter relationships are not explicitly considered in the JBA models - the probability of any size rupture occurring is estimated from the number of past offset occurrences on the shears and from the observation of ruptures occurring between the shears in a given recorded time period.

Using the assumptions of zero offsets in 40,000 years, the TERA probability (frequency), estimates of an offset occurring under the GETR Building range from 1.4 x 10-6 per year to 7.6 x 10-5 per year. With the assumption of zero offsets

-7 in 128,000 years, the offset probability estimates range from 4.5 x 10 per year to 3.5 x 10-5 The upper probability estimates-of 7.6 x 10-5 per year per year.

-5 and 3.5 x 10 per year are obtained under the assumption that the relative error (standard deviation over mean) on the mean rate of earthquake occurrence is 10%. As the relative error on the mean occurrence rate increases, the offset

-6 probabilities decrease; the lower probabilities of 1.4 x 10 per year and 4.5 x 10-7 per year are obtained under the assumption that the relative error is 100%.

Section B As a rough comparison with TERA's assumptions of 10% and 100% as bounding values for the relative errors, the available data from Sieh's trench excavation data (Document 1) indicate a best estimate of the relative error as calculated by GE in Document 3 to be 50% with roughly 95% confidence that the relative error is greater than 23"..

We should be careful in makina these comparisons between TERA's relative errors and those from Sieh's trench since TERA utilizes a Bayesian model dnd Sieh's values correspond to classical statistics estimates.

For a relative error of 50%, TERA's probability estimates for an offset occurrence are 5.6 x 10-6 per year (zero

-6 offsets in 40,000 years) and 1.8 x 10 per year (zero offsets in 128,000 years).

For a relative crror of 25% (the closest value to 23% that TERA gives), the probability estimates are 2.0 x 10-5 per year (zero offsets in 40,000 years) and 6.9 x 10-6 per year (zero offsets in 128,000 years). The TERA report concludes, based on its

-6 evaluations, tnat JBA's best estimates of 1.2 x 10 per year and 1.0 x 10 per year for the rupture probability (for Approaches 1 and 2) and its

-6 conservative estimate of 7.2 x10-6 per year (for Approach 2) are reasonable.

Sensitivity Evaluations in Documents 1, 2 and 3 Documents 2 and 3 provide additional sensitivity studies on the variation in rupture probabilities resulting from changes in the parameter values used in the JBA models. The table below summarizes the different parameter values that were quoted as being used in Documents 1, 2 or 3 and which produced offset occurrence probabilities lying in the range of 1.0 x 10 per year to 1.7 x 10-5

-6 per year

-7

-7 (the low offset probabilities of 5 x 10 per year in Approach 1 and 4 x 10 per year in Approach 2 oiven in the JBA report for a low confidence of 10% are ignored here).

Se'ction B

,7_

Parameter Values Producing Offset Probabilities From 1.0 x 10-6 per year to 1.7 x 10-5 per year Tx (B-1/B-3)(ft) 52 210 420 Tx (B-2)(ft) 88 140 280 r (B-1/B-3) (f t/yr).99 x 10-4, 2.18 x 10,=

r (B-2) (ft/yr',

.66 x 10-4,1.45 x 10-#,=

N (B-1/B-3) 8 16 32 128 160 1280 7000 N (B-2) 5 27 54 85 270 850 7000 2.4 x 10f,, 2.7 x 10, 3.9 x 10, 5.3 x 105, 6.2 x 10,1.4 x 10,

5 5

5 6

t (B-1/B-3) (yrs; 3

6 6

1.6 x 10 1.9 x 10, 2.0 x 10, 8.4 x 10', 3.1 x 105 3

5 6

6 6

t3 (C-2) (yrs) 1.8 x 10, 6.1 x 10, 6.2 x 107,1.3 x 10,1.4 x 10,1.6 x 10

• 6 1.9 x 10, 2.0 x 10. 5.6 x 10, 3.1 x 10, 9.9 x 105 5

c/t 0.2, 0.5, 0.7, 1.0.

