ML20128P893
| ML20128P893 | |
| Person / Time | |
|---|---|
| Issue date: | 02/18/1993 |
| From: | Kennedy J Office of Nuclear Reactor Regulation |
| To: | Office of Nuclear Reactor Regulation |
| References | |
| PROJECT-679A NUDOCS 9302250181 | |
| Download: ML20128P893 (6) | |
Text
.. _ _ _. _.
- ..A i
" g[pa nseg.Io,
- UNITED STATES -
f,
[
p, NUCLEAR REGULATORY COMMISSION g
l W ASHtNG T ON. D. C. 20$55 2
k...*/
February 18, 1993 a
t Project No. 679 ORGANIZATION: Atomic Energy of Canada, Ltd., Technologies (AECLT)
SUBJECT:
MEETING WITH AECL TECHNOLOGIES TO DISCUSS THEIR COMMENTS ON THE l
STAFF'S DRAFT SECY PAPER ENTITLED, " ISSUES PERTAINING TOLTHE l
ADVANCED REACTOR (PRISM, HHTGR, AND PlVS) AND CANDU 3 DESIGNS i
AND THEIR RELATIONSHIP TO CURRENT REGULATORY REQUIREMENTS" On February 2, 1993, members of the U.S. Nuclear Regulatory Commission's (NRC)
Advanced Reactors Project Directorate (PDAR) met with representatives of AECL Technologies (AECLT) to discuss AECLT's comments on the staff's draft SECY paper entitled, " Issues Pertaining to the Advanced Reactor (PRISM, MHTGR,.and PlVS) and CANDU 3 Designs and Their Relationship to Current Regulatory Requirements." AECLT provided their comments on the draft SECY_ paper in a i
letter to the staff dated January 25, 1993. includes a list of attendees. Enclosure 2 provides the January 25, 1993, response from AECLT.
Robert Pierson, PDAR Director, stated that the purpose of the meeting was to discuss AECLT's comments in their January 25, 1993, letter (Enclosure 2).
In:
these comments, AECLT noted that the NRC's new schedule for completion of the CANDU 3 preapplication review aushes submission of the CANDU 3 design certification application to tie end of the 1995-96.timeframe. Mr. Pierson-questioned AECLT what was meant by that-statement. AECLT indicated that their design certification application would most likely be submitted in mid.to late 96 because they need approximately 2 years from the preapplication review to incorporate the staff's findings into the final design.
AECLT's comment letter indicated that their intent to go forward with an.
application for certification of the CANDU 3-design is 1ndependent of any schedule for building a CANDU 3 reference plant in Canada. Mr. Pierson indicated that, although a CANDU 3 reference plant would help the staff during the design certification stage, it is not a requirement.
Mr.- Pierson also.
indicated that because the staff-is attempting to treat the CANDU 3 as an evolutionary design, it would not require an entire plant prototype.; However, that does not mean that the staff would not require testing of certain aspects of the_ plant design, such as on-line refueling, if the staff deemed it necessary.
The cover letter on Enclosure 2 response states, "AECLT 'is looking to the-preapplication' review to resolve all-issues identified during the review, in the sense that the PSER identifies the information which AECLT must k[i j
240003 p
susm num cm n 9302250181 930218 PDR PROJ 679A PDR
AECL Technologies February 18, 1993 provide in the CANDU 3 design certification application in order for the staff to successfully complete its review of that application." The PDAR staff pointed out that providing AECLT with complete issue resolution is~not the purpose of the preapplication review.
The intent of the preapplication review in the U.S., and to provide direction to the preapplicant regardinis to staff would need to resolve the issue during design certification.g what the To that end, the staff has already identified severai technical and policy issues for the CANDU 3 design. The staff will attempt to resolve issues to the greatest extent possible, but the preapplication review does not guarantee that the issue will not be revisited at the design certification stage.
l the staff cannot guarantee that.311 issues will be identified in theFurthennore, preapplication review.
In addition, the preapplication review will not go into the detail the staff would expect at the design certification stage. to Enclosure 2 contains AECLT's specific comments regarding the six issues identified as applicable to CANDU 3:
accident evaluation, source term, containment performance, operator staffing, positive void reactivity, and control room design.
At the beginning of Attachment 1, AECLT states, "The ace' dents listed by the staff are not excluded from consideration in CANDU, but have little consequence because of the redundant shutdown systems."
PDAR staff reiterated the fact that even though the CANDU 3 design has twoThe safety-grade, independent, and diverse shutdown systems, the staff is still interested in understanding the behavior of the design should both shutdown systems fail.
intent to evaluate the consequences of severe accidents in both theThe sta evolutionary LWRs and in the CANDU 3 and advanced reactor desigas.
In the draft SECY paper commented on in Enclosure 2, the staff states "The CANDU 3 preapplicant in their current safety analyses, has exc of the consequences o,f events with frequencies of less than 10',luded analyses
/ year from the safety evaluation."
AECLT indicated that this statement was inaccurate, because they evaluate a single loss-of-coolant accident coincident with a loss of emergency core cooling (estimated frequency of 7.6E',).
The staff maintains that except for this event, AECLT has not provided consecuence s lY5es for events beyond 10',.
However, the staff did agree to clarify the previous statement in the draft SECY paper to more accurately reflect AECLT's philosophy.
In Attachment 1 to Enclosure 2, under Accident Evaluation, AECLT stated that limits are not placed on the scope of severe accidents that may be considered in designing the CANDU 3.
The staff takes issue with this statement because AECLT maintains they do not have to evaluate the consequences of events involving a failure to shut down due to the low likelihood of the event happening.
