Resolution of Generic Safety Issues: Task II.E.2: Emergency Core Cooling System (Rev. 3) ( NUREG-0933, Main Report with Supplements 1–34 )
The objectives of this task were to: (1) decrease reliance on the emergency core cooling system (ECCS) for events other than LOCAs; (2) ensure that the ECCS design basis reliability and performance were consistent with operational experience; (3) reach a better technical understanding of ECCS performance; and (4) ensure that the uncertainties associated with the prediction of ECCS performance were properly treated in small-break evaluations.
ITEM II.E.2.1: RELIANCE ON ECCS
This TMI Action Plan48 item called only for the collection of ECCS operating experience. Risk reduction would require that conclusions and recommendations be made and acted upon. Since the stated purpose was to decrease the reliance on ECCS for events other than LOCAs, it was assumed that this item would ultimately lead to the implementation of some hardware modifications.
The ECCS of PWRs and BWRs was being actuated for events other than LOCAs. Reliance on the ECCS for events other than LOCAs should be evaluated to ensure that: (1) the ECCS design basis reliability and performance were consistent with operational experience; and (2) a better technical understanding of ECCS performance could be reached.
In accordance with Item II.K.3(17),98 licensees were requested to submit a report detailing dates and length of all ECCS outages for the previous 5 years of operation, including causes of the outages. This report would provide the staff with a quantification of historical unreliability due to test and maintenance outages, which was to be used to determine if a need existed for cumulative outage requirements in the TS. The requested report was to contain: (1) outage dates and duration of outages; (2) cause of each outage; (3) systems or components involved in each outage; and (4) corrective action taken. Test and maintenance outages were to be included in the above listings covering the 5-year period. The licensees were requested to propose changes to improve the availability of ECCS equipment, if needed.
This issue was covered under Item II.K.3(17) which was implemented as part of NUREG-O737.98 Thirty out of 36 Technical Evaluation Reports (TERs) were expected from Franklin Institute by September 3O, 1982; at the time of this evaluation, 9 had been received. RRAB/DST/NRR was to issue SERs to DL/NRR for the 30 plants by November 15, 1982 and the task was to be closed out by DL/NRR by December 31, 1982. By December 31, 1982, Franklin Institute was expected to issue the remaining 25 TERs, and SERs were to be issued for these plants by RRAB/DST/NRR by February 15, 1983. The final 35 actions were to be closed out by DL/NRR by March 31, 1983.
ITEM II.E.2.2: RESEARCH ON SMALL BREAK LOCAs AND ANOMALOUS TRANSIENTS
This TMI Action Plan48 item was intended to focus research on small breaks and transients. It included experimental research in the loss-of-fluid test (LOFT) Semiscale, BWR full integral simulation test (FIST), and B&W Integral Systems Test facilities, systems engineering, and materials effects programs, as well as analytical methods development and assessments in the code development program. Most of the experimental work for small-break LOCAs (SBLOCAs) was completed in FY 1982 with data analysis to be conducted in FY 1983. Since October 1982, the LOFT project had been supported by an international consortium, of which NRC was a member.
The primary goal of the small-break and transient research was to improve operator performance during off-normal events. The research on analytical methods development and assessment was directed toward improving existing computer codes, development and application of advanced computer codes for SBLOCA and other accident analysis, and development of a fast, easy to use, engineering analyzer capability.
Part of the program was to examine SBLOCAs and anomalous transients; specifically, the ability of typical process instruments to provide accurate and sufficient information to operating personnel. Advanced control room and diagnostic instrumentation was used in LOFT as part of the augmented operator capabilities program to assess operator needs to mitigate the consequences of LOCA and transient sequences.
Only reduction in operator error during LOCA and transient sequences was assumed. It was also assumed64 by PNL that SBLOCAs or transients leading to a LOCA, typically via a stuck-open pressure relief valve, represented the initiating events applicable to this issue. Using Oconee-3 as the representative PWR, these initiators were an S3 LOCA and T1, T2, or T3 transient coupled with relief valve closure failure (Q). This applied primarily to PWRs; however, the same approach was used for BWRs.
For PWRs, it was assumed that operator errors involved: (1) failure to align suction of high pressure recirculation system to the suction of the low pressure recirculation system; and (2) failure to open both containment sump suction valves in the low pressure containment spray recirculation system at the start of recirculation. For BWRs, it was assumed that the operator failed to manually initiate the automatic depressurization system (ADS). Operator error in such sequences was assumed to be reduced by one-third as a result of a combination of operator training and improved instrumentation.