t (yrs) 4000 8000 15000 62,000 80,000 g

t* (yr) 40,00C 6?,000 128,000 160,000 195.000 Definition of Terms T - Total cumulated offset on existing shears x

r - Strain rate on existing shears N - Number of offsets on existing shears t - Time period for offsets on existing snears s

o/E - Coefficient of variation of the distribution for time between offsets t - Time since last offset g

T* - Age of soil beneath reactor building In addition, one example was given for Approach I where the strain rate (r) was increased to 300 x 10-4 ft/yr for shears (B-1/B-3) and 200 x 10-4 ft/yr for shear (B-2) producing an offset probability of 2.8 x 10-5 per year. Because of relationships among the parameters, not all parameters in the above table are independent of one another. This table shows that a broad range of input

)

l assumptions produces offset probabilities which only vary by a factor of 10.

This lessens the importance, from a probability analysis standpoint, of exact inout

- parameter values.

Section B In addition to the above sensitivity analyses, Document 3 produced examples of

-4 parameter values giving offset probabilities of 1 x 10 per year.

For Approach 1, 1 x 10-4 per year probabilities correspond to:

1) c/t=0.000218 well below the lower 90% estimated confidence bound on c/t of 0.23, 2) t*=8,000 years--an extremely short age, and 3) to an unrealistically high confidence of 1-10-90 (.999...ninety six times). For Approach 2,1 x 10-4 per year probabilities correspond to:

1) t =

g 80,000 years, an extren,ely long time period; 2) t*=21,000 years and N (on B-1/B-3)

=8 and N (B-2)=5, values all at unfavorable low ends of their ranges; 3) ts (for B-1/B-3)=84,000 years and ts (f r B-2)=56,000 years, very low values for these time periods; and 4) for an unrealistic confidence of 1-10-H3,

Conclusions The probabilistic analysas presented in the JBA reports (Documents 1, 2 and 3) are methodologically sound. The TERA rt Jel (Document 5) presents an alternative probabilistic model which is not as empirical and is more traditional; the TERA model does require more data and more assumptions to be made on rupture para-meter relationships. As pointed out in the reviews, available data 'are sparse requiring sensitivity studies to be performed to gain any confidence in the rupture offset probabilities which are estimated. A wide range of sensitivity studies on variation of parameters werepformed A the JBA prchabilistic models, which included a variety of sensitivity evaluations performed in the reviews of the models.

The TERA model extends the parametric sensitivity analyses by developing a different alternative probabilistic model to compare with the JBA models.

Based on the sensitivity analyses and the alternative model, the probability of a surface rupture offset occurring beneath the reactor building has been shown to lie between 1 x 10-6 per year and 1 x 10-5 per year (to order of magnitude przcision).

The highest surface rupture offset probabilities calculated were 1 x 10-4 per year and corresponded to assuming values for parameters at the conservative cnd of their ranges. An undiscovered shear under the reacter building

Section B.

could also give a surface rupture probability of 1 x 10~4 per year if t* = 8,000 years is used as the period in which no offsets have been observed; values of t*

greater than 40,000 years would again give probabilities less than 1 x 10-5 per year.

The analyses and reviews which have been performed therefore indicate that the probability of offset occurring under the reactor building is expected to lie in

-6

-4 the range of 1 x 10 per year to 1 x 10-5 per year with 1 x 10 per year being a conservative upper bound. The probability results for the GETR are credible and should be used to supplement the deterministic evaluations in making a final

decision, i

r

.i s

,i e

o Etc,

o

1. j

'o, UNITED STATES

[$,",,,

'h NUCLEAR REGULATORY COMMISSION g \\%i C

WASHINGTON D. C. 20555 8

j

'eee*

GETR SECTION C ENGINEERING SEISMIC DESIGN PARAMETERS

_1 SECTION C ENGINEERING SEISMIC DESIGN PARAMFTFRt As set forth in Section A the complex geol

'c and seismological setting of the site necessitates the consideratior. of the effects of the following in evaluating the GETR structural design:

1.

the maximum vibratory ground motion; 2.

the maximum credible surface displacement on the Verona fault; a set of parameters for the appropriate combination of surface displacement 3.

and vibratory motion; and 4.

potential landslide.

Licensee Proposal The licensee has proposed the following sets of parameters for the reevaluation of the GETR facility:

1.

A maximum effective ground acceleration of 0.69 anchored to Re'gulatory Guide (R.G.)

1.60 spectra, for vibratory motion induced at the site by an event on the Calaveras fault.

2.

An effective ground acceleration of 0.4g anchored to R.G.1.60 spectra, for vibratory motion induced at the site by an event on the Verona fault.

3.

A surface displacement of one meter on identified shears and no offset at or beneath the GETR facility.

In addition, the licensee ha: reevaluated the GETR using the following sets of parameters although he considers them to be very conservative:

1.

A maximum effective ground acceleration of 0.8g anchored to R.G. 1.60 spectra.

2.

An effective ground acceleration of 0.39 anchored to R.C

.s spectra, com-bined with a surface rupture offset of one meter beneath che GETR.

Staff Position The geological and seismological design bases for the engineering design of the GETR are based on the evaluatt,s provided by the Geosciences Branch, Section A of this safety evaluation, and the recommendations of the NRC consultants, Dr. N. M.

Section C Newmark and Dr. W. J. Hall, Appendix A.

Based on these evaluations it is the staff's position that the following seismic design parameters, which differ somewhat 'from those proposed by GE, are appropriate for the GETR:

1.

The R.G.1.60 spectra anchored to 0.75g as the maximum effective vibratory ground motion at the site. This is set by motion on the Calaveras fault.

2.

A surface displacement of one meter of reverse-oblique net slip along a fault plane which could vary in dip from 10 to 45 degrees and which could occur on a Verona fault zone strand (splay) beneath the GETR during a single earthquake event.

3.

An effective vibratory grourid mdtion of 0.69, anchoring the M.G.1,60 spectra, together with a fault displacement of one meter as described in 2. above.

Structural Analyses As part of its structural reevaluation the licensee has provided a number of reports regarding the effects of combined vibratory ground motion a,d surface offsets. These reports are listed below.

Review of Seismic Design Criteria for the GETR Site (EDAC Report 117-254.03)

(Submitted May 8, 1980)

Probability Analysis for Combined Surface Rupture Offset and Vibratory Ground Motion (JBA Report 11-014-01)(Submitted April 23, 1980)

Additional Investigations to Determine the Effects of Combined Vibratory Motions and Surface Rupture Offset Due to an Earthquake on the Postulated Verona Fault (EDACReport 117-253.01)(Submitted May 8,1980)

Additional Investigations to Detennine Effects of Vibratory Motions Due to an Earthquake on the Calaveras Fault (EDAC Report ll7-253.02)(Submitted May 8,1980)

Additional factors to be considered in the evaluation of the adequacy of the GETR Reactor Building to resist postulated seismic events (EDAC Report 117-254.02)

(Submitted May 8, 1980)

Upon completion of the staff's and its consultant's review of these reports, the staff will issue its evaluatior, of the acceptability of the GETR seismic design.

Section C )

The seismic input parameters presented ia this evaluation will provide the bases for the structural evaluations of the GETR facility.

As noted in Section A, Part III.2 of this safety evaluation, the licensee still must perform a landslide investigation program. The impact of this program's findings on the structural analysis will be determined once the program has been completed and reviewed by the NRC staff.

Dated:

May 23, 1980

)

a 1

l i

9

APPE" DIX A NAT.HAN M.

N EW M AR K CONSULTING ENGINEER!NG SERVICES 1211 CIVIL ENGINEIRING 3UILO:NG UMSANA lL:.;NCIS 01:01 14 April 1980 i

Mr. Chris Neisen Operating Reactors, Branch No. 4 Divisien of Operating Reactcrs Nuclear Regulatory Coranission Washington, D. C. 20555 1

Mr. W. Burkhardt Fuel Reprecessing and Recycle Branch Divisien of Fuel Cycling and Mat'erial Safety Nuclear Regulatory Cc=ission t

Willste Building 7915 Eastern Avenue Silver Spring, Maryland 20910 Re:

Seismic Evaluation of Vallecites 3ite Contract NRC-03-72-150 Gentlemen:

We are enclosing two copies for each of you of the report on the Seismic Evaluation of Vallecitos Site by N. M. Newmark and W. J. Hall dated 14 April 1950.

Respectfully submitted, U.S.H4 W. J. Hall f. %. Wk enclosuras N. M. Newmark Distribution:

i l

2 - Mr. Chris Nelson, NRC l

2 - Mr. W. Burkhardt, NRC 3 - Dr. N. M. Newmark 3 - Dr. W.'J. Hall b

luS o L'~

o,O O R303(

a l

l

NATHAN M.

NEWMARK CONSULT!NG ENGINEERING SERvfCES 1211 C3VIL ENGINEERING BUILDING URB ANA. ILL]NQl3 61801 14 April 1980 SE!5MIC EVALUATION OF VALLECIT05 SITE by N. M. Newmark & W. J. Hall Natnan M. Newmark Consultinc Encineerino Services 121i civil Engineering Suiicing Urbana, Illinuis 61301 I.

CCNDITICNS CCNSIDERED AND BASIS OF E'/ALUATION The purpose of tnis rescrt is to define a raticnal basis for seismic evaluation of the General Electric test reactor and other facilities located near Vallecitos, California. Tne :ajo# facilities censidered are within three miles of the Calaveras Fault and very close to, or possibly just over, a fault identified as the Verona Fault.

After discussion with a number of persons and a review of reports, dccuments, and letters from NRC, the U.S. Geological Survey, and the TERA Corporation, studies for Diaolo Canycn, and recogni:ing the lack of correlation of damage to structures and ecuipment in relation to peak accelera-tien (including tne 6 August 1979 Coyote Lake earthquake and the 15 October 1979 !:cerial Valley earthquake), in the lignt of our judgment and experience we reccmmend the use of the criteria described nerein for the seismic evalua-tion of the site and for the review of structures and ecuicment in structures at the site.

It is noted that these recommendations are the writers' sole responsibility anc do not represent the official views of NRC or tne USGS.

It is considered that4 occur en :ne Calaveras Fault an DUPLICATE DOCUMENT I Entire docume entered into system under:

MY 036[

ANO No. of pages:

b

2 site. A considerably smaller magnitude of earthouake, in the range of magni-tude from 5 to certainly no more than magnitude 6 might occur on the Verona Fault.

It appears that the site conditions involve a fairly thick layer of sediment or sedimentary material, which in general for the same magnitude of earthquake would be expected to have a smaller peak ground acceleration than would occur in competent crystalline rock. However, the peak ground velocities would not be greatly different in the sediment from that expected in rock. The transient displacements, hcwever, would generally be expected to be larger in the sediment.

In general, the most sericus conditions arising from earthquake induced motions at the site come from earthquakes with the source close to the site rather than frem more distant earthquakes along the fault system.

This has been taken into 3ccount in the reccamendations made herein.

Although the writers in general would prefer to use probabilistic approaches to seismicity and also to seismic design, censidering probabilis-tically the response and the strength of structures and equipment subjected to dynamic motions, the recer=endations given herein are based on essentially deterministic criteria using NUREG-CR 0098 spectra with an SSE acceleration defined as anchoring the spectrum in accordance with the procedures in that report. To draw the NUREG spectrum, it is cur recernendation that a peak velocity of 43 in./sec and a peak displacement of 36 in. be used when scalec to 1 g.

For 0.75 g, then, the peak velocity is 36 in./sec and the peak displacement is 27 in. The acceleration used to anchor the near source SSE.