The staff has been clear on the issue of severe accidents, and specifically on failure to scram events on the CANDU 3 design. The staff has requested AECLT to perform consequence analyses of severe accidents, which
s AECL Technclogies February 18, 1993 would include events with failure to scram which, for CANDU 3, includes postulated event sequences resulting in substantial core damage. requested that the es.
The staff agreed to provide such a request.
On the issue of source terin, AECLT said in Enclosure 2 that *...we do not agree that a pressurized heavy-water reactor is so fundamentally different from LWRs that it should require a different methodology for establishing source terms than that which NRC is now in the process of establishing for evolutionary LWRs." The-staff's policy issues paper did not state that a different method for each design was contemplated.
It is the staff's intention to apply the methodology currently being developed in NUREG-1465,
" Accident Source Terms for Light Water Nuclear Power Plants," (a draft report for comment) to the CANDU 3 design to the extent-possible.
Regarding containment, AECLT did not believe that this issue should apply to CANDU 3.
The staff pointed out that the issue does ap)1y to CANDU 3 secause the CANDU 3 containment is not an essentially leak-tigit-structure as previously defined for light-water reactors. Currently, the staff accepts up to approximately 0.5 percent per day design leakage rate out of LWR containments.
The CANDU 3 containment design has a test acceptance leakage rate of 2 percent per day, and use 5 percent per day leakage in the safety analysis. AECLT and the staff agreed that this issue does in fact apply to CANDU 3.
The draft SECY paper identified o>erator staffing as an issue for the CANDU 3 design. AECLT does not believe tais is an issue because they intend to meet the NRC staffing requirements at the design certification stage.
The staff agreed to propose removal of CANDU 3 from the applicability matrix in the draft SECY paper.
On the issue of positive void reactivity, AECLT stated that they do not agree with the staff's issue statement in the draft-SECY 3 aper, in addition, in Attachment I to Enclosure 2, AECLT states, "...the my question is whether the reactivity shutdown systems are reliable enough to reduce the frequency of extremely low value (i.e.,10'pilure g all reactivity shutdown systems to an reactivity insertions with a f to 10' peryear)." Pie staff reiterated that this is not the key (ssue. The issue is whether or not the staff should accept a design in which the positive void reactivity increases dramatically during certain events.
In order to evaluate the positive reactivity insertion events in a CANDU 3, the staff intends to require the analysis of the consequences of events which could lead to large reactivity insertions. As
-discussed previously, this would include events involving a failure to shutdown.
Finally, on the issue of the control room and remote shutdown area design, AECLT believes that the staff's proposed recommendations preclude the evaluation of the approach used in CANDU 3 for control room and remote-
,g' AECL Technologies February 18, 1993_
q
.i shutdown area design.
The staff intends to :>ropose the issue to the.
Commission for guidance, but at this time, t1e staff's recommendation to the Commission is to-require a seismically and electrically qualified main control room.
Original-signed by:
Janet L. Kennedy, Project Manager Advanced Reactors Project Directorate Associate Directorate for. Advanced Reactors and License Renewal Office of Nuclear Reactor Regulation
Enclosures:
1.
List of Attendees l
2.
Ltr. 01/25/93 AECLT to NRC w/attch, cc w/ enclosures:
See next page t
i
- TNMk Distribution:
RMeyer,-RES BPM:PDAR:ADAR Central file JLKennedy LA:ADA
/
PShea
" ":: x d;::: -
PDR JNDonohew-
-g / g 3-g/g93 TEMurley
'WVpshaw FJMiraglia :-
PShea SC:PDAR:ADAR
):PDAR:ADAR DMCrutchfield ACRS-(10)
@iersen
'RCPierson 0GC:
THCofth i
4/I7 /93 dt/t7/93 THCox
.EJordan EDThrom PDAR R/F BUpshaw-CANDU R/F;
.0FFICIAL RECORD COPY-1 Document Name:
SUMRY2.2-h
AECL Technologies february 18, 1993 CANDU Project No. 679 cc:
Louis N. Rib, licensing Consultant AECL Technologies 9210 Corporate Boulevard, Suite 410 Rockville, Maryland 20850 Bernie Ewing, Manager Studies and Codification Division Atomic Energy Control Board P.O. Box 1046, Station B 270 Albert Street Ottawa, Ontario, Canada KlP SS9 A.M. Mortada Aly, Senior Project Officer Advanced Projects Licensing Group Studies and Codification Division Atomic Energy Control Board P.O. Box 1046, Station B 270 Albert Street Ottawa, Ontario, Canada KlP SS9 Project Director - CANDV-3 AECL CANDU 2251 Speakman Drive Mississaugua, Ontario, Canada L5K 182 L. Hanning Muntzing Newman & Holtzinger, P.C.
1615 L Street, N.W., Suite 1000 Washington, DC 20036 Steve Goldberg, Budget Examiner Office of Management and Budget 725 17th Street, NW.
Washington, DC 20503 Mr. A.D. Hink Vice President / General Manager AECL Technologies 9210 Corporate Boulevard, Suite 410 Rockville, Maryland 20850
ENCLOSURE 1 MEETING WITH NRC AND AECL TECHNOLOGIES TO DISCUSS COMMENTS ON THE NRC's DRAFI SECY PAPER REGARDING KEY POLICY ISSVES FOR THE ADVANCED REACTOR AND CANDU 3 DESIGNS FEBRUARY 2, 1993 ATTENDEES Hamn Affiliation Janet L. Kennedy NRC/NRR/ADAR/PDAR Edward D. Throm NRC/NRR/ADAR/PDAR Louis N. Rib AECL Technologies Ray W. Durante AECL Technologies Robert L. Ferguson AECL Technologies Michael H. Fletcher AECL Technologies Robert T. Curtis AECL Technologies Robert C. Pierson NRC/NRR/ADAR/PDAR Herbert Feinroth AECL Technologies A.D. Hink AECL Technologies Jack N. Donohew NRC/NRR/ADAR/PDAR h-
ENCt.05URE 2 A
ATC.
