ML20040E169
| ML20040E169 | |
| Person / Time | |
|---|---|
| Site: | Limerick  |
| Issue date: | 02/01/1982 |
| From: | Schwencer A, Thadani A Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML19271A825 | List: |
| References | |
| NUDOCS 8202030208 | |
| Download: ML20040E169 (6) | |
Text
b 02/01/82 UNITED STATES OF AMERICA NilCLEAR REGULATORY C0 tit 1ISSI0tl BEFORE THE AT0 HIC SAFETY AND LICENSING BOARD In the flatter of
)
)
PHILADELPHIA ELECTRIC COMPANY
)
Docket Nos. 50-352
)
50-353 (Limerick Generating Station,
)
Units 1 and 2)
)
AFFIDAVIT OF ALBERT SCHWENCER AND ASH 0K C. THADAtlI REGARDING SCHEDULE FOR STAFF REVIEW 0F PRA 1.
fly nane is Albert Schwencer.
I an Branch Chief, Licensing Branch 2, Division of Licensing, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Commission.
In this capacity, I have supervisory responsibility for the review of various applications for construction and operation of nuclear power reactors.
Included among the applications assigned to my Branch is the operating license application for Limerick Generating Station, ifnits 1 and 2.
2.
t1y nane is Ashok C. Thadani.
I an Branch Chief of the Reliability and Risk Assessment Branch, Division of Safety Technology, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Connission.
This Branch has been assigned the responsibility for the review of the Probabilistic Risk Assessment ("PRA") submitted by the Applicant for the Limerick Generating Station.
3.
The purpose of this affidavit is to respond to a concern raised by the Atomic Safety and Licensing Board regarding the schedule for the 8202030208 820201 DR ADOCK 05000
. 4 Staff's review of the Applicant's PRA. Tr. 169-173.
During its inquiry into the proposed contentions on the PRA, Judge llorris focused the Staff's attention (Tr. 170-171) upon the following excerpt from the Commission's " Statement of Interim Policy: Nuclear Power Plant Accident Considerations linder the flational Environmental Policy Act of 1969," 45 Fed. Reg. 40101 (June 13,1980):
However, it is also the intent of the Commission that the staff taka steps to identify additional cases that might warrant early consideration of either additional features or other actions which would prevent or mitigate tha consequences of serious accidents. Cases for such consideration are those for which a Final Environmental Statement has already been issued at the Construction Pennit stage but for which the Operating License review stage has not yet been reatned.
In carrying out this directive, the staff should consider relevant site features, including population density, associated with accident risk in comparison to such features at presently operating plants.
Staff should also consider the likelihood that substantive changes in plant design features which may compensate further for adverse site features may be more easily incorporated in plants when construction has not yet progressed very far.
The Board then requested the Staff to file a statement of how its schedule for review of the PRA takes into account the above concern reflected in the Commission's Policy Statement.
Tr. 172.
The following constitutes the Staff's statement.
4 The Applicant was requested to prepare a PRA by letter of tiay 6, 1980 from Darrell G. Eisenhut (Director, Division of Licensing, NRC) to Edward G. Bauer, Jr. (Vice President and General Counsel of the Applicant, Philadelphia Electric Company). The Applicant submitted its PRA on liarch 17,19%, along with the Final Safety Analysis Report and the otner portions of the application. After completion of the Staff's
" acceptance review," the application (including the PRA) was formally
. docketed on July 27, 1981. Additionally, in response to the Staff's request of July 6,1981, for further information, the Applicant submitted fault trees on Septenber 24, 1981.
5.
The Staff's present schedule for preparation of the SER (including its review of the PRA), which updates and supplements the information provided at the Special Prehearing Conference (Tr. 169-170),
is set forth below:
09/30/82 - Draft PRA NUREG issued.
02/04/83 Draft SER issued.
08/06/83 SER issued.
SSER (Supplemental SER) issued.
10/17/83 6.
The PRA NUREG will be prepared by Brookhaven National Laboratory (BNL), acting as consultant to the Staff.
An explanation of the Staff's review process is necessary for an understanding of how it intends to use the results of BNL's review of the PRA. The Staff's position on whether the Limerick facility can be operated without undue risk to the public health and safety is a composite of positions reached by its various review branches.
Each of these branches has responsibility for the review of certain plant systems, aspects of an applicant's organization, or site considerations. The review conducted by BNL is expected to result in a NIIREG/CR (contractor) report which will include an assessment of the methodology used and will identify dominant accident sequences based upon BNL's review of the PRA.
BNL's responsibility is to advise the Staff as to the dominant accident sequences so that the Staff can determine whether any additional design features or other actions are
.. necessary to compensate for adverse site features. These deteminations will be reflected in the Staff's SER. The Staff's position with respect to the Applicant's PRA will not, therefore, be reached until the issuance of the SER (and the SSER), unless the Staff's initial evaluation (draft SER) indicates that significant improvements and/or modifications are necessary.
7.
The Applicant concludes in its PRA that:
"The Limerick site-specific best estimate CCDF [ Complementary Cumulative Distribution Function, which relates the expected frequency (number of accident events per year) to the consequence (e.g., number of early or latent fatalities, property damage)] curves are slightly below the WASH-1400 curves for both early fatalities and latent fatalities for all calculated consequences."
Executive Sunnary, p. 11.
The Applicant has assuned certain design improvements (e.g., an anticipated transient without scran ("ATWS")
prevention /nitigation systen "at least as good as the Alternate 3A modification identified by the NRC Staff in NUREG-0460") (PRA, p. 3-48) in performing its analysis.
The Staff will review the Applicant's assumptions to verify that they reflect either 'urrent design configuration or fim connitments to design inprovenents.
8.
The Staff's schedule for the PRA review for Limerick is consistent with its overall plan for implementation of the Connission's June 13, 1980 Policy Statement. As indicated in SECY-81-25, "Perfornance of Probabilistic Risk Assessment or other Types of Special Analyses at High Population Density Sites" (January 12, 1981, copy attached),
Limerick is one of three sites (the other two being Indian Point and Zion) which have "Substantially Above Average" Site Population Factors
- ("SPF").
SECV-81-25 clearly states, however, that the determination of l
whether any special mitigation features will be required for Limerick (and, if so, what features) will be made after completion of the Zion / Indian Point studies, as part of the overall long-range progran.
SECY-81-25, p. 7, para. 3.
Thus, the Connission is aware of the Staff's establishnent of priorities with respect to the completion of the review of the PRA's and the reconnendation of any special nitigation features which may be deened necessary for these three sites. The Staff has proceeded with the review of the PRA for Limerick on a schedule consistent with this assignnent of priorities. The Commission was also advised by the Staff that if probabilistic risk assessment for a site indicated frequencies of severe core damage in the range of 10-4 to 10-5 or lower per reactor year, "any action to require modifications should await the consideration of other reviews and studies so that any backfits would be appropriately ccordinated with other possible requirements."
SECY-81-25, p. 9.
The Limerick Applicant's mean value for the frequency of core melt is 1.5x 10-0 per reactor year.
PRA, p. 3-100.
Part of the Staff's review of the PRA will focus on verifying the reasonableness of this core melt probability.
In view of the relatively low apparent value l
of this estinate and the higher priority of the review of the Zion and Indian Point (both operating plants) studies, the Staff assigned a somewhat lower priority to the Limerick PRA review. However, the schedule is still fully compatible with the development of the SER and could not, in any event, have been expedited much more than a few months considering the availability of the Applicant's subnittals.
. 9.
For the reasons discussed above, the Staff believes that the schedule it has established for the review of the Limerici PRA and the determination of what, if any, special nitigation features it will recommend is consistent with the Commission's Policy Statement. The Applicant's operating licenses for Limerick nay, of course, hr:
conditioned in such manner as this Board concludes is necessary to provide " reasonable assurance that the health and safety of the public will not be endangered by operation of the facility." 10 C.F.R.
S 50.35(c). The Staff's review schedule will not, therefore, preclude the imposition of any design or other changes determined to be necessary.
/
'h
% E5rt Schwencer fl Ashok C. Thadani Subscribed and sworn to before me 3
this / Srday of 7,.<2%
, 1982.
GnMa1 K._>h,J nota'ry pub 1fc p fly Commission expires:
d /,// [ M f
f
7 w
January 12, 1981
/
,2a r,
\\,....../
ELD FILE COP OELD FILE COPY
/
POLICY ISSUE (Information)
For:
The Comissioners From:
William J. Dircks Executive Director for Operations
Subject:
PERFORMANCE OF PROBABILISTIC RISK ASSESSMENT OR OTHER TYPES OF SPECIAL ANALYSES AT HIGH POPULATION DENSITY SITES
Purpose:
To provide the Comission with an initial list of sites prioritized by population density; and to inform the Comis-sion of the nature, extent, and priority of probabilistic risk analyses (PRA) or other types of special analyses that the staff is performing or planning for all operating reactors and applications for reactor licensees. This subject relates to the Statement of Interim Policy on Nuclear Power Plant Acci-dent Considerations affirmed on May 15, 1980, and published in the Federal Register on June 13,1979 (45 FR 40101).
It also relates to the development of a transition policy between the old and new siting criteria requested by the staff require-ments memorandum dated June 30,1980 (Enclosure 3). flo action is requested of the Comission at this time. Complete response to item 2 of the June 30 memorandum will be developed after review and discussion of this paper and the Indian Point and Tion reviews with the ACRS.
Backaround: During the Comission's discussion on April 16, 1980, concerning SECY-80-131, " Accident Considerations Under NEPA", there were expressions of interest in plants in early stages of construction that might be candidates for special risk studies. This was in addition to an expanded treatment of accident considerations for l
cases undergoing liceasing review and resulted in language being added tc the Statement of Interim Policy that the staff should
" Consider the likelihood that substantive changes in plant design features which may compensate further for adverse site features may be more easily incorporated in plants where construction has not yet progressed very far."
Contact:
M. L. Ernst, NRR 492-8016 fG
o
. During tha discussion of SECY-80-131 on April 16, 1980, Limerick was identified for a special PRA study, following the Indian Poirtt/ Zion special risk studies (NUREG-0660. TMI Action Plan, Task II.B.6). By letter dated May 6,1980, the staff requested the Philadelphia Electric Company to perfom within 120 days a ~
preliminary risk assessment of the Limerick facility. However, j
the results of this study will not be available from the licensee until at least March.
Consumers Power Co. has initiated a risk study at Big Rock Point to support the utility's desire to not backfit certain of the TMI requirements because of a relatively low power level and low site population density.
In addition, Crystal River was evaluated by the NRC under Phase I of the IREP (Interim Reliability Evaluation Program); and reviews of Calvert Cliffs 1, Arkansas 1. Millstone 1, and Browns Ferry I were recently initiated as Phase II of IREP.
One purpose of the IREP studies is to develop a common methodol-ogy for the evaluation of all operating plants under NREP (the National Reliability Evaluation Program). The industry is also interested in developing a comon methodology for the application of PRA, as is evidenced by the EPRI-sponsored review of Sequoyah and the NSAC-spons:: red review of Oconee. Currently, arrangements have been made among the NRC, AIF, IEEE, and the ANS for the cooperative development of a common PRA methodology that would draw upon all of the above experience.
If this development is successful, the results likely will be used for the NREP in lieu of the methodology being developed under IREP.
In addition to IREP/NREP, other programs are current 1'y underway or soon will be underway which will directly or indirectly impact the assessments of the safety of operating plants. These include the SEP program; implementation of the Bingham amendment (Sec-tion 110 of the NRC FY-80 Authorization Act (Public Law 96-295));
continued examination of the Indian Point petition, including special accident mitigation features; degraded core rulemaking; evaluation of shutdown heat removal capabilities; and a number of other TMI related subjects and possible rulemakings.
At an appropriate time a coordinated, disciplined approach to the review of operating plants must be adopted that considers all of these progr m s in an integrated manner and achieves an effective and efficient utilization of the considerable resources required by these activities. This point should be weighed in any consid-eration to do additional interim studies.
Discussion:
In the development of an appropriate course of action the staff addressed two basic questions:
6
~w---..-,ev1-- - <.. -, -.
--.------w
-r-,----,-n,..w--
m----
,-,-----,-er,w--,----
-.-t-
' 1.
If additional studies should be performed at sites having high population densities or other possible adverse charac-teristics, which sites should receive priority consideration?
2.
At such sites, what type of additional studies would be appropriate?
To arrive at decisions regarding these questions, the staff con-sidered several criteria as described in detail in Enclosure 1.
Basically, however, the staff's considerations centered upon the likelihood that such interim studies would result in sufficient, near-term improvement in the safety of operating reactors to warrant the comitment of priority resources.
Site Identification
.The first question involves the identifica-tion of plants which might be candidates for any special studies.
The only adverse site feature considered by the staff in this analysis was population density, since little infonnation was readily available on some potential risk factors, such as liauid pathways and evacuation effectiveness. ' Criteria for external hazards are being considered in siting policy, to be followed by rulemaking; and site-specific meteorology likely would not sub-stantially affect the results of the analyses.
There are many ways that population density might be considered, such as number of people nearby, total population out to 10, 20, 30, or 50 miles, or some process that weights nearby people higher (because of less opportunity for atmospheric dispersion) but still considers exposures to persons more remote from the site. Trial application of these alternatives by the staff indicated that the general results of the analyses (i.e., sites identified as higher risk) are not sensitive to the chosen analysis method to any significant extent.
The method employing the Site Population Factor (SPF) was selected, since it was considered by the staff to provide a fairly reasonable numerical representation of radiological risk, assuming a given airborne release of radiological materials. The SPF methodology weights population by proximity to the site in accordance with the typical average atmospheric dispersion factors at various distances from a site; i.e., site specific meteorology is not used. The resultant SPF values were weighted by plant power level (thermal MW).
A discussion of this process and the results of the screening for all reactor sites are provided in Enclosure 2.
In general terms, most of the reactor sites in the United States have a power level weighted SPF value that is within a factor of four of the median weighted SPF for all reactors, which is a value of 206 out to a radius of 30 miles (equivalent to a unifonn population d'ensity of
, less than 100 people per square mile out to 30 miles). However, 8 sites have "Above Average" power lavel weighted SPF values, i.e., values four to eight times greater than the median value; and three sites -- Indian Point, Limerick, and Zion -- have. s "Substantially Above Average" values (10-15 times higher than themedian).
Special studies are already being perfomed on Indian Point,.
Limerick, and Zion. The staff believes that the 8 "Above Average" reactor sites are the only additional ones that might be construed to have " adverse" population densities, and of these, only Bailly and Millstone 3 are in the early stages of construction -- Bailly being about 1% complete, and Millstone 3 about one-third complete.
The staff is aware that the criteria used to separate the reactors into groups are somewhat arbitrary. For example, Pilgrim 2 (which is awaiting a CP) is not listed in the "Above Average" group, since it has a power level weighted SPF value only waewhat greater than 3 times higher than the median. Howc'.er, Bailly (which is in the "Above Average" group) has a power level weighted SPF vaTiie only 20% higher than Pilgrim 2.
Generally speaking, the risk represented by population density is linearly proportional to the populatior. Therefore, unless there is some other associated factor, such as evacuation efficiency, that is nonlinear with increasing population, determination of a point at which an increasing population suddenly becomes an
" adverse" site feature is necessarily arbitrary. Nevertheless, the staff has concluded tlat the categories chosen are reasonable, since:
1.
Choice of a weighted SPF smaller than a f)ctor of four above that of the median site as a demarcation would result in very little difference in risk between the median site and the
" adverse" sites.
2.
The " median" site, upon which the analyses were based, has almost a factor ci two less population than the " average" site, thus the chosen demarcation point is only about a factor of two higher than the average site.
3.
The errors in perfomance of risk analyses are of the same order of magnitude or perhaps greater than the factor of four; therefore, the significance of the conclusions of any compre-hensive risk analyses performed at these " adverse" sites would I
be somewhat lost in the uncertainties of the analyses, if a lower point of demarcation were chosen.
(
1 h '"
e*-
=
y' k
+
d' l
+ -
5
~
4.
All sites (except Haddam Neck) that have a power level weighted SPF within a factor of four of the median site also have less than an average of 500 persons per square mile at a 30-mile radius, and Haddam Neck has less than 500 persons per square mile at all distances up to and including 20 miles.1 To illustrate the lack of a clear demarcation line for identifying sites with " adverse" population characteristics, three of the 11 sites that have a weighted SPF greater than a factor of four above the median (Seabrook, Three Mile Island, and Waterford) have a population density less than 500 persons per square mile at any distance up to at least 50 miles.1 The staff believes that the method chosen to identify sites with potentially adverse population characteristics is reasonable.
However, one must not conclude from this analysis that the sites so identified actually are the ones that contribute predominately to the overall risk from nuclear power reactors.. Factors other than population density have a major influence on risk.' No consideration was given to plant design in this paper; and varia-tions in plant design have been demonstrated to have a major influence on risk -- potentially much greater than population differer.ces. This fact has been spotlighted in the recent Indian Point and Crystal River risk studies, Types of Risk Studies - The second question involves the type of special risk studies that might be worthwhile to conduct, if it is determined that special studies are warranted because of adverse site characteristics. Possible kinds of studies are:
1.
Complete risk studies using WASH-1400 methodology.
2.
Truncated WASH-1400-type evaluations that analyze only selected potential dominant accident sequences, based on judgment supported by experience.
3.
IREP-type analyses, which are principally targeted at core melt sequences (no in-depth review of containment or of releaseconsequences).
4.
CRAC-type aralyses, which are aimed at consequences only (i.e., assume given releases and pathways, but use regional orsite-specificmeteorology).
5.
Plant-specific analyses of special mitigation features (such as filtered vented containment), which would require a reasonable knowledge of accident sequences and potential release pathways to provide some judgment as to overall risk reduction.
2Per the 1970 census.
M-mm I mm um
_ - =
=
i 6-1
)
All of the above possibilities involve PRA to some degree, since PRA (at least as the term is comonly used) can involve varying degrees and sophistications of probabilistic analyses of accident sequences, mitigation, and/or exposure of people. A discussion of some of the considerations involved in conducting PRA ts found in Enclosure'1.
The staff considered the above types of analyses and has the following technical conclusions:
l 1.
CRAC-type of analyses using regional or site-specific meteorology and demography are a useful way of estimating the risk of accidents, provided that accident sequences and release pathways can reasonably be assumed.
If such cannot be assumed, then these analyses still give interesting relative information to permit comparisons of one site to i
another, assuming the same reactor on each' site. While such relative analyses might appear to be useful, they would not yield good plant / site-specific risk profiles, and the use of SPF values (which do not use site-specific N 2eorology)likely would display the relative comparisons of shes sufficiently accurately to determine whether population represents an
" adverse feature" without using the sophistication of the i
CRAC code. However, running CRAC calculations wculd perhaps improve the public's understanding of site-specific radio-logical risks at a nominal expenditure of manpower.
2.
Design-specific risk assessments (IREP, or using WASH-1400 methodology) are very useful ways to estimate risk; hcwever, l
they are strongly dependent on a full knowledge of final plant design, operating experience, and operating procedures.
Performance of such analyses without full knowledge of the above would be so subject to error that they likely would not warrant the expenditure of resources. Also, the results of such analyses could be misleading, and the studies likely l
would have to be repeated at some later date when such detailed
(
inforntion becomes available.
The IREP analyses provide a good core-melt modeling cf plants, l
which would give a useful perspective on plant safety. While I
containment and other release mitigation features also are l
important to safety and should not be neglected, such full risk studies would be much more demanding on resources than the IREP-type studies.
vented containment)yses of mitigation features (such as filtered Plant-specific anal i
3.
are useful and are being done for Indian Point and Zion. These studies will be extended to other plants i
under a program to be developed after the results of the 7/IP
+ + - - - - - - + - -
=w~,
,r c,,---vv=,,.,r-
,-~-+,----,v.
..4r..----.-n y
. studies are available. While knowledge of accident sequences and release pathways are necessary to evaluate fully the safety benefits of such features, such knowledge can be some-what more deterministic in nature than would be required a for the more probabilistic-oriented risk studies. Considera-tion of requiring the use of such features on reactors would likely be dependent on the specifics of reactor design, pcci-dent sequences, and release pathways. Whether the site nas an " adverse" population density would probably be of somewhat lesser consideration in any technical decision to implement such a requirement.
Current and Future Staff Actions - The course of action for per-formance of risk studies currently being followed by the staff for sites with " adverse" population features as well as for all other plants is summarized below:
1.
For all future environmental statements, and for those currently in preparation, there will be an expanded discussion of accident risk similar to that provided in the statement for the Summer plant.
2.
CRAC studies for all reactor sites are underway by the staff in connection with the technical basis for rulenaking on Part 100. The results of these studies, which use regional or site specific meteorology, will be compared to SPF data, which use a single, typical or average meteorology.
3.
Using population density (SPF calculations) weighted by power level as a way of identifying sites with " adverse" population features, Indian Point, Limerick, and Zion are identified in the "Substantially Above Average" category. Risk studies using WASH-1400 methodology are being performed by the licensees for all of these plants, as are studies of special mitigation features for Indian Point and Zion. The results of these studies will be reviewed by the staff.
No other actions are being taken at this time; but special mitigation features will be studied for Limerick in the future, subsequent to completion of the Z/IP studies, as part of the overall long-range program.
4.
Using population density (SPF calculations) weighted by power-level as a crude way of identifying additional sites with potentially adverse features, eight sites (Bailly, Beaver Valley, Fermi, Millstone, Seabrook, Shoreham, Three Mile Island, and Waterford) are identified in the "Above Average" category. Of these, Bailly and Millstone Unit 3 are the only ones in the early stages of ccnstruction.
4 -- -,..- -.,.-..
,m, n -. -
For Millstone the staff intends to:' identify and advise the licensee of special design considerations to be c:gnizant of PRA studies performed on plants with similar systems;g of in the final design of the plants based on the result perform ariskstudyunderNREP(theNationalReliabilityEvaluation Program) on a relatively high priority basis; and perform appropriate long-range studies of special mitigation features, dependent en the results of the Indian Point and Zion stud,ies.;
The NREP reviews likely would be perfomed prior to the issuance of the OL, provided that the methodology for such reviews is developed in a timely manner.
The special problems associated with Bailly will be addressed within the context of: outstanding petitions filed under 10 CFR 2.206; and the staff reconsnendations on applying NTCP requirements (NUREG 0718) to prevent CP holders for which onsite construction has been minimal.
5.
For the remaining six sites in the "Above Average" category, the licensees will be advised of special design considera-tions, and the NREP and appropriate studies of special mitiga-tion features will be perfonned in a manner similar to Bailly and Millstone Unit 3.
However, any OL issuance decisions likely would not be contingent on the performance of such studies, particularly for the NREP review.
