ML20235T115

Comments on Plant Containment Event Tree.Overpressure Failure & Corium/Wall Failure Discussed.Understands That Sandia Begun New Analyses to Address Comments Received from Util & Peer Review Group

ML20235T115
Person / Time
Site:	Peach Bottom
Issue date:	09/19/1986
From:	Boyer V PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC
To:	Morris B NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
Shared Package
ML20235T023	List: ML20235T021 ML20235T068 ML20235T085 ML20235T115 ML20235T139
References
FRN-52FR7950, RTR-NUREG-1150 NUDOCS 8710090265
Download: ML20235T115 (9)

Text

- - _ _ _ _ _ _ _ _ - _

l

. .s PHILADELPHIA ELECTRIC COMPANY g

2301 MARKET STREET P.O. BOX 8699 PHILADELPHIA. PA.19101 (215)841 4500 ((p {g ]ggg V.S.DO YcR SR. VICE PRESIDENT

' H UCl.E A R POWE R Mr. Billy M. Morris, Director Division of Reactor System Safety U.S. Nucicar Regulatory Commission Washington, DC 20555

Subject:

Comments on Peach Bottom Containment Event Tree

Dear Billy:

We appreciate the cooperation of M. L. Ernst in arranging for our attendance at the Peer Review Meeting on the Peach Bottom Containment Event Tree development held at Sandia on July 16, 1986. The meeting was most beneficial and provided us with an increased awareness of both the program and its status. As a result of the meeting and review of the draft report material, we have generated a series of recommendations and comments (see Attachments) which are forwarded to you, Sandia and the review team for consideration.

Recent activities have indicated that the draft report has already played a large part in the public and regulatory perception of Mark I containments. The impact of the revised draft is also expected to be significant. Therefore, we would like to see the analysis done

. properly and the results presented in a manner which directly addresses Mark I performance issues. Discussion with the Peer Review Group indicated that the results are currently controlled by two assumptions: overpressure f ailure (pressure and location) and corium/ wall failure. Increasing the containment failure pressure by about 10% and decreasing the likelihood of corium/ wall failure would virtually eliminate calculated early containment failures. These factors should be addressed directly, as follows:

Overpressure failure - The analysis which is relied upon to estimate containment pressure capability is an clastic analysis of a Brown's Ferry containment. The clastic analysis, in effect, provides an extreme lower bound to the Brown's Ferry pressure capability by identifying a pressure at which no significant yielding would occur.

Under higher pressures, support would be provided to the drywell theshell by the massive concrete shield wall, making (assumed) failure at

" knuckle" extremely unlikely. Use of the very conservative lower pressure not only gives an incorrect picture of ultimate capability, but also reduces the impact of leaks which are more important at higher pressures. To aid in understanding of Mark I behavior at elevated pressures, the industry has commissioned Chicago Bridge and Iron, designer of the Peach Bottom Containment, to perform a more detailed analysis to estimate the ultimate failure pressure and likely location (s). This additional information should allow you to make a more accurate assessment of Mark I capabilities.

G710090265NUREG BM ppgt#

PDR 1150 0 i

l 1'

l l

J i

Corium/ wall failure - The previous evaluation was dominated by the assumption that vessel failure would lead directly to containment l failure due to molten core debris contacting the drywell shell. The l j

assumption was based on a " vote" by experts, several of whom areThe on the Peer Review Group of the Containment Event Tree analysis.

manner in which this'is treated in the Event Tree masks its importance and eliminates t.he interaction between the other Event Tree nodes and the "models" which the experts used in their voting. In effect, containment performance was determined primarily by this group's vote q

and the remainder of the Containment Event Tree acted only to mask the importance of the loosly documented " expert judgement". Because of the l i

importance of the assumptions in this area, changing opinions and the l interrelationship between failure "models" with other Event Tree nodes, the issue of corium/ wall failure should be treated separately from the containment Event Tree. The questions asked should be structured to address the issues and the assumptions used by the expert panel should be documented in support of their votes. In this manner, the value of the Event Tree in understanding containment behavior will be enhanced, and the issue of corium/ wall failure can be directly addressed.

We understand that Sandia personnel have begun new analyses to address comments received from Philadelphia Electric and Peer Review Group and look forward to continued participation in the review process.

ARD/cw/09028601 p p;2+,5 +

Copy to: V. Stello - NRC R. Bernero - NRC F. Eltavila - NRC C. Reed - CECO.

A. Benjamin - Sandia M. Corradini - Univ. of Wisconsin

r .

.4

+

Attachme.nt 1 Overview Coments on the Peach Bottom Containment Event Tree

1. The Peach Bottom CET represents a significant degree of detail in the Investigation of severe accident progression. Due to the complexity, its review and use by personnel other than the developers !s difficult. The icvel of detall also limits extension of the results to other BWR's. Therefore, it would be useful to condense and sunnarize the detailed thinking on the dominant failure paths to enhance the use of the CET as a decision aid. The CET could be' mode tractable if it were dlsected into a few major parts, each with it's own endpoints (i.e. sequence definitions) and conclusions. One method would be

- to break the tree 'into two sections, one being the present fonn  ;

ending at, but not including, the containment failure node with  !

the remaining being secondary containment behavior. This would provide a deeper understanding of the effect containment failure mode assumptions have on the dominant severe accident scenarlo probabilities and source terns. Additional suggestions are presented in conment 5.

2. The analysis would be nede nere useful in support of regulatory decisions if a wider spectrun of sequences (e.g. loss of nakeup, LOCA, etc.) were included. This enhanced understanding of the containment behavior would, for example, demonstrate that it would be extremely effective, regardless of assumptions, in an accident'such as the one at TMI Unit 2.

3. The two najor contributors to calculated containment failures at Peach Bottan appear to be:

- Assuned overpressure failure In the drywell .

l

- Assened melt-through of the drywell shell .  !

Neither failure mode is supported to the degree necessary to 1 warrant the-level of confidence impiled by a central estimate. )

Uncertainty is also significantly different than the current estinates would Indicate.

Specifically:

- Overpressure Fallure - The value used is based on an assunption of a free-standing elastic pressure vessel. Long  !

before the failure would occur, the drywell shell would come in contact with the concrete shield wall, which would provide substantial support. The existing estinate is probably best viewed as a lower bound. Use of this conservatively low pressure not only gives an incorrect picture of ultimate strength but also reduces the impact of leaks which are more important at higher pressures.

- Assured Melt-Through of the Drywell Shell - It has been reported that recent MELCOR nodels nay predict that noiten noterial nelt-through of the drywell shell is significantly ,

I less likely than the predictions used in the CET. We agree with the Peer Review Group that requantification is necessary.

If an " expert" panel is to be used in quantifying this area L of the CET, a group of questions should be selected to l directly address the issue, including statenents of assunpticas which are to be used. This should be treated separstely from the CET.

The importance of these issues was highlighted by the results of a sensitivity studt which Indicated that changes in the above two assumptions result in hnving essentially all early containment  !

failures to late failures (with potential for recovery). l

4. The plant-specific nature of the CET and its quantification should be emphasized in the conclusions.

5. Additional efforts which would be of value include:

- Variation of other assumptions used in quantification for input to an uncertainty analysis and aid in understanding the Impact of CET assumptions.

- The relationship of containment failure modes to the questions could be provided, le.ading to an importance ranking of the questions.

- Similarly, Identification of the dominant paths (e.g.

cutsets) for the tree would be a valuable addition.

- Plctoral representations of portions of the tree would be a valuable aid in understanding.

Inherent within the structure of a containment event tree is the  !

dependence of nodal issues or functions on previous nodal conclusions. While the construction of the Peach Botton CET appears to have taken this type of dependence into account the results may be skewed due to failure assumptions of previous conclusions. One exanple of this might be the conclusion that given the failure of the energency diesels no CRD Injection could occur. This assunption has a substantial impact on vessel melt-through and contalnnent failure mode. Given the operation of CRD by ranote diesel, vessel failure and subsequent DW liner burn - l through will not occur. Thus, previous argunents for enhancing the use of the CET by condensing and sunnarizing could be useful j for recognizing these dependencies. ,

6. The consensus of the Peer Review Group was that the optimistic and pessimistic walk-throughs are of little value as an uncertainty estimate. This is a valid conclusion.

l 7. The phenonenological uncertainties associated with sone CET nodal points were bounded and quantified by the use of an " expert" panel. Documentation of the CET provides little Information regarding the process used in quantifying values given by the panel. The manner in which the " experts" were chosen is unclear, vendor or nenufacturer representatives could provide views based ,

upon equipnent tests or capabilities not accessible to others. 4 Substantiation of the probabilities chosen by the panel would not only provide the desired doctnentation but would also give an appreciation of the current understanding and uncertainty between members of the panel.

8. Source Tenn - The marriage of the CET with the five supporting Source Tenn Code Package (STCP) runs is of interest. Blnning of the neny CET end states and the process used to interpolate / extrapolate the STCP results are important and should be well docunented.

9. Contalnnent Leakage One of the principal uses of the CET is to evaluate the like11 hood of certain offsite release paths. Sone paths are associated with larger consequences while others are fai rly snell . The current CET recognized leak paths but does not fully consider either the nuTher of paths nor their potential for pressure reduction.

Anong the possible leak paths are:

MSIV Icak to condenser Drywell head (included now with low likellhood)

Wetwell hatch i Bellons failure (wetwell path)

HPC1/RCIC exhaust Various vent options Leakage through core-wall failure for few cases postulated The leak paths will result in sone release but typically allow deposition and retention / delay thereby reducing the consequences. Without a thorough assessnent it's difficult to say how probable a leak path might be. Experlence with Integrated leak rate tests and judgenent would suggest that the current CET underestinetes the likelihood of leak (no overpressure) fallure substantially.

ARD/cav/07288610

COMMENTS ON APPENDlX A Conment Page A-2: A value of IE-8 as a cutoff appears much too low given the state of art and the phenomena that are not known at such a level.

O Conment Page A-8: "Inerted containments carry an overpressure of nitrogen" Response: This is not true at all inerted plants although Peach Bottcm does carry an overpressure.

o Conment Page A.9: Peach Bottom has 4 diesel generators.

o Conment Page A.9: The ccmnon mode failure rate assuned for Peach Bottcm batteries is unrealistically high.

o Conment Page A.11: HPCI will operate at containment pressures up to 150 psig; RCIC up to 40 psig. RCIC can switch back to CST If water is available (CST refflied). HPCI can switch back as well, but the procedure is scmewhat trore complex.

Ccmnent Page A.12: CRD flow (without operator actions,

. reactor pressure high) is either 110 gpm (scramed) or 55 gpm (no scram). Greater flows exist at lower reactor pressures.

O Ccmnent Page A.13: All 4 RHR punps (nominal flow 10,000 gpn) can function in LPCI, RHR, suppress 1on pool cool 1ng and conta1nment spray nodes. Each of the four LPCS punps is rated at 3125 gpm.

Each of the 4 diesels serves one RHR and one LPCS punp.

o Conment Page A.14: Peach Bottom does not have condensate booster punps .

Comment Page A.15: HPSW does not serve the drywell coolers.

Reactor Building Closed Cooling Water and Emergency Service Water would fulfill this function during loss of offsite power. (Ref.

Question 15, as well)

Conmont Page A.16-Peach Bottom has five ADS valves each with a local accumulator and backup from Instrunent nitrogen, ' nstrunent air and safety grade nitrogen systems.

O Conment Page A,18 (case 2): Operators are aware that reducing pressure will allow on SORV to reseat. Reluctance is " highly unilkely".

Conment Page A.18 (case 3): Concern expressed is conjecture.

Depressurization is directed by procedure and practiced during training.

Ccmnent Page A-21: "CHPCI C RCIC) are inherently capable of maintaining level." Failure of punp seals and NPSH are not the concern of the high temperature limit. Lube oil cooling and subsequent bearing damage are of potential concern with temperatures over 2400 F.

,- -2 o Conment Page A.38- There appears to be some mis-understanding about the venting procedure. For example, there is no need to take specific action to deflate the seals on the 18 in. valves prior to opening them. It is agreed that the definition of

" successful va.nt" is significant. The effect of venting which is not " successful" is not treated in the CET.

