ML20205J781

Source Term NRC Perspective

ML20205J781
Person / Time
Issue date:	03/12/1985
From:	Bernero R NRC
To:
Shared Package
ML20205J677	List: ML20205J673 ML20205J692 ML20205J718 ML20205J743 ML20205J760 ML20205J769 ML20205J781 ML20205J794 ML20205J820 ML20205J852 ML20205J894 ML20205K095 ML20205K127 ML20205K168 ML20205K183 ML20207A118 ML20207A145
References
FOIA-87-7 NUDOCS 8704010414
Download: ML20205J781 (22)

Text

.

... ~

SOURCE TERMS .

AN NRC PERSPECTIVE ROBERT M. BERNER0 USNRC MARCH 12, 1985 -

'l; I

pop; 9 9- 7 6/40 8704010414 870330 PDR SIOLL 8 -7 ,

- 1 l

l B

There is a long history in nuclear power plant regulation of the use of accident release or source term information. In the late 1950s the first major reactor accident release and consequence study, WASH-740, was completed. Within a few years a similar model of accident releases was incorproated in the regulatory process through the Technical Infonnation Document TID-14844, which was adopted in Part 100 as an essential part

, of the licensing process. First it was used in the siting calculations but later, as the regulatory process has evolved, the TID-14844 release assumptions have entered many other parts of regulatory practice. It is useful to note here that the deterministic, design-basis-accident regulatory process has never really been completely separate from what we now call severe accident analysis. Ever since the early 1960s we ,

~

have used the design basis accidents and the single failure criterion _

to ensure systems sufficiently reliable to prevent core melt, but we have nevertheless postulated the releases of a core melt, from TID-14844, at crucial points to ensure sufficient capability for containments, radiation resistance for equipment, and other things important to our ,

defense-in-depth regulatory approach.

/ It is not surprising then that in the early 1970s, when the Reactor Safety Study effectively superseded WASH-740, that there was use of its probabilistic risk analysis (PRA) results in the regulatory process. In 1978 EPA joined NRC in the publication of NUREG-0396 which used the accident release characteristics of WASH-1400 to develop the technical basis,for emergency planning around nuclear power reactors. Our present requirements for emergency preparedness in 10CFR Part 50 derive from that -

document.

i

1

_g_

a

. l In 1980 the Commission adopted the policy of using WASH-1400 based estimates to describe the risk of severe reactor accidents'in each Environmental Impact Statement. Other regulatory uses of these PRA analyses grew, such.as in the determination of priorities for regulatory action and estimating costs and

benefits of proposed changes.

And now in the mid 1980s we have a new body of knowledge of the risk of i

l severe reactor accidents. This new work is being published for wide public ,

j and scientific review in the NRC Source Tenn Reassessment, NUREG-0956, as well as the IDCOR reports and other sources. A major independent review of the science of fission product transport has just been released by the American Physical Society. As concluded by the American Physical Society Study Group, there has been considerable progress in the scientific basis and the ability to predict these accident releases. It is indeed appropriate that we consider the careful use of this new ability to revise our regulatory process.

A Integrated analysis of risk are needed to measure the significance of our new understanding of accident releases. In advance of a more rigorous

, quantitative balance, it is useful to do a qualitative reexamination of the J

two plants from the Reactor Safety Study in light of our better current knowledge of accident frequency or probability and our new knowledge of  ;

accident releases. Taking Surry first let us examine four major classes of severe accidents. Transients leading to core melt, including station i blackout sequences, are estimated today to be about the same frequency as estimated in WASH-1400 or even somewhat higher. Operational experience  !

and analysis since WASH-1400 raise this frequency but many post-TMI i corrective actions have reduced it. If we look at the releases calculated ,

, , - - , - - - - - - - , - - , - , . , _ _ . - - - - - - - , _ , - , , , _ . ,_ y y , - ---,- -

for the blackout core melt in BMI-2104 we find the releases of iodine and cesium to be more than an order of magnitude lower and the tellurium a factor of three lower. It is reasonable to project that the consequences. of these sequences will be shown as substantially lower than calculated in WASH-1400.

The likelihood of interfacing system LOCA, Event V, was substantially reduced some years ago by direct regulatory action. And now the consequences, as shown in BMI-2104, are estimated to be lower by as little as 40 percent or as much as an order of magnitude, depending on where the piping ruptures. .

It is worth noting that our experience in evaluating this event in quite a few plants has shown that the consequences can vary over a very broad

range which is controlled by plant specific features. For example, the calculations in Surry don't reveal the significant uncertainty regarding the possible production of more volatile forms of the lanthanides and actinides, as will be noted later in Peach Bottom, but this may be the result of a combination of plant and sequence characteristics that may not exist in a similar plant. The uncertainties in fission product transport behavior, predicting deposition and reemission in the piping, as difficult 1

as the uncertainties in predicting the location and form of system rupture ,

in each possible system of each clant and the accumulation of water in the J

release path. WASH-1400 treated this as an unscrubbed release and thus produced a single sequence which constituted 50 percent of the large release i category. ,

i .

The small LOCA frequency is indicated here as about the same as in WASH-1400.

Reactor coolant pump seal experience tends to increase this frequency but post-TMI corrective measures tend to reduce it. The consequences of small i

LOCA are estimated to be substantially lower. For large LOCA in PWRs the

l 4  :

l

)

estimated frequency can fairly be estimated to be lower than that given in I l

WASH-1400 because of the continuing work in fracture mechanics of piping.

The calculated consequences of large LOCA from BMI-2104 are only slightly lower than the Reactor Safety Study.

What picture of Surry's risk can we obtain from this tentative reexamination?

The risk for Surry was dominated by transients, small LOCA, and Interfacing Systems LOCA;- large LOCA was much lower due to its low probability. Our .

revised picture of risk shows substantially lower consequences on two of the dominant classes and lower frequency on the third. The overall risk is probably going to come out substantially lower. Surry, a PWR with subatmospheric containment probably falls in the lower end of the risk range for PWRs considering their containments, better than an ice condenser but not quite as good as a large dry. And so one can expect substantially lower risk for the reactors with large dry and subatmospheric containments with no clear idea yet for those with ice condenser containments.

i If we turn to the boiling water reactor, Peach Bottom, the severe accident l sequences are arranged ir. tive e.ajor sets; ATWS, Loss of Containment Heat

/ Removal, Other Transients, Small LOCA, and large LOCA. Taking ATWS first -

l we estimate that the frequency is lower considering both the experience since WASH-1400 which would increase our estimate of it and the effects of the ATWS rule which reduce it. The consequences calculated in BMI-2104 really don't appear to be different from WASH-1400 and we have now identified an uncertainty which could increase the consequences. The core concrete  ;

1 interaction, if it achieves very high corium temperatures in the bottom of the containment, can promote chemical reactions which reduce the oxidation l state of the lanthanides and the actinides. This would lead to the i

formation of more volatile forms of these elements which include lanthanum, plutonium, and other isotopes with significant radiotoxicity. )

i The Loss of Containment Heat Removal sequence is separated from the other transients because more realistic sequence analyses which consider current operating procedures and training virtually eliminate this sequence. Also its estimated consequences would be lower. Operating experience since WASH-1400 indicates that the frequency of other transients leading to core melt would -

be somewhat higher than WASH-1400 predicted. Their consequences have not yet been analyzed in Peach Bottom but comparison to similar sequences in this and other plants suggests their consequences will be lower. The same can be

, said for small LOCA sequences and their frequency should be about the same.

The fracture mechanics analyses available now suggest lower frequency for Large LOCA in both PWR and BWR. However, the problem of intergrar.ular stress corr:sior, crann; dr.ngs ir. ar. ancertainty whicn makes one reluctant to predict ar.jtnis; o.: ab: t the sar,e frequency for Large Lan ir. E ns as in' W45n-1400. Ano nere in tne Large LOCA sequence or other sequences with early j containment failure the uncertainty regarding the release of lantnanides and

/

actinides is significant. The estimated releases of iodine and cesium are I

generally lower but adverse behavior by the lanthanug group could cancel out those reductions or even result in a net overall increase in consequences.

i And so the picture for Peach Bottom, a BWR with Mark I containment, is

not so clear; there are some downward trends but the possibility of higher releases of lanthanum and similar species requires more careful scrutiny.

Due to currently running test and analysis programs, a much better understanding

6-of this salient uncertainty should be available as 1985 goes on.

A BWR with Mark II or Mark III containment was not analyzed in WASH-1400,

, but Mark II units are being studied and BMI-2104 analyzed the Mark III unit, Grand Gulf.- From all the results available to date, the overall risk of Mark III units appears to be lower than for Mark I units, with

. the Mark II units falling between.

And so the overall pattern we see forming here is that reactor risk is apsarently substantially lower for some containment types with the recognition tnat tne picture is not clear for otner containment types.

There are many regulatory areas affected by accident releases or source -

terms. They are listed here in three general categories. The first

, category covers those regulatory areas where overall risk is estimated.

By Commission policy each. NRC-published Environmental Impact Statement contains a realistic description of the severe accident risk of that reactor, using the best available knowledge. Although, there are only -

) , a few such statements remaining to be issued on existing plants, we are 1

virtually compelled to use the new methodology in them. In a similar way our work in setting safety priorities and weighing the costs and benefits of proposed changes should take these new estimates of release into account. In most cases they are expected to lead to lower priority or less justification for change, but in some cases the opposite may be true since the revised methodology has identified many changes in the form and place of deposit for radionuclides. And in our work on severe i accident issues with IDCOR regarding existing plants and with applicants .

for review of future plant designs, due consideration of our revised J

l o .

knowledge of accident releases is necessary.

A second category of regulatory areas affected by source terms includes offsite matters, regarding reactor siting and emergency planning. A few years ago the staff prepared a rulemaking package for reactor siting that

would have removed the design basis accident dose calculation from the i

process and used population criteria directly. That work in 1981 and 1982 was based on WASH-1400 risk models and showed that our current siting ,

criteria are sufficiently conservative. The Commission directed the staff te delay the sitir; rulenaung until new source tem.information became available. In emergency ;.lanning tne 10-mile and 50-mile radii

! for ;-eplar.tir.; an: tne other aspects are based on the WASH-1400 risk portra a'. Tnere nas beer. an unfortunate tendency in some quarters to interpret the 10-mile planning zone as a 10-mile evacuation zone, ignoring the basis and advice for graded response to accidents which is

clearly spelled out in the 1978 and 1980 emergency planning documents, 2

NUREG-0396 and NUREG-0654. Now, with revised estimates of reactor accident .

releases indicating significant reductions in the likelihood of early fatality, we may consider a shorter radius of preplanning, less need  !

i 1

! of exercise demonstration, and a much greater emphasis on graded response, concentrating on the population in the first mile or two from the reactor.

And the new estimates are important for collateral issues such as the policy i

regarding distribution of non-radioactive KI for thyroid blocking. Not long ago the staff recomended to the Comission that distributing KI for thyroid blocking is not worthwhile; that recomendation was based on the WASH-1400 source terms which included substantial releases of radioiodine.

Newer understanding shows that the radiciodine releases are much lower,

, with the iodine in the less volatile and more soluble form of Cs!.

4

,-,m.- ,,--,r--- , - - . , - .,- - - - ~ , , , - - - , - -,n,- , - - , - - ,, , - - - , -

, . , b

_ . _ ~ . _ ._ _ __ _________._____________ _ _ _

4 , 8-The third category of regulatory area affected by source terms includes ,

i the performance standards for much of the reactor's safety equipment. I l

Perhaps the most important is the containment performance. In the current regulatory structure the leak tightness required is given in the technical specifications for the plant, it is tested regularly using a costly procedure i spelled out in Appendix J to Part 50, and the whole requirement is controlled i

) on the basis of the original siting dose calculation which is dominated by the large volatile iodine release of TID-14844. While the regulatory process has long dealt with this type of leak tightness for containment, we have only i

recently begun to attend to the various paths for containment bypass which can greatly reduce the effectiveness of containment in severe accidents.

If we turn to other safety equipment in the plant we find evidence of our i

l 20-year focus on volatile iodine as the principal radionuclide in an accident release. Equipment qualification for radiation exposure is based i

! cn the TID-14844 source term. Equipment design and performance specifications 4

l are based on the TID-14844 source term and the focus on iodine release is  ;

1 evident in things such as the extensive use of charcoal filters.

i

/ If we are to use this new ability carefully we should understand its significant aspects and limitations. We can say that considerable progress ,

has been made but we must not forget that severe accident release estimation

$ is not an exact science. Point estimates of releases are not useful alone, j they must be considered in the context of their probabilities and their uncertainties. Consideration of both is a necessity in making policy. The l

i risk description of WASH-1400 was complex, the synthesis of many possible r

l

.g.

I '

j

.i accident sequences, each with a different set of outcomes and releases.

Now with more detailed understanding of the core melt processes and fission product transport, the results are even more complex. We cannot use a simple measure or factor of reduction to gauge our new understanding of accident releases, consequences, or risk. Rather, it takes some integrated analysis with the revised methodology to determine i

what a revised regulatory basis should be. Those integrated analyses are going on right now. The industry IDCOR group has done some of them ,

and the NDC and its contractors are working on others. The work is not complete but, even taking due account of the uncertainties, it appears that significant reductions in risk estimates and a sound basis for many regulatory changes is demonstrable.

4 Let us single out one area of possible regulatory change and consider the

prospects, emergency planning. In 1978 the NRC and EPA joined in the i publication of NUREG-0396, " Planning Basis for the Development of State and f Local Government Radiological Emergency Response Plans in Support of Light Water Nuclear Power Plants." In that document the WASH-1400 model to describe .

LWR risk was used to generate a description of how severe accident risk

, s varies with distance around a typical nuclear power plant. Many factors involved in public protection were considered, the desire to avoid altogether

)

thedirect..no[stochasticeffectsofradiation,thedesiretoreduce .

possible exposures to radiation involving significarit risk of stochastic effects, the socio-economic and health costs of mass evacuation, the i

! shielding effect of typical homes and business buidlings, etc. The authors

) weighed these factors and came up with a body of recomendations which included one recommending preplanning for radiological emergencies for a i

i

,i .

radius of about 10 miles around each nuclear reactor site. One of the most significant technical factors in the choice of the 10-mile planning radius is illustrated here in Figure 1-11 taken frcm NUREG-0396. If you examine this illustration of the conditional probability of exceeding certain whole l l body doses as a function of distance, you can see the relationship for the I

principal doses of interest. The whole body dose of 200 Rem is usually taken as the threshold of life-threatening dose, that is, of early fatality.

The 50 Rem whole body dose is usually taken as the threshold of non-stochastic effects, that is, clinically detectable effects of radiation injury. And the 1-5 Rem doses are the ones recommended by EPA as the appropriate threshold for protective action regarding stochastic effects, that is, the risk of latent cancer resulting from radiation exposure. From this curve you can see that the effects of atmospheric dispersion and the limit to the amount of radioactive material in the release combine to control the lethal reach of the release. As illustrated, life-threatening doses would be possible even close to the plant in only one core melt out of ten, and dispersion effectively limits the outer radius of life-threatening dose to about 10 miles. Remember this is the WASH-1400 model based on the Surry plant analysis, which is quite pessimistic about containment performance l

! and fission product. With that model for reactor risk, NRC and FEMA joined.

l in the publication of NUREG-0654, in 1980 to present guidance for the i fonnulation of emergency response plans. In this figure, taken from NUREG-0654, NRC and FEMA presented the guidance illustrating the two planning zone radii, 10 miles for the plume exposure pathway and 50 miles for the flood chain or 1

ingestion pathway. This figure also illustrates an example response area 1 for the plume exposure pathway, an action zone of constant radius around the plant and a downwind action sector of larger radius. Action might be sheltering, evacuation, or sheltering followed by relocation. Note that

. . . 11 the example response areas shown'in this figure within the 10-mile .

radius, selected on the basis of accident severity, weather conditions prevailing at the time, etc. In severe cases, action for the plume exposure 4

pathway could even go beyond the 10-mile preplanning radius. ' And so the existing rules and guidance already incorproate the principle of graded response to accidents, action graded to match the circumstances.

What change then might we expect if current technical work indicates significantly lower releases for most nuclides in most sequences, albeit with some lingering uncertainties? If we look again at Figure 1-11 from NUREG-0396, we can focus on the 200 Rem curve. Better containment per-formance, or reduced frequencies for each size of release. will simply j

move the whole curve down in the probability scale. If one could demonstrate a reduction of two orders of magnitude by this means alone, one might advance the argument that the combined probability of core melt (typically i .

about 10 / year) and the probability of consequently exceeding a 200 Rem i dose would yield a maximum probability of about 10-7 per year for exposing ,

anyone near a reactor to a life-threatening dose. That might be a sufficient  ;

argument to reduce the level of effort for offsite emergency preparedness

\

/ but note that it would not change the planning radius, it would still ,

i leave the knee of the curve out near 10 miles.

1 \

On the other hand greater fission product retention in the plant in most

sequences will limit the quantity of radioactive material in.the plume and thereby limit the outer radius at which it can still cause a 200 Rem

]

i exposure. The curve will droop and its knee might come in a few miles.

I i

12

,t .

It would take a really dramatic change in the plume inventory to bring the curve down far enough for the planning radius to be only one or two miles.

4

But there is an interesting prospect in this now. In their report the American Physical Society Study Group noted that our improved understanding of accident releases shows both improved containment performance expectations-and greater fission product retention in plant. The two together are likely to produce a very interesting new curve. We do not have that curve yet and ,

it is like the technical basis for any oth!r regulatory change that would be based on the new source term information. We need to finish our homework, to complete the scientific work with due consideration of the uncertainties,

~

and then we will have a sound basis for regulatory change. We are almost

! there, and 1985 is filled with the work of finishing loose ends in research and analysis. We appear to be working on a solid scientific basis and heading along the right track with due care. For success to be solid and clear in the eyes of the general scientific community and the public we must persevere in our important interactions on this work for a little longer. .

l -1 1

i -

J 1

l i

1

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

i

!' 1960 1970 1980 1990 1 SEVERE ACCIDENT i RESEARCH PROGRAM BNL ACCIDENT REACTOR SAFETY SOURCE TERM CONSEQUENCE STUDY STUDY REASSESSMENT STUDY J WASH-740 WASH-1400 NUREG-0956

\

1 g i TID-14844 WASH-1400 SOURCE TERM RELEASE ASSUMPTIONS RELEASE TABLES ANALYSIS METHODOLOGY i v v V 1 SITING REGULATIONS EMERGENCY PLANNING 10CFR100 ENVIRONMENTAL IMPACT PLANNED REGULATORY REGULATORY GUIDES STATEMENTS IMPLEMENTATION DESIGN BASIS ACCIDENT PRA ANALYSES j ASSUMPTIONS 1

) -

SOURCE TERMS IN THE REGULATORY PROCESS 1

l .

j

.q . .

l ,

TENTATIVE REEXAMI. NATION OF SURRRY CHANGES SINCE WASH-1400 ACCIDENT SEQUENCE SEQUENCE i

SEQUENCE FREQUENCY CONS._EQU.ENCE.S .

J-TRANSIENTS INCLUDING BLACK 0UT ABOUT THE SAE SUBSTANTIALLY l PERHAPS SOMEWHAT LOWER t'

HIGHER LOWER SLIGHTLY LOWER, I INTERFACING SYSTEMS LOCA L VERY PLANT DEPENDENT LA UNCERTAINTY SMALL LOCA ABOUT THE SAE SUBSTANTIALLY LOWER I LOWER SLIGHTLY LOWER LARGE LOCA l

,1

~

i s,

3 i ,

TENTATIVE _ REEXAMINATION OF PEAC.H_ BOTTOM i

CHANGES SINCE WASH-Ill00  ;

i -

ACCIDENT SEQUENCE. SEQUENCE

! SEQUENCE lRE00ENCY CONSEQUENCES l

! ATWS LOWER ABOUT THE SAME BUT UNCERTAIN l -

LA RESULTS COULD

{ INCREASE 1

LOSS OF CONTAINENT MUCH LOWER LOWER I HEAT REMOVAL i

OTHER TRANSIENTS HIGHER LOWER i SMALL LOCA ABOUT THE SAE LOWER ABOUT THE SAE GENERALLY LOWER LARGE LOCA BUT UNCERTAIN LA RELEASES COULD CANCEL.0R EVEN

' INCREASE i

,8

• REGULATORY AREAS AFFECTED BY SOURCE TERMS I. OVERALL RISK e REACTOR RISK DESCRIPTION ENVIRONMENTAL IMPACT STATEMENT .

SAFETY PRIORITIES COST BENEFIT ANALYSES e SEVERE ACCIDENT ANALYSIS ASSESSMENT OF Ex1 STING PLANTS i -

USE IN FUTURE PLANT REVIEWS

. l a

sr .

O e

9

• v e

REGULATORY AREAS AFFECTED BY SOURCE TERMS II. OFFSITE e REACTOR SITING DBA DOSE CALCULATION -

POPULATION CRITERIA e EMERGENCY PLANNING RADIUS OF PLANNING RADIUS OF EXERCISE RADIUS OF ACTION TIMING REQUIREMENTS K1 FOR THYROID BLOCKING i

e

1 REGULATORY AREAS

' ~

AFFECTED BY SOURCE TERMS Ill. ON SITE o CONTAINMENT PERFORMANCE -

REQUIRED LEAK TIGHTNESS CONTAINMENT BYPASS 1

e EQUIPMENT QUALIFICATION FOR RADIATION EXPOSURE e EQUIPMENT DESIGN AND PERFORMANCE FILTERS SPRAY SYSTEMS SPACE COOLERS

~

HABITABILITY e

e o

e h

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

n WHAT IS THE SIGNIFICANCE 0F CURRENT RESULTS?  ;

e CONSIDERABLE PROGRESS HAS BEEN MADE SINCE Tile REACTOR SAFETY STUDY BUT SEVERE ACCIDENT RELEASE ESTIMATION IS NOT Aii EXACT SCIENCE s

.)

s CAREFUL CONSIDERATION OF UNCERTAINTIES IS MANDATORY; e THERE IS NO SIMPLE ANSWER OF A, UNIFORM QUANTITATIVE REDUCTION OF RELEASE, CONSEQUENCES, OR RISK (DIFFERENT '

, NUCLIDES, SEQUENCES, PLANTS AND CONTAINMENTS) e THE REVISED METHODOLOGY YlELDS VARIABLY LOWER RELEASES'THAN CURRENT REGULATORY BASES FOR iX)ST ACCIDENT SEQUENCES. IN u

/ GENERAL, THIS SHOULD LEAD TO SIGNIFICANT REDUCTIONS IN RISK ESTIMATES AND A CLEAR BASIS FOR MANY REGULATORY CHANGES

\

\ v

=

1 l

• • ~  :

.lG. 1-11 FROM NUREG-0396 I

1, . a a a s s sig a a a e a i . eiisog i G

=

=

ee "

i

" 1 REM SREM - i 0.1 . .

E . .

.g 4

~ '

to -

3W -

=t 53 w

• • 50 REM EE -

. "4 8*=

gs 0.01 -

. s  : -

5 g

5 -

=

g - 200 REM W

r 0.001 1000

~

1 10 100 l l

1 DISTANCE (MILES) l S

4 3

f i

s

- - .. . , , . .. . - , . . , - . - ..-,-,n., - -. . , , - . , , - . - - , , - , . , , - - , - . , .

l

)

• O  %

l ..

i-a -

= us E!

l- *I en 3 -3...E Ih.g i( I E8Ea!W -

g 2.=

0 E" I.gE r5 1

tie!=, -55.3 18 W ,.

f;5

, f  !

• • IE!Ei5 EIf I ,E 3

.1 -

=

9 '%

/ g)T ~

N

/

/

% .I 1

i

';1

$ / '~

=

/ \ -

ii i

S 4

=

E

/ ,

.i

/ .g.,

4

- ---- s -

i x- +e 6 \

,l---l / v y

-/ -

=!!k=J \3x --

l 6 _ ,. ,. . .

, /

/

~

l, .

r g) }

h}

5 : I 1

k.5 2 a 5 u\tn t g _- _

=

'S 1\- \  !

E e S

-__ - , . - , __i

F , .

'/  %

" UNITED STATES

• E\ <n NUCLEAR REGULATORY COMMISSION M WASHINGTON, D. C. 20555

. E;i'g [N.. /

\; APR 4 198 5

, C.

Docket Nos.: 50-443 and 50-444 3

Mr. Robert J. Harrison President & Chief Executive Officer Public Service Company of New Hampshire Post Office Box 330 Manchester, New Hampshire 03105

Subject:

Seabrook PSA Review

~'

The purpose Seabrook of this letter Probabilistic Safety isAssessment to apprise y(ou PSA)of the status review of the effort and ourNRC decision to terminate this review.

A staff sumary and evaluation (Enclosure 1) of the salient features of the draft review report from our cor.trr.ctor, Lawrence Livermore Laboratories (LLNL), is provided along with the LLNL report (Enclosure P.) constituting the Phase I review of the Seabrook PSA.

Our review of the Seabrook PSA did not identify any safety issues which merit immediate action. The largest contributor, Station Blackout, comprising 4 of the top 22 accident sequences is an issue currently being generically pursued as Unresolved Safety Issue A-44. Any recomendation for action addressing these sequences would be forthcoming through the resolution of this issue. Overall, the review did not identify a discrepancy or error which is estimated, at this point, to significantly change the quantitative results of the PSA. Areas of disagreement and questions are documented in the LLNL draft review report (Enclosure 2).

The review has been impeded by circumstances and problems in several areas. The PSA was submitted to the NRC voluntarily by Public Service of New Hampshire (PSNH) during a period of financial problems surrounding the completion of the Seabrook plants. Since the PSA has not been tied to a specific licensing action, we understand a decision was made to terminate the contract with your consultant at the completion of the PSA. Also with completion of the PSA, you reduced the manpower effort in this area. You did provide resources to support our plant visit in late August 1984, but.,did not provide any further support in terms of supplying all documentation requested, timely answers to questions from our contractor and the staff, and, having severed the contract with your consultant, could not provide an avenue for answers or documentation from the authors of the PSA. We acknowledge these decisions were not made in a spirit of non-cooperation but were in the main dictated by financial circumstances; nonetheless, it affected our ability to provide a complete review.

' Q1.R ' Sl' l W4/9hN1---(54$  % ,

2

o .

~

From another source, the PSA itself, problems arose regarding the contractor's ability to perfonn a review wtitch provided verification of the methods, assumptions, and results of the PSA. More importantly, the review was fragmented due to the structure and documentation of the PSA which resulted in the inability to assess the impact from areas of disagreement on the perception of plant safety, overall risk, and core melt frequency as reported in the PSA.

The Seabrook PSA estimate of overall core melt frequency is about 2 x 10-4/

reactor-year. However, a very large number of sequences contributing to the overall core melt probability with the single most dominant sequence contributing less than 15% to the total and the top 22 sequences contributing approximately 50% to the total. The Seabrook PSA included consideration and quantification of 58 initiating events. 'These were collapsed to initiating

• event groups and the contribution to core melt frequency by these initiator groups. A rough aggregation of accident sequences with similar character-istics to better identify dominant contributors to ccre melt frequency  ;

was performed. Sequences initiated by loss of Offsite Power overwhelmingly dominate core melt frequency followed by Fire-initiated and Small LOCA sequences with significantly smaller contributions.

Since neither the submittal nor the review of this PSA resulted from a requirement, and since issues meriting insnediate attention or regulatory .

action'have not been identified, and impediments still exist that affect the ability to review this PSA, we are terminating the review at this l

' point. The Reference 1 report and this letter is being transmitted to you for your information. Should circumstances change, we will be happy to respond. to the need for further specific and supported review efforts at-that time.

Sincerely,

$ %'f

, George,. . Knightdk, Chief Licens'ing Branch'No. 3 Division of Licensing

Enclosure:

As stated cc w/ enclosure:

See next page

. . . . . . . . .. . . . . ~

m

t . .

o ,

Seabrook Mr. Robert J. Harrison President and Chief Executive Officer '. .

Public Service Company of New Hampsfiire Post Office Box 330 Manchester, New Hampshire 03105 Thomas Dignan, Esq. E. Tupper Kinder, Esq.

John A. Ritscher, Esq. G. Dana Bisbee, Esq.

Ropes and Gray Assistant Attorney General 225 Franklin Street Office of Attorney General Boston, Massachusetts 02110 208 State Hosue' Annex

' Concord, New Hampshire 03301-Mr. Bruce B. Beckley, Project Manager Public Service Company of New Hampshire Resident Inspector ,

Post Office Box 330 Seabrook Nuclear Power Station Manchester, New Hampshire 03105 c/o US Nuclear Regulatory Comission Post Office Bo'x 700 Dr. Mauray Tye, President Seabrook, New Hampshire 03874 Sun Valley Association 209 Summer Street .

Mr. John DeVincentis, Director Haverhill, Massachusetts 01839 Engineering and Licensing i

Yankee Atomic Electric Company-Robert A. Backus, Esq. 1671 Worchester Road

. O'Neil, Backus and Spielman Framingham, Massachusetts 01701 116 Lowell Street Manchester, New Hampshire 03105 Mr. A. M. Ebner, Project Manager United Engineers & Constructors Ms. Beverly A. Hollingworth 30 South 17th Street .

7 A Street . Post Office Box 8223 Hampton Beach, New Hampshire 03842 Philadelphia, Pennsylvania 19101 William S. Jordan, III Mr. Philip Ahrens, Esq.

Diane Curran Assistant Attorney General Harmon, Weiss & Jorcan State House, Station #6 20001 S Street, NW Augusta, Maine 04333 Suite 430 Washington, DC 20009 Mr. Warren Hall Jo Ann Shotwell, Esq. Public Service Company of Office of the Assistant Attorney General New Hampshire Environmental Protection Division Post Office Box 330 One Ashburton Place Manchester, New Hampshire 03105 Boston, Massachusetts 02108 Seacoast Anti-Pollution League D. Pierre G. Cameron, Jr. , Esq. Ms. Jane Doughty General Counsel 5 Market Street Public Service Company of New Hampshire Portsmouth, New Hampshire 03801 i Post Office Box 330

Manchester, New Hampshire 03105- Mr. Diana P. Randall i 70 Collins Street Regional Administrator - Region I Seabrook, New Hampshire 03874 U.S. Nuclear Regulatory Comission

! 631 Park Avenue King of Prussia, Pennsylvania 19406

• • • • • • • • • 'd* e pea me w - e **ew*= * * "* .**44 a****

• e ,

.Mr. Calvin A. Canney, City Manager '

,, Mr. Alfred V. Sargent, City Hall Chairman 126 Daniel Street Board of Selectmen ,

Portsmouth, New Hampshire 03801 Town of Salisbury, MA 01950 l Ms. Letty Hett Senator Gordon J. Humphrey Town of Brentwood U. S. Senate RFD Dalton Road ~

Washington, DC 20510 Brentwood, New Hampshire 03833 (Attn: Tom Burack)

Ms. Roberta C. Pevear Senator Gordan J. Humphrey Town of Hampton Falls, New Hampshire 1 Pillsbury Street Drinkwater Road Concord, New Hampshire 03301

• Hampton Falls, New Hampshire 03844 (Attn: Herb Boynton)

Ms. Sandra Gavutis Mr. Owen B. Durgin, Chairman Town of Kensington, New Hampshire Durham Board of Selectmen RDF 1 Town of Durham East Kingston, New Hampshire 03827 Durham, New Hampshire 03824 Charles Cross, Esq.

Chairman, Board of Selectmen Shaines, Hardrigan and Town Hall McEaschern South Hampton, New Hampshire 03827 25 Maplewood Avenue Post Office Box 366 Mr. Angie Machiros, Chairman Portsmouth, NH 03801 Board of Selectmen for the Town of Newbury Newbury, Massachusetts 01950 Mr. Guy Chichester, Chaiman Rye Nuclear Intervention Ms. Rosemary Cashman, Chairman Committee Board of Selectmen c/o Ryo Town Hall Town of Amesbury 10 Central Road Town Hall Rye, New Hampshire 03870 Amesbury, Massachusetts 01913 Jane Spector Honorable Richard E. Sullivan Federal Energy Regulatory Mayor, City of Newburyport Commission Office of the Mayor 825 North Capital Street, NE City Hall Room 8105 Newburyport, Massachusetts 01950 Washington, D. C. 20426 Mr. Donald E. Chick, Town Manager Mr. R. Sweeney Town of Exeter New Hampshire Yankee Division 10 Front Street Public Service of New Hampshire Exeter, New Hampshire 03823 Company I 7910 Woodmont Avenue Mr. William B. Derrickson Bethesda, Maryland 20814 Senior Vice President Public Service Company of New Hampshire Post Office Box 700, Route 1 l Seabrook, New Hampshire 03874 l

.  ?

C' ENCLOSURE [

. Staff Summary Review of the Seabrook PSA '

~

8ACKGROUND , -

~

Lawrence Livermore National Laboratory (LLNL) conducted a review of the Seabrook Station Probabilistic Safe'y t Assessment for the Reliability and Risk Assessment Branch (RRAB), Division of Safety Technology. This probabilistic safety assessment (PSA) was performed by Pickard, Lowe, and Garrick Inc. for Public Service of New Hampshire (PSNH) and Yankee Atomic

, Electric Company (YAEC). The' PSA was also provided to the NRC for its

~

information.

The review of the Seabrook PSA was performed by a project team composed of personnel from NRC staff, LLNL staff, subcontractors and

~

consultanIs'.'

} ..

. OVERALL EVALUATION OI THE SEABROOK PSA ' ~

An examination and review of the dominant sequences identified in the PSA was performed in light of the various co3cerns t' hat have been identified in ~

Chapters 3 and 4 of Enc.losure 2 for the internal and external events. This examination was necessarily 1.imited by the inability to reconstruct and reevaluate the event trees with consideration of these concerns, and to then compare the new results to the PSA results. , It w3) not possible'to perfor'm this evaluation and comparison because of the lack of information and answers to technical questions. '

i Nevertheless, the overall findings of the LLNL review resulted in the judgment that the dominant sequences presented in the PSA generally appear to be

~

4 0

reasonable (although conservative) in a quantitative sense. That'is to say 4

that one could expect to find that the quantitative results of,a hew

~

evaluation would.not find the probability of core melt to be significantly larger than described in the ' S4, P because of the generally conservative quantitative approaches and a'ssumptions incorporated in many places in the l PSA. A summary of the sequences and review findings is provided in the following section. No significant omissions were found in terms of an overall contribution to the frequency of core melt. Several modeling errors were found that indicate an incomplete or different understanding of interactions between plant sytems or human beings (operators) and plant systems; these are described in the internal events section of Enclosure 2.

While LLNL considered it likely that a reevaluation of the sequences where these differences were identified would not significantly affect the overall ,,

core melt frequency, it may provide different qualitative results regarding .

the understanding of dominant contributors and therefore different insights. However, the significance of these differences could not be completely assessed. 0

SUMMARY

The PSA results for core melt probabilities were 1.6E-4 per reactor year (RY) for internal events and 6.2E-5/RY for external events,, for a total of 2.3E-4/RY.

External events were dominated by contributions of 2.9E-5/RY from seismic events and 2.6E-5/RY for fires. The scope of the review did not include a review of containment response or offsite consequences nor extensive requantification.

R e

g = - - . - - - . - - - - - - - - , - - - - . .- ----e-, - - - , -- - - - - - - - - - - + --r- g, , , - - - - - y,--e-.--.---

3- .

4 . .

A very large number of sequences contribute to the overall core melt frequency The ,Seabrook PSA included consideration and quantific,ation of 58 # vents.

• Tables 1, 2, and 3 display the' accident sequence contributions to core melt frequency grouped by' initiating event categories and provide a summary of the top 22 sequences.

However, by collapsing these sequences to categories of sequences with

, similar characteristics, the resultant sequence-frequencies and contributions to overall core melt frequency are presented in Table 4.

• e eq G

$ 0 m

• 9 *

• s.

.h 0

1

. 1 e

e e

,- .-w - . , - ,

. , . -e- , - . . -.

.. e .

o TABLE 1 CONT'RIBUTIONSOFACCIDENTSEQUENCES o ,,

['

. GROUPED BY INITIATING -~ EVENT TO CORE MELT FREQUENCY, -

.u- , i

I.nf tiator, Group Percent '

Contribution Contributors Percent Contribution

.. Loss of Co'olarit ~11 i Small LOCA- -

,{ .8

.Inventorf . ;,.

_e 0thers 2 3 Transient Events a e Reactor trip (excluding .e Turbine trip

• 6 support system '4 faults) 31 o Partial loss of feedwater 4 e Steamline break -

4 e Excessive feedwater 3 e MISY closure 2 e Others 8 Comon Cause e Loss of offsite power

, ' Initiating 58 '

29 Events e Seismic events .

12 e Fires 11 e Other support system faults 3 e Other external events 3 Total 100

. 100 ....

. Core Melt Frequency:

Mean 2.3 x 10-4 per reactor year l

s 6

1

, , . - - - , - , _ _ , . , , , - - . , - - - -,-,----,-,a m- - - - - - - - - , - - - - , - - - . - - - - - - -,-----,m,----.,------w--,,- ,, . - , - - r.--, , . - - -- , - - - , , - -

d TABLE 2 .

i CONTRIBUTIONS OF SEQUENCES GROUPED BY' INITIATING EVENT TO FREQUENCY OF RISK '

i ,

SIGNIFICANT RELEASE CATEGORIES AND CORE HELT i .

• L.

• g Release Category frequency Centributleas .

levents per reactor yearl -

g,,,p,yg 4 lattiattag

Event !altlatnag Event Large Small tag, Fretutact

6reep Baseest Overp'r essure l'L' Centributleas

, Coatalement Coatalament g g,3,4 ,gg, Overpressure - Centalament levents pu Sy ass By as' Ne vapertsattem g,,,,g .

IY T4r vaperiaatten 55 ****' 7'I

.. s selease T3, selease 53  ;

Less of La'rge(OCA* * * * *

• Caelaat small LOCA * * *

• 1.1-6 1.4 6

Inventary laterfaclag Systems LOCA .1.8-6

• 2.5-6 1.7-5 2.0-5

• 0 0 0 0

. Steae Generator Tube Ampture *

• 0 1.3 6 1.0-7 8.2-7 *

• 6.5-7

, 1.7-6 i '

! General teacter Trip

• Transients Turtlae TrtP * *

• 6.16 6.7-6 1.3 5  :

Less of Main Feeesater *

• 3.9 6 3.3-6 1.0 5 1.0 6 s,3-6 * **

1 , Partial Feedwater Less a 3 * .*

l.1 5 j  ; . Encessive Teeevater * *

• 5.0-6 2.5-6 786-

? . Less of Condenser Vacous * *

• 2.5-6 2.7-6 5'.7-6 M5IV Closure * * *

• 8.9 7 1.1 6 Less of Prieary flow = a *

• a.9 6 5.0 6

Steam Line treak * * *

• l.2-6 1.5 6 2.4-6 Mala 5 team tellef Opens * *

• 6.9 6 7.36

)

• 5.0-7 1.4-7

- 7.e-7 Cesnea Cause Less of offsite Fever *

• l' Initiating 6.8-6 5.5-5 4.9-6 ' 1.5 6 6.9 5

'Less of Sae DC Sus * * * *

• Events ' Less of Service Water * *

• 1.7 6 2.3 6

- support Less.of Campeneet Coelleg ~* * * *

< 2.5-6 0 2.5-6

, Systee -

o 1.4 -6' 0- -1.4-6 Favits

• 2

• i

- Esternal ^ 3elsetE Events (total) 5.8-7 1.7-5 Events Fires Itetall 5.3-7 4.0'-6 2.9-6 2.8-6 2.8 5 i 6.6-7 5.3-6 2.0-5.

- Fleed (totall . * * .

2.3-7 1.1-6 2.5 5 Truct Krash *

• 1.4-6

• 3.9-6 1

1.5-7 1.4-6 1.4-7

• 4

• 1.8 6 Total ,

2.4-6 1.8-5 1.0-5 8.0-5 5.8 ,5 6.0-5 2.3-4

.\

1 *Less than 15 centributies.t'srelease category frequency, f

Holt: Empemential metatlee is ladicated la abbreviated fere; l.e.

1.1-6 e 1.1 a 10-6, e

i . .

, e - .

3

• 1 ea . ,.

,- , _ _ . - .-...s.

1 .

- - - ^ '

TABLE 3

SUMMARY

OF ACCIDENT SEQUENCES WITH SIGNIFLCANT* RISK AND CORE MELT FREQUENCY'C0tiT smeet I of 2 Sequence aanklag Instlatlag Additlemal Systes failures / Sequence .

E at "u== Actla5 Resultlag Dependent failures frequency tatent- Early j -

. leer reacte ye4ri yC u.aism meaith

! alst ' asst

' Less of Offsite Opsite AC Pomer, lie Recovery of AC Power Campement coollag, high pressure makeup Power

~ '

Before Core Samage 3.3-5 1 1 *

IECCS). reacter coelaat pusy seal LOCA. ,

4 "6

contalament flitraties and heat removal. -

Less of Offsite Service lister. les Recovery of Offsite easite AC power, component coollag, high g.2-6 Power Feuer 2 2

• i -

and low pressure makeup (ECCS) reacter *  !

i cpolant pump seal LOCA. coatalament filtratten and heat removal.

, Small LOCA tesidual Meat Removal tiene.

• i 8.g 6 3 * * ,

j Centrol teen liene Component coollag, high and low pressure Fire 8.7-6 4 3

• makeup (ECC5), reacter coelaat pump seal LOCA. contalament filtratten and he%t * '

reeeval. -

i

• l

Less of Mals 'Selid State Protectles system Reacter trip. emergency feeduater high 1

! Teeesater - 8.3-6 . 5 4

• i and low pressure makeup (ECCS), contay. -

1 aeat filtraties and heat removal.

I Steam Llae aperator Fellere to Estah11sh Long Tere treak laside 5.6-6 6 *

• Contalement .

lee:t Bemova) - . .

/ -

Seacter trip #

7-- L'Ceeling-

• Illgh and low pressure askeup (ECCSI. 4.6-6 7 5 * '

s reacter coelaat pump seal LOCA. costala-

• seat filtration and heat removal. -

Lcss of effsite Trela. A ensite Feuer. Train 8 Service i Power teater. Ile Recovery of AC Power Sefore Trala S easite power, component coollag. 4.4-6 8 6 *

- high and low pressure makeup (ECCS). -

Core temage reacter coelaat peep seal LOCA. contala-f

. . meat flitratten and heat removal, Ecss of effsite Trala B easite Power. Trata A Service

Pomer ifater.*to Recovery of AC Peuer Befera Train A easite power. compement coollag. 4.4-6 3 7

• Core Samage high and 16w pressure mateep (ECC5).

i reacter coelaat pump seal LOCA. Centala-meat illAraties one heat removal.

t PCC Area Fire liene Component colillag. high and l'ow pressure. 4.1-6 le 8 *

} makeup (ECCS). reacter Coelaat pump seal i

' LOCA. contalement flitratlee, and heat removal.

'megligl' ele centrlhettee to rist.

Molt: Taponeettal motetten is ladicated in abbreviated fore; 't.e. .

3.3-5 = 3.3 a 10 . -

Ogesp122283 s 1

i .

4 ..

i 9 .

) .

TABLE 3' } continued) i ,

l .

i - 5 meet 2 et 2 t , Sequence taaklag

} lastlassag Additlemal System Failures / Sequence ,

i Eweat ha Actfees Resultlag Dependent Failures Frequency E*"**"

{ to I*'II f

, i (per reacter year) Mealth Ilealth 9

alsk alst i partial Less of .Campement Caeltog esala Tceewater  ; .. . Nigh and low pressure askeup (ECCSI. reacter 3.84  !! , 9 *

{ , coelaat pump seal LOCA. contalement filtra. ,

ties, and heat removal. .

Catte Spreadlag neae Component coollag, high and low pressure

. Reen Fire - 3.54 12 le *

• makeup (ECCS). reacter coolant pump seal .

, ., LOCA. contalament filtration. and Mat removal. .

Less of One DC s,

Emergency Feee ater. No Recovery of Sleed and feed coeltag. Trata A , 3.24 * *

Bus Emergency or Startup Feeduster 13 costalement filtration and heat removal. -

Reacter Irly Operater Failure to Estahlf sn Long Ters mese.

Neat Aenewal.

. 3.04 14 *

• Turtlee Trip t- ; 1 t Coeling Nigh and low pressure makeup (ECCS). 2.84 15 11-

• i reactor coolaat pump seal LOCA. Centale.

meat filtraties, and heat removal.

~

j Less of Service me Camponent coeltag. high and law pressure

Water 2.3 4 16 12

• askeup, reacter coelaat ppap Seal LOCA.

, costalament filtration, sad heat removal. ,

! partial Less of Operator Failure to Establish Long Term Mone. 2.34 * *

Feee ater -

Neat Aseeval 17 4 Turbine sulldlag easite AC power. Ile ascovery of AC peser Offsite power, component coollag, high l Fire Defere Care Samage

. 2.34 18 13 *

' . and low pressure askeup (ECCS). reactor j

• coelaat pump seal LOCA cantalament filtra-

, , ties. and heat removal. '

e i Small LOCA Traja ) Safety Features Actuatten.

j Trata A high and law pressure askeup and 2.24 19 * *

• Trels A Sesidual lleet Bemewal residual heat removal; train 8 costalament

{ ,

filtraties and heat removal. .

! e Small 19CA Trata A Safety Features Actuelles.

1 Trala 8 high and law pressure makeup and 2.24 20

• i Trels a tesidual uset aseaval

-

• residual heat removal; train 8 costalement

• . filtratten and. heat removal.

Ivehtee Trip teacter Trip. Failure to Manually Scres

Functional laatllity to provide adequate 1.94 25 * *

~

Reacter and to Effect Emergency Beratles high pressure makeup.

I:Lerfacing deae. Low pressure makeup, residual heat

. Syste'es LOCA , 1.4 6 27 14 1 reeeval, costalement lselatles and filtratlea.

l

, *Negllylble contributlep to rist.

• j asr. e.. . . -

L_ _ _ _ _ _ _._ _ __ - ._ - - - - - - - - - -

3 * . .

. ~ .

TABLE 4

'. Number of Contribution Initiating . .Aggre' gate Sequences. - Total Frequency /

Event Group Frequency /RY to Overall CMF in Top 22 in Top 22 Small LOCA' $2x10 5 , 9% 3- $1x10 5 Loss of Offsite $7x10 5 30% 4 $5x10 5 Power Seismic $3x155 13% 0 i

j Fire $3x10 s 13% 4 $2x10 5 i Component Cool- six10.s 4g 3 sixio.s

! ing Water System j - Failure .

!L ATWS ~1x10.s 4% 2 six10 5

, Transient (other $1x10 5 4% 3 six10 5 i

than LOOP) followed ,

by Failure of.Long .

! Term Heat Removal

• 1 -

Interfacing Systems s2x10 5 +1% 1 $2x10.s .

LOCA s Other Transients $4x10 5 19% 2 $5x10 5 Other External O Flood, Truck Crash $6x10.a ~3% 0 e

O O  %

9 6 C

i .

4 .

l ,

}

SEQUENCES INITIATED BY INTERNAL EVENTS

. The exteilt and type of internal event initiators and their treatment is generally reasonable and consistent with those considered in other PRAs.

The event tree models in most cases correctly represented the accident sequence phenomecology assumed in the PSA; however, several areas of disagreement with the assumed phenomenology were identified. There is also a concern that the requirement to have each event on an event tree

. independent of the others has resulted in large and very. complex trees which are difficult to follow and analyze. In addition, the large number of

~

sequences, on the order of 100 times as many as in previous PSAs, effectively

~

fragmented many accident scena,rios which could' be simply described as single

• p ,

sequences into a large number of sequences,.so that the usefulness of the -

. event tree sequences as a means to obtain engineering ins'ights was lost. I

! Although many deficiencies in these trees are described in the text of -

Enclosure 2, it was not possible to provide a pr'eliminary assessment of the o i quantitative effect on the PSA results, primarily because of the complexity of the trees in the PSA and their use of proprietary codes to perform the

+

quantitative evaluations. -

LOSS OF OFFSITE POWER .

Sequences initiated by a loss of offsite power, taken'c'ollectively, dominate the overall core melt frequency. Of the sequences comprising 30% of core melt frequency, 4 sequences are in the top 22 sequences with.a 22% ,

i

. .I i

i

-_ ~ -. ... . _ - _ - _ _ _ . _

- - . - - . -_. .~ . _ _ _ _ .- . - - . -

t

j contribution to CMF, the highest of which, contributes 15%.

, In this sequence, loss of offsite power is followed by failure of onsite AC power (2 of 2 diesel generators failing) with.a resultant Reactor Coolant Pump Seal failure (RCP Seal LOCA) due to loss of Component Cooling Wat and High Pressure Makeup. The other three sequences initiated by loss of offsite power involve total Service Water System failure and combinations of one train of Service Water failure with failure of the opposite train of onsite -

AC power. These sequences also lead to a RCP seal LOCA, though the

, individual sequences have lower frequencies due to the differing system failures in the accident progression. The frequency assumed for loss of offsite power is generally consistent with those in other PSAs, but two assumptions in these sequences cause concern. The first is that the RCP seal LOCA will occur immediately upon loss of all AC power and that the leak rate is 20 gpm per pump. Based on previous analyses (Millstone 3 PRA, for 4

example) it is not reasonable to assume that a leak will occur immediately.

A more realistic time frame is on the order of 30 to 60 minutes before the' seals fail. At that time, previous analyses assume that the leak rate will rapidly accelerate to a much higher rate (s300 gpm) and subsequent total failure since the seals will )e in a degraded condition'under high mechanical and thermal stress. The extremely low flow rate assumed in the j

Seabrook PSA extends the decurrence of core uncove'ry and dafnage to a much later time than is' considered realistic. Under the assumption of a higher flow rate, core damage is more realistically expected to occur about two hours after power is lost an'd unrestored. Justification for the assumptions in the PSA was not provided and the overall effect of the concerns regarding

7 ' ' .' .

l them have not yet been quantitatively assessed. LLNL considered it likely

. . that a reevaluation of these sequences where differences were idehtified

. would not significantly affect overall core melt frequency, h'owever, it may .

result 'in a different understanding of the tiaing of core melt and the resultant distribution of sequences over plant damage states.

i .

SMALL LOCA ,

Of the sequences initiated by a small LOCA comprising 9% of overall core

] . melt frequency, 3 sequences appear in the top 22 with a contribution of about 4% to CMF. The highest frequency sequence involves. failure of Residual Heat Removal (RHR) as a source of low pressure injection following manual depressurization with the other two involving combinations of a Safety Features Actuation (SFA) Signal train failure and failure of an RHR .',,

train. The SFA failure affects containment filt' ration and heat removal. O f ' .,

j . concern are the two separate values assumed for a small LOCA, representing breaks that can be isolated and those which are nonisolable. It has been l 1 ..

[ generally recognized in previous PSA analyses that isolable breaks do not i

significantly contribute to overall small LOCA frequency due to the amount l i

1

of time available for the oper,ator to isolate them prior to the need for emegency core cooling. Therefore, the concern is with the nonisolable break

] frequency assumed in the Seabrook PSA which is lower than those found in

! various other PSAs and PSA reviews. This frequency is based on the ability to isolate a random RCP seal LOCA at a given plant. The Seabrook plant does i

! not have primary loop isolation valves, thus an RCP seal LOCA would be i .

I 1

• 4 E

\. .

L .

8-

~

, cons,idered nonisolable. Since this was not considered in the PSA frequency

~

on nonisolable small LOCAs, it appears tha this initiating event has been underestimated 'by as much' as a factor of four when compared to'other data

• sources (e.g., ANO-1 IREP anal'yses).

Component Cooling Water System Failure (CCW)

.The other category of sequences of interest involve failure of the CCW -

System. These three sequences, which involve resultant RCP seal failure,

, have an aggregate frequency of *1x10 5/RY and contribute 5% to overall CMF.

)

This value of 1x10 5/RY is somewhat lower than those dete*rmined for other PRAs for similar plants (e.g., Zion, Indian Point).

It has not yet been determined whether the particular configuration of the CCW system at Seabrodk has design features which would explain this difference, one aspect I. .

of the PSA worth noting is that while the st'udy considers a total loss of '

the CCW System as an initiating' event, it does not consider loss of a single train. The basis provided is that if a single train is lost, the reactor will not immediately trip and the operator can procee8with an orderly shutdown, thus it is not an initiating event, s

Anticipated Transients Without Scram ATWSsequencescontributeU%tooverallcoreme1[ frequency. Two sequences, with collective frequency of *1x10.s/RY, were among ,the top 22. sequences.

One sequence is initiated by a loss of main feedwater with subsequent mechanical failure of the Solid State Protection System (8.3x10 8/RY) and 1

l

. 1

. g. *

.' /

L . .

l - -

the other is initiated by a Turbine Trip followed by failure of automatic

{

.and manual scramming of.the, reactor (1.9x10 s/gy), .

, Using the new AWS rule to prdvide guidance and information, some problems with the AWS event tree were identified, in areas such as operator i recovery and credit.for bleed and feed.

The PSA gives credit to the possibility of operator action to effect s.anual

i. reactor scram following automatic scram failure. This action, however, is not modeled explicitly on the tree; it is applied directly to the failure of I~

RPS leading to A WS. It is valid to consider this type of recovery, but an

) action of this import should have been included explicitly on the tree'.

i .. ., .

It is also important to note that this recovery action can only be applied ., ,

to electrical failures..of the RPS, so that RPS failures should have been * *

. ~

l . divided into electrical and mechanical failures as stated in the ATWS rule, d'

l The assumption that it is necessary for the operator to shut down the l reactor after the initial phase of the A WS is reasonable and consistent with the A WS Rule. However,',the Seabrook PSA assumes that this action must

{ be taken within ten minutes, which appears to be conservative. Once the initial phase of the AWS is over, the power equ11} bra.tes at the secondary heat demand and the plant will operate safely for extended periods of time.

This was supported by many analyses and a simulator run performed on the '

) Seabrook simulator during the plant visit of August 29-31, 1984. It appears l

q i

t D

, -wn.wn- ,,.,-,-m.., --,-,._w,,- ---------,7,._w-, r, ,- ,---m. - - , ,..-.-_n,,.,a.= ,n, n a ,. y.. , , ,---,,.,m- .

t . -

1 . .

j . , .

j that the time frame is more on the order of 60 minutes or more, except when a primary safety valve sticks open or the ATWS tree is entered.from a LOCA initiator.

j In this case, a 20 minute time frame is more appropriate than the

• 4 10 minute operator action tisi assumed in the Seabrook PSA. ,

l The PSA also assumed that it is possible to mitigate an ATWS by using bleed-l and-feed with HPI alone if emergency feedwater fails. This would

• theoretically provide boration to shut down the reaction simultaneously with

) ,

bleed-and-feed cooling. This method has not been considered in most other i

) ' PSAs, and takes an inordinately large amount of credit fo'r the ability of HPI to provide flow at operating pressure. Also, there would be much greater j amounts of heat to be removed through the PORVs with makeup flew than for a normal

• bleed-and-feed scenario' It is not clear that heat removal .

and reactor shut down could be accomplished 'under these conditions without ,

the emergency feedwater system.' Therefore, it is more appropriate to i

conclude that all sequences with failure of emer,gency feedwater would lead a '

to core melt and may increase the ATWS sequence frequency. This may be '

offset by the other conservatisms assumed in evaluating ATWS events. The .

overall quantitative effect on ATWS and overall ore melt frequency has not yet been evaluated.

l i

i Transients with Failure of Lono Ters Heat Removal , ,

l The contribution to overall core melt frequency from this category of 1

1 sequences comes from a large number of sequences, only'three of which appear e

e 1

u _ _ _ __ _ . _ _ ..- - - - - - . ~ - - - - ~ - - - - - - - - - - ~--- - - - - ' - - - - -

- . - . ~ .-- - .- _.._ -... .- ._ . _. . - - - - ___- _

s' l .

. ~ .

1 in the top 22 sequences. Failure of long term heat removal (high and low i .

pressure makeup in the RHR mode configuration) is dominated by common cause

, failures'(e.g., maintenance unavailabilities) and independent ~ hardware

• ~

failure's. These failures involve failure of ' pumps to start and valves in j RHR trains and heat exhanger valves failing to open. ,

?

1 J

i The category of "other transients" is comprised of an extremely large number

, of sequences of the remaining sequences analyzed as contributors to overall J

' core melt frequency. This category has not been decomposed to obtain i

)'

i

- insights in this initial review. . ,

The functional success criteria used in the PSA were generally found to be reasonab1e,'with some exceptiops. These criteila, however not clearly 3 .,.

I stated in many cases, included both conservative and optimistic examples and

, ,in general appeared to be inadeguately documented.

l l

1 The review of the failure rate data used in the PSA consisted of a comparison of the individual component failure rates with other snurces and l 2

a review of system failure probabilities and unavailabilities. The data

, l

values presented were found to be reasonably consistent with other data sources available to the review. A comparison of.. system failure probabilities with other sources of similar data revealed that these values were reasonably

!. consistent with the other sources.

1

1 i

i l .

I

1 5',  ?

12 - -

'~

\

Cons,ideration and treatment of dependencies and common cause failures in the

~

. I PSA were evaluated in the review in three categories: common cause Initiating events.* int'ers'ystem dependencies, and intercomponenl. dependencies.

• The methodology used in the analysis appears reasonable and appropriate. No l important omissions in the treatment of dependencies were identified by the review. The treatment of common cause data was of some concern because of the exclusion of passive components and the use of very low beta factors (i.e. , .

factors to account for common cause failures) for some components although l no instance was identified that would significantly change the results.

EXTERNAL EVENTS -

The external events considered in the PSA are earthquakes, fires, aircraft I acciderits, internal .and external flooding, extreme winds, and turbine missiles ..

l It is important to note that sequences initi'ated by the various external evNts , .

I j (not including LOOP) were not s'Ignificant contributors and that only fire initiated sequences appeared in the top 22 sequences. This is not entirely 0

j consistent with other PSA findings (such as those for Zion, Indian Point, and Millstone 3).

1 1

l l -

The methodologies used in the detailed assessments are generally reasonable and consistent with the state-of-the-art; however, there were notable disagreements in several areas.

l 4

)

l i

I e

I e

--e m . -,m---,--r--,---..--.,~r--- we- - - - , , , - - _ _..- - - , --.-,._,-.---.-,wi,e,,--.--..e._- m, w,yrr -., ,,,,.---,-i.

~

l .

i

,s . '

i (3.5x10 8/RY), and Turbine Building Fire foliowed by loss of all AC power (2.3x10 */Ry).

All of these sequences result in loss of component cooling.

l

. RCP seal,LOCA, loss

• of. ECCS, and failure of containment filtration and heat .

removal'. The fire analysis performed for the' PSA, based on current fire i .

protection guidance, appears accurate and valid and the frequencies of the

fire induced initiating event which include system failure appear reasonable.

I The contribution to core damage due to fires at the various locations analyzed .

falls within the range of those calculated from other fire assessments at f . nuclear power plants (IE-4/RY to IE-7/RY).

1~

j The analysis of fire propagation for determining the loss of safety related

~

l functions is rigorous and explicit and the considerations of fire phenomena,

! ' ~

i material properties, fire deteption and suppression, operator action, and '

I modeling uncertainty at each fire location were reasonable. *" .

i '

4 s

There is a concern, however, about the manner in which the fire induced l initiating events are processed through the plant matrix. It appears that c

1 these initiating events, which already include component or system failures, l 1

l are being incorrectly combined with auxiliary and front-line event trees

that have not explicitly considered these same failures. This concern has l yet to be verified and evaluated. . ,2 INTERNAL FLOODING ' '

The PSA treats internal flooding primarily qualitatively, with a l .

O

C A

• 4 assigned to key parameters in the analysis. A review of individual parameters in the analysis and a comparison with the interim Seismic Hazard Characterization P'rogram lead to the qualitative conclusion.that the .

hazard analysis' results may be optimistic and the uncertainty underestimated. _

The methodology used in the PSA for determining the seismic fragilities is 1

j appropriate and adequate to obtain a rational measure of the capacity of ,

the structures and equipment.

Based on a preliminary review of the l

results of the PSA, the mean frequency of core melt value of 2.89E-5 per year appears to be high relative to the optimistic hazard curves used in the analysis. Calculations indicate that the capacities of the key components at j the SSE value are low and generally less than values determined for other PSAs such as Limer.ick and Millstone. In addition, comparing fragility parameter .',,

l 1

valuesofS'abrook.andotherPWRs(newandold),thecapacityvaluesofequiE.

e ment considered also appeared to be low for Seabrook. Based on experience with

past PSA reviews and information gained during the site inspection, the 4

capacities of0the dominant contributors, though they have not been quantita-I tively reevaluated, are likely not to be as low as indicated.

1

~

i FIRES i . . -

Fire induced sequences contribute $13% to overall-core, melt frequency with 4 ,

, sequences appearing'in the top 22 sequences contributing $2x10 8/RY (9%) to CMF. These sequence initiators are Control Room fire (8.7x15 8/RY), Primary Component Cooling Area Fire (4.1x10 '/RY), Cable Spreading Room Fire l .

_ _ _ . _ _ _ _ ___ _ -_ .-__-._. _ .-_- _ L__ __. .___ _ L __ . _ _ _ _ _ _ . . ~ .

. _ . -- - -. - - . - __ - --_ =- _

l

. ~... j

= *

. . }

k C

- 1 s ,

EARTHQUAKES

j .

, SEISMIC HAZARD .

AND SEI,SMIC FRAGILITY ,

Though ' sequences initiated by seismic events ' contribute $135 to overall core i

melt frequency, none of these sequences appear in the top 22 and the highest j individual sequence frequency is $4x10 */RY. The occurrence of a seismic j event could initiate a sequence at the Seabrook plant in any of several .

t ways.

Failure of the offsite power transformers (Reserve Auxiliary '

i Transformers and Unit Auxiliary Transformers), or switchyard equipment. -

would result in offsite power to the plant being lost. Also, at higher

~

accelerations, a failure of the reactor internals could cause the control i

rods to jam and not position for reactor shutdown. It is also possible that 1 -

anearthqualIecouldcausethepnchorboltstyingdownthesteamgenerators

' i 1

) or reactor coolant pumps to fail, thereby permitting the equipment to tilt,'",,

j ,possibly resulting in a break at the primary cooling system piping. Other i

failures such as instrument buses, that would cause a transient type event.

l

! nuld occur at accelerations higher than those that would already have a

i caused a loss of offsite power and would result in the same sequences.

4 -

a The methodology used in the evaluation of the frequency of the seismic 1

j hazard at Seabrook is consistent with the state-of.lthe-art of commercial PSAs. However, there is disagreement with numerous applications.of the

).

methodology in the PSA. In particular, the procedure 'used to perform the -

uncertainty analysis failed to document the choices made and the uncertainty 4

l -

4

• j .

1

, - - . - - , _ , - - - - - . , , _ - . _.--.-n----.-__...,,-- - , - -

i .

.- l

- - l

, A=

. ,, ,7 quan,titative analysis performed for a turbine building and switchgear room flood. The qualitative analyses consider all internal flood sources for

~ .

l .

each defined location, in'cluding floods caused by fire protect'io'n~ equipment

• and sources from adjacent locations.

In all these locations.it was i

concluded that the risk due to flooding was insignificant.- The quantification of the turbine building flood appears to be reasonable and adequate as was

/

the qualitative treatment of flooding in other locations.

a 0

1 .

i .

i 1

8 1

g s

l s i

1 . I o 1 I

'm r

,l l

f *1

)

I i

f

g-y  %, UNITED STATES

,; a NUCLEAR REGULATORY COMMISSION c f WASHINGTON, D. C. 20555 NAY 8

% no . . .o# 1985 Docket Nos.: 50-443 and 50-444 i

Philips Ahrens, Esq. --

State of Maine Department of the Attorney General State Mouse Station 6 Augusta, Maine 04333 ,

Subject:

Meeting Notice; Docket Nos.: 50-443 and 50-444 My letter of April 15, 1985, same subject, has raised some questions prompting me to clarify what will be my committed notification actions. When I schedule a meeting that is less than a calendar week away from the date on my meeting notice, I will take the necessary actions to advise your office of the scheduled meeting as well as other parties to the Seabrook proceedings.

Sincerely,

\

Victor Nerses, Project Manager Licensing Branch No. 3

• Division of Licensing cc: See next page j i

k

%Y y a ;; gg-7 -

Lt.u.

I RUN DATE: 07/15/85 OFFICE OF NUCLEAR CEACTOR REGULATION 67, PAGE:

TECHNICAL ASSISTANCE PROGRAM SUPPORT SYSTEM

SUMMARY

OF FY 85 COSTS FI] 8: A3778 TITLE: REVIEH OF PRA FOR SEABROOK NPP PROJECT MANAGER: LYON, H. DIVISION: DOSI BRANCH: RSB COSTS REPORTED FOR EACH MONTH CUMULATIVE COSTS THROUGH EACH MONTH BUSINESS DOE ACTUAL BUSINESS LETTER DOE ACTUAL FINANCIAL DOE LETTER FINANCIAL REPORT INFO SYSTEMw VOUCHER DOE REPORT INFO SYSTEME VOUCHER OCT 7,487.00 7,487.51 7,487.51 OCT 7,487.00 NOV 19,395.00 19,394.74 7,487.51 7,487.51 19,394.74 NOV 26,882.00 26,882.25 26,882.25 DEC 18,662.00 18,661.99 18,661.99 DEC 45,544.00 45,544.24 45,544.24 J 7,3 0.00 18,557.80 18,557.80 JAN 45,544.00 64,102.04 64,102.04 FEB 7,107.00 7,107.03 7,107.03 FEB 52,651.00 MAR 71,209.07 71,209.07 0.00 0.00 0.00 MAR 52,651.00 71,209.07 71,209.07 APR 0.00 0.00 0.00 APR 52,651.00 MAY 71,209.07 71,209.07 0.00 0.00 0.00 MAY 52,651.00 71,209.07 71,209.07 JUN 0.00 0.00 0.00 JUN 52,651.00 71,209.07 71,209.07 JUL 0.00 0.00 0.00 JUL 52,651.00 71,209.07 AUG 0.00 71,209.07 0.00 0.00 AUG 52,651.00 71,209.07 71,209.07 SEP 0.00 0.00 0.00 SEP 52,651.00 71,209.07 71,209.07 COMMENTS:

~

~

CESTIMATED DOE VOUCHER BASED ON ORM ACCOUNTING REPORTS ~~ ~ -

Fe18 7

%3 w

_ _ _ = _ _ _ _ _ _ _ _ _ _ _ _ __. _

.I I ~ '. .

RUN DATE: 07/15/85 CFFICE CF NUCLEAR REACTOR CEGULATION PAGEs: 661 TECHNICAL ASSISTANCE PROGRAM SUPPORT SYSTEM SUf91ARY OF FY 85 COSTS FID 0: A3712 TITLE: SFTY EVAL OF CORE-MELT ACC: INDIAN POINT-OR & ZION PROJECT MANAGER: LYON, M. DIVISION: DOSI BRANCH: RSB COSTS REPORTED FOR EACH MONTH CUMULATIVE COSTS THROUGH EACN MONTH BUSINESS DOE ACTUAL BUSINESS DOE ACTUAL LETTER FINANCIAL DOE LETTER FINANCIAL DOE REPORT INFO SYSTEMw VOUCHER REPORT INFO SYSTEMN VOUCHER OCT 3,953.00 3,954.21 3,954.21 OCT 3,953.00 3,954.21 3,954.21 NOV 3,453.00 3,452.82 3,452.82 NOV 7,406.00 7,407.03 7,407.03 DEC 2,234.00 2,233.93 2,233.93 DEC 9,640.00 9,640.96 9,640.96 JA7 513.00 514.30 514.30 JAN 10,155.00 10,155.26 10,155.26 FEB 17.00 17.10 17.10 FEB 10,170.00 10,172.36 10,172.36 MAR 17.00 17.10 17.10 MAR 10,187.00 10,189.46 10,189.46 APR 17.00 17.10 17.10 APR 10,204.00 10,206.56 10,206.56 MAY 0.00 0.00 0.00 MAY 10,204.00 10,206.56 10,206.56 JUN 0.00 0.00 0.00 JUN 10,204.00 10,206.56 10,206.56 JrL 6.00 0.00 0.00 JUL 10,204.00 10,206.56 10,206.56 AUG 0.00 0.00 0.00 AUG 10,204.00 10,206.56 10,206.56 SEP 0.00 0.00 0.00 SEP 10,204.00 10,206.56 10,206.56 COMMENJin CESTIMATED DOE VOUCHER BASED ON ORM ACCOUNTING REPORTS Y

__ _