6.
For all other plants, activities similar to those identified in 5., above, will appropriately be undertaken. However, the details of the NREP and special mitigation features pro-grams have not been developed, and the priority given to such reviews will be based on a numbec of factors other than i
population density, such as licensing status, age of the plant.
l and the initial results of the program to identify deviations from the regulations of particular significance (the " Bingham" amendment).
7.
Emergency response studies are currently underway to detennine the effectiveness of evacuation, sheltering, and other protec-tive measures at all reactor sites. These studies are related to requirements associated with the new emergency planning rule and for the rulemaking on Part 100. Sensitivity studies for s
these efforts will provide the relationships to assess the influence of emergency response on risks from high population sites.
Backfit Deteminations - If, as a result of any of the studies identified above, additional measures appear to be required to reduce risk, the applicant will be requested to proposa plant Nhat would be provided would be a fairly complete description of the potentially dominant accident sequences, the safety systems used to mitigate the accidents, and identification of the inter-dependencies between major safety systems themselves and their support systems. These functional interdependencies are critical paths whose failure could result in significant accidents.
. ~
. modifications, including analyses to demonstrate the degree of safety improvement. In the near tem, calculational refinements to " fine tune" the PRA analyses will be discouraged, since the basic intent (until development of a comon PRA methodology for the NREP) should be to compare systems and not to try to demon--
strate a new level of absolute risk for nuclear power plants.
The time frame for implementing any needed proposed modifications would depend on the magnitude of the deviations from the currently
" accepted" risks and input from parallel NRC actions on Class 9 events, severe c' ore damage protection, siting, emergency prepared-ness, ATWS, the systematic evaluation program, and establishment of safety goals and criteria. If a plant already satisfies our current design criteria with respect to redundancy and diversity and the TMI requirements, experience demonstrates that the estimated probabjlity of severe core aamage will likely be in to 10-5 :1Y, in which case we believe that any
/
the range of 10-action to require modifications should await the consideration cf other reviews and studies so that any backfits would be appro-priately coordinated with other possible requirements. Any plants that are found to have a higher probability for severe core damage, or which substantially exceed other currently
" accepted" or "nomal" levels of risk, will be reexamined against our current detenninistic criteria and required to correct the deficiencies or otherwise reduce the risk in a reasonably expeditious manner.
4[
5 Willidd J. Dircks Executive Director for Operations
Enclosures:
1.
Discussion of Decision Criteria
?.
Prioritization of Sites i
with Regard to Popula-tion Density 3.
Memo from S. Chilk to W. Dircks dtd 6/30/80, "SECY-80-153, Advance Notice of Proposed Rulemaking on Reactor Siting" DISTRIBUTION Comissioners Commission Staff Offices Exec Dir for Operations ACRS ASLBP ASLAP Secretariat
- es 0 9 t
- .O.
t ENCLOSURE 1 t
e
t' Discussion.of Decision Criteria Various NRC staff proposals recently have recannended a number of approaphes,
for determining candidate sites for possible future risk studies. Selection i
criteria hava included Class 9 considerations, high population densities, unique emergency preparedness considerations, ar.d/or power level considerations. provides a proposai for a priority listing for any such near-term studies. However, an even more basic question is whether jgyc more ad, hoc risk studies need or should be performed, or whether future plant reviews utilizing PRA should await the development of the National Reliability Evaluation Program (NkEP), utilizing a more disciplined and agreed-upon methodology. This enclo-sure focuses on the considerations that affected staff decisions regarding the performance of any new near-term PRA studies at high population density sites.
The decision criteria used by the staff were:
. Conformance with regulations
. Improve NRC decision making on licensing actions
. Enhance public understanding
. Effective use of PRA resources 1.
Conformance with Regulations - Ine first criterion is that the program must conform to NRC's regulations; recognizing, however, that policy and regula-tions can be changed. The NRC's regulations address the use of PRA in the Commission's statement of interim policy and guidance (not a " regulation",
1 per se) concerning its position on the disclosure of accident risks under NEPA.1 145 FR 40101, June 13, 1980.
i
.(
This interim policy states that events or accident sequerices that lead to releases shall include, but not be limited to, those that can reasonably be expected to occur. Such analyses shall include in-plant and external causes of accidents that can result in inadequate cooling of reactor fuel and melting of the reactor core. The environmental consequences of releases (resulting from such accidents) whose probability of occurrence has been estimated shall also be discussed in probabilistic terms. Health and safety risks to individuals and population groups shall be discussed in a manner that fairly reflects the current state of knowledge resarding such risks, including associated uncertainties. Among other things, the Statement of Interim Policy states:
"It is the intent of the Comission in issuing this Statement of Interim Felicy that the staff will initiate treatments of accident consideration., in accordance with the foregoing guidance, in its ongoing NEPA reviews, i.e., for any proceeding at a licensing stage where a Tinal Envirorrnental Impact Statement has not yet been issued. These new treatments, which will take into account significant site-and plant-specific features, will result in more detailed discussions of accident risks than in previous environmental statements, particularly for those related to conven-tional light water plants at land-based sites.
It is expected that these revised treatments will lead to conclusions regarding the environmental risks of accidents similar to those that would be reached by a continuation of current practices."
"Houcver, it is also the intent of the Commission that the staff t9 s steps to identify additional cases that might warrant early cons.ideration of either additional features or other actions which would prevent or mitigate the consequences of serious accidents.
Cases for such consideration are those for which a Final Environ-mental Statement has already been issued at the Construction Pennit stage b4 for which the Operating License review stage has not yet been reached.
In carrying out this directive, the staff shculd consider relevant site features, including population density, associated with accident risk in comparison to such features at presently operating plants. Staff should also consider the likeli-hood that substantive changes in plant design features which may compensate further for adverse site features may be mdre easily j
incorporated in plants where construction has not yet progressed very far."
~
. Conclusions that may be drawn from the above statements regarding the present regulations are:
The Interim Policy Statement is silent on the performance of PRA -
for operating reactors.
For reactors currently under construction but not in the OL review stage, PRA studies might be warranted. Such studies, if performed, should consider site features compared to such features at presently operating reactors, and should consider substantive chaeges in plant design features which might compensate further for (any) adverse site features.
The Interim Policy Statement does not address the type of studies that should be performed.
2.
typrove NRC Decision Making - The second criterion relates to improving the ability of the NRC to make reasonable plant / site decisions regarding substantive changes in plant design features. The first aspect of this criterion considered by the staff concerns verifiability, and the second deals rith the treatment of uncertainty and other aspects of performing PRA which affect the evaluation of the results.
As will be seen in subsequent discussions, the verification process would be substantially easier, if the PRA methodology used for these risk studies were reasonably' consistent with the techniques used in the Reactor Safety Study. This would enhance verifiability by permitting a reascaable compar-ison of the resultant consequence curves to obtain a relative risk canpar-ison to typical reactor / site combinations. Obviously, modification of the O
4 methodology (models, data, assumptions) can change the results for both the subject plant and the reference plant. However, the staff would.be principally interested in changes in specific plant features that would provide substantive safety imprcyements (i.e., lower risk) 4t sites tsith adverse characteristics, and not " improvements" based on refinements in analyses.
If the analyses are not reasonably constrained to the WASH-1400 data base and techniques, these studies (which should basically be comparative in nature) would begin to take on the character of absolute risk studies, in addition to making verifiability more difficult. The net result of such studies would be the expenditure of additional liRC resources to verify the analyses, plus increased controversy over the assumptions, data base, and methodology at the expense of a clear focus on design improvements (com-pared with a " typical" reactor) to compensate for an adverse site charac-teristic. As a matter of interest, this is the nature of the Indian Point, Zion, and Limerick studies being performed by the industry. A great deal of effort is going into phenomenological studies and the refinement of data b's6s and asy.saptions with an eye toward demonstrating that the overall risk a
from nuclear power plants is substantially lower than the estimates in WASH-1400.
The final aspect of the second criterion relates to the ability of the NRC to utilize such PRA as a tool to assist decision making. The use of risk studies to determine whether additional design features should be required does not mean that the staff is abandoning established requirements.
For
. instance, risk analyses perfomed for power plants at low population sites should not be used to permit single train safety systems, even thtce]h the studies might show that the risks would be below those of the reference -
plant. Thus, at the present time the results of any such studies would more likely be used to add new requirements, not delete present ones.
Since only high population density sites presently are being considered, the implicit concern is one of population exposure as well as individual exposure. Therefore, the results of these studies would likely be pre-sented as complementary cumulative distribution functions (CC0F) which show the probability of exceeding consequences (early fatalities, latent cancers, property damage) of a given magnitude as a result of radicactive releases. The use of severe core damage probabilities alone likely would be considered insufficient for this assessment, since such analyses would not include, for example, population densities and mitig2 tion measures such as containment. Thus, core melt calculations alone would not gemit l
a direct evaluation of risk to the populatir4..
1 In the absence of any previously sanctioned acceptance criteria for societal risk, it is very likely that the results of these risk studies would be com-pared with the CC0F's presented in the Reactor Safety Study (WASH-1400) for reference light water reactors to decide whether further action is required to reduce risk. However, such comparative analyses using truncated rtL studies would have several weaknesses, as discussed below:
a.
Completeness - Because of resource constraints, the studies envisioned likely would not treat external avents, sabotage, fire, etc., and i
9 i
c likely would not include a full scope of detailed fault tree analyses; i.e., they likely would be characterized as truncated studies. These omissions might not be sericus, since an absolute risk detennination would not be the objective -- only a comparative systems evaluation.
However, the limited analyses would introduce some uncertaiisty, since such studies would not include a complete risk modeling of all plant systems, subsystems, and components. This weakness should be limited by careful attention to inclusion of all the major unique plant differ-ences between the subject plant and the reference plant, and peer judgment when the studies are reviewed.
b.
Uncertainty - The results would represent probabilities at about the 50 percent confidence level (ignoring a certain lack of statistical rigor in the analyses and certain bounding assumptions that are made) and would not address the potential that the uncertaintfes in the analyses could be significantly different between the plants compared. '
This could distort the results compared with what might be achieved at higher confidence levels. Dr. Okrent (Science, April 25, 1980) has suggested that the risk be assessed at a high level of confidence to ensure better treatment of the data. However, because of the likely limited nature of any new studies, the substantial effort required to address propagating errors and to carry uncertainties forward, the effect of phenomenological anumptions, and the fact that many of the data are point estimate values er judgments, it is not, believed that detailed uncertainty analyses would be justified. The problems asso-l ciated with uncertainty could be alleviated to some degree by careful i
l
7-attention to the performance of sensitivity studies and by using the same methodology, assumptions, and data base as is used in WASH-1400, with significant deviations clearly defined and justified.
c.
Dependence on Final Design - 5ystems level PRA is not useful unless it is performed on the as-built or final design of the plant. While generic analyses such as WASH-1400 (which uses two plants as representa-tive prototypes) give a useful perspective regarding 'a particular design, small differences from the reftrence design in an as-built plant can have substantial impact on the dominant accident sequences. For this reason, the performance of systems PRA on plants in the CP review stage or in the early stages of construction would have significantly fewer benefits than PRA performed during the OL stage or for an operating I
reactor.
In a similar vein, a substantial contribution to risk origi-nates from the specific operating and maintenance procedures that are in effect at the plant. Since these will not be final until shortly before operation, the maximum benefit of performing a PRA would not be achieved until the plant is near operational.
d.
Realism - Past experience has shown that a considerable amount of judgment is used in the risk studies. These include human error, treatment of common cause failures, physical phenomena related to containment failure, and equipment operability under adverse condi-tions.
In these areas, the staff's evaluations would likely tend to be conservative, particularly for these truncated studies. Such conservatism could result in the requirement of more marginal safety improvements.
w,,m-p.
-m e
i
. 3.
Enhancement of Pcblic Understandin? - Whether or not any of the possible programs to perform additional PRA studies would substantially enhance public understa' ding is debatable. However, one certainly would not want to embark on a program that would not be understandable to the public, unless such a program likely would result in the identification and imple-mentation of significant safety improvements.
On the one hand, the public is aware of the PRA tool, especially since publication of the various TMI reports and the recent considerations regarding the interim acceptability of Indian Point.
In this regard, it would be difficult for the public to understand any decision not to utilize PRA as an aid to making decisions in controversial areas, such as those associated with sites of high population density.
On the other hand, there are such substantial uncertainties in the perfor-mance of PRA that any result (regardless of how realistic or how conserva-tive) will be very controversial. Therefore, in the face of such contro-versy the public will be unsure as to the mer:ts of the results, which would be a disservice to public understanding.
Therefore, it would appear that to optimize enhancement of public under-standing, the scope and nature of any study that is performed should be such as to minimize the effects of uncertainty in the results.
4.
Effective Use of PRA Resources - As noted in the Background section of this Commission Paper, there are a substantial number of PRA's underway at the present time by both the NRC and the industry. These analyses L
i
~
~
-9 are being performed princi isy to evaluate the acceptability of systems design and siting of specific plants, and to develop a standardized PRA methodology.
There are not many trained people in the field of PRA -- indeed, half of NRR's professionals currently assigned to PRA are involved full-time in IREP, and the others are essentially fully comitted to the Limerick, Zion, and Indiar: Point review:. The same situation exists in industry -- in fact, the utilities involved in the IREP program are taking full opportunity of this program to train their people in PRA.
Therefore, we are essentially at the pcint where the impleme..'.stion of any new studies would begin to seriously dilute ongoing PRA efforts, including the development of a ccamon methodology. While this fact should not itself preclude the initiation of a new PRA effort, one must assure that such an effort would likely have a safety payoff that would warrant the further dilution of the longer-term program.
Additionally, one must be careful that the usefulness of PRA is not over-played. There is a danger that if ad hoc studies are required by the NRC but are not carefully controlled and monitored, the results could subse-quently be questioned and perhaps found to be seriously flawed.
If thir.
were to happen on a few controversial cases, then PRA could receive a sub-stantial black eye and might be perceived to be not very useful as a decisional tool.
If this were to happen, a potentially useful tool might be dismissed out-of-hand for future use, only because of a premature and
. incautious use. Again, the potential safety payoff of any early use of PRA in controversial cases should be sufficiently great to warrant taking the risk of perhaps performing seriously flawed studies.
The scope of all of these studies vary considerably, as can be seen from.
The IREP studies are the least manpower intensive, but they are limited essentially to the evaluation of severe core damage accident sequences. More detailed studies that include accident consequences and external events (such as seismic) require 5 to 10 times more manpower than the Phase II IREP. The performance of the so-called short term Zion and Indian Point studies (not listed in Attachment 1) required less manpower than the IREP studies, since they did not model the entire plant but only focused on unique differences in plant design compared with the Surry Analysis in WASH-1400.
i
PRORABILISTIC RISK ASSESSMENT STilDIES CURRENTioY ltNDERWAY NSSS/CONTAINilENT PARTIES LEVEI,OF ESTIHATED PLANT TYPE INVOLVED EFFORT (MY)_
COMPLETION DATE SCOPE OF EFFORT
- P C
A E
Oconee R&W/ Dry NSAC 15-20 Adr,1981 X
X X
8 Utilities Consultants l
Sequoyah W/ Ice Condenser EPRI 12-16 Phase I -
X X
X Utility Dec 1980 Consultants Phase II -
Dec 1981 Limerick CE/ Hark II Utility 7-8 Sept 1980 X
X i
HSSS Vendor Consultants Zion /Indinn Pt W/ Dry Utilities e'30 Oct 1980 X
X X
l NSSS Vend'or (3 units at 2 sites)
Consultants
. Crystal River B&W/ Dry HRC-RES 6-7 Sept 1980 X
Consultants (Inittal IREP Study) l Calvert Cliffs 1 CE/ Dry NRC-RES 3-5 June 1981 X
Arkansas 1 B&W/ Dry Consultants (per plant)
H111 stone 1 GE/ Mark I Ilrowns Ferry 1 GE/ Hark 1 (INEP Follow-up Studies)
- P Accident Probabilities C
Accident Consbquences Plant Availability A
External Events E
(* <
e'.
8 e
g G
O e
S e
ENCLOSURE 2 e
O
.4
=e e
e 5
J
.q-Prioritization of Sites with Regard to Population Density 1.
Introduction In comparing and evaluating the population around nuclear power reactor sites, the staff has long recognized that the population characteristics of a site, that is, its density and distritution, art a relatively crude measure of the consequences associated with the accidental release of radioactivity. The residual risk from an accident would depend not only upon the population den-sity of the site, but also upon many other factors, such as reacter design, onsite and offsite management and technical support rescurces, external hazards, liquid pathway considerations, meteorological conditions at the time of the accident, and effectiveness and nature of public protective actions taken.
In addition, the risk is not uniform for all members of the population regard-less of distance from the site, but would be higher for those persons relatively close to the site, and would generally decrease with distance away from the site.
I An analysis has been carried out to obtain a first-order prioritization of sites based upon population density and distribution. The discussion that follows outlines the rationale and methodology used and gives the results of this analysis.
2.
Methodology In carrying out this analysis, the following assumptions and methodology were used:
-a 2-(a) All sites where a reactor was either in operation, under construction, or where a construction permit was presently under active review were evaluated. This involved a total of 93 sites.
(b) The population data used were taken from NUREG-0348, based on.the 1970 census. The population data for the Fermi site as reported in NUREG-0348 are in error and were corrected for this analysis by a special l
computer run of the 1970 census tape.
(c) Although it is well-known that individuals closer to the reactor are at a higher level of risk, given an accident, than those more renotely located, the precise quantification of the variation of risk with distance is still somewhat uncertain. For the purpose of this analysis, the distance weighting given by the Site Population Factors (SPF), as given in WASH-1235, were used.
Further, population beyond 30 miles was neglected, because the consequences at distances within 30 miles were considered to dominate any considerations of overall societal impact, and beyond 30 miles the potential population exposure differences from site to site become less sharp. Preliminary analyses carried out by the staff have indicated that somewhat differing weighting schemes, or the factoring in of population out to 50 miles, does not change the resulting prioritization of sites to a significant degree.
(d) The power level of the largest reactor at the site was multiplied by the SPF value to account, in a first-order way, for the variation of reactor fission product inventory from site to site. Only one reactor at a site was considered, even where multiple reactors exist or are contemplated,
, because the probability of an accident involving more than one reactor simultaneously was considered negligible. Although it can be argued that the population around a 4 reactor site is at a higher level of risk than those around a single reactor site, the prioritization of sites is l
intended to give a measure of the relative consequences, given that an accidert has occurred. The number of reactors at a site presumably effects only the probability of an accident. Also, it could be argued that a multi-reactor site would have some attributes that would reduce risk, compared to a single-reactor site, because of greater management and technical resources that can be applied to reducing either the likuli-hood or consequences of an accident. Using the above methodology, the reactor power level times the SPF value was calculated and tabulated for aach of the 93 sites considered. The results are discussed below.
3.
Results The reactor power level times SPF (P x SPF) was calculated for each of the 93 sites. The resulting values ranged from a high value of 2980 to a low value of 6.
The median value is 206; and the median site has a population of less than 100 persons per square mile, which is almost a factor of two less than the population of the average site. The sites are not listed in numericC order, since this would imply a greater degree of precision th.ar, u v rranted by the uncertainties in the analysis. Also, as pointed out previously, the residual risk at a particular site cannot be measured in tenns of consequences alone, since plant design and other factors are important contributors to risk. Therefore, we decided to place each site
. into one of five groups or categories. The variation within a given group was selected to be sufficiently small so that each site within that group is considered to have about the same ranking.
In selecting the groups we decided to use the median value and factor of two varia-tion about the median to demarcate the " average" group boundaries. The l
other groups were chosen as indicated below.
Group No.
Title Range I
Below Average PXSPF less than one-half the median value (PXSPF < 100)
II Average PXSPF between one-half and twice the median value (PXSPF from 100 to 400)
III Slightly Above PXSPF between twice and four Average times the median value (PXSPF from 400 to 800)
IV Above Average PXSPF between four and eight times the median (PXSPF from 800 to 1600)
V Substantially Above PXSPF greater than eight times Average the median (PXSPF > 1600)
Within each group the sites have been listed in alphabetical order, as shown in the following tables.
Group V - Substantially Above Average l
1.
Indian Point 2.
Limerick 3.
Zion
=
, Group IV - Above Average 1 1.
Bailly 5.
Seabrook 2.
Beaver Valley 6.
Shoreham 3.
Fermi 7.
Three Mile Island 4.
Millstone 8.
Waterford Group III - Slightly Above Average 1.
Byron
- 11. Peach Bottom 2.
Catawba
- 12. Perkins 3.
Cook
- 13. Pilgrim 4.
Cherokee
- 14. Perry 5.
Erie 15.
Salem 6.
Forked River
- 16. Sequoyah 7.
Heddam Neck
- 17. Susquehanna 8.
Hope Creek
- 18. Rancho Seco 9.
McGuire 10 Turkey Point
- 10. Midland 20.
Zimmer Group II - Average 1.
- 21. Palisades 2.
Bellefonte
- 22. Phipps Bend 3.
Black Fox
- 23. Prairie Island 4.
Braidwood
- 24. Quad Cities 5.
Browns Ferry
- 25. River Bend 6.
Calvert Cliffs 26.
Robinson 7.
Clinton
- 27. San Onofre 8.
Brunswick
- 28. Shearon Harris 9.
Davis-Besse
- 29. Summer
- 10. Duane Arnold 30.
Surry 11.
Fort Calhoun
- 31. St. Lucie 12.
Fitzpatrick
- 32. Skagit
- 13. Ginna
- 33. Trojan
- 14. Hartsville
- 34. Vogtle
- 15. LaSalle
- 35. Watts Bar
- 16. Maine Yankee
- 36. WPPSS 3/5
- 17. Marble Hill
- 37. Vemont Yankee
- 18. Nine Mile Point
- 38. Monticello
- 19. Oconee 39 Yellow Creek 20 Oyster Creek 1Bailly and Millstone Unit 3 are the only plants in Group IV that are in the early stages of construction.
. l Group I - Below Average j
1.
Allens Creek
- 13. Kewaunee 2.
Big Rock Point
- 14. Lacrosse 3.
Callaway
- 15. North Anna 4.
Comanche Peak
- 16. Palo Verde 5.
- Cooper,
- 17. Pebble Springs 6.
Crystal River
- 18. Point Beach 7.
Diablo Canyon
- 19. South Texas 8.
Dresden
- 20. WPPS5 2 9.
Farley
- 21. WPF55 1/4
- 10. Ft. St. Vrain
- 22. Wolf Creek
- 11. Grand Gulf
- 23. Yankee Rowe
- 12. Hatch 9
6 a
I i" :';
- 9'.
. d O
e o
ENCLOSURE 3 t
...5.: