ML20086L254
| ML20086L254 | |
| Person / Time | |
|---|---|
| Site: | Limerick  |
| Issue date: | 02/06/1984 |
| From: | Boyer V PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC |
| To: | Eisenhut D Office of Nuclear Reactor Regulation |
| References | |
| RTR-NUREG-0974, RTR-NUREG-974 NUDOCS 8402080022 | |
| Download: ML20086L254 (11) | |
Text
- -
4:
PHILADELPHIA ELECTRIC COMPANY 23O1 M ARKET STREET P.O. BOX 8699 PHILADELPHIA. PA.19101
{
V. S. BO Y E R SR. VICE PRFDDENT NUCLIs A N POWE R February 6, 1984 Mr. Darrell G. Eisenhut, Director Division of Licensing U.S. Nuclear Regulatory Commission Washington, DC 20555 Subj ect:
NUREG-0974 Supplement No. 1:
Draft Environmental Statement Related to the Operation of Limerick Generating Station, Units I and 2, Docket Nos.
50-352 and 50-353
Dear Mr. Eisenhut:
We have reviewed the subject supplement to the Limerick DES.
A summary of our major comments and concerns are provided below and detailed comments are attached.
OVERALL We have performed and presented in the Severe Accident Risk Assess-ment (SARA) a full scope probabilistic risk assessment o' the Limerick Generating Station.
This analysis included the effects of internal and external initiating events and an extensive analysis of uncertainties. We concur with the staf f 's conclusions that
" accident risks from Limerick are expected to be a small fraction of the risks the general public incurs from other sources".
- However, it is Philadelphia Electric Company's position that inappropriately large conservatisms have been incorporated in the DES analysis. We believe that the SARA analysis provides a mre realistic assessment of the risks.
SOURCE TERMS AND UNCERTAIhTf l
Some of the source terms in the DES are higher than any in prior environmental statements, probabilistic risk assessments, or source term analyses. They are also higher than those estimated in the l
SARA. The use of such large source terms contributes to calculated consequences that are essentially upper bounds.
It would, therefore, appear to be unnecessary to include additional quantitativa uncertainty or sensitivity analysis.
(
8402080022 840206 08'
/
PDR ADOCK 05000352 D
PDR l
s...
SOURCE TERMS AND' UNCERTAINTY (continued)
Since any quantification of uncertainty must include consideration of the level of conservatism in the analysis, we view the DES.
presentation of quantified uncertainties as inappropriate and unnecessary. :Riis derives from our belief, reinforced by the SARA analysis, that the DES results are upper bounds and that while the NRC's NEPA policy statement requires identification of major uncertainties it does not require quantitative uncertainty analyses.
MEAN'VS. MEDIAN NUREG-0880 " Safety Goals for Nuclear Power Plant Operation" specifies the use of median values for probabilistic risk calcu-lations. The SARA analysis presented the results as medians due to our agreement with the view that median values are more appropriate than means. The SARA analysis also provided point estimates for completeness. Most previous staff environmental statement assessments of risk (Susquehanna, Fermi 2, Byron) have utilized median frequencies.
The Limerick DES departs from this practice.
NUREG/CR-?S28 The staf f has used the NUREG/CR-3028 point estimates, which we believe are incorrect, for internal events in the DES analysia.
Four issues, identified to the staff at our meeting of September 26,1983 (reference 1 of attachment), have not been addressed by the DES.
These issues are: Recovery of Feedwater, Loss of Offsite Power Initiator Frequency, HPCI Restart Failure Probability, and Failure of Manual Depressurization.
We believe that the Severe Accident Risk Assessment (SARA) represents a realistic assessment, including uncertainties, of the public risk due to potential accidents at the Limerick Generating Station and fulfills the requirements of NEPA in regard to their realistic portrayal. An alternative to the use of SARA in the FES would be to use the conservative source terms shown in the DES but to delete the quantitative display of l
l uncertainties from the FES and to explicitly note the conservatism in that approach.
l Enclosure See Attached Service List l
l l
oc: Judge Lawrence Brenner (w/ enclosure)
Judge Peter A. Pbrris (w/enclos' ire)
- Judge Richard F. Cole (w/ enclosure)
Troy B. Conner, Jr., Esq.
(w/ enclosure)
Ann P. Hodgdon, Esq.
(w/ enclosure)
Mr. Frank R. Rtrnano (w/ enclosure)
Mr. Robert L. Anthony (w/ enclosure)
Mr. Marvin I. Iewis.
(w/ enclosure)
Ms. Phyllis Zitzer (w/ enclosure)
Charles W. Elliott, Esq.
(w/ enclosure)
Zori.G. Ferkin, Esq.
(w/ enclosure)
Mr. Thtrnas Gerusky (w/ enclosure)
Director, Pennsylvania Bnergency Managenent Agency (w/ enclosure)
Mr. Steven P.-Hershey (w/ enclosure)
Argus Iove, Esq.
(w/ enclosure)
Mr. Joseph H. White, III (w/ enclosure)
David Wersan, Esq.
(w/ enclosure)
Robert J. Sugaunan, Esq.
(w/ enclosure)
Martha W. Bush, Esq.
(w/ enclosure)
Spence W. Perry, Esq.
(w/ enclosure)
Jay M. Gutierrez, Esq.
(w/ enclosure)
Attrnic Safety and Licensing Appeal Board (w/ enclosure)
Atanic Safety ard Licensing Board Panel (w/ enclosure)
Docket ard Service Section (w/ enclosure) l t
i s '.I ~?','<,Y) Z sM l'p ?.T 4 2 Q: n.,s. Nh-l As i P..,'4 *yl,'f. -b.a O C i 00.%** b. a r* :
.'"I.'
a
^ ~
~ *
- s t
e g-,
s.
- + ;.,
2 o
L-I l
l
\\
COMMENTS ON NUREG-0974 SUPPLEMENT NO. 1 DRAFT ENVIRONMENTA'. STATEMFHT RELATED TO THE OPERATION OF LINERICK GENERATING STATION UNITS 1 AND 2 l
DOCKET N05. 50-352 AND 50-353 i
CNAPTER/SECTION
SUMMARY
AND CONCLUSION Page v Item (1):
We concur with the conclusion of this item and believe that the substantial conservatisms introduced 'by the staff strongly reinf orce that conclusion.
ENVIRONMENTAL CONSEQUENCES AND MITIGATING ACTIONS Page 5-4 Table 5.11a:
The assumed power level of 3458 MWth and 5-5 is not consistent with the actual full power license I!mit of 3293 MWth.
i Table 5.Ita: The values for radioactive inventory differ from those appropriate for Limerick by ratios ranging from.49 to 1.92.
The appropriate Limerick inventory is as shown in Table 7.1-24 of the EROL, Table E.8 of the PRA, or Table 10-1 of i
SARA (See also comments on pages 5-16 and N-2).
l Page 5-15 Section 5.9.4.5(2),
top line:
It should be noted l
t hat t he source terms are conservat ively treated in the DES analysis contrary to the " improvements" suggested by this statement.
Section 5.9.4.5(2),
third full paragraph, third l
j sentence:
This is a very conservative assumption in the source term development, considering t he advances made in this area since publication of l
WASH-1400.
It is recognized that this is currently an ongoing subject of research but this approach is j_
overly conservative and ignores the Limerick l
! j i
f
Ep;cific'scurcs tara enzlysis of NUREG/CR-3028, tho LGS-PRA,- or SARA.
It is noted t hat the source
+-
terms are more extreme t han those cc.isidered possible -in the SARA sensitivity analysis or used in previous staff ES analysis.
k Page 5-16 Section 5.9.4.5(2),
first full parrgraph, first sentence:
The use of mean values is inconsistent
- with previous DES /FES analyses by the staf f which-have used median f requencies.
The representation 4
of seqbence. frequencies f or LGS is t heref ore more conservative than other ES analyses.
Section 5.9.4.5(2), f irst f ull paragraph, third and fourth sentences:
The use of an effective peak ground acceleration of 0.4g as the dividing line between severe earthquakes and earthquakes of low to. medium severity is conservative. At t hat level
- of acceleration there would likely be only a small numher of, if any, bridge or overpass failures, and complete. disruption of the road network is not I
expected.
The " Late Reloc" mode of evacuation (see comment on p. 5-21,' third paragraph), in which it is assumed that people are relocated after 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, is excessively conservative for earthquakes in the 0.4g acceleration range. 'The SARA figure of 4
0.61g (SARA page 10-14) for severe disruption of the road net wo'rk is 'more realistic.
The staff' provides no justificatic"n for the assumption that
.i n the event of a severe earthquake, it would be impossible to evacuate anybody for at leust 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />.
This topic was discussed in our, August 29, 1983 submittal (PEco reply to Geosciences Question 07).
Section 5.9.4.5(2),
third full paragraph, second sentence:
Inventory is inappropriate for a BWR (See comments on Table ll.o above and page N-2 below).
Page 5-17 Table 5.11c:
Changes made from the SARA analysis are significant
-and result in increased conservat ism of the staf f results.
Some source terms in the DES are higher than any in the RSS, e.g.,
94-96% iodine enmpared with a i
maximum of 90% in saquence BWR2 (RSS Table VI 2-1);
86-87% cos tum compared wit h a maximum of 50% in the RSS.
For previous BWR,FES's, rebaselined WASii-1400 source terms were ut ilized with the largest source term was TC hf ' with 45% iodine, 67% cesium and 64%
kf tellurium).
Further, steam explosion source terms t,.
containing 40-50% ruthenium can only come cbout in I
I
the cv:nt that 50% er muro of the coro is involved In a steam explosion. Such steam explosions were not considered in earlier FES's (see, for example, Susquehanna FES Table 6.1.4-2).
Therefore, the LGS has been treated more possimistically than other BWRs in previous FES's.
In summary, the use of the RSS methodology in the DES is very conservative.
Page 5-18 Table 5.11d:
From inspection of the frequency values it is apparent that changes in the containment event trees for hydrogen burn, steam explosion and SGTS operation from the LGS-PRA er HUREG/CR-3028 have been made.
The changes and their bases have not been provided.
For Class II and Class
- III, the DW and WW failure modes were combined with the WW failure since the staff assumed a pool decontamination factor of one for saturated pools.
This yields a
conservative evaluation.
The frequency differences are substartial and are due to the staf f not correcting the BHL assessment (HUREG/CR-3028) of the LGS-PRA and the select ion of
.4g as the breakpoint (see comments or page 5-16 and H-3).
Correction of the values per comments previously provided (see reference il and using the more realistic
.61g would reduce risk estimates substantially and reinforce the conclusions of the DES.
Page 5-21 Section 5.9.4.5(2), second paragraph:
The use of a single delay time (2 hr) is not as realistic as the model used in SARA (Table 10-8, with a 1, 3, or 5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> delay time).
The DES (p. H-2) criticizes the SARA model because it is "not site specific".
However, tho staff has used the some model in other l
FES' (e.g., Susquehanna, HUREG-0564).
l Section 5.9.4.5(2),
third paragraph, first sentence:
Limiting the ability to effectively l
evacuate is not unreasonable, but assuming no l
evacuat ion f or the populat ion around the site is overly conservative.
1 Page 5-23 Section 5.9.8.5(2),
third paragraph and ** note:
l The DES utilized mean frequencies for the various l
accident sequences and categories rather than l
medians.
The mean for a
positively skewed distribution such as the log normal corresponds to a confidence level well above the median and could correspond to confidence Irvels as high as 80-90 percent.
The median which represents the value l
where there is an equal likelihood of being greater ;
?
i-
U 3,
+
_u
=~
cr,1ses to censid:rsd to.'bs* a lmoro cpprcpristo e
measure of risk.
Most previous environmental assessments of risk (Susquehanna, Fermi 2,
Byron) utilizedi_ median frequencles in t heir analysis
~
rat her t hen means.. While thE impact of this is small for. Internal initiaters it is large for the seismic initlators where ' t he ; mean is almost a f actor of 20 higher than the median.
s' Page 5-25 Figures.5.4b to 5.4g and Table 5.11gt T he.. result s
.through shown are' more severe t han present ed in t he LGS S ARA.
5-31 This is* due 'to t he higher f requencies, higher s
release frautions, and more. pessimistic emergency response. ' The major d'i f f erence is in t he CCDF f or
~
early fatalities (Figure 5.4e) which lies above the upper 95% curve from SARA. The represented upper confidence levels (cross hat ched areas) are not substantiated and are unprecedented for any assessment of risk.
It is difficult to believe t hat these are plausible outcomes at any but the most extreme (much greater t han 99%) confidence levels.
The early f atality risk associated with Limerick, i
based ore tnis analysis, is dominated by extreme seismir events of rare occurrence.
It is particularly importent to point out that the postulated seismic event in eddition to inflicting damage on the Limerick site would cause substantially higher fatalitios directly to the population (through falling
- debris, building structural
- failures, etc.) t han are at t ributed to Limerick related causes.
Regardless of Limerick operation, substantial risk to the population exists from large magnitude seismic events.
This issue can be addressed if the risk to the population associated with a large magnitude i
seismic event is displayed along with the incremental risk associated with such an event during Limerick operation.
It would then be l
cpparent t hat the incremontal risk of Limerick i
operation would be negligible.
F Page 5-48 Section 5.9.4.5(7)
Probability of Occurrence of Accident: No mention of the differences between the PRA and the BHL review (HUREG/CR-3028) is made.
The points made at the 9/26/83 meeting (reference
- 1) regarding the BHL reassessment would lower the BHL core melt frequency by 7.3 x 10-5 l'
l Section 5.9.4.5(7)
Erobabilitv of Occurrence of Accident second paragraph, fifth sentence:
The i
l criticism cf the use of 95% and 5% is a matter of >
Lo r
tem:ntics.
The 95% and 5% limits cro uncd in SARA e
to represent a
range of frequencies which incorporates a large psrt of the uncertainties.
Section 5.9.4.5(7):
Probabilliv of Occurrence of Accident, second paragraph, sixth sentence:
This i
statement is not clear as to intent.
The staff's t
analysis uses only a point estimate.
The EROL (SARA) does consider the uncertainty in seismic probability in detail in making its assessment.
The DES provides no adequate justification for its use of point estimates, particularly in the seismic events.
Page 5-49 Section 5.9.4.5(7),
first full paragraph:
A comparison is made of the DES core melt probabill'ty with publ i s hed PRAs.
This la inconsistent and presents an unfair comparison because the values for o t her plants are f rom the published PRA's not from staff or staff contractor reviews.
Section 5.9.4.5(7),
second full paragraph, fifth sentence:
The staff has disregarded the Limerick spealfic analysis contained in HUREG/CR-3028, the i
LGS-PRA, or SARA with regard to source terms development.
Page 5-52 Section-5.9.4.5(7),
fifth full paragraph The meteorological sampling scheme used in CRAC is mentioned as a source of uncertainty.
CRAC2, which was used in EARA, was specifically developed to reduce this uncertainty (see Procedures Guide p. 9-92).
Page 5-56 Figures 5.4n, o, p, q, r:
The sincertainty bounds, to 5 contradict the text on page 5-54, which states that "the risk uncertainty bounds could be well over a factor of 10, but not as large as a factor of 100.
Within these uncertainty bounds,
- however, the uncertainties associated with the probability-Integrated values of consequences (the risks) are likely to be less.
No explanation is provided of what the uncertainty bounds represent.
They appear to be an arbitrary factor of 100 in all cases.
It is expected that the point estimate represents something greater than a 50% confidence level and may be greater than the 90% confidence level in some cases.
It is believed that the r
weight of evidence supports a realistic estimate of risk considerably lower than the DES values and that the DES point est imates are much nearer the I
upper est imate t han the median. f M**c--e*+we r-wgi. r w m w -
swgw---%y._,
eS,..-
4-.,wy-_.7.,,
EVALUATION OF THE PROPO*ED ACTION Page 6-1 Table 6.1, the reduction in generating costs of $34 million/ unit / year presented and discussed in Section 6.4.2 (page 6-3) underestimates the operating savings attributable to Limerick.
As presented in EROL Teble E320.1-1 (Revision 15, August 1983), the operational savings should be
$188.0 million in 1986 for single unit eporation.
APPENDIX H General Risk important source terms ut ilized by the staf f are highnr t han those used in any other assessment to which the staff compares results.
Tho source mpact on t he risk. While terms have a substantial 8 the staff acknowledges that the source terms are probably conservative (".
. source terms used in the staff analysis cannot be much higher in t he maximum, but could be substantially lower.";
page 5-51) the impact of this is not assessed nor considered.
Page H-1 Second paragraph, fifth sentence: More is known about fission product release and transport t han was kntown at the tims of the Reactor Saf ety Study; where the values used were very conservat ive in the absence of knowledge.
The staff conservatism tends to make the analysis an upper bound rather than a realistic point estimate.
It should also be pointed out that the source terms are higher than past ES analyses by the staff.
Page H-2 Second full paragraph:
It is extremely conservative to use a DF of one for a saturated pool.
The applicability of a DF of greater than one for a saturated pool has been well established by experiment (reference 2, 3, and 4).
APPENDIX _d General The DES assumptions for amargency response modeling are a limited set of the possible responses.
SARA considered a more comprehensive set.
APPENDIX N Page H-1 Comparisons of CRAC and CRAC2 in the international benchmark exercises produced result s that are quito close.
However, the meteorological sampling scheme used in CRAC leads to much greater uncertainties than in
- CRAC2, see Figure 9-17 of the PRA Procedures Guide. l
\\
/
Pcgn H-2 S;c:nd full p;ragrcph: W3 havo twa esnc rno with the calculation of core inventory.
The first is that the use of 105% power is inconsistent with the application to a probabilistic analysis.
A power level of 105% is strictly a design basis assumption and is not a best estimate of the actual power level for Limerick.
- Secondly, the DES used an inventory developed for a PWR by a code (ORIGEN) which was designed for another purpose.
The LGS-PRA ut ilized the code RADCINDER f or a BWR core (See response to Quantion E.01(b) page Q-126, Vol. I of t he LGS-PRA).
Page N-5 First paragraph, sixth sentence:
The staff criticizes the SARA because there are "too many variations in the optimistic direction for nonse"ere eart hquaka condit ions,"
as one of the factors leading to suspicions that "the upper estimates of the overall CCDFs in the EROL are biased".
- However, optimistic sensitivity studies in SARA were used to fix the lower estimate and had no ef f ect on t he upper est imate.
First paragraph, sixth sentence:
The staff crit icizes the SARA analysis f or not varying source terms to encomnass some of the high values of t ha release fractions used in the staff analyses.
- However, as noted in the foregoing discussion of the DES source terms, we believe t hat the DES source terms are highly conservative.
We also believe t hat the largest source term in SARA,
- VRH20, is conservative because radionuclide retention in the primary coolant system and reactor enclosure was neglected.
Therefore, we consider that, if crit icism is to be leveled at the SARA and EROL source terms, it should be because these scurce terms are too high rather than too low.
Comment References (t)
R.
E.
Martin, dated:
December 9,
- 1983,
Subject:
Summary of Severo Accident Risk Assessment Review Meeting held on September 26, 1983.
Wi t h at tachment s.
(2) Rastler, D.
M.,
1981.
Sup_pression Pool Scrubbina Fnctors for Pogtulated Boilino Water Reactor Accident Conditions, General Electric Comnenv, NEDD-25420, Class 1.
(3) Marble, W.
J.,
et. al.,
1982.
" Retention of Fission Products by BWR Suppression Pools During Severe Accidents", paper presented at $
i
?
the ANS M;;t ing en Thermal Roscler S2foty, August 30t h - September 2, 1982, Chicago, Ill.
(4) SARA, Philadelphie Electric Company,
- 1983, Section E2.6.1 and Tablo E-4.
g.