AECLTechnologies g a. %
9210como*a'e Bouleva d w, p,,,,,,,, q Genev penager Sune410 Rockde Maryland 20050 usa AECLT 93016
$$^oo^j7C' Fae (301)417 0740
""'03"2 Project 679 January 25,1993 Mr. Dennis M. Crutchfield Associate Director for Advanced Reactors and License Renewal Oflice of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington. DC 20555 Re:
Commission Papers on Policy Issues and Schedules Conceming the Preapplication Reviews of Advanced Reactors and CANDU 3 Designs.
Dear Mr. Crutchfield:
His letter is in response to your letter of December 16,1992, which provided AECL Technologies (AECLT) with two NRC Staff papem conccudng the preapplication resiew of the CANDU 3 design. One paper was SECY-92-393 conceming " Updated Plans and Schedules for the Preapplication Reviews of the Advanced Reactor (MHTGR, PRISM, and PIUS) and CANDU 3 Designs." De other was a draft SECY paper entitled " Issues Pennining to the Advanced Reactor (PRISM. MIITGR, and PIUS) and CANDU 3 Designs and Beir Relationship to Current Regulatory Requirements." We have reviewed the two papers and address each in tum.
SECY-92-393 establishes a revised schedule for completion of the preapplication resiew of the CANDU 3 design.. According to SECY-92-393, the draft Preapplication Safety Evaluation Repon (PSER) is to be issued in June 1994 and a final PSER in December 1994i Ris represents a significant change of twelve months over the earlier scheduled completion of June 1093 for the draft PSElt Be June 1993 date prosided time for AECLT to address any issues raised in the PSER and to submit its application for certification of the CANDU 3 design in the early part of the 1995-96 timeframe. AECLT has chosen the 1995-96 timeframe because it would allow a certified CANDU 3 design to be available to U.S. utilities in the L
L expected timeframe for the placement of new orders for nuclear power plants. De new l -
schedule of December 1994 pushes submission of the CANDU 3 design certification
.n r1 i 1
C 0 0 0 7,f) i 13 9gyvaoin
_ %,=
l 679A PM
L Mr. Dennis ht Cmtchfield Page January 25,1993 application to the end of the 1995-96 timeframe, with little to no allowance for contingency or slippage. Herefore, it is critically imponant that the new schedule in SECY-92-393 be faithfully adhered to and not extended.
In providing the Commission with relevant background on the CANDU 3 design. SECY 393 indicates that Atomic Energy of Canada Limited (AECL) is " negotiating to stan constmetion in a Canadian province which could serve as a prototype for the CANDU 3 design in the U.S." and that AECL "would re evaluate its design cenification plans in the U.S. if Canadian constmetion plans did not materialize." Similarly the dran issues paper states that "a CANDU 3 reference plant is a key element in [AECLTs) plan for standard design certification." Bese statements require clarification in two aspects.
First, as the draft issues paper makes clear, and AECLT fully endorses, the CANDU 3 design is an evolutionary heavy water design deriving from CANDU designs opemting in Canada and elsewhere, for which there is over 200 reactor years of full power opemting experience.
Consequently, a prototype CANDU 3 is not required for design certification. Also, as the draft issues paper makes clear, while a reference plant built in Canada would greatly benefit the Staffs resiew of the CANDU 3 design, building such a plant is not necessary for cer'ification of the CANDU 3 design. ilather, what is ofimponance is the relevant opemting experience of the CANDU plants from which the CANDU 3 design evolved.
Second, as to potential availability of a reference plant in Canada, AECLT is pleased to inform the NRC that on December 21,1992, the Govemment of Saskatchewan and AECL signed a Memorandum of Understanding (MOU). De MOU provides, among other things, for completion of the design and engineering for the CANDU 3, including the contribution of
$20 million in matching funds by the Govemment of Saskatchewan to those being contributed by AECL Dese funds are in addition to the approximately 5100 million already spent by AECL over the past 5 years. In this and other respects, the relationship between the Govemment of Saskatchewan and AECL is sirnilar to that between the Depanment of Energy and the Advanced Reactor Corpomtion in the U.S. Fitst-of a Kind-Engineering efron. He MOU represents further progress in the advancement of CANDU technology as emtxxiied in the CANDU 3. His progress notwithstanding, it is imponant to understand that our intent to go fonved in the 1995-96 timeframe with an application for cenification of the CANDU 3 design in the United States is independent of any schedule for building a CANDU 3 reference plant in Canada.
He draft Policy Issues Paper discusses the present scope of the CANDU 3 pre-application review, indicating that the StalT has resised the scope of the issues considered at the preapplication review stage, limiting them "to those which could affect the licensability of the proposed desigrt" AECLT is looking to the pre-application review to resolve all issues identified during the review, in the sense that the PSER identifies the information which
Mr. Dennis hi Crutchfield Page January 25,1993 AECLT must provide in the CANDU 3 design certification application in order for the Staff to successfully complete its review of that application.
AECLT is especially pleased to have the opportunity to comment on the draft Policy Issues Paper prior to its being finalized for submission to the Commission. AECLT would like to address the six substantive issues identified in the draft parrr as relating to the CANDU 3 design; specifically, Accident Evahiation, Source Temi, Containment PcTformance, Operator Stafling. Positive Void Reactivity and Control Room Design. In formulating these comments, AECLT has followed the Staffs distinction between Advanced Reactor issues and CANDU 3 issues and, consequently, addressed only those comments specifically applying to the CANDU 3 design. '1hese issues are discussed in detail in Attachment 1 to tids letter.
If you have any questions regarding this letter or the attachment, please do not hesitate to call.
Very tmly yours, J
A. D.Ilink Viec President / General Manager AECL Technologies Attachments: As stated cc:
Janet Kennedy, NRC CANDU 3 Project Manager
ATTACHMENT 1 AECLT SPECIFIC COMMENTS ON CANDU 3 DESIGN ISSUES AND l
TilEIR REIA110NSillP TO CURRlWF REGULATORY REQUIREMENTS i
In this attachment, AECLT addresses the six issues which the draft Policy Issues Paper identified as pertaining to the CANDU 3 design; specifically, Accident Evaluation, Source Temt Containment Performance. Operator Staffmg, Positive Void Reactivity and Control Room Design. Additionally, AECLT comments on the effects of the proposed Category 2 classification issues.
For cach of the six issues, the drafl Policy issues Paper characterizes and discusses AECLTs approach to the CANDU 3 desigrt Based on the characterization and discussion, the papx proposes a recommended resolution of the issue. AECLT believes that the Staffs current evahtation does not give credit to the CANDU design approach.
1.
The basis for A7WS in the first place was the single line of defense in LWRs against some accidents. CANDU chose to address the ATWS question by having redundant shutdown. '1he accidents !!sted by the Stafr are not excluded from consideration in CANDU, but have little consequence because of the redundant shutdown systems.
1hc whole point of redundant shutdown is to provide real safety, as opposed to providing analysis of events without shutdown. 'lhis recognition is lacking.
2.
Events which would lead ta core melt in conventional LWRs, namely LOCA/LOECC,-
do not do so in CANDU because of the presence of the moderator. To use the consequences of a severe accident to challenge the design, without examining the defenses that have to fail before those consequences occur, removes the incentive from the designer to reduce the frequency of those consequences.
AECLT corrects the chameterization and the evaluation in the discussion section, as necessary. In addition, AECLT comments on the recommended resolution of the issue and, where AECLT differs with the recommendation, offers an attemative approach for considemtion.
A. ACCIDENT EVALUATION ISSUE: Identify appropriate event categories, associated frequency ranges, and evaluation criteria for events that will be used to assess the safety of the proposed designs.
1
4 4
AECLT COSD1ENT: AECLT does not believe that the dmil Policy issues Paper accurately chameterizes AECLTs approach to accident evaluation. The drafl paper states:
"The CANDU 3 preapplicant, in their current safety analyses, has excluded analyses of the consequen.:cs of events with frequencies of less than 10'/ year from the safety evaluation. Events which would be excluded from considemtion, based on the cal DU 3 design characteristics and system reliabilities, would include anticipated transient without semm (_ATWS), unserammed loss-of-coolant accidents (LOCAs),
delayed scram events, and other events which could afTect reactivity insertion (fo example, from control system failures). As a result of the positive void tractivity coeflicient associated with the CANDU design, events involving even a relatively short senur. delay could result in a core dismption accident."
AECLTs approach, which is based on design resiew guides accepted by the Atomic En Control Board (AECB), is summarized in the points which follow. We request that the dmfl paper be revised in accordance with these points.
For the CANDU 3 design evaluation, event sequences and their End States are 1.
determined by systematic review without regard to End Sette frequencies. The Conseptual Probabilistic Safety Assessment (CPSA) considers End State freque low as 10"/ year.
Reactivity insertion events are not excluded from consideration.1hree systems are 2.
provided for such events:
1he Group 1 Regulatoty System with Mechanical Control Absorber Rods for 1)
Anticipated Transients.
1he Group 2 Shutdown System I with rapid shutdown rods for Accidents.
2) 3)
The Group 2 Shutdown System 2 with mpid liquid poison injection for Accidents.
1he CPSA gives the end state frequency for the large LOCA with a failure to shutdown at 10*/ year.
The CANDU 3 Safety Analysis has no absolute frequency cutoff. As tu'ed in the 3.
Conceptual Safety Report, Appendix C, Section 5.2 Category B Events:
"lhe events in Category B are those for which the firquency of the event can be calculated using probabilistic tools to obtain a realistic assessment of the risk involved (public and economic risk)...
2-
e "lhe stepped curves in Figure 2 will be uscd as the acceptance criteria for Category B analyses. These are intended to be used as event based criteria, to provide a measure of the acceptability of the consequences of a given event, which is a function of its likelihood of occunence. As in the previous probabilistic assessments, events with frequencies less than im wents per year are not considered to be of high enough frequency that they generally need to be considered. In.those_ cases where them.Eigum_2.will be_extmnolated.as necessam" 4.
1he current CANDU 3 Safety Analysis focuses prinurity on identifying design requirements and recommendations for design improvement and assessing the adequacy of the safety systems.1he Safety Analysis establishes categories of events, along with evaluation methods and acceptance criteria for each event.
5.
The events analyzed are selected because of their impact on the conceptual design, regardless of their frequency. For example, for the containment design, the event analyzed is a large loss-of-coolant accident with emergency core cooling uruvailable (LOECC).
With respect to the treatment of ATWS, unsemmmed LOCAs and delayed semm events, see the discussion below in the section conceming Positive Void Reactivity.
6.
The drafl Policy issues paper also states:
"The CANDU 3 approach which limits the scope of severe accidents examined appean to be inconsistent with the provisions of 10 CFR 52.47."
This statement is not accumte. As discussed above in Point No. 3, limits are not placed on the scope of severe accidents that nuy be considered in designing the CANDU 3.
In the dmfl Policy issues Paper, the Stafl' proposes to develop "a single _ approach to a 7.
advanced reactor designs during the preapplication resiew." Although omitting mention of its applicability to the CANDU 3 design, it appean that the apprr ;h is to apply to the CANDU 3 design as well. Assuming that to be the case, the pap:r should be corrected.
8.
The fint bullet in the Reccmmendations section states:
" Events will be selected detenninistically and supplemented with insights from probabilistic risk assessments of the specific designs."
l l
AECLT believes that the criteria which will be used in detenninistically selecting the events should be identified. Also, we would like to know whether the Staff will p
.. -. ~ - - - -. -
continue the historical requirement for conservative analysis for Design Basis Accidents (DBAs) and best estimate analyses for bcyond DBAs. %c NRC Staff recommendations for "deterministically" selecting events for analysis appears to be arbitmry and contrary to the spirit of NRCs existing safety goals.
As we discussed in our recent comments on the Advance Notice of Proposed Rulemaking conceming severe accident requirements, we think that assumptions and acceptance criteria should be established for severe accidents, including event cutoff frequencies and consequence acceptance limits. Attachment 2 provides a copy of the relevant AECLT comments on the ANPR, Comment #4 and Comment #8.
B. SOURCE TERM ISSUE: Should mechanistic source terms be deseloped in order to evaluate the advanced rmetor and CANDU 3 designs?
4 AECLT COMMENT: he NRC Staff position, as stated on page 7, is that:
"De CANDU 3 will also be different from LWR designs in certain respects. He.
coolant contains significant amounts of tritium. Following failure of a pressure tube, r
there is no heavy-unlled reactor vessel to contain releases (there are large volumes of water in two concentric low-pressure tanks; modemtor and shield unter).
q Consequently the timing of releases is expected to be different from. LWR's.
l Berefore CANDU 3 also wanants a separate evaluation of source terms."
l NRC Staff then recommends that the CANDU 3 source terms used for preapplication resiew be based upon mechanistic analyses and recommends specific guidelines for performing these analyses.
AECLT does not object to the NRC approach to evaluate our proposed source tenn during the preapplication review. _ We have a specific comment on the guidelines which is noted below.
i However, we do not agree that a pressurized heavy water reactor suchsCANDU 3 is so fundamentally different from LWR's that it should require a di{erent methodolog) for establishing source terms than that which NRC is now in the prm vfista6iishing for evolutionary LWRs. Specifically, the NRC Staff notes on page 7 ofits draft letter that NRC Staff is now in the process of"... developing for LWR's a revision to the TID-14844 source temi (NUREG-1465, " Accident Source Temis for Light Water Nuclear Power Plants," draft report for comment, June 1992)." AECLT requests that NRC give consideration to applying the same methodology which is developed for Advanced LWRs to Heavy Water Reactors during design certification resiews. In this way, the CANDU 3 will be judged on the same basis as the other water reactor designs currently under licensing resiew by the NRC..;
In support ofits position, AECLT notes that differences in design beturen PWRs and BWRs (for example the absence of a steam generator in BWRs) must be accounted for in source term calculations ~ and the methodology being developed by NRC must allow for these differences in design. We believe that the differences in design between CANDUs and PWRs, insofar as they affect fission product tmnsport after an accident, are of a similar nature. We note that CANDU's use the same type of zircaloy clad umnium oxide fuel as LWRs; hence the foel behavior aspects of source term calculations should be the same. Also,-
the fact that tritium levels are higher in CANDU reactors because of the use of heavy water can be easily accounted for in the source term methodology.
We recognize that NRC has not completed its development and implementation of new source tenn standards for LWRs and therefore we agree with the approach recommended by the Staff for the preapplication review. However, we request that NRC approach the design -
cenification review using the new standards now being developed by NRC for application to evolutionary LWRs.
With respect to the specific guidelines for development of mechanistic CANDU 3 source temis for the preapplication review, we believe the meaning of the term " credible severe accident" is unknown; we trquest that NRC clarify this term.
C. CONTAINMENT ISSUE: Should advenced reactor designs be allowed to employ attemative approaches to traditional " essentially leak-tight" contairunent structures to provide for the control of fission product release to the environment?
AECLT COMMENT: he CANDU 3 design utilizes a traditional dry containment very similar to those in use on licensed LWRs and proposed for advanced LWRs. Herefore, AECLT does not believe the issue, as stated by NRC, is relevant to CANDU 3.
However, in the discussion section NRC states that "...for evolutionary LWRs, the StatT, in SECY-90-016, proposed to use a conditional containment failurt probability (CCFP) or deterministic conatinment performance goal to ensure a balance between accident prevention and consequence mitigation. During the evolutionary LWR reviews, a great deal of careful resiew was necessary to assure that a probabilistic CCFP would not be used in a way that could detract from a balanced approach of severe accident prevention and consequence mitigation. For advanced designs and the CANDU 3, limited experience exists in the analysis and evaluation of severe accidents which could lead to significant difliculty and uncertainty in assessing a CCFP. For this reason, the Staff recommends that a deterministic containment performance goal be adopted for the CANDU 3."
l
. t S
AECLT COMMENT: AECLT agrees with the Staff's conclusion that "a positive void
- coefficient should not necessarily disqualify a reactor design." All reactors an: subject to the -
insertion of positive reactivity under certain transient or accident conditions. The specific tmnsient or accident varies with the reactor type. For the CANDU 3 the total void worth is benveen S2 and $3. While all such insertions raise significant concems, thekcy question is whether the reactivity shutdown systems are reliable enough to reduce the frequency of reactivity insertions with a failum of all reactivity shutdown systems to an extremely low value (iA 10' to 10" per year).
in order to evaluate positive reactivity in ertion in the CANDU 3 design, the Staff proposes.
to:
"[ require the analysis of] the consequences of events (such as ATWS, unsemmmed LOCAs, delayed scrams, and transients affecting reactivity control) that could lead to core damage as a result of the positive void coeflicients."
AECLT has several observations regarding this recommendation; including, conceming the meaning of the tenns "ATWS" and "unscrammed LOCAs" as the terms might be applied to the CANDU 3. The terms were developed in the early days of LWR resiews and have specific meaning and special significance for those resiews.
1.
ATWS: If the ATWS defmition is retained with respect to the CANDU 3, its significance is diminished because of CANDU 3's multiple shutdown systems; namely, the Group 1 Regulating System and the hvo independent, diverse and redundant Group 2 Shutdown Systems.
For a PWR, Anticipated Transient Withota Scram (ATWS) is a faulted response to an anticipated initiating event n: quiring control elemmt assemblies (CEA's) insertion for reactivity control. The initiating event is defined to be the occurrence of a tmnsient requiring reactor trip for reactivity control coupled with failure of a trip to occur due to either mechanical failure of the CEAs to insert or the failure of both the Reactor Protection System (RPS) and the Altemate Protection System (APS) to genera *e a trip signal.
Although 10 CFR 50.62 defined a prescriptive solution for the ATWS scenario in temis of prevention and mitigation, the success criteria for the event is given in NUREG-(M60, Volume 3.
For the limiting ATWS scenario, the criteria relating to the pressure boundary integrity and functbmlity of the valves required for long term cooling are of primsy interest.-
The concem is that if the peak pressure in the RCS exceeds Level C stress limits (approximately 3200 psia)..a breach of the primary coolant pressure boundary will occur and the Safety Injection System check valves will be jammed closed. This l
would result in a LOCA with no RCS makeup available.
I !
m 7
m m;
. se -
J lWe are concemed'that NRC will evaluate CANDUj3 using different criteria than the criteria '
i eing used to evaluate cornainment _of evolutionary LWR n: actor designs.TWe do'not see a b
T good reason for this, and request that NRC consider instead evaluating the CANDU 3 =
containment using the same approach and criteria used in evaluating the containment designs :
for evolutionary LWRs.
-S With regard to the specific accidents to be used by NRC to evaluate containment
' perfonnance, we have the following comment:
"The Staff proposes _to postulate a core damage accident as a containment challenge event and require that containment integrity is maintained for a period of---
approximately_24 hours after the onset of core dmnage." =
We believe that a more specific definition of a " core damage accident" is needed. In this regard, we note that three types of events can lead to_" core damage accidents": (1) reactivity, events; (2) loss of heat sink events at high pressure; and (3) loss of heat sink events at low pressure. For the CANDU_3 design, the_ event frequencies of types (1)'and (2) events will be less than 10' - low enough to be able to be considered " incredible."1Thus, for the CANDU -
3 design events of these types should not have to be considered when evaluating challenges tol containment. Type 3 events comprise the " core damage accident" used as the_" containment.
' challenge" for the CANDU 3 design.
F. OPERATOR STAFFING AND FUNCTION ISSUE: Should advanced reactor designs be allowed to operate with a staffing complement 4
that is'less than that currently required by LWR regulations? -
AECLT COMMENTS: The draft Policy Issues paper. states:
"The CANDU 3 preapplicant has not proposed a specific number oflicensed opemtors,:
but the Stafl's expectation is that CANDU 3 will' meet the current LWR'staffmg requirements."
y AECLT does not understand why this issue was addressed to the CANDU 3 in the issues; I
~
matrix but not in the text.
g IL POSITIVE VOID REACTIVITY COEFFICIENT ISSUE: Should a design in which the overall inherent reactivity tends to increase under.
specific conditions or accidents be acceptable?
j
.i !
Y
...L
.L.
~
0 l
AECLT COMMENT: AECLT agrees with the Staffs conclusion that "a positive void coefficient should not necessarily disqualify a reactor design." All reactors are subject to the insertion of positive reactivity under certain transient or accident conditions. He specific transient or accident varies with the reactor type. For the CANDU 3 the total void worth is between $2 and $3. While all such insertions mise significant concems, the kcy question is whether the reactivity shutdown systems are reliable enough to reduce the frequency of reactivity insertions with a failure of all reactivity shutdown systems to an extremely low value (ir,10' to 10'* per year).
In order to evaluate positive reactivit.t insertion in the CANDU 3 design, the Staff proposes to:
"[ require the analysis of) the conseqtences of events (such as ADVS, unscrammed LOCAs, delayed scrams, and transients affecting reactivity control) that could lead to -
core damage as a result of the positive void coefficients."
AECLT has sevemi observations regarding this recommendation; including, conceming the meamng of the terms "ABVS" and "unscrammed LOCAs" as the terms might be applied to the CANDU 3. He terms were developed in the early days of LWR reiiews and have specific meaning and special significance for those resiews.
1.
ATWS: If the ATWS definition is retained with respect tr 'he CANDU 3, its significance is diminished because of CANDU 3's multiple shutdown systems; namely, the Group 1 Regulating System and the two independent, diverse and redundant Group 2 Shutdown Systems.
For a PWR, Anticipated Tmnsient Without Scram (ATWS) is a faulted resix)nse to an anticipated initiating event requiring control element assemblics (CEA's) insertion for reactivity control. He initiating event is dermed _to be the occurrence of a transient requiring reactor trip for reactivity control coupled with failure of a trip to occur due to either mechanical failure of the CEAs to insert or the failure of both the Rextor Protection System (RPS) and the Altemate Protection System (APS) to generate a trip signal.
Although 10 CFR 50.62 defined a presenptive solution for the ADVS scenario in tenns of prevention and mitigation, the success criteria for the event is given in NUREG-N60, Volume 3.
For the limiting ATWS scenario, the criteria relating to the pressure boundary integrity and functionality of the valves required for long term cooling are of primary interest.
He concem is that if the peak pressure in the RCS exceeds level C stress limits (approximately 3200 psia), a breach of the primary coolant pressure boundary will occur and the Safety injection System check valves will be jammed closed. His-would result in a LOCA with no RCS makeup available.
Using the same definition for CANDU 3, an ATWS would be a faulted response to an anticipated initiating event requiring the Group 1 Mechanical Control Absorber rods (MCA) to be inserted for reactivity control. If the MCAs fail to insert, either of the two Group 2 shutdown systen.. remain poised to shutdown the reactor without a severe pressure transient.
2.
Unserammed LOCA:
For an "unscrammed LOCA", defined as a large LOCA requiring the insertion cf the 3
shutdown rods of the Group 2 Shutdown System (SDSI) for reactivity control, the Group 2 Shutdown System (SDS2) will insert poison into the moderator and will shutdown the reactor without resulting in a severe pressure transient.
Severe Accident End State Producing Positive Reactivity Insenion: For the CANDU 3 3.
design, the severe accident End State producing positive reactivity insertion is shown by the CANDU 3 Accident Analysis to be a Failure to Shutdown when reactor shutdown by the Group i Regulating System and the two Group 2 Shutdown Systems has failed to occur. Consequences could include a mismatch betwren power production and the heat sink, resulting in severe fuel overheating and core damage As discussed above in our accident evaluation comments. the CPSA gives an End State Frequency of a large LOCA with failure to shutdown of 10* per year.
Acceptance criteria for severe core damage End States have not yet been established Staff proposes to take the frequency of positive reactivity insertion events into accoun analyzing the phenomenon in the CANDU 3. Specifically, "The Staffs review of these analyses will take into account the overall risk perspective of the designs. [A requirement to change designs] will depend on the estimated probability of the accidents as well as the severity of the consequences."
AECLT agrees that consideration of the significance of positive reactivity insertion events should take into account the overall risk perspective of the designs. AECLT notes that acccotance criteria for such events have not yet been established. AECLT recommends tha they be established during the preapplication raiew of the CANDU 3 design. In this in its recent comments on the Severe Accident ANPR, AECLT provided its siews on a l
selection process for severe accident events. See Attachment 2.
l CATEGORY 2 CLASSIFICATION The draft Policy Issues Paper identifies two issues for which the Staff recommends no departure from current regulations; namely, the Control Room Design and SSC Safet Classification issues. AECLT notes with concem that implementation of this recommenda
-8
)
l
~
a w
m will arbitmrily' cut off review of new and innovative design approaches in these areas.
AECLT asks that this recommendation be reconsidered.LWe believe that safety principles.
should govem and'that new designs should be allowed to demonstmte how they meet suc safety principles. Following such examination, the adequacy basis can be developed.
CONTROL ROOM AND REMOTE SilUTDOWN AREA DESIGN Can current requirements for a seismic Category I/ Class IE control room at:d ISSUE:
altemate shutdown panel be fulfilled by a Remote Shutdown Area, and a non-seismic Category 1, non-Class 1E control room?
AECLT does not believe that the draR Policy Issues Paper accumtely AECLT COMMENT:
characterizes AECLTs approach to control room and Secondary Control Area (SCA) design.
He draft paper states:
"The main control room is not designed to be operable following an earthquake, tomado, fire, or loss of Group 1 (non-essential) electric power, but the operator must remain available to proceed to the secondary control area."
This statement is inaccurate. AECLTs approach to control room and SCA design is summarized in the points which follow. We request that the draft paper be revised in -
accordance with these points.
A CANDU 3 plant does not employ a " remote shutdown area" of the type connoted in 1.
the present NRC regulations and incorporated in current U.S. reactors. The CAND has a secondary control area (SCA) that is, in fact, a second control room. The SCA'-
duplicates the control consoles available in the MCR for the control and monito the Group 2 systems. The design basis for the man-machine interface in the SCA is a duplication to the fullest extent practical of control locations, layouts,~ and capabil present in the MCR. He plant design basis requires diat plant operators remain in MCR if it is available and functional. he MCR is used to operate the plant safely, under normal conditions and most accident conditions. L Suflicient control a instmmentation are provided in both areas to shutdown the_ plant, achieve cold :
shutdown conditions, and maintain it in a safe condition under accident conditions including Loss-of-Coolant-Accidents. However, the MCR is designed for the effects of earthquakes and tomadoes to the extent of providing the operating Staff with _ _ _ :
protection from physical hann. Should the MCR become uninhabitable, contro plant would be shifted to the SCA. The plant is designed such that all actions required to be accomplished while the plant operators shificontrol _to the SCA are accomplished automatically. The route from the MCR to the SCA is qualified to allow its use in the event of earthquakes or tomadoes..
~
~
=------
1.
Regarding operability following a fire, AECLT wants to emphasize that no control 2.
room will remain operable following the control room fire required to be postulated by NRC fire protection requirements. However, the CANDU 3 design, which separates the plant into Group 1 and Group 2 areas, provides a significantly improved capacity to respond to fires; in that, a fire in any Group 1 area (including the MCR) will not prevent safe shutdown using Group 2 systems from the SCA and, likewise, a fire in any Group 2 area will not prevent shutdown from the MCR using Group 1 systems.
It is incorrect to characterize Group 1 systems as "non-essential" because it implies 3.
that Group 1 systems are "non-safety-related". The CANDU 3 design applies a graded level of design standards commensumte with the safety function to be performed in contrast to U.S. practice which applies extensive, safety-grade requirements to stmetures and systems that are safety-related and few, if any, requirements to those that are non-safety-related. The NRC Staffs statement in the draft SECY papers This further implies that Group 1 power would be lost given a loss of offsite power.
is not correct. There are two redundant Group 1 diesel generator sets (as well as two redundant Group 2 diesel generator sets).
De separation of the CANDU 3 plant into Group 1 and Group 2 areas provides an 4.
enhanced capability to respond to other hazards that could render any control room inoperable. These hazards include sabotage, aircraft crashes, extemally-genemted missiles, smoke, and toxic gas. The CANDU 3 design also provides enhanced emergency planning capability by providing a redundant area for monitoring and control of essential plant parameters throudiout all plant conditions from normal operation to cold shutdown.
In the draft Policy Issues Paper, the Staff discusses its reasons for recommending no depanure from current regulations regarding control room design.
AECLT believes that the NRC Stafi's evaluation approach to this issue appears to be more prescriptive regarding control room design than we believe is required by GDC-2 and G
- 17. The GDC permit a graded application of standards to stmetures, systems, and components commensurate with, as GDC-1 states, the importance of the safety functions perfomled. De importance of a control room in a plant that has essentially two control rooms is diminished from that in a plant with only one control room and a remote shutdown panel. Nevatheless, in either plant, the necessary functions need to be identified and the appropriate standards applied. The crux of this issue is the control room design envisio GDC-19. 'Ihe CANDU 3 design vis-a-vis GDC-19 is discussed as follows.
The NRC Staff states that it is reluctant to approve any design that would increase the frequency of evacuation of the control room during design basis accident conditions or hamper the control or monitoring of upset conditions as the event progresses. AECLT is general agreement with the NRC Staff position and believes that the CANDU 3 design satisfies the objective except for the low probability seismic and tornado events. As 1
~^
discussed above, AECLT feels the CANDU 3 design adequately addn:sses the concems identified by the StafT regarding this issue and provides benefit to public health and safety, AECLT requests that the StafT reconsider its no depure recommendation.
-. I
"--------n._..___..__
4 ATTACIIMENT 2 -
Excerpts from AECLT Comments on NRC Advance Notice of Proposed Rulemaking Conceming Acceptability of Plant Performance for Severe Accidents
[ref. AECI.T letter to NRC dated December 21,1992]
General Comment No. 4 4.
Specifically, in the mie and implementing guidance the following matters should be addressed:
A.
Selection Process for Severe Event Seauences Considered irtthe Desien. He selection process should be based on event frequency. De process would establish the frequency limits to: (1) define the events requiring design changes to mduce their frequency, (2) define the events that require features to mitigate the event's consequences and (3) define events that need not be considen:d in the design.
B.
Comequence Limits: For each event sequence defined by A(1) and A(2) above (e.g. reactivity events, loss of heat sink at High/ Low Pressure), acceptable consequences for the event frequency should be defined on an overall basis (e.g. containment stress and leakage, radiological consequence limits). In addition, a phenomenon acceptance criterion should define the acceptable consequences for each individual phenomenon (e.g. hydrogen, molten fuel, non-condensable gas) associated with the event consistent with the overall acceptance criteria and the design featums that produce the phenomenon.
C.
Phenomenon Accentance Criteria: For each phenomenon acceptance criterion.
systens/ features should be identified which provide the means to mitigate the consequences of the phenomenon.
D.
System / Feature Desien Criteria: For each system'fcature, design criteria should be established for capacity, load combinations, environmental conditions vs time, and reliability. The reliability criteria should include: redundancy, diversity, power supply, separation (from each other and from systems / features whose failures are involved in the severe accident event sequences), and environmental qualifications.
E.
System' Feature Demonstration Reauirements: For each system' feature, the demonstration analysis / test requirements should be defmed. Rese should y
I include assumptions, acceptance criteria, analytical methods, and test requirements.
Cicneral Comment No. B 8,
As discussed in 3 and 4 above, a severe accident mie should specify a cut-off event frequency such that events below this frequency need not be considered in the design and for which funher analysis is not required.
NUREG'CR-5368, " Reactivity Accidents" reported the results of analyses of light water reactor reactivity events performed by Bmokhaven National Laboratory. For that efibrt, Brookhaven categorized potential event sequences as being wonhy of further analysis, or not. One of the screening criteria used to determine the -
importance of a sequence for funher analysis was whether the sequence required too-many low probability events to occur in combination. Brookhaven established a screening methodology with which low probability events could be eliminated from further considemtion, Event sequences with a frequency of less than IE-7 per mactor year were considered
" incredible" and not recommended for further study.
AECLT believes that the generic severe accident mie should codify similar screening criteria.
2
-i