Resolution of Generic Safety Issues: Item C-8: Main Steam Line Leakage Control Systems (Rev. 1) ( NUREG-0933, Main Report with Supplements 1–34 )
Dose calculations by AAB/NRR in 1975 indicated that operation of the main steam isolation valve leakage control system (MSIVLCS) required for some BWRs could result in higher offsite accident doses than if the system were not used and the integrity of the steam lines and condenser was maintained. The calculations performed by AAB at that time assumed non-operation of the MSIVLCS and took credit for cold-trapping of iodine and volatiles in the steam lines and condenser. In addition, long holdup times and release either through stack filters via the waste gas treatment system or leakage from the steam system were assumed. Leakage from the main steam condenser system would be small because normal operation requires that leakage be maintained at a low level. The integrity of these systems is not assured during earthquakes since they are not designed for SSE. However, it was believed that the probability of failure due to earthquake of both the fuel and these systems was small. By contrast, the MSIVLCS draws a negative pressure downstream of the MSIVs to collect leakage past the valve seats and processes the collected leakage through a safety grade filtration system for release to the environment. Relatively little cold-trapping or holdup time occurs when the MSIVLCS is used. Therefore, the calculated doses for releases through the MSIVLCS would be greater than the calculated doses for releases through the steam system, unless the integrity of the steam system is lost.
This NUREG-04713 item was initiated to investigate whether the MSIVLCS recommended in Regulatory Guide 1.9621 was desirable. After its identification, the issue was categorized to be of little or no significance to plant risk (i.e., Category C) and minimal staff effort was devoted to it. However, new concerns arose from operational experience that indicated a relatively high MSIVLCS failure rate and a variety of failure modes at some BWR plants that resulted in the initiation of Issue 16 (Section 1 of this report). New data413 on the magnitude and frequency of MSIV leakage at BWRs renewed concerns for the viability of the MSIVLCS design. In addition, the question of backfitting MSIVLCS to BWRs was raised.219 The evaluation of Item C-8 incorporated all of the concerns outlined above.
The AAB calculations for accidents with a TID source73 indicated a potential increase in offsite releases of iodine by two to three orders of magnitude for proper operation of a MSIVLCS, when compared to the calculations of releases assuming the steam system is intact and MSIV leakage is eventually released through the condenser. Therefore, use of the MSIVLCS prescribed by Regulatory Guide 1.9621 could increase the overall risk to the public. Additionally, the AAB calculations assumed a relatively low rate of MSIV leakage. The new data413 collected by OIE revealed a high frequency of measured MSIV leakage at some operating plants which could be in excess of the TS limit of 11.5 SCFH by more than two orders of magnitude. Leakage of this magnitude is beyond the design capacity of most MSIVLCS and, as a result, it was believed that the public risk associated with excessive MSIV leakage could be higher than previously assumed.
(A) Plants having MSIVLCS would provide procedures and train their operating staff to use the more efficient steam and waste gas treatment system (SWGTS), if available, as the first option following a major accident. The MSIVLCS would be treated as a backup system to be used only if the normal treatment system is not available.
(B) Install MSIVLCS at all the "grandfathered" BWRs and train and equip the operating staff to treat the MSIVLCS as a backup system as in (A) above.
(C) "Fix" MSIV leakage characteristics and continue to use the MSIVLCS at those plants which have (or will have) them as the first choice of treatment for MSIV leakage following a major accident.
(D) "Fix" MSIV leakage and use the MSIVLCS as a backup system at those plants which have (or will have) them, as in (A) above.
(E) "Fix" MSIV leakage, install MSIVLCS at all "grandfathered" BWRs, and train and equip the operating staff to treat the MSIVLCS as a backup system as in (A) above.
(F) Disable all MSIVLCS and accept MSIV leakage at its existing magnitude and frequency.
In the analysis of this issue, the following major assumptions were made:
(1) Frequency of core-melt event in BWR = 3.8 x 10-5/RY (Grand Gulf PRA) NUREG/CR-2800,64 Appendix B.
(2) Failure/demand probability of MSIVLCS (i.e., will not function properly when needed) = 5 x 10-2, when MSIV leakage is less than 100 SCFH, and 1.0 when MSIV leakage is greater than 100 SCFH.
(3) Failure/demand probability of SWGTS = 5 x 10-2 (i.e., the SWGTS will not be available if desired to prevent direct leakage to the environment). Unavailability of the non-seismic portions of the SWGTS due to seismic events was assumed to be covered by the 5 x 10-2 failure probability.
(4) The SWGTS is not available for use for 26% of the core-melt scenarios. (Examination of the Grand Gulf PRA indicated that 26% of the core-melt scenarios were either initiated or exacerbated by the loss of offsite power, which is required to operate the condenser and SWGTS.)
(5) If neither the MSIVLCS nor the SWGTS is available, MSIV leakage is released directly to the environment. (The potential to contain MSIV leakage in the steam line until the SWGTS is available for treatment was not considered.)
(6) Of the 50 expected BWR plants, 25 will have a MSIVLCS and 25 will not provide one unless required to do so.
(7) All plants in the population have an average remaining life of 30 years.
(8) The partitioning efficiency of the MSIVLCS is 99% (i.e., reduces releases by a factor of 100).
(9) The partitioning efficiency for the SWGTS is 99.9% (i.e., reduces releases by a factor of 1000).
(10) Maximum MSIV leakage was assumed to be about 3000 SCFH, based on the maximum reported MSIV leakage observed at Browns Ferry Units 1, 2, and 3 (IE Bulletin No. 82-23).220
(11) The probability of failure/demand of an MSIV to close is 10-3 and the probability of MSIV isolation demand failure is 5 x 10-5 (WASH-1400).16
(12) Average MSIV leakage and the frequency of occurrence per test are as indicated in Table 2.C.8-1. This table was derived from the data provided in a memorandum from OIE413 in which the results of an industry survey of BWR MSIV leak rate tests were discussed. The derivation of this table is discussed later under the frequency/ consequence estimate.
|MEAN MSIV LEAK RATE (SCFH)||RELATIVE FREQUENCY||MEAN MSIV LEAK RATE (SCFH)||RELATIVE FREQUENCY|
Since none of the BWR core-melt release categories assume immediate direct environmental releases which bypass the containment wet well and suppression pool and, in some instances, other containment or auxilary systems which would mitigate releases, it was believed that basing the consequences for this issue on the consequences derived for BWR Category 1 through 4 releases was not appropriate. The Accident Evaluation Branch (AEB), therefore, provided the results of consequence analyses of core-melt accidents with large MSIV leakage414,415 using the CRAC I Code.64 For these consequence estimates, the population and meteorology of the Perry site were used, along with some characteristics of the Browns Ferry steam lines. The Perry site has an average population density within a 50-mile radius of the plant of about 320 persons per square mile, as opposed to the assumption of a uniformly distributed population with a density of 340 persons per square mile used in other generic issue risk estimates. Thus, analyses were for a hybrid BWR plant with a 3834 MWt power level.
A direct release consequence analysis was performed to simulate an accident sequence in which releases occur immediately downstream of the first non-seismic Category 1 component (turbine stop valve) in the main steam line. WASH-140016 BWR-2 release category fission product source terms were assumed. A two-hour delay prior to initiation of fission product release from the core and a 0.27 hour delay in the main steam lines were used. A nominal low-energy ground level release to atmosphere at the turbine stop valve was assumed. MSIV leakage was assumed to be about 3000 SCFH. Computed average consequences were 5.2 x 106 man-rem and 45 early fatalities within 50 miles of the site.415
An industry survey of MSIV performance was performed for the years 1979 through 1981. The results of this survey413 and additional information provided were utilized to develop the MSIV leakage rates and frequencies indicated in Table 2.C.8-1. MSIV leakage was demonstrated by testing to vary from less than the TS limit of 11.5 SCFH to as great as about 3500 SCFH. Since most MSIVLCS are designed to accommodate a maximum leakage of about 100 SCFH, the leakage data was divided into three groups, i.e., leakage less than or equal to 11.5 SCFH, leakage between 11.5 SCFH and 100 SCFH, and leakage greater than 100 SCFH. The frequency (percentage) of all tests with measured leakage within the three groups was determined from the data. For the first two groups, MSIV leakage of 11.5 SCFH was assumed for those valves which "passed" the leak test, and a median leakage of 55 SCFH was assumed for those valves which fall into the group with leakage greater than the TS limit but not greater than 100 SCFH. For the third group a weighted average was determined.
