ML20247D915
| ML20247D915 | |
| Person / Time | |
|---|---|
| Site: | 05000605 |
| Issue date: | 09/08/1989 |
| From: | Scaletti D Office of Nuclear Reactor Regulation |
| To: | Murley T NRC |
| References | |
| NUDOCS 8909150132 | |
| Download: ML20247D915 (10) | |
Text
- _ - - -
ye g N-, gdFQsivg'o[
.. UNITED STATES "8'
n NUCLEAR REGULATORY COMMISSION j'ip'f qIl WASHINGTON, D C. 20555 g September 8.-1989' R ~
' Dbcket 'No. ' STN S0-605 m
l : MEMORANDUM FOR: T. Murley* B. Grimes P. McKee J. $niezek* F. Congel A. Thadani.
F. Miraglia* J. Roe C. McCracken S. Varga* . C. Grimes J. Larkins*'
D. Crutchfield* B. Boger W. Lanning G. Lainas E. Adensam .T.; Martin, EDO G. Holahan M. Virgilio Operations Ctr.**
C. Rossi W. Travers' F. Gillespie L. Shao B. D.-Liaw W. Bateman L. Reyes, RII**
THRU: Charles L. Miller, Director gns
. Standardization and Life Extension
. Project Directorate .
Division of Reactor Project - III, IV, V- and Special Projects FROM: Dino Scaletti, Project Manager Standardization and Life Extension Project-Directorate .
Division of Reactor Project - III, IV, Y and Special. Projects
SUBJECT:
- DAILY HIGHLIGHT - FORTHCOMING MEETING WITH GENERAL ELECTRIC COMPANY TO DISCUSS THE STAFF'S REVIEW OF THE-PROBABILISTIC RISK ASSESSMENT FOR THE ABWR. . DISCUSSIONS WILL ALSO FOCUS ON REFINEMENTS IN THE INTERNAL EVENTS PRA THAT HAVE RESULTED FROM CONSIDERATION OF THE ABWR DESIGN SAFETY ENHANCEMENTS. AN EXTERNAL EVENTS PRA OVERVIEW WILL ALSO BE PROVIDED. (SEE ATTACHED LIST OF DISCUSSION TOPICS)
DATE: 8. TIME: September 12-13, 1989 9:00 a.m. - 4:30 p.m.
LOCATION: General Electric Co.
175 Curtner Avenue San Jose, CA.
s am:u GC&
m m kDO;) O cv ny Gi }\
4 i TV
____i_________ _ - -
.~
n PARTICIPANTS'* : NRC GENERAL ELECTRIC P.'Niyogi, RES .J. Quirk T. Pratt, BNL J. Duncan E. Che111ah..RES J. Fox R. Fitzpatrick, BNL' C.-Hsu, BNL D. Scaletti, NRR Dino Scaletti. Project Manager Standardization and Life Extension Project Directorate Division ~of Reactor Project - III, IV, Y and Special Projects
- Meetings between the NRC technical staff and applicants or licensees are open for interested members of the public, petitioners, interveners, or other parties to attend as observers pursuant to the "Open Meeting Statement of NRC Staff Policy", 43 Federal Reg ~ister 208058,6/28/78. However portions of this meeting may be c15ied t6~ifie p6511c to protect General Electric Company proprietary information. Members of the public who wish to attend should contact D. C. Scaletti at-(301)492-1104' .
I 4
i
a e, .
- ATTACHMENT A following list of questions have been raised during the design certification review of the event trees and fault trees shown in Appendix 19D (Probabilistic Evaluations)of the ABWR Safety Analysis Report
- 1. In most of the currently available BWR PRAs, the loss of offsite power sequence with successful recovery of offsite power within 30 minutes (i.e., TM sequence in Fig. 190.4-4) is transferred to the MSIV closure (i.e.,isolationevents) event tree. Please provide the basis for transferring it to the reactor shutdown tree (i.e., Fig. 19D.4-1) instead.
- 2. Should not the event tree top event, Q (Feedwater), appearing in the reactor shutdown event tree (Fig. 19D.4-1) be replaced by "Feedwater and PCS"? Otherwise, a branch should be added to the uppermost sequence (with an end state of OK) to determine the success or failure of the top event, W. Note that condenser problems (hardware or others) can lead to a manual shutdown.
- 3. Please provide the basis of not crediting automatic depressurization for the safety function, X, in the reactor shutdown event tree (Fig.
19D.4-1). Also, why no credit is given to LPCS (low pressure core spray) for the safety function, V, in the same event tree?
- 4. Does ABWR have a design feature which allows the operator to utilize RCIC in steam condensing mode to transfer reactor decay heat to the ultimate heat sink? If yes, why is no credit given to such a feature in evaluating the sefety functionW(containmentheatremoval)?
- 5. In essentially all of the event trees shown in Fig.190.4-1 through Fig.
19D.4-14, failure of the W function (long-term heat removal) is assigned a probability of failing to run RHRA or RHRB or RHRC rather than failing to start and run RHRA or RHRB or RHRC, if the preceding V function (RHR injection or condenser) is a success. This would be correct if one of .
the RHR pumps was successfully started and run to accomplish the mission of the V function, and then switched to a long-term heat removal mode.
Note, however, that success of the V function can also be achieved, as indicated in Table 19.3-2, by using one condenser pump and one condenser i transfer pump. In such a case, the approach taken in the ABWR FSR will underestimate the failure probability of W since the RHR pump has to be started and then run throughout the mission time. Also, can one low pressure RHR pump alone always accomplish the missions of both the V and l the W functions for all the transients including a large LOCA?
1
- 6. In both the non isolation event tree (Fig.19D.4-2) and the isolation / loss of feedwater event tree (Fig. 19D.4-3), the uppermost sequence (with an end-state of OK) should branch out at the top event, W, since success of Q (feedwater) alone does not automatically warrant success of W. The same comment also applies to the 10RV event tree (Fig.
19D.4-11).
t ___
( - -
p
- 1 .
1 ;
1 3 . 7.<
- Inallbutoneeventtree-(themediumLOCA'eventtree, Fig. 19D.4-13 l is the only exception), the event' tree top. event, X, is assigned a single
' failure probability of 2.CE-03. Sometimes, it is referred to as " ADS or manual depressurization," and at other times, simply " manual ~
depressurization.". Which of these:is correct? Should not the probability
'be different if.it. includes failures of both ADS and manual depressurization? Please also explain why only for the medium LOCAs, the value is much smaller (6.1E-06).
8.- .In the non-isolation event tree (Fig 19D.4-2), what is the basis of crediting RCIC.if there are stuck-open SRVs and loss of feedwater?
9.- In Table 190.4-1 through Table 19D.4-17, the branch-point 'value of the safety function V (LPFLA or LPFLB or LPFLC available)wasassigneda value of 7.92E-02, with the source of the data given as Table 19D.4-1.
No such data, however, can be found in Table 19D.4-1. Also, for the loss of offsite power event trees, failure of V (LPFLA or LPFLB or LPFLC or
-one condensate and one condensate transfer pump) is given a value of 2.6E-02. Again, no such data can be found in the tables. Please explain how this value was calculated. 10. In the isolation / loss of feedwater event tree, should the event tree top event heading, RCIC(U), be replaced by Uhr, which includes both HPCF ane: RCIC?
. 11. For isolation / loss of feedwater events, successful RHR operation using the PCS requires reopening of the MSIVs and the recovery of feedwater if it is initially lost. In Fig. 190.4-3, which event tree top event takes into consideration the reopening of MSIVs? Also, will the chance of reopening the MSIVs be smaller if there are stuck-open SRVs?
- 12. In the loss of offsite power and station blackout event tree (Fig.
generators 19D.4-4),)the (7.76E-04 is used probability of failing to sort out stationallblackout three diesel sequences (i.e.,BE2, BE8, and BEO) from the loss of offsite power sequences (i.e., TE2, TE8, and TEO). Note, however, that "all DG not fail" could mean: (1)oneDG is available, (2) two DGs are available, or (3) all three DGs are available. In Figs. 19D.4-5 and 19D.4-6, the unavailability of Uh (HPCF B or C with a probability of 1.58E-02) was computed based on the assumption that two diesel generators are available. If only one DG is available at the onset of loss of offsite power, this unavailability 1 could become larger. It appears that some kind of weight-averaging ,
should be applied to modify this value based on the probabilities of {
having either one or two DGs when the loss of offsite power occurs.
Also, in Fig.19D.4-4, the failure probability of opening SRVs following an ATWS event was taken to be 1.0E-06. For ATWS events, a large number (15) of SRVs need to be opened for pressure relief, and, hence, the failure probability of opening the required number of SRVs can be expected to be larger.
- 13. In the TE2 and TE8 event trees (Fig. 19D.4-5 and Fig. 19D.4-6), all the branch point probabilities for W(RHR) were computed incorrectly, because they are inconsistent with those shown in Table 19D.4-1 which are applicable to the case of loss of offsite power. i l
l l
i
= _ _ _ _ _ _ - - _ - - - - - - - - - - )
jw .
o m 14, in all of the loss _of offsite, power event trees (Figs.119D.4-5, 4-6,L and i W .4-7),-the failure probability of. HPCF (Uh) is taken to be the same-irrespective of the offsite power ' recovery time and regardless of
^
whether there are stuck-open SRVs. . Can the heating up of suppression
? pool' for a prolonged period of time due to stuck-open SRVs adversely affect the availability of'HPCF7
'15. LPlease provide the basis of not considering stuck-open SRVstin'the. h I.- stationblackouteventtree(BE2, Fig.19D.4-8).
~16. In the'same event tree cited above-(item 15),: the failure probability of W(RHRA or RHRB or RHRC) is taken to be' 5.19E-04,' which does not' correspond to-that (1.59E-03) shown' in Table'19D.4-1:for the case of '
loss of offsite power. Are the' values shown in the column under the heading-of " Loss _ of Offsite Power" in'. Table 19D.4-1 also applicable to station blackout? If not, please explain.
- 17. . 'In 'the station blackout event. tree (BE8, Fig.19D.4-9), why does the sequence with success of RCIC need to be branched out for testing the.-
success of HPCF? ~ According to the- success criteria ' listed in Table -
'19.3-2, successful core cooling using a high pressure-system can be
. achieved.by using either RCIC or one train.of HPCF for all transients including loss of offsite power. Furthermore, both HPCF and LPFL-require ac power.which, in this. case, is not available for nearly 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />.- Please explain why_both HPCF and LPFL are included as. event tree top events. .
19.- For 10RV transients, there is no immediate automatic scram signal, and the operator may be_ required to manually scram the reactor and start the makeup system before the suppression pool' temperature exceeds the heat capacity temperature limit. Please provide the. basis of not' including
" timely manual scram" as.an event tree top event in the :IORY event tree
_(Fig.19D.4-11).
-20. Does the event. tree top event, Uh, appearing in the 10RV event tree include both RCIC and HPCF or just HPCF? If it- includes RCIC, the branch-point value shown in the figure appears to be inconsistent with that shown in Table 19D.4-1. If it does not' include RCIC, please explain'why RCIC is not credited.
- 21. Please explain why feedwater (Q) was not credited as a viable means of core' cooling in the small LOCA event tree (Fig.19D.4-12). Note that, according to the success criteria shown in Table 19.3-2, feedwater can be used to successfully cool the core in the event of a small steam LOCA.
- 22. Please explain why HPCF is given credit in the large LOCA event tree (Fig. 19D.4-14) despite the high degree of depressurization caused by the large LOCA.
- 23. Please provide justification of not considering vapor suppression in the large LOCA event tree.
~
'24. In constructing the ATWS event tree (Fig. 19D.4-15), no distinction was made between ATWS events with MSIV closure (isolation) and those with bypass available (non-isolation), although the former is generally more severe and limiting. Please explain why the same branch-point probabilities were = used in quantifying the ATWS sequence frequencies despite differences in the success criteria, such as the time available for the operator to inhibit ADS or the unavailability of normal heat removal system for containment heat removal (see Table 19.3-3).
- 25. Some of the headings depicting the event tree top events in the ATWS event tree, such as those for ADS inhibit, RCIC and feedwater or HPCF, are misplaced. Also, the heading for long-term heat removal (W) is inconsistent with that shown in Table 19D.4-18 or Table 190.3-3.
- 26. It appears that the low core-damage frequency (9.1E-09/RY) found for ATWS sequences is mainly driven by the low initiating event frequency (9.34E-09/RY),whichwasobtained by taking scram failure probability (C) to be 1.0E-08. Please explain in detail how this scram failure probability was calculated. From the fault tree developed for a single control rod drive (Fig. 19D.6-17a, Figure 1), the probability of failure to insert an individual control rod can be estimated to be roughly 3.0E-06. No explanation, however, is given as to how this probability is used to generate the probabilities of the basic events shown in the Figure 1).
Also, no probability data is given for the eventfaultRPS tree of control rod drive system initiate scram) appearing in the fault trees for reactivity control (Fig. 19D.6-16b).
- 27. In Table 19.3-3, the time available for the operator to initiate one train of St.C is given to be 10 minutes for both isolation and non-isolation ATWS events. Should not the time available for the former be shorter because the suppression pool is heated up sooner?
- 28. For an ATWS event which is initiated or accompanied by closure of all MSIVs or loss of condenser, can adequate core coolant inventory be maintained by RCIC alone (as indicated in Table 19.3-3)? For some BWRs of current design, such an event requires HPCI or a combination of HPCI and RCIC.
- 29. In quantifying ATWS sequence frequencies branch-point value ,
wasusedforW(containmentheatremoval)thesame regardless of whether there )
are stuck-open SRVs. Was suppression pool heating due to stuck-open ;
SRVs taken into account in estimating the failure probability of W? I
- 30. Is there any reason why the event tree top event " ADS inhibit" in the ATWS event tree is placed before "Feedwater or HPCF" and "RCIC" although it appears more logically correct to place it after the latter top events?
- 31. Was any functional event tree or fault tree developed to analyze the unavailability of feedwater, condensate, and condenser system? How was the unavailability of feedwater (Q), for example, evaluated for different transient initiators?
l L-_-_____-____--_____ _ _
y- -
k 32. -In_ the event tree quantifications, the frequency of a particular-
- accident sequence was obtained by multiplying .together the initiating event frequency and the branch-point probabilities of the failed safety
' functions, such as U, V, or.W, appearing in the sequence description.
This approach :is proper if the branch point probabilities were evaluated by_ properly accounting for the common-mode failures among .the event tree top events by linking together-the relevant fault trees. Were these fault-tree linkings done in the ABWR analyses to obtain the upper-bound L ~o f. minimal cut sets for safety function failures, such as UV, QUV, or UVW7 If not, please' explain how the branch-point probabilities were calculated for. the _ individual safety functions, such as. U, V, or W.-
.' 33. - Were all the system failure probabilities.(except for RCIC) listed in Table 19D.4-1 obtained by quantifying the fault trees shown in Section
.19.D.67 Were the. probabilities of failing all ECCS systems computed by linking the high pressure and low pressure system fault trees? If so, which mode of the low pressure system was used? Also, were these values actually used in the. event tree quantifications?
-34.- Were the fault trees for the support systems, such as electric power system, service water system and instrumentation system, individually quantified? .Are the results of such fault tree quantifications (in -
terms of minimal cut sets) available for comparison.with BNL
. calculations?-
-35. A preliminary calculation by BNL using the fault trees and the basic event input data shown in Section 19 D.6 indicates that the failure probability of. the RCIC system is roughly 0.128. This is almost a factor.of three larger than those shown in Table 19.D.4-1, which are footnoted to be based upon operating plant ~ performance. Please justify the use of the latter values in the accident sequence quantifications.
- 36. What modifications to the fault tree input data were made to obtain the system failure probabilities corresponding to loss of offsite power (lastcolumnofTable19.D.4-1)? Was the failure of switchgear taken into consideration when the failure probability of the W function (for i example, in Fig. 19.D.4-7) was calculated?
- 37. Please briefly describe the possible impacts of omitting the development of system fault tree for plant air system on the frontline and the support systems.
- 38. It was noted that a very small fraction of the failure data shown in Table 19.D.6-2.through 19 D.6-7 are inconsistent with those shown in the relevantfaulttrees(forexample, DIV2 MUX, HMV14BHW and HXV032CQ in Table 19 D.6-2). Which values were actually used in the fault tree
.quantifications?
- 39. The break areas for the various LOCAs (large, medium, and small)are defined to be significantly larger than those used in, for example, the Limerick PRA. Do the initiating event frequencies used in the event -
tree quantification reflect these changes in the definition of break '
sizes?
g . _ - _ _ - - - _ _ _ _ __. - - - _ _ -
w, 40.: Please provide all background triformation used for the quantification of f :the split fractions..(branch probabilities) cf U. containment event trees Figures 19D.5-4 through 190.5.15). Please also provide theithermal
. hydraulic definitions of the following, terminologies' used during the' quantification:.
p >
. : core damage.
L . . core melt ' '
p
.. -core melt arrest in the vessel
. core melt arrest in the containment
- 41. . Please provide detailed thermal-hydraulic, structural
- and experimental.
evidence to . justify the effectiveness of core melt arrest after vessel failure, while ensuring that the containment maintains its structural
' integrity..
42.- Please provide detailed thermal-hydraulic calculations showing that-the variations of accident progression within each accident class are relatively small compared with the variations between accident classes to be binned together.
- 43. Please provide detailed information showing actual available time for evacuation in Table 19E.3-6 for the consequence calculation.
'44. .Please prcvide the following information:
.5.1. The structural details of the core and RPV (e.g., the surface I
'l
. area, thickness, elevation, mass and the diameter of the upper and lower grid plates, core barrel, bottom and upper heads, etc.)
.a. The normal water level and water inventory in various tanks, and ,
systems. (e.g. RCS). I
- b. Radial power distribution
- c. SRV flow area
- d. Details of control blades and box 1
- e. Total structural mass below the reactor vessel in the dry well
- f. Mass fraction of iron in the containment concrete (e.g. reinforcing, etc.)'
- g. Please. provide a set of MAAP input files and the code listing with the modifications for the ABWR analysis.
- h. Type of concrete and its composition
- 1. Pleas 3 provide a set of CRAC input files used for the consequence calculations.
- 45. Please provide the following document:
- a. NED0 24011 PA
- 46. Please provide clarification to the following:
- a. In Section 19.1.2, the probability of containment failure due to loss ;
of heat removal is given as 1.8E-9. This probability corresponds i only' to the CET sequence number 4 in Figure 19D.5-12 and does not l reflect-the overall performance of the ABWR containment. Please ]
clarify the corresponding paragraph in Section 19.1.2.
1 f
F \
,,y.. -
I L'
- b. Please clarify the units'of the societal risk and individual risk on
- p. -:l,' page 19.1-1..
L c. The numbers.in Tables 19.3-4 are not consistent with that of Figures 19D.5-4 through -15. . Please clarify.
d.' : What is the' status of- the containment in the temperature. range between 500 deg. F and 1000 deg. F (Page 19.3-10)? Please provide
.the basis for the estimated leak area vs. pressure table (page i =19.2-4).
- e. ' Table.19.3-4 does not include the frequency of the suppression pool bypass sequence. . Please clarify.
-f. Please clarify the process and provide the rational for the interface between the CET analysis and the release category definition.
S l
l w______ _ _ _ _ _ _ _ _ -- _ _
l-
.j ,
- i. ..u.. ...
l..
~2~ September 8. 1989 l
PARTICIPANTS *: NRC GENERAL ELECTRIC lr.
e P. Niyogi, RES- J. Quirk T. Pratt, BNL' J. Duncan E. Che111ah, RES 'J. Fox R. Fitzpatrick, BNL C. Hsu, BNL l D. Scaletti, NRR l
/s/
Dino Scaletti, Project Manager Standardization and Life Extension Project Directorate Divis1orr of Reactor Project - III, IV, Y and Special Projects
- Meetings between the NRC technical staff and applicants or licensees are open
-for interested members of the public, petitioners, interveners, or other parties to attend as observers pursuant to the "Open Meeting Statement of NRC Staff Policy", 43 Federal Register 208058,6/28/78. However portions of.
this meeting may be closed to theT>uS11c to protect General Electric Company proprietary information. Members of the public who wish to attend should contact D. C. Scaletti at (301)492-1104.
DISTRIBUTION:
CentralaFile? B. Grimes.
NRC~PDR"~ Receptionist PDSNP R/F NRC Participants ACRS (10) V. Wilson D. Scaletti L. Thomas E. Hylton H. Vandermolen GPA/PA OGC E. Jordan M. Cunningham CV PDSLE etti:sg CMiller 9/7/89 9/g /89 i
i
,n(