Based on the above assumptions and using the dominant accident sequences, the reductions in core-melt frequency were calculated64 to be 5.2 x 10-6/RY for PWRs and 1.8 x 10-7/RY for BWRs.
The reductions in public risk were calculated to be 15 man-rem/RY for PWRs and 0.5 man-rem/RY for BWRs. Assuming 90 PWRs with an average remaining life of 28.8 years and 44 BWRs with an average remaining life of 27.4 years, the total public risk reduction was 41,000 man-rem for all forward-fit and backfit plants.
Industry Cost: It was estimated that upgrading operator training and installing upgraded equipment would cost $0.5M/plant. It was assumed that equipment installation was primarily in the control room, with no increase in radiation exposure, and that only backfit plants were involved. Therefore, assuming 47 PWRs and 24 BWRs, the industry cost was estimated to be $36M. This cost was applied to backfit plants only since the changes resulting from this program would presumably be incorporated into the initial design and licensing of the forward-fit plants.
NRC Cost: This item was an ongoing program; therefore, sunk costs had already been taken in FYs 1980, 1981, and 1982. It was estimated that 20% of the FY 1983 LOFT budget was earmarked for the SBLOCA program. This represented approximately $3.1M. In addition, it was assumed that $0.2M would be required to establish new criteria for reactor instrumentation and operator training. NRC annual review was estimated to require an additional 1 man-day/RY. At a rate of $2,270/week and using the remaining plant life assumed above, this cost was about $1.7M. Therefore, the total NRC cost was estimated to be approximately $5M.
Total Cost: The total industry and NRC cost associated with the possible solution was $(36 + 5)M or $41M.
Based on an estimated public risk reduction of 41,000 man-rem and a cost of $41M for a possible solution, the value/impact score was given by:
Based on a potential public risk reduction of 41,000 man-rem, a value/impact score of 1,000 man-rem/$M, and a reduction in core-melt frequency of less than 10-5/RY, this issue was given a medium priority ranking (see Appendix C). The test program called for was completed by the staff and showed that the ECCS will provide adequate core cooling for SBLOCAs and anomalous transients consistent with the single failure criteria of 10 CFR 50, Appendix K. Ongoing thermal-hydraulic research was aimed at defining the degree of uncertainty in the ability of existing analytical models to simulate those transients on full-scale LWRs and not at proving capability. Thus, this item was RESOLVED and no new requirements were established.817
ITEM II.E.2.3: UNCERTAINTIES IN PERFORMANCE PREDICTIONS
Small-break LOCA analyses performed by LWR vendors to develop operator guidelines had shown that large uncertainties may exist in system thermal-hydraulic response due to modeling assumptions or inaccuracies. It was necessary to establish that these assumptions or inaccuracies were properly accounted for in determining the acceptability of ECCS performance pursuant to Appendix K of 10 CFR 50.
The reason behind this TMI Action Plan48 item was that, historically, the SBLOCA analyses were never reviewed by the NRC in the depth and detail with which the large-break analyses were reviewed. One of the obvious lessons of the TMI-2 accident was that SBLOCAs are much more likely to occur and, therefore, a highly detailed re-review of the small-break analyses might have been appropriate.
SBLOCAs do not automatically result in rapid depressurization of the primary system. The more complicated blowdown makes it more difficult to predict ECC injection flow rates, water level, and many other parameters as a function of time. Moreover, there are many more possible locations for the break. In addition, the possibility of unexpected thermal-hydraulic phenomena cannot be ruled out. Since the SBLOCA analyses must conservatively bound a plant's response to all possible small breaks, all of these effects should be understood as well as possible.
The proposed solution in NUREG-066048 called for NRR to issue instructions to holders of approved ECCS evaluation models to evaluate the uncertainty of small-break ECCS performance calculations; NRR was to evaluate these uncertainties. If changes were needed in the existing analysis methods to properly account for these uncertainties, recommendations were to be made to the Commission to adopt such changes. Ultimately, the adoption of these changes would result in changes to the analyses upon which plant TS were based. This could result in some restrictions on power level under certain circumstances.
According to WASH-140016 estimates, small breaks (2 in. to 6 in. diameter) are expected to occur at a rate of 3 x 10-4 event/RY; very small breaks (0.5 in. to 2 in. diameter) are estimated to occur at a rate of 10-3 event/RY. Should such an event occur, it was estimated (based purely on judgment) that there may be a 10% chance of the actual peak cladding temperatures exceeding the temperatures predicted by the 10 CFR 50, Appendix K calculation due to the modeling uncertainties mentioned above.
However, in addition to the modeling conservatism, the 10 CFR 50, Appendix K calculations assumed the worst case single failure. Moreover, the small break analysis was very seldom limiting; usually the calculated small break peak cladding temperatures are about 400ºF below the 2200ºF Appendix K limit. Finally, a plant does not normally operate with the LOCA parameters (FQ, MAPLHGR, etc.) at their limits.
Because the specific worst-case single failure varied for different plants, it was not practical to use fault trees to calculate the probability of such a failure. However, some perspective was gained by examining the following estimated failure rates from Appendix II of WASH-1400:16
PWR HPSI 1.2 x 10-2/demand
PWR Emergency Power 10-5/demand
BWR HPCI 9.8 x 10-2/demand
BWR Emergency Power 10-6/demand
The frequency of a system failure severe enough to approximate the Appendix K single failure assumptions was estimated to be, at most, 10-1/demand. Given a small LOCA, a modeling uncertainty, and something approximating the worst-case single failure, the actual peak cladding temperature would be greater than that calculated by the analyses. However, there was still considerable margin to significant core damage because:
(1) The small-break analysis is rarely limiting. Usually there is about a 400ºF margin between the calculated small-break peak cladding temperature and the 2200ºF limit.
(2) Most plants operate well within their LOCA limits (i.e., are not "LOCA-limited").
(3) To get severe damage, a significant amount of cladding must achieve a temperature significantly higher than 2200ºF. The case of the hottest point of the core barely exceeding the temperature limit does not automatically imply severe damage.
These three considerations were summed by assuming that there was, at most, a 5% chance of significant core damage given a small LOCA, a model problem, and a near-worst-case single failure. Putting all this together, the frequency of events with significant core damage was estimated to be, at most, about 7 x 10-7 /RY.
If cladding temperatures rise significantly above 2200ºF in a large portion of the core, the likely result would be a bed of debris. It was assumed that there was a 10% chance of a core-melt and a 90% chance of widespread cladding failure but no fuel melting. Neither of these fit readily into the WASH-140016 Release Categories. The core-melt case was approximated with 5 x 106 man-rem (which is greater than or approximately equal to the consquences of PWR-1 through PWR-7 and BWR-1 through BWR-4), and the non-core-melt case by 120 man-rem (which bounds PWR-9 and BWR-5).
Industry Cost: It was estimated48 that 15 staff-years and $1M of computer time would be required to perform the studies. In addition, 3 staff-months per operating plant were needed to implement procedural and TS changes. Since there were 70 plants operating, the estimated total direct industry cost was $4.25M (The 57 plants under construction at the time of this evaluation would not require implementation costs since the new analyses would displace analyses which would have been required in any case.)
In addition to the direct cost, there was an indirect cost due to the effect of further restricting operating parameters. Using the earlier assumptions that there was a 10% chance of finding a non-conservatism and a 5% chance of being SBLOCA-limited, and assuming further that at least a 1% power reduction resulted under such circumstances, the indirect costs averaged at least $5,500/RY. There were 43 operating PWRs with a cumulative experience of 350 RY and 27 BWRs with a cumulative experience of 260 RY. Adding the 36 PWRs and 21 BWRs that were under construction and assuming a plant life of 40 years, there were 4,470 RY remaining. Thus, the indirect cost was $24.6M and the total industry cost was $(4.25 + 24.6)M or $28.85M.
NRC Cost: It was estimated that 15 staff-years and $100,000 would be necessary for the staff to review the studies; in addition, the 70 backfit plants would require one staff-month each. (Again, the 57 plants under construction would not need a significant amount of extra review effort since the new reviews would displace the reviews of other analyses that would have been submitted.) Thus, NRC costs were estimated to be about $1.2M.
Total Cost: The total industry and NRC cost associated with the possible solution was $(28.85 + 1.2)M or 30.05M.
Based on an estimated public risk reduction of 1,565 man-rem and a cost of $30.05M for a possible solution, the value/impact score was given by:
Based on the safety importance and value/impact score above, this issue had a low priority ranking. In addition, RSB/DSI/NRR had noted that much of the technical concern of the issue was automatically being investigated98 in the implementation of Item II.K.3(30) which was in progress at the time of the initial evaluation of the issue in November 1983. In order to prevent duplication of effort and because the work on Item II.K.3(30) was in progress, the issue was given a low priority ranking (see Appendix C). In NUREG/CR-5382,1563 it was concluded that consideration of a 20-year license renewal period did not change the priority of the issue. Further prioritization, using the conversion factor of $2,000/man-rem approved by the Commission in September 1995, resulted in an impact/value ratio (R) of $19,230/man-rem which placed the issue in the DROP category.