Examination of the data revealed that, of MSIVs provided by three different manufacturers, one particular type of valve dominated the extreme leakage incidents; about 60% of the MSIVs in service were provided by this manufacturer.
One licensee had embarked on an improvement program for its MSIVs which were of this particular type. It was expected that the improvements planned for these valves would result in the valves being similar in design and operation to the valves of the other two manufacturers. Therefore, it was assumed that, if MSIV leakage improvements were made, all valves would be expected to have MSIV leakage characteristics and a frequency distribution the same as that indicated by the 1979-1981 data for the other two manufacturers' valves. The results are depicted as "current" and "after fix" in Table 2.C.8-1.
Consequences of a direct release of 11.5 SCFH, 55 SCFH, 500 SCFH, and 1500 SCFH MSIV leakage were determined by ratioing the average consequence calculated by AEB for the direct release of a 3000 SCFH leak to the above leakages. The risk analysis also considered the low probability event of a core-melt accident in which one or more main steam lines were not isolated. For this case, a direct release consequence of 100 times the average consequence calculated by AEB for a 3000 SCFH leak was assumed.
A simplified event tree was developed using the afore-stated assumptions and consequence estimates. The event tree included the probability of core-melt accidents, the probabilities of various levels of MSIV leakage, and the probability of failure of MSIVLCS and SWGTS. The redundant series configuration of MSIVs was also considered in the event trees. The simplified event tree was utilized for a spectrum of MSIV leakage rates to determine the probability of core-melt releases to the environment directly through the MSIVLCS and through the condenser and SWGTS.
A specific consequence was determined for each event tree path by ratioing the average consequence of the 3000 SCFH direct release determined by AEB to the specific MSIV leakage assumed for that path. When releases were found to occur through either the MSIVLCS or the SWGTS, the consequence was reduced by the appropriate assumed partitioning factor.
The probabilities for the specific paths through the event tree were multiplied by the consequence in man-rem for that specific path and the products summed to determine the total risk for the event tree. The probabilities and consequences for the basic event tree were adjusted, as necessary, to determine the total plant risk for operation of BWR plants as they existed and for the total plant risk following each of the possible solutions. The analysis revealed that BWR plant risk was dominated by those event tree paths in which high (greater than 100 SCFH) MSIV leakage was assumed.
It should be noted that the simplified event tree does not account for "cascading" leakage in a main steam line which has two MSIVs in series. This would represent a leakage reduction. In addition, for those scenarios in which a LOOP (26% of all core-melt accidents) is assumed to occur, MSIV leakage was assumed to be directly released to the environment if the leakage was greater than 100 SCFH (the MSIVLCS design capacity). In reality, there is a rather large probability that the leakage could be contained within the steam line until such time that offsite power is recovered and treatment is again possible through the condenser and waste gas treatment system.
The risk associated with large MSIV leakage was determined for seven cases as follows:
CASE 1 - Those plants (25) which have (or will have) a MSIVLCS were assumed to treat it as a safety system and, thus, would operate it in preference to the normal treatment systems in response to a major event. It was assumed that this represented the existing state of operating plants and, thus, Case 1 was used as the base case. The total risk calculated for Case 1 was 1.33 x 104 man-rem.
CASE 2 - Those plants (25) which have (or will have) a MSIVLCS treat it as a backup system to the SWGTS and, thus, only fall back on MSIVLCS in the event that the normal treatment system is not available following a major event. The total calculated risk for this case was 1.33 x 104 man-rem.
CASE 3 - All plants (50) have a MSIVLCS and treat it as a backup system to the normal treatment system. The total risk for this case was 1.21 x 104 man-rem.
CASE 4 - MSIV leakage is "fixed" and those plants (25) which have (or will have) a MSIVLCS continue to regard it as a safety system and, thus, will operate it in preference to the normal treatment systems in response to a major event. The total risk for this case was 2.37 x 103 man-rem.
CASE 5 - MSIV leakage is "fixed" and those plants (25) which have (or will have) a MSIVLCS treat it as a backup system to the SWGTS and, thus, will only fall back on the MSIVLCS in the event that the normal treatment system is not available following a major event. The total risk for this case was 2.34 x 103 man-rem.
CASE 6 - MSIV leakage is "fixed" and all plants (50) have a MSIVLCS and treat it as a backup to the normal treatment system. The total risk for this case was 1.45 x 103 man-rem.
CASE 7 - Disable the MSIVLCS at all plants which have (or will have) it. Existing MSIV leakage is accepted. Following a major event, MSIV leakage would be treated only with the normal SWGTS when available. The total risk for this case was 1.43 x 104 man-rem.
The above risk estimates were then applied to the possible solutions with the following results:
(A) Plants which have a MSIVLCS would be required to develop procedures and train their operators to use the more efficient SWGTS, if available, as the first option following a major event. The risk reduction afforded by this solution was determined by subtracting the total risk of CASE 2 from the total risk of the base case (CASE 1) and was (1.33 x 104 -1.32 x 104) man-rem = 100 man-rem.
(B) Install MSIVLCS at all BWR plants which were "grandfathered" and provide procedures and operator training to treat the MSIVLCS as a backup system to the normal treatment system as in (A) above. The potential risk reduction for this solution is the difference between Case 1 and Case 3 and was (1.33 x 104 - 1.21 x 104) man-rem = 1.2 x 103 man-rem.
(C) "Fix" MSIV leakage and continue to use the MSIVLCS at those plants which have (or will have) it as the first choice of treatment for MSIV leakage following an accident. The potential risk reduction for this solution is the difference between Case 1 and Case 4 and was (1.33 x 104 - 2.37 x 103) man-rem = 1.09 x 104 man-rem.
(D) "Fix" MSIV leakage and use the MSIVLCS as a backup system at those plants which have (or will have) it as in (A) above. The potential risk reduction for this solution is the difference between Case 1 and Case 5 and was (1.33 x 104 - 2.34 x 103) man-rem = 1.10 x 104 man-rem.
(E) "Fix" MSIV leakage, install MSIVLCS at all "grandfathered" BWRs, and train and equip the operating staff to treat the MSIVLCS as a backup system as in (B) above. The potential risk reduction for this solution is the difference between Case 1 and Case 6 and was (1.33 x 104 -1.45 x 103) man-rem = 1.18 x 104 man-rem.
(F) Disable the MSIVLCS at all plants which have (or will have) it. The potential risk reduction for this solution is the difference between Case 1 and Case 7 and a reduction in risk of (1.33 x 104 - 1.43 x 104) man-rem or -1,000 man-rem, a risk increase.
Industry Cost: Industry costs for the various solutions were estimated as follows:
Solution (A) - Training and Procedures for Using Normal Treatment System First
|(a) Develop Procedure||$50,000/plant|
|(b) Control Room Display||25,000/plant|
|(c) Operator Training||60,000/plant|
Therefore, the total industry cost is $(135,000)(25) = $3.38M.
Solution (B) - Install MSIVLCS at All BWR Plants - Procedures and Training forUse as Backup
|(a) Training & Procedures||$135,000/plant|
|(b) Procure, Install & Maintain MSIVLCS|
|(i) Procure & Install||$500,000/plant|
|(ii) Maintain, Surveillance (10 man-wks/yr x $2,000/man-wk x 30 yrs)||$600,000/plant|
Total Industry Cost = (50)($0.135M) + 25($1.1M) = $34.25M
Solution (C) - "Fix" MSIV leakage and Use MSIVLCS at Those Plants Which Have
(or Will Have) It as First Choice
|(a) Valve Modifications (Licensee Estimate)||$350,000/plant|
|(b) Licensing Submittal and Review||$150,000/plant|
60% of all MSIVs would be modified
Total Industry Cost = (0.6)(50)($0.5M) = $15M
Solution (D) - "Fix" MSIV Leakage and Use MSIVLCS at Those Plants Which Have(or Will Have) It as Backup System
Total Industry Cost = Cost of Solution A + Cost of Solution C
= $3.38M + $15M = $18.38M
Solution (E) - "Fix" MSIV Leakage, Backfit MSIVLCS to "Grandfathered" Plants and Use as Backup System
Total Industry Cost = Cost of Solution B + Cost of Solution D
= $34.25M + $15M = $49.25M
Solution (F) - Disable MSIVLCS at All Plants Which Have (or Will Have) It
|(a) Disable MSIVLCS = 1 man-week||$2,000/plant|
|(b) Maintenance and Surveillance of MSIVLCS - Discontinue||-$600,000/plant|
Therefore, the total cost saving is (-$598,000)(25) = - $14.95M.
NRC Cost: It was estimated that a total of 5 man-years of professional staff and consultant efforts would be required to perform accident analysis of various options, perform the necessary trade-off studies, develop and justify recommended new requirements, review and approve the requirements, and implement the requirements. At a cost of $100,000/staff-year, it was estimated that the NRC cost to complete this issue was $500,000.
Based on the data above, the value/impact scores for the various solutions are as follows:
No quantitative assessment of uncertainty bounds was made. However, in the course of the analysis, a number of parameters were noted which would be expected to introduce rather large uncertainties. Some examples are as follows:
(b) It was assumed that two MSIVs in series, which would both be expected to have high leakage, would pass the large leakage expected for each valve. In actuality, this would result in a cascading leakage path and a reduction in through-leakage by a factor of 2 or 3.
(c) For those scenarios in which high MSIV leakage is expected and the SWGTS is unavailable due to the loss of emergency AC power, it was assumed that MSIV leakage would be released directly for the duration of the accident. In reality, direct releases might be prevented by isolation of the steam line with non-safety valving. Also, the SWGTS would have a high likelihood of being recovered upon recovery of AC power and releases could be again intercepted before the accident had run its full course. This simplified approach thus introduces large uncertainty for this accident sequence which was found to be one of the dominant risk sequences for this issue.
Thus, although a quantitative uncertainty estimate was not made, it appeared that the uncertainties in the risk calculations would be rather large and would in all likelihood be plus or minus one or two orders of magnitude. A variation of the calculated risk reduction by plus or minus an order of magnitude could move the calculated value/impact ratios from the low range to the high range.
The resolution or non-resolution of this issue would not affect the core-melt frequency for BWR plants. However, it was believed that that this issue should be addressed in terms of the third aspect of the Commission's interim safety goal, i.e., the probability of the loss of a defense-in-depth layer; in this case, the containment. The issue deals with the effects of large MSIV leakage up to 3600 SCFH. This is the equivalent of the release of about 25% of the containment volume per day. For core-melt transients (the dominant BWR risk events), the interdiction effects of the pressure suppression pool would be bypassed and effluent from the reactor would pass directly to the MSIVs. For these reasons, large BWR MSIV leakage might be considered a loss of containment integrity.
From Table 2.C.8-1, the frequency of high MSIV leakage is about 0.25/demand. For two valves in series, the expected frequency of high through-line leakage would be (0.25)(0.25)/demand or 0.0625/demand, if the failures (i.e., leakage) in each valve are independent. There are 4 steam lines per reactor, therefore, the frequency of one of the 4 steam lines experiencing high through-line leakage is 4 x 0.0625 or 0.25/reactor/demand. MSIV leakage rate tests are performed prior to each fuel cycle and following the fuel cycle. Prior to the fuel cycle, large leakage is not permitted and the valves are repaired as necessary to pass the leakage test. Therefore, if the fuel cycle is assumed to be approximately one year, the average frequency of a steam line having a high through-line leakage is [(0 + 0.25)/2]/RY or 0.125/RY. Thus, the probability of loss of a defense-in-depth layer (the containment) is about 1.2 x 10-1. This is a high value for containment unavailability.
Although the value/impact scores indicated a medium priority, this issue was given a high priority based on the consideration of the potential uncertainty and the defense-in-depth posture. It was recommended that: (1) the issue be redefined to stress the magnitude and consequences of MSIV leakage and to reflect existing testing methods; and (2) leakage control systems be evaluated only as one of the possible means for controlling MSIV leakage.
In resolving this issue, the staff considered five alternatives: (1) take no action; (2) require the addition of standard capacity LCS at plants without an LCS; (3) require increased capacity LCS at all plants; (4) disable existing LCS