ML20085D788

Presents Results of Review of Testimony on Commission Question 5.Testimony on Question 5 Is as Complete as Possible But Suffers from Same Deficiencies Re Relative Risk

ML20085D788
Person / Time
Site:	Indian Point
Issue date:	07/01/1983
From:	Amico P APPLIED RISK TECHNOLOGY CORP.
To:	Gleason J, Paris O, Shon F Atomic Safety and Licensing Board Panel
Shared Package
ML20085D753	List: ML20085D788
References
NUDOCS 8307290108
Download: ML20085D788 (7)

Text

.

AD Applied Risk Technology Corporation l P.O. Box 175, Columbia, MD 2104S (301) 964 3'769 July 1, 1983 James P. Gleason, Esq. , Chairman Mr. Frederick J. Shon Dr. Oscar H. Paris Administrative Judges Atomic Safety and Licensing Board -

U.S. Nuclear Regulatory Commission Washington, D.C. 20555 -

Dear Administrative Judges:

This letter report is intended to present the results of my review of testimony on Commission Question 5 of the Indian Point hearings. I have -

chosen this means of reporting to you on this question because the natu 9 and extent of the testimony does not warrant a large formal report as was submitted for Commission Question 1.

In the area of completeness of the testimony on this question, it is complete insofar as it is possible to be. The caveat which applies to_

this statement is that since the Question 5 testimony makes use of the testimony from Question 1, it obviously would in general suffer from the same deficiencies as those cited in the Qbestion 1 report. It appears, .,

however, that within the context of these deficiencies the testimony '

presented by all parties represents as much a statement of relative risk as is possible. Thus, further testimony specifically directed to the answering of Question 5 would not serve to illuminate the issue further.

Once again, it must be made clear that this is not meant to infer in any way that the deficiencies in Question I testimony are unimportint. ' , -

Thus, changes in the Question 5 conclusions as a result of changes in tne referenced Question 1 testimony could have an impact. With this clarified. I will briefly sumarize the testimony of each of the parties on Question 5.

L_I_CENSEES The licensees concluded that Indian Point risks are in the range of risks from other nuclear power plants. Specifically, latent fatality risk is low and available information suggests that latent fatality risk may not vary greatly among nuclear power plants,

_ the absolute risk of early fatalities is even lower than that for latent, which reduces plant-to-plant vari-ability, 8307290109 830727 PDR ADOCK 05000 G

= _ _ _ - - -_ .

both of these risk r;easures meet the NRC's proposed safety goals, reductions in the source term would reduce both of these risk measures, and could effectively eliminate early fatality risk.

The plants meet the individual and societal risk safety goals by large margins. Although these safety goals do not represent an " average risk" from nuclear power plants, they do essentially represent the upper end of the risks you would want to have to assure that nuclear power plants contribute very little to the overall risk. Since the IP plants meet these goals by such a large margin, it is not unexpected that these plants do not pose undue risk and are among the average plants in risk.

When compared directly with other plant PRAs which have been performed, IP is the lowest for individual early fatality. risk and in the middle of the PWRs for interfacing systems LOCA frequency. They are among the icwest of the plants for societal risks, both early and latent fatalities. These comparisons are made only for risk from internal initiators, since the other risk studies did not treat external events.

It is important to note that there are significant differences in the nethodologies of the various risk studies due to advances in PRA over ..

the last 10 years. These differences in the sophistication of the PRAs a s not studied; thus, the comparisons made do not take them into consideration. For example, it is likely that the interfacing systems .

LOCA frequencies of the other plants would be lower if improvements were considered in the numbers. ~ -

Deterministic considerations support the conclusion 'th'at Indian Point is act above average in contribution to risk. These considerations consist .

of special design features, such as: ' ~ ~

the containment building is structurally very strong and very large to withstand both seismic events and high internal pressures; the geometry assists in cooling debris and circulating hydrogen and steam five containment cooling fans and four spray pumps is a highly diverse containment cooling system two diverse containment recirculation sumps are available three gas / turbine generators provide additional backup power

_ confirmatory signals are provided to key valves in case they are in the wrong position

containment weld channel pressurization and isolation valve seal water systems exist maximized flexibility to interconnect service water and component cooling water trains to supply any cooling loads is provided, although_the disadvantage of this capability due to a potential disabling pipe break was not considered in the IPPSS.

Attempting to use NUREG/CR-2239 (the Sandia Siting Study) te produce accurate estimates of risk from existing nuclear power plants is not valid. This study does not address real plants, and the differences between plants could severely affect risk estimates. The early fatality results are limited by overly conservative emergency response assumptions. NUREG/CR-2239 does not have a release category which corresponds to the IPPSS 2RW (late release) category. It uses only the SST 1, which represents category 2 and has higher release fractions and -

a much earlier release time. This also carries over into a much more conservative emergency response scenario, which assumes no emergency response outside ten miles for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> whereas the IPPSS assumed 90 percent sheltering. This results in auch higher dosages and early fatalities, as well as a greater need for supportive treatment. Thus, NUREG/CR-2239 does not model reality and comes up with higher consequences as a result. Even the staff re-analysis of the IPPSS, which used more conservative assumpti)ns than the IPPSS, showed much lower consequences than the Sandia study, especially for early fatalities. The only usefulness of this study is to illustrate the sensitivity of risk to certain assumptions. For example, ,

source term reduction effects all risks, b0t more -

markedly early health effects, and the effect is more pronounced for higher population density sites * '~ "

, emergency response reduces early effects, which are very sensitive to delay time, but does not greatly affect latent risk shielding and early relocation beyond ten miles is very j effective in reducing early effects the use of evacuation to ten miles in combination with sheltering from 10-25 miles is just as effective as evacuation to 25 miles NRC STAFF I

The NRC staff concludes that there is no reason to believe that Indian Point risk is well outside the range of other U.S. plants. The Indian Point Task Force Report (NUREG-0715) supports this evaluation. It was i

the conclusion of NUREG-0715 that the Indian Point design was about 10 times better than the average reactor but that the site was 10 times worse than the average site.* Thus, Indian Point does not pose undue risk. However, NUREG-0715 conclusions were based on internal events only, and thus to some extent it is superseded by the IPPSS.

Of the 13 other PRAs performed, only 4 include externally initiated events and only Zion approaches the IPPSS in completeness. Comparing internal events only, the core melt frequency at IP is about average for the 13 plant PRAs, even-when accounting for the uncertainty-bounds. If the core melt frequencies are compared for only the severe release sequences, IP is among the lowest of the plants analyzed. Comparing the frequencies of severe release sequences is a better measure of comparative off-site risk than comparing overall core melt frequencies.

However, the comparison of internal events only is not comprehensive, and for ' Indian Point, external events are the dominant contributors to risk and their analysis is the most uncertain part of the study. It is -

not really possible to compare plants just by internal events and rank ,

them absolutely, although there is some correlation and inferences can be drawn. Further, the PRAs used for comparison were not performed using consistent methodologies and assumptions. Thus, the direct comparison of one with the other may well be invalid.

As with the licensees, the NRC staff concluded that deterministic considerations support the conclusion that Indian Point does not pose undue risk. Specifically, the following features were sited: _.

only one large pipe is susceptible to interfacing systems LOCA, as opposed to 3-6 in most other plants

~ ._

three gas / turbine generators a7d three diesel generators are available for emergency AC power production, as opposed to only two diesel generators in most other plants the containments are among the better ones at mitigating consequences, although one cannot be completely confident

that the containment heat removal systems can continue to l operate after core melt; even if this is so, the

! percentage increase in risk would not be large; the IP containments are superior to most other large dry

• Both NUREG-0715 and NUREG/CR-2239 (the Sandia Siting Study) show that the site characteristics of Indian Point would make it one of the highest risk sites in the country if all other things were equal.

That is, if the plant design was average in preventing severe acci-dents.

l

containments at delaying overpressure failure; for the likelihood of containment failure, the Indian Point containments can be compared with other plants with the following results:

for before core melt failures overpressure; below average interfacing systems LOCA; below average-earthquake; unknown -

ir.ternal missiles; average (low risk) external missiles; average (low risk) leakage; unknown (low risk) steam generator tube rupture w/open S/RV; average for after core melt failure pressure spike; below average reactor vessel missiles; average ,'

Comparing Indian Point with the NRC's proposed safety goals is a useful exercise, although there are reasons why these quantitative design- -

objectives are not requirements for Indian Point. The Indian Point site meets the safety goal for early fatalities for the after fix case '(which '

! represents the most realistic evaluation) regardless of emergency

response and supportive treatment assumptions, although the margin by

,' which the goal is met is less than the uncertainties associated with the evaluation. It meets the safety goal for latent cancers for the after fix case by a very large margin. The margin is so large that it is hignly unlikely that the goal could possibly be exceeded. The Indian l

Point units meet the safety goal for total core melt frequency for the after fix case, although the margin is very small and the uncertainties are very large. -

1 l INTERVEN0RS The intervenors conclude that it is not possible to establish a true range of risks from other nuclear power plants for comparison with Indian Point. Since this is so, only the potential for very high consequences should be considered. This would indicate that Indian I

4

Point may very well pose undue risk to the public. Action should be taken, possibly even shutdown, to protect the public from this risk.

Many risk studies have shown that all plants do not have similar " risk p rofil es . " Design differences, external events, site characteristics and emergency response foctors all have been shown to effect plant risk.

Risk profiles can change with time, and the particulars of the changes cannot always be predicted, or even the direction of the change. Site specific risk studies have made us able to understand how site risk characteristics influence risk. This is not as true for the influence of design / operation risk characteristics. Uncertainties in how site factors affect risk will effect all sites in the same manner, thus not changing relative risk rankings. Design / operation experience is not as extensive, and there are many uncertainties involved. Common-mode failures (both internal and externally induced) have limited data available ar:d thus of great uncertainty. The differences among risk dominant sequences for various plants mean that these uncertainties will -

effect each plant differently; thus, relative risk could be altered.

There are many reasons why comparisons should not be made between IPPSS and WASH-1400, RSSPAP, and IREP and that these studies do not define the

" range of risks" for nuclear power plants. Major among these is that external events are not included, the state of the art of PRA has changed and thus results are not comparable or consistent, WASH-1400 release categories were used in RSSPAP and IREP (not plant specific),

RSSMAP analysis was not exhaustive but rather guided by WASH-1400 results, and IREP did not deal specifically with risk. SomeofthePRAi used for comparison were in~d ependent of the above programs, but comparisons are still not valid. Comparison with Big Rock Point is academic since the plant is so small. Limerick excludes external ~ events and also suffers from an attempt to make it comparable to WASH-1400.

Zion can be compared to IPPSS because methodology is identical ~, but"this ~

does not by itself establish a " range of risks." It is true that if the omissions from WASH-1400, IREP, and RSSMAP were evaluated (external events, etc.), the sequence frequencies would increase given that the methodology stays consistent. They could not decrease.

Comparisons can be more reliably drawn between site consequence characteristics then accident frequency, based on what we know new. At some point, the magnitude of consequences is such a large value that the risk is unacceptable, regardless of probability. NUREG-0350 suggests that it might be appropriate to examine the high consequence-low probability part of the risk spectrum by itself, as opposed to looking at expectation values. The Sandia siting study evaluated the effects on risk specifically of site characteristics by evaluating 91 reactor sites throughout the country assuming each site had an identical reactor.

Thu.s, all other things being equal, the study shows that for release category SST 1, which is the most representative to use, Indian Point falls in the upper group of plants for risk due to early fatalities,

1 early injuries, and latent cancers. Using the mean results in each categcry, Indian Point ranks second in early fatalities, first in early injuries, and first in latent cancers. Even when the Sandia results are corrected for the actual plant power level, but still assuming all other things equal, Indian Point 3 was ranked second, first, and first while Indian Point 2 was ranked third, second, and second. If a comparison is made using category SST 2, Indian Point would not generate early fatalities substantially different from any other site. Also, considering just the high consequence accidents (maximum calculated values), IP 2 and 3 rank 7th and 6th in early fatalities and are not in the top ten for latent cancer fatalities. The " worst-case" results from the Sandia study are useful because they give a crude, upper-bound _

estimate of consequences. They also show that implementation of '. / 3. A emergency response beyond 10 miles car substantially reduce these __.-

consequences.

Comparisons with safety goak is. nct germane to Question 5 because. it does not address the acceptelitity of risk. The safety goals are incomplete because they onb fccus on certain risk measures. Also, the goals were intended as an astessrant tool, not as a true licensing limit, and they are subject to change after a two-year evaluation i period. The Commission stated that during this two-year period, the goals should not be used in the licensing process.

Since risk comparisons are not valid, it is necessary to deal simply with the fact that the Indian-Point consequences due to its siting are extremely high. Because of this, serious consideration should be given to mitigation systems at IP~to protect the public from the uncertain risk posed by IP. Thus, these systems should be installed without regard to cost if the plants are to be kept running. -

This concludes the discussion of each party's testimony.

This letter report completes, to the best of my knowledge, all the deliverables requested by the board regarding my review of the Indian Point hearings. If any additional material is required, or if there are any questions or comments regarding any of the reports submitted, please call me at (301) 964-3769. It has been a pleasure to be of service to the ASLBP in these matters, and I thank you for this unique opportunity.

Sincerely, C f }l/ '

Paul & ico cci J. Hard/ASLBP

- en -, - . - . .,--- , , , g