Coment Page A.39 (case 1): Appropriate procedure changes are being made.

Conrrent Page A.40: See conments concerning venting on page A.38.

Conment Page A.42: Although a delta P of 25 pst is necessary to open the valves, only 5 ps! is needed to keep them open.

Pnetmatic supply to the SRV's does not isolate on containment isolation signals. Passive backup is continually available from Nitrogen bottles. Valves can be made operable at higher containment pressures by manually increasing the Nitrogen supply pressure (in reactor building or outside reactor building). The instrtment nitrogen system is normally maintained at 100 psig.

l the passive backup supply assures 85 psig without operator action.

C Conment Page A.44 - The Peach Bottom Station Blackout Procedure directs the operator to change HPCI and RCIC suction to the CST to avoid overheating. The procedure also directs load shedding, o Conment Page A.45: Peach Bottom vacuun breakers have been nodified to minimize this type of failure. The calculated results should be viewed as conservative, o Conment Page A.53: "No weight to wetwell failure" Response: The report used by the " experts" as a decision basis is not appilcable due to its assumptions. Wetwell failures are much more likely than assuned. (See NUREG/CR 3653) o Comnent Page A.54: Fission product retention in the reactor building is much different than for a knuckle failure.

Conment Page A.58: Suppression pool failure below the water line is very unlikely. The lower half of the suppression pool is thicker and better supported than the top.

Page A.67,68,69: Provide example of where the conditional failure probability of coolant injection, post containment failure is included. CRD can suffice nuch earlier if minor opera *.or actions are included (P.67).

Conment Page A.78: The Torus Room is designed to maintain water level over the Tee quenchers and LP punp suctions in the event of torus failure.

Corment Page A.89: Deflating the seals is not necessary.

Conment Page A.117: Peach Bottom does not have cable tray sprays.

(Applies to several subsequent questions)

.~.

..* ' o Cornrent Page A.165: Question 82 Case N1 With early failure of containment sprays, no credit is given for f late containment sprays. Case H1 seems ill defined.

Response: This 15 a pessimistic approach which provides no credit for AC power restoration or operator actions.

o Conments Page A.173: Debels coolability is dismissed as not possible.

Response: Quantification by discussion with two Individuals is as inappropriate as identifying published opinions to the contrary as optimistic, with the caveat that it has not been reviewed by the authors of this report. The question deserves better support for its quantification. If expert opinion is to be relled upon, a more rigorous approach is needed.

Conment Page A.177,178: "The IDCOR Task 23.1 assessment was that the drywell temperature would rise sufficiently to cause loss of seal Integrity" Response: Task 23.1 did not use penetration seal failures as the failure node of containment but they were Identified to possibly fall at elevated temperatures.

l l ARD/ Jet /07298601 1

1 i

- ------.-._______.___,.m

4 .r. l-C044ENTS ON APPENDIX B

. Ccmrent Page B.6' (Question 28):

~

The location of containment failure due to overpressure is appropriate for calculation'rather than expert judgement. If expert' Judgement Is to be used, the group should include.the authors of.the several analyses which-have been done'on Mark I containments as well as the designer of

.the containment. The effect of the concrete shleid wall should be included in the expert discussion,.as well as other plant specific details, o Corrment Page B.7 (Question 29) See conment on question 28.

. Ccmnerit Page B.7 (Question 32): For this plant spec'Ific' analyste plant specific offsite power recovery estimates should

.be used. :This should include the effect of the eleven unit hydroelectric plant with blackstart capability which supplies one of the offsite power feeds and the capabilltles of the Owner's emergency organization to marshall offsite resources.. Is there a reference for the ASEP non-recovery probability?

Conment Page B.19 to B.24 - The hand-calculation is a useful

~

adjunct to the expert's voting 'and should be encouraged. The support for the expert Judgement which is relled upon for questions 68,76 and 77 is not included. These define the drywell melt-through probability.

ARD/ Jet /07318601-l l

E-_ ____ _ _ _ _ . _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _