ML19330C096
| ML19330C096 | |
| Person / Time | |
|---|---|
| Site: | Big Rock Point File:Consumers Energy icon.png |
| Issue date: | 06/23/1980 |
| From: | Rolonda Jackson Office of Nuclear Reactor Regulation |
| To: | Crutchfield D Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML19330C094 | List: |
| References | |
| TASK-02-04.A, TASK-02-04.B, TASK-02-04.C, TASK-03-06, TASK-2-4.A, TASK-2-4.B, TASK-2-4.C, TASK-3-6, TASK-RR NUDOCS 8008070546 | |
| Download: ML19330C096 (37) | |
Text
_-
k
.,?.,
UNITED STATES ys et g
NUCLEAR REGULATORY COMMISSION
[
s wasmwaTom. o. c. :csse g.g j
%[-;R/
AN 2 3 lcA MEMCRANDUM FOR:
D. Crutchfield, Acting Chjef Systematic Evaluation Program 3 ranch
/
THRU:
James P. Knight, Assistant Director for Ccmponents and Structures Engineering, CE FRCM:
Robert E. Jackscn, Chief Geosciences 3 ranch, CE
SUBJECT:
INITIAL REVIEW AND RECCl44ENDATICt.S FCR SITE SPECIFIC SPECTRA AT SEP SITES We have been working for tne past two years with the SE? 3 ranch and their c:nsultants in order to provide preliminary rec:mendations regarcing site specific spectra to be used in the SEP for evaluation of the seismic design adequacy of the selected plants.
The 3 ranch recommendaticns are attached, however, it shculd be noted that they are subject to the limitaticns described in tha sections entitled
" Purpose and Sc: pef' and"Recc::::endations." These recomendations were prepared by Dr. Lecn Reiter based primarily en documents submitted in the Site Specific Spectra Program. We expect that our evaluation of items still forthc: ming in the Site Specific Spectra Program may result in the follcwing:
1.
It is likely that t!:ere will be further changes in the return periods associated with the reccmmended spectra for the various sites. These return periods will still be able to be described as dof the order of 1000 or 10,000 years", which is the cresent descriptien of the spectra and the level implicitly accepted by NRC in recent licensing decisions.
2.
There will be no major change in the relative levels of seismic ha:ard between sites.
3.
There will be little or no change in the "ceterministic c:mcarisens a
for the varicus site used to waluata the acce:tacility of the spectra rec: mended in the attached review.
4 There 's a preliminary indication that a reducticn in spectra at inter-medi;;e and icw frequencies may be called for at rock sites (Dresden, Ginna, Hadaar Neck and Millstone). Probabilistic predictions of eak velocities at neu sites may also be affected.
~
go 0807 0 540
JUN 2 31980
-2 While it is difficult to predict the oute:me of an innovat10e program that is still in progress it is our best estimate, based on the above, that this subsequent evaluation will not result in very large changes in spectra recomended for use in ne evaluation of the SE?.
We recomend that you utili:e these spectra in your reanalysis of the SE?
facilities. We further rec: mend that a minimum spectra be established as discussed in tne recort. This rec:mmendation is based en the innovative nature of the Sita Specific Spectra Program and the need for continued review and maturation cf the program. Tne site specific spectra provided are generally less than wculd result frem a literal application of Appendix A to 10 CFR and the current Standard Review Plan thicugnout the frequency range of interest for nuclear power plants.
Since follow up work and sensitiYity studies are c:ntinuing, we will monitor progress and provide a final recemendation in llacember 1980 upon comoletion and review of these ele :ents of the program.
, kC'&
hs44 Robert E. J Json, Chief GeosciencesAranch Division di Engineering
Enclosure:
As stated cc: w/ enclosure R. 'lollmer D. Eisennut G. Lainas H. Levin D. Allisen G. Lear L. Heller J. Greeves F. Schauer G. Bagchi D. Bernreuter, LLL L. Wight, TERA GSB Personnel p
B
Initial Review and Recomendations for Site Specific 5pectra at SEP Sites purpose and Scope This review presents initial recommendations for Site Specific Spectra to be used in the reevaluation of SEP plants.
It is based upon review of the following items.
(1) Draft Seismic Hazard Analysis: TERA - Lawrence Livemore Laboratory (LLL), 3 volumes, August 1979.
(2) Peer Review Coments to above reports, Individul coments by Dr. O. Nut:11, Dr. L. Sykes, Dr. D. Venezianc Dr. A. Ang, (LLL Review Board); Fugro, URS Blume Assoc., Dr. A. Corr.of i, Mr. R. Holt, Ccmonwealth Edison (licensee sponsored reviews); Dr. L. Abramson (NRC, Applied Statistics Branch) Fall-Winter 1979.
(3) Response to Peer Review Site Specific Spectra Project (SSSP), TERA, May 1980.
(4) Draft Seismic Hazard Analysis: SSSP Sensitivity Results, TERA-LLL, May 1980.
(5) Attenuation Panel Feb.1980, and coments on the panel meeting by Dr. O. Muttli, Dr. M. Trifunac, Dr. R. McGuire, Dr. N. Donovan.
(6) Letter Report evaluation of Attenuation Panel by TERA, April 4,1980.
(7) Letter Reports on Ossippee Attenuation Model by TERA, May 22, May 29, 1980 (S) Interim Sumary of assessment of conservatisms by TERA, May 30, 1980.
(9) Evaluation of Ossippee Attenuation Models and alternatives by LLL, May 23, 1980.
(10) Seismic Hazard Evaluation for SEP plants (Draf-) N. M. Newmark (May 30,1980).
~
In addition to these documents there have been many discussions and telephone conversations with individuals at TERA, LLL, reviewers, attenuation panel members and Ors. Newmark and Hall.
Following is a list of other items and reviews which will be forthcoming and could have an impact upon the results.
1.
Review of t!.2 Draft Seismic Hazard Analysis by the USGS.
2.
Aoditional Review and coments by Ors. Newmark and Hall.
3.
Review of all submissions by the licensees on their recomendations for site specific spectra (several have been reviewed).
4.
Comparison of SSSP results with other eastern U. S. hazard analyses.
5.
Feedback meeting with original expert group.
6.
Recomendation frcm TERA-LLL and possible reanalysis based upon utilization of in'put from sensitivity results, attenuation panel and feedback meeting.
Recomendations it is recomended that the following spectra presented in the Sensitivity Results (May 1980) be used as site specific free field spectra.
Eastern U. S. (Yankee Rowe, Connecticut Yankee, Millstone, Ginna, Oyster Creek)
"1000 year" spectra assuming no background and Ossippee Attenuation.
Central U. 5. (Dresden, Palisades, Lacrosse, Big Rock Point)
"1000 yr a spectra assuming no background anc Gupta-Muttli Att.enuation, j
?
e.
These spectra account for gross site conditions (soil or rock) and do not take into' account any specific conditions which may result in amplification i
(Lacrosse, Yankee Rowe, Palisades).
It is also recomended that a minimum be established for which no spectra be allowed to go below. It is suggested that this minimum be the median (50th percentile) representation of real spectra for a magnitude 5.3 earthquake.
This minimum exceeds the "1000" yr spectra for Big Rock Point, Lacrosse and Palisades at frequencies greater than 2 to 3 Hz.
The rationale for these recomendations are discussed below.
General Coments The SSSP was conceihed as a multi-method approach for detemining site specific spectra (Bernreuter,1979). It encompassed probabilistic approaches at predicting peak acceleration, peak velocities and unifom hazard spectra for different....-.
return periods and a empirical approach which includes calculation of 50th and 84th percentile spectra from ensembles of real data at different magni-tudes, site conditions and distance ranges. The probabilistic approach utilized is basically that suggested by Cornell (1968) which has been modified to formally incorporate " expert" judgements. This approach is explained in detail in the documents referenced abche and in Part 1 of the Executive Summary by TERA Corp.
The difference between so called " deter ninistic" approaches (for example, thatfoundintheStandardRehiewPlan*)andprobabilisticapproachesaredescribed i
below. In the deterministic approach (Figure 1) local (fault) and regional
- Although this approach is c:xnmonly called " deterministic it is better described as "judgemental-empirical." A true deterministic approach would involve using the principles of physics to calculate ground motion due to a rupturing fault.
4 (tectonicprohince)sourceregionsarespecifiedgeometrically(Step 1).
The largest earthquake associated with each source is then defined frem historical seismicity and/or geological estimates, and it is assumed to occur at a location in each source closest to the site in consideration (Step 2). The resultant ground motion (usually peak acceleration) at the site from each of these sources is then estimated utili:ing magnitude-acceleration or intensity-acceleration relationships (Step 3). The largest of these is then considered the controlling ground motion and it determines the assumed earthquake loacing at the site (Step 4). In the current flRC practice this earthquake loading (Safe Shutdown Earthquake) usually is peak acceleration used to anchor the standardized Regulatory Guide 1.60 spectrum. This method does not take into account the frequency of earthquake occurrence and allows no description of uncertainty.
In the probabilistic approach described in Figure 2, earthquake sources are detemined (Step 1) as in the deterministic approach. Historical seismicity is then used to determine an earthquake recurrence model for each source (Step 2). This model is usually determined from a linear regression analysis relating earthquake size (magnitude or intensity) to frequency of occurrence. These recurrence models are terminated at the largest earthquake expected from each source. Most probabilistic models assume that earthquake occurrence follows a Poisson process or that these earthquakes occur randemly with respect to time and space within a gihen source. The ground motion (peak cr spectral parameter) at the site frcm the different earthquakes at different distances is estimated using a set of magnitude (or intensity) - ground motion relationships that explicitly incorporate the dispersion of the data about such relationships'(Step 3). Finally.. integrating the effect of different size earthquakes from different locations in different sources with the
-5 recurrenceinformationfromStep2,theprobabilitiesthatgi$enlevels ofgroundmotionwillnotbeexceededwithingihentimeperiodsare calculated (Step 4).
The deterministic approach is strongly controlled by the choice of input parameters (source configuration, intensity-acceleration relationship, response spectraetc.). Sizeable changes in characterizationof safe shutdown earth-quakes for Nuclear Power Plants in the past 5 to 10 years hahe resulted from staff adoption of the Regulatory Guide 1.60 spectrum and the Trifunac-Brady (1975) intensity-acceleration relationship. Probabilistic prediction can also bedrihenbythechoicaofinputparameters. In the eastern U. S. these input parameters or their statistical representation cannot in many cases be unambiguouslyderihedfromtheexistingdata. Theinnohativeapproachof the SSSP was to canhas expert opinion as to what' the choice of these input parameters were, what range they might be expected to assume and what credibility could be attached to them. Each experts input was treated separately,.
spectra were computed for each expert at each site than a trial synthesis was performed ccmbining all the experts at each site based upon their own self-ranking. Theinputparameterscoheredfourareas: (1) the configuration of seismic source zones in the central and eastern U. S. (2) the largest earthquake expected in each of these zones (3) the earthquake actihity rate and recurrence statistics associated with each zone and (4) methcds for predicting ground motionintheeasternandcentralU.S.fromanearthquakeofagihensizeata gihendistance.
Responses were receihed from 10 of the 14 expert polled. (The questionnaires were lengthy and required seheral days to answer in a cceprehensihe manner).
These responses were almost exclusihely directed at the first three areas.
The significant lack of resporise in areas of ground motion made it necessary Additional 1
for TERA-LLL to dehelop its own ground motion dets mination scheme.
approaches were presented in the sensitihity results and an additional special " Attenuation Panel" was convened to discuss this difficult problem.
Inadditiontothegroundmotionproblem,theextensihepeerrehiewconducted for the initial draft report identified other problem areas. The most significant of these were related to the way each expert's :enation was treated and the assumed dispersion of the data. These subjects were also treatedinthesensitihitystudiesmentionedabohe. Specific discussions on each of these problem areas follow.
Soecific Corwents Ground Motion Determination The problem is to quantitative 1y predict groi;nd motion east of the Rockies when there is practically no strong notion data recorded in this region. The existing data base (most Western U. S.) was recorded in areas where seismic wahe attenuation and, to sema extant, seismic sources are different.
Amethodmustbedehelopedtopredictthismotiontheoreticallyormakeuseof the historical (non-instrumental) felt reports frcm the eastern U. S. in The initial conjunction with strong ground-motion data from the western U. S.
results(August 1979)utilizedfeltreportsfrcmthewell-documentedSouthern Illinois Earthquake of 1968 and the assumption that ground motion associatec with a gihen felt effect (site intensity) and epicentral distance will be the same in both east and west. Thesensitihitystudies(May1980) examined the affects of assuming that the ground motion associated with a giYen felt
effect and given earthquake si:e will be the same for both east and west.
The studies accomplished this result for three felt-effect predictions; the 1968 Southern Illinois Earthquake, the 1940 Ossippea New hampshire earthquake, and a modification of the Gupta-Nuttli (1976) relation based upon seYeral central U. S. earthquakes. While the attenuation panel had mixed feetings there seemed to be scme preference for this latter assumption. In conjunc:icn with the sensitiYity studies, the existing data set was also =cdified to preYentunduedependenceuponasingleearthquakeandtoeliminate strongmotionrecordsthatwerebelieYedtorepresentonlypartofthe actual shaking. In addition, studies of seYeral other earthquake suggested a difference in attenuation of ground motion between the northeastern and central U. S.
At distances greater than 100 kilometers, the affects of shaking appear less attenuated in the central U. S. when compared with that in the northeast. As a result of these considerations, we recommend that the 1980 model based upon the Ossippee earthquake be used as a basis for determining ground motion in the northeastern U. S.; while the 1980 model based upon the Gupta-Nuttli relationship be used as a basis for determining ground motion in the central U. S.
The Ossippee attenuation was calculated seYeral ways. In the original SSSP Sensitivity Results (May 1980) an aYerage distance was first computedforeachintensityleYelandthenaregressionwasperformedtreating distance as the independent parameter and site intensity as the dependent A significant difference was obserYed when the aYeraging was emitted parameter.
and the regression performed directly on the data (TERA Letter Reports, May ~42 and May 29,1980). It is not imediately clear which approach is =cre appropriate. Conceptually it appears better to avoid the aYeraging step.
We
feel,howeher,th:tatthistimetheoriginal_techniqueusingthe aheragingstepshouldbeused. The reasons for this are (LLL Letter, May'23,1980): (1) bis method is analagous to that used by Gupta andNuttli(1976)toderi0etheirattenuationrelationship.(2)the second method would predict ground motion significantly less at most distances than that proposed by the theoretical model of Nuttii (1979) while the original method falls much closer to his model.
The attenuation panel reccmmended greater use of such theoretical relationships for determining 'round motion. Initial calculations show that when these theoretical relationships are incorporated into SSSP methodology peak accelerations for return periods of 1000 years appear to be simi'iar to the Gupta-Nuttli and original Ossippee attenuations.
While some small differences between central and northeastern attenuation can be expected we feel that at this time, reliance upon results produced utilizing a particular regression technique on one earthquake in the northeast which are significantly less than theoretical and empirical results for the central U. S. is imprudent. Clearly,howeher,determinationofa proper attenuation relationship is an area that requires additional work.
Zoning The initial treatment of experts input to configuration and credibility of seismic source zones allowed for the existence of a background zene consisting of 4
l
d
.g.
theunion(enehelope)ofalltheexpertszonesinaparticularregion.
The extent to which this background zone was used depended upon the experts general lehel of belief (credibility) in the existence of these zones. As a result, this leads to tying one expert's results to others and the allowance of specific numbers of the larger earthquakes normally associated with a seismic zone being allowed to occur anywhere within the background. Various rehiewers criticized this approach and some alternatihes were suggested.
Thesensitihitystudiescomputedspectrabasedupontheoppositeextremei.e.
the assumption that each expert had 100% belief in his zone and no background need exist. These two computations bound the problem.
For SEP sites, the latter assumption results in a reduction in estimated seismic hazard. Ifasitewerelocatedinthemiddleofanactiheseismic zone such as New Madrid the assumption of no background wouid result in an increase in estimated seismic hazard. There are many arguments that may be made as to how this problem may be treated correctly. It seems clear that neither extreme is correct and some better way of accounting for credibility is warranted. TERA-LLL has argued that a true representation of credibility in such a ecmplex problem may be very cumberNme computationally and prohibitihely expensive. It is our recernendation that, barring such a computaticn spectra intermediate between these two assumptions be used at this time. As shown below the actual difference between spectra ecmputed using the two extreme assumptions is not large and any error in estimating the intermediate spectra will not hahe a significant effect.
- Discersion of Data In The August 1979 report the dispersion assumed about the final ground motion prediction was assumed to be log normal with #=0.9 (base e). In addition the distribution was truncated at i 2e. This size of the dispersion was deter-mined combining disparsions normally encountered in determining site intensity frcm earthquakesize(epicentralintensity)andinconYertingthisintensitytogrcund motion. These indihidual dispersions can be considered as due to randomness found in nature. Severalrehiewersarguedhoweverthattreatingtheseerrorsas independentanddisregardingtheircrosscorrelationischerlyconserhativeand that it increases the total dispersion beyond that resulting frem true randomness.
Where ground motion records due exist, e.g. Western U. S., the dispersion associated with ground motion frem a given size of earthquake can usually be described with e=0.6 to 0.7.
Data points do not normally extend out beycnd limits of 3 3v. These criticisms are considered valid and its recommended that the dispersion defined as 0 0.7, truncated at 1 37 be accepted.
Extension of the truncation point beyond 37 will not have a significant effect upon the results.
i SynthesisCurhes Some alternate methods were suggested to synthesize the results of the harious
)
~
expert judgements. The SSSP utilizes a self-ranking system. In the opinion of TERA Corporation, alternate methods would not have a significant effect upon the synthesized c::rhes. By inspection it appears that the synthesis curhes represent a median or scmewhat higher than median representation of the individual spectra ccaputed for each expert. It is recenri: ended that this synthesis be used to describe the hazard.
l
-U-Integration of Recomendations In the sensitivity studies, uniform hazard spectra are presented for all the ground motion models recomended above,1.e. Ossippee (1980 model) for north-eastern sites and Gupta-Nuttli (1980) for central U. S. sites.
All spectra are computed assuming no backgrounc andc=0.912c truncation.
These spectra are approximately equal to the recomended spectra of 750.7 1 30 truncation with a zoning assumption intermediate between a back-ground and no background because: 1) The decrease in peak accelerations and peak velocities computed fer representative individual experts from cs0.9 (; 2c-) to <rs 0.7 (13e-) is on the average about 7 to 10~. for the Gupta-Huttli and Ossippee attenuations; (2) the ir. crease in peak accelerations and peak velocities from no background to background is on the average about 15 to 20'.' for the August 1979 attenuation (the only comparison available).
Although there is some preliminary indication of attenuation mocel dependence for the background-no background ccmparison these approx 1mations are considered adequate given the precision of the spectra and the size of the differences.
Adecuacy and Conservatism of the Recomended Soectra While the "1000 year" spectra are recommended it is not possible to state with any certainty that the true return period (inverse of annual risk of exceedence) is 1000 years. Generally these estimates are believed to be conservative for the following reasons.
12 -
- 1. Strong motion data sets are in many ways biased toward high values.
Non-triggered instruments or low-lehel records receihe little attention.
This is also true at great distances and for longer periods where noise may be contributing significantly to obserhed motion.
- 2. The assumption that earthquakes occur randomly within a given seismic source zone is conservative for large zones of low to moderate level seismicity such as those around most SEP sites. While the sources of central and eastern U. S. earthquakes remain hidden,most seismologists conclude that damaging earthquakes will eventually be associated with specific faults.
- 3. The uniform spectra represent ecmposite risk from different source zones which may effect different frequency ranges. Under certain situations, exceeding the spectra at different frequencies implies the simultaneous occu.rence of earthquakes in more than one source zone,
- 4. The assumption that intensities from large earthquakes attenuate at the same rate as intensities frcm small earthquakes is conservative.
Some non-conservative aspects of this and other studies are:
- 1. The strong-motion data set used mixes accelerograms recorded in the true free field with those recorded in the basements of buildings. Many engineers feel that the effect of large foundations in these buildings is to reduce high frequency motion.
- 2. The probabilistic spectra represent the chance of being exceeded more than once in a given return period. The probability of being exceeded twice or more, however, is small when compared to the proba' ility of being exceeded u
1 only once.
l 1
1 1
s Based upon consideration of 411 of the aboUe and their estimated relati$e weights,we consider the true return period associated with these spectra to be longer than 1000 years. TERA in a recent reassessment of conservatism (Letter,May 30,1980) conclu' des that those spectra presented in the Sensitivity Results as "1000 year spectra" can be conserhatihely represented as 5000 to 10,000 year loads. Additional work will better define what the return periods are.
At the present time howeher, we believe that there is no way of indicating what these true return periods are or establishing rigourously defined conbidence i
limits. In the past there has been implicit acceptance et :lesign spectra that wereassumedtohaYereturnperiodsoftheorderof1000or10,000 years. It is our_ judgement that these spectra fall within this description.
The most important quality of these spectra is that, although no great confidence can be attached to the absolute probabilities (i.e. return periods),the systematic incorporation of. expert opinion and uncertainty and the wide ranging sensitihity
^
tests indicate greater stability when estimating relative hazard probabilities attheselehelsofgroundmotion.Thiswculdapplytoestimatingtheequivalent levelsofprobabilitiesofexceedenceatdifferentsitesandsmallrelatiYe differences in probabilities of exceedence at the same site. Thus,while we are not sure that the "1000 year spectra" really represent 1000, 5000 or 10,000 year return periods at all the sites we have greater confidence that they represent approx-l imately equivalent levels-of hazard whatever the true return period is. This is based in large part upon the relative consistency of effects associated with the sensi-tivity tests (SSSP SensitiYity Results, May 1980) and the synthsizing of wide
' ranges of expert judgement with respect to each region.
i e
w
- 14 Comoarison of Soectra with " Deterministic" procedures In order to further evaluate the adequacy and reasonableness of the racommended design spectra several comparisons with non-probabilistic techniques were performed.
Ccrearison with scectra determined usino the tectonic crovince accroach (accendix A1.
In this approach (Figure 1) the largest historical earthquake that has occurred in the host province is assumed to occur near the plant while the largest historical earthquakes in adjacent provinces are assumed to occur in these provinces at locations closest to the site.
The ground motion at the site from these earthquakes is estimated and this determines the seismic input to design. Tectonic province boundaries and earthquake sizes were estimated based upon recent licensing decisions.
The configuration of the New M.adrid Zone was also used assuming the more recent suggestions of Nuttli and Herrmann (1978). The assumptions for each site are listed in Table 1.
Earthquake size is also given in terms of magnitude (m ) and these are based upon recent individual determinations b
of the magnitudes from intensity data and the general relationship proposed by Nuttii and Herrmann (1978).
Utilizing these events, a series of theoretical and empirical equations were used to predict the peak accelerations and velocities at each site.
In order to deal with differences in these equations, selected results representing the most appropriate theoretical and empirical relationships were averaged to arrive at final estimates of peak acceleration and velocity. Table 2 shows the con-trolling (largest) peaks estimated at each site. These are compared with the peak accelerations and velocities associated with the recommended uniform ha:ard (probabilistic) spectra.
15 -
The unifonn hazard peak accelerations reach or exceed the deterministic peak accelerations at all sites except Palisades, Lacrosse and Big Rock Point.
This is a reflection of the fact that these 3 sites lie in areas of low seismicity and estimated seismic hazard in the central stable region. The unifann hazard peak velocities exceed the detenninistic peak velocities except at Dresden where it is less. This is a reflection of the fact that probabilistic techniques take into account larger than historical earthquakes. Sensitivity studies show that these have the largest effect upon peak velocities. This is reflected in the deterministic procedure for Dresden where the proximity of the New Madrid zone has a significant impact.
In general it can be said that the 1000 year uniform hazard peaks bracket the deterministic peaks. Differencesbetweenthetwosetsofhaluesresult from the ability of the uniform hazard approach to overcome the artificial constraints often posed by the "tec'.snic prohinced approach. Thus, while the tectonicprohinceapproachwouldrequireBigRockPointandHaddamNeckto utilize similar seismic input for design purposes, the probabilistic methodology takes into account the real difference in seismicity and perceihed earthquake hazard at these sites.
The deterninistic peak accelerations and velocities are converted to response spectra using the amplification factors suggested by Newmark and Hall in NUREG CR-0098. Figs. 3 thru 11 compare the recommended unifonn hazard spectra with
~
50th and 84th percentile deterministic spectra.
In the central U.S. the recom-mended spectra generally fall below or at the 50th percentile. In the eastern United States the uniform hazard spectra are approximately l
equiialent to the 84th percentile deterministic spectra. While the l
deterministic peaks are generally lower than the predicted peaks, use of the 84th percentile amplification factors usually more than compensate for the differences. AgaintheuniformhazardspectramoreadequatelyreflectperceiYed relatiYehazard. The"tectonicpro0ince"approachcanbemadetoachieYe conserhatisminthiscasebyutilizingconserYatiheamplificationfactors.
Figures 12 and 13 show the uniform spectra compared to Reg. Guide 1.60 spectra anchored at 0.1 and 0.2g.
Following suggested Standard ReYiew Plan procedures for new plants that is utilizing the trend of the means of Trifunac and Brady (1975) to anchor the Reg. Guide 1.50 spectra, would result in design spectra anchored at between 0.12 and 0.20. The specific acceleration used would 9
depend in large part upon.the applicants submittal and the rehiewer's conserYatism.
For the central U. S. the recommended spectra are mostly below the Reg. Guide spectrum anchored at 0.lg while for eastern U. S. the recomended spectra are at oraboYetheReg.Guidespectrumanchoredat0.1g. The aYerage recomended spectrumwouldberoughlyequihalenttotheReg. Guide 1.50Spectrumanchored at a peak acceleration of about 0.1. Theobservationthattheaheragepeak 9
acceleration associated with the recommended spectra (Table 2) is about 0.15g
~
'illustratestheoftendiscussedconserUatismoftheReg.Guidespectrum. It was conservatiYelyderiYedfromearthquakesofdifferentsizesrecordedat different distances and different site conditions.
Comoarison with Real Soectra A more applicable comparison can be found in Figures 14 and 15. Here the recommendedspectraarecomparedtothe50thand84thPercentileleYelsof ensemblesofresponsespectraderiYedfromstrongmotionrecordsrecorded at nearby distances (usually 27 km or less) from earthquakes of magnitude
5.3 + 0.5 in the western U. S. and Italy. At these distances differences in regional attenuation are not pronounced. At periods less than 0.3-0.5 seconds the recommenced spectra fall in between the 50th and 84th percentile except for Palisades, LaCrossse and Big Rock Point which are slightly below the 50th Percentile. Differences again can be related to real differences in earth quake hazard.
There can be some concern however in that the recommended spectra may fall below some minimum level of ground motion from a nearby magnitude 5.3 (In-tensity VII). While Intensity VIII or larger earthquakes have been restricted in historical time in the central and eastern U.S. to five or six locations, Intensity VII earthqua'Kes have occurred in sufficient numbers and at sufficient local. ions such that we believe that they could occur anywhere in the U.S. at varying levels of certainty.
It is prudent therefore to establish such a minimum level although a direct uniform hazard assessment would more accurately reflect relative earthquake hazard.
It is recommended that this minimum be set at the 50th percentile of the plotted real spectra.
While the 84th percentile has been used in deterministic techniques it is not suggested that it be used as a minimum since it is more a reflection of the dispersion of data resulting from the magnitude and distance range needed to gather an adequate number of records for statistical treatment.
As indicated above use of the 50th Percentile would~ have a small effect upon Lacrosse, Palisades and Big Rock Point.
i t
Conclusions Baseduponrehiewoftheindicateddocumentsandthecomparisonwith" deterministic" proceduresmentionedabohe,webeliehethatthesite-specificuniformhazard responsespectrasuggestedrepresentanadequaleleheloffreefieldground
~
motion for use in th'e reehaluation of the SEP plants. TheEaryinglehelsofthese spectra more accurately reflect true hariations in real seismic hazaro than tnose derihedutilizingthe"determini' tic"tectonicprovinceapproach. We also s
believe that it is prudent to establisn some minimum lehel below which no spectra be allowed to fall. It is recommended that this be the 50th percentile of real data from a nearby magnitude 5.3 earthquake as shown in the comparative plots.
- - Utilization of this minimum would have a small effect upon Palisades, Lacrosse and Big Rock Point. These spectra do not take into account specific site amplification factors that may be present at Lacrosse, Palisades or Yankee Rowe nor do they reflect consideration of additional studies still ongoing in the SSSP program. Those spectra presented were computed for 5% damping.
9 y
Table 1 Controlling Earthquakes used in the Tectonic Province Approach Local Earthquake (Host Province)
Site
-(Averace Epicentral Distance 10-15 km)
Distant Earthcuakes (other than nost erovinces Yankee Rowe mb 5.3 (Intensity VII) mb6.0 (Intensity VIII) from White Mt. zone (80 km)
Haddam Neck mb 5.3 (Intensity VII) mb S.0 (Intensity VIII) from White Mt. Zone (130 km)
Millstone mb5.3 (Intensity VII) mb 6.0 (Intensity VIII) from White Mt. Zone (140 km)
Oyster Creek mb 5.3 (Intensity VII) mb 6.0 (Intensity VIII) from White Mt. Zone (375 km) mb 5.8 (Intensity VIII) from Southern Valley and Ridge (550 km:
Ginna mb5.3 (Intensity VII-VIII) mb 5.75 (Intensity VIII)(from55 km)
Clarenden-Linden Fault Drasden mb 5.3 (Intensity VII-VIII) mb 7.5 (Intensity XI-XII)from New Madrid Zone (280 km)
- mb6.7 (Intensity X) from Wabash Zone (200 km)
Palisades -
mb5.3 (Intensity VII-VIII) mb7.5 (Intensity XI-XII) from New Madrid Zone (315 km)
- mb6.7 (Intensity X) from Wabash Zone (300 km)
Lacrosse mb5.3 (Intensity VII-VIII) mb7.5.(Intensity XI-XII from New Madrid Zone (600 km)
- mb6.7 (Intensity X) from Wabash Zone (530 km)
Big Rock Pt.
mb5.3 (Intensity VII-VIII) mb7.5 (Intensity XI-XII) from New Madrid Zone (760 km)
"mb6.7 (Intensity X) from Wabash Zone (650 km)
- Controlling event based upon Nuttii and Herrmann (1978) interpretation of Mississippi Embayment Seismic Zoning.
Table 2 Comparison of Predicted Peak Accelerations and Velocities Based upon Probabilistic*
and Deterministic ** Techniques 2
Site Peak Acceleration (cm/sec )
Peak Velocity (cm/sec)
Probabilistic Deterministic Probabilistic Deterministic 1.
Yankee Rowe 195 123 22 11 2.
Hadden Neck, 202 123 20 9
3.
Millstone 184 123 18 9
4.
Oyster Creek 161 123 18 9
5.
Ginna 169 132 17 10 6.
Dresden 124 132 16 20 7.
Palisades 102 132 15 12 8.
LaCNsse 91
'132 14 9
9.
Big Rock Point 81 132 11~~~~
g s
- Probabilistic values are those associated with TERA-LLL's synthesis for the 1000 yr return period. Attenuation model used for sites 1-5 was 1980 Ossippee for sites 6-9 1980 Gupta-Nuttli. While explicit values assumed no background and a dispersion of 7=0.9 + 2 7 This is estimated to be equivalent to intermediate background and a dispersTon of C=0.7, + 3 7
- Deterministic values were comouted using Table 1 and averages of results from the following suites of predictive equations.
Local Events - all sites, suite (a)
Distant Events -_ northeastern sites (1,2,3,4), Suite (b),
central sites (6,7,8,9) Suite (c) intennediate site (5) Suite (a).
The suites of equations are:
- a. ' Herrmann (personal comunication,1980), TERA-LLL Aug,1979, TERA-LLL 1980 Ossippee, TERA-LLL 1980 Gupta-Nuttli.
b.
Herrmann ~(personal communication,1980), TERA-LLL 1980 Ossippee c.
Herrmann (personal comunication,1980), TERA-LLL Aug. 1979, TERA-LLL 1980 Guota-Nuttli,
i raut.T 1
a.;a. ww SITE FIXED DISTANCE R
,$ ~ ~
r, 8
FIXED MAGNITUDE M r
AAEA souncs STED 2 N'
SELECT SOURCES GOVERNING EARTHGUAKE h
nu ucwuoE a aces. snares
. :r-FIXED
_ _ J * * :,
PEAK
'. *:.,oara ACC8 CRATION l*.
I I
Ot3TANCE STEP 6 "3
LOADING AT
~ ATTENUATION THE SITE FIGURE I DETERMINISTIC APPROACH TO LOADING AT THE SITE TEIA CCRoCRAECN Dec. 8,1978
h FA4JLT..
Lwe Senes)
Loc or
- 51TE re.cr
~.
Eatramu
>=
F i
I a
p b
AntA sounCs 7
mAcninu m STEP 1 STEF 2 SOURCES RECURRENCE h
UNCERTANTY 1.0
. N ATTEWATION
- KAGNIT1 JOE M PEM 3
ACCELERATON
,,(*,
CDF DATA *
\\
0 0
0 ACCELERATON STED 4 STED 3 PROSABILITY OF ATTENUATION NON-EXCEEDENCE WITHIN A TIME PERIOD t
)
FIGURE 2 CURRENT APPROACH TO HAZARD MAPPING FOR PEAK VALUES
=\\
b+
TERA CORPCRATICN Dec. 8,1978
h
~. -..
)
if f
@s 100.0 X
/\\
A..
Y D
- i
.- r e,
0 o,.
so s a
/
g 10.0, f,
~
/
m w
'o
/,
n
=
E j,
4" e-3 "oE f
/
IC v
f
/
~
.<o
~
s
~
j
.p 0.!
0.01 0.1 1.0 10.0 PERICC-SEC.
84% - Deter ninistic soectra using 84% amolification factor from NUREG CR - 0098.
5C% - Deterministic spectra using 34% amplification factor fecm NUREG CR - 0098.
Y - Recommended probabilistic spectra.:
Figure 3
-- n
~
1:
HA00AM, NECK,(5%),
1000.0 l
S
. g>o Y
e 100.0 v
s D
- J
- J 4
- 2. _ g gs 8
~ _
~ O' Ec %
y
/'
10.0
'/
~
m
/
~
5/
/
/'
0o s
e g
W t
e
.o$
/
d
//
c:
j,
./
g,g 3
j /'
[
/
- <o g
-,o 9
c.i.
0.01 0.1 1.0 10.0 WRIOC-SEC.
84% - Deterministic soectra using Sa% amplification f actor from fCREG CR - 0098.
50%
Deterministic seestra using 50% amolification
-f actor from NUREG CR - 0098.
l l
H - Recomended probabilistic spectra.,
i i
Figure 4
1 MILLSTONE (5%)
~
4
. g, 100.0 i
X
/N
!N N/
N i
% g, 8
_ m.
e4 M
. C-f.%.
a g
.0 10 i
F i
-@c r,
=
e W
3 C
.f 8
,.e W/
/
/
\\:::
e j/
N
,..o
/
=
4
. g.
01 0.01 0.1 1.0 10.0 PERICOSEC.
84% - Deteministic spectra using 84% amolification factor frem NUREG CR - OC98.
50% - Deterministic soectra using 50% amolification factor from NUREG CR - 0098.
M - Recomenced probabilistic spectra.:
Figure 5 l
t
~---
4 i
OY, STER, CREEK,(5Q
.f,.
4 s
o*
100.0 8
N N
sg D
n-o 8
_ _m.
o*-
m, y
W
/
10.0
/
~
p
~_
m w
./0.g
/
/;
g C
/
g
.f l
/
b
/
I *O.
w y
~
f
- <o
~
g
~
s
. p.
0.1 0.01 0.1 1.0 10.0 PERICC-SEC.
84% - Deterministic spectra using S4% amplification facter from NUREG CR - 0098.
50% - Jeterministic spectra using 50% amolification f actor from NUREG CR - 0098.
0 - Recomended procacilistic spectra.
t Figure 6
GINN4 (E%),,
1000.0
~$,.
E g>g a100.0,
/
/
/
s 1
v j
~
- r s%
8
- 9 7, e
" o*
5:9
~
E
.0,
/
/
J
[10 g:.
w
. /o
,f 0
e i
y
/
p
\\/
f g
,.e i/
/
J f
i
/
\\
\\
NE N
,g
.g
/
/
/
LI 0.01 0.I 1.0 10.0 PERIOO-SEC.
84% - Deterministic scectra using 53% amplificatien f actor from NUREG CR - C098.
50% - Deterministic spectra using EO% amolificacion f actor from NUREG CR - 0098.
G - Recomended procabilistic spectra.,
Figure 7 l
m
l
- - - ~. - -
D.RESD.EN, (5%)...
,,n o>%
0100.0,
/
-/
l
\\
1 s
2 y
_W f
v i
e%
~
3
//~
~0
- i w
C
///
.g@%
/
3
/
-l w
l W.a:
~
is N
. to w
0 i
/ //
e y
/'/
'/ /
\\
4 d
/
c-
// /
g
/
x-5 i ~r j
t y
N NE
.e3
. g.
0.1,
0.0 t 0.1 1.0 10.0 PERICC-SEC.
84% - Deterministic scectra using S4% acclification f actor from NUREG CR - 0098.
50% - Deterministic soectra using 50% acclification f actor from NUREG CR - 0098.
D - Reconnended probabilistic spectra.
t Figure 8
i P,ALISADES,(5%),,
1000.0 O
J,.
(4
. g.
y100.0,
- /
/
/
s
- E v
?.e D
y
.a
- g 9
- IcL
.g@
3
/
lI, /
/
10.0,
,/
S
'/
w
'o
/,
n
=
k
~
//
is n
/, '
.f
,e
. /
/
/
5 b
,i
- <o
~ p 9
0!
0.01 0.1 1.0 10.0 PERICO.5EC.
84% - Deterministic s;ectra using 84% amplification factar frem NUREG CR - 0098.
50% - Deterministic spectra usins 50% amplificatior f.acter frem NUREG CR - 0098.
P - Reccernended probabilistic spectra.
Figure 9
LACROSSE (5%.),
1000.0 Oss
~
o>
/
/
/
0 S 100
?
.0!
/\\
D
- 4
- -/.,,
_f y
//
10.0,
~
g
//
/
w
'O 0
/
~%
~
/
~
E
'/
~
q?g I
I IC b
b j
. M,
. g-0.1 0.01 0.1 1.0 10.0 PERICC-SEC.
Sa% - Deterministic scectra using 84% amplification factor from NUREG CR - 0098.
50% - Detem.inistic scectra using 50% acclification i
f actor frem NUREG CR - 0098.
i L - Recorrenced cr:bacilistic spectra.,
Figure 10
~_
i.
BIG ROCK,P0, INT {5X),
1000.0 c#
I
. g>4 100.0 N
N i
D O
_t-g 8
" C' r _ _ E.,,
J j,
s/
10.0 y
x
~
N
/ /
~
m
. /o
-w f.
0
' t h
/
~
//
4*
e b
//
"o?
E
/
/
ic t
/
N N.,o
',a 7
0i.
0.01 0.1 1.0 10.0 PERICC.SEC.
84 - Deterministic s;:ectra using S4% amplification factor from NUREG CR - 0098.
50". - Deterministic spectra using 50". amplification f actor from NUREG CR - C098.
3 - Recommended probabilistic spectra.
I Figure 11
E.U.S. Reconnended Probabilistic Spectra and Regulatory Guide 1.60 Spectra 1000.0 3
- N"
- g
/
d 10C.0 s >i
~'Q wf C2 j
/$
g
-<fa s
~ o,-
/
a
.'/
10.0 f./
8
'o i
w g
e g
y o
w
'o j
.M /
/
/
i
(
h hi
,o f
/
'o
/
/
/
41 0.0t 0.1 1.0 10.0 PERICC-SEC.
Y - Yankee Rowe 0 - Oyster Creek H - Hacdam Neck G-Ginna M-Millstene 0.1 - R.G.1.60 anchored at 0.lg l
0.2-R.G.1.60 anchored at 0.2g
\\
Figure 12
)
C.U.S. Recommended Probabilistic Spectra and Regulatory Guide 1.60 Spectra 1000.0
~
o.
/
-/
/
a 5 100.0 b
\\
\\
e c.2
~
m 2,9 c
~
y' g
77.)*'
b s
C
. O' a
IC O.
/
,/ //
~
m
.... ~
W
,'/
. 'q,
~
f
~
,j i
o
~
Q
,'j e?>
~
d
,y;
./
- /
l.C SO
- ?'
g 7
/
/
/
Al 0.01 0.1 1.0 10.0 PERICO-SEC.
D - Dresden P - Palisades L - Lacrosse B - Big Rock Point 0.1 - R.G.1.60 anchored at 0.19 0.2 - R.G. 1.60 anchored at 0.2g Figure 13
~ _...
i Recocrnended Probe.bilistic Spectra at Rock Sites and Recorded Spectra at Rock Sites
~
x
~R,. s N 100.0
~
b 5
\\
2 o
g s
' o,-
a
,/
l0.0 I
/
/
// /
\\
~
3
///' /
% 1Y.L
"'Ro N
j
~
'A. ' '/ /
~
- f 1 '/
os r
h
.f
^
j e
//
? I//
/
,,c
/[d'/
N NE E
i
- <o
/I, /'
&g
-f
/
/
/
0.1 0.01 0.1 1.0 10.0 PERIOO.SF.C.
0-Dresden H - Haddam Neck G-Ginna M '4illstone 34% - 84% scectra frem nearby Mac. 5.3 +.5 event.
50%-50% spectra from nearby Mag.'5.3 + 75 event l
Figure 14
Recomended Probabilistic Spectra at Soil Sites and Recorded Spectra at Soil Sites l'XC.0 c#
S y,;:
e.
100.0 j
f D
a
- r s
~
y
---8 o,-
a
//
/ //
j 10.0 l
Y f,
r
- Q o0
//
W
/,
/
d
/
8 f
.C
=
/
l.C
\\
N
! / /'
- /
/~
r%
b' f
-?
c
~
i 0.1 0.01 0.1 1.0 10.0 PERICO-SEC.
y - Yankee Rowe i
0 - Oyster Creek P - Palisades L - Lacrosse B - Big Rock Point St.% - 84% spectra from nearby Mag. 5.3
.5 event 50% - SC% spectra from nearby Mag. 5.3,.5 event i
l
\\
Figure 15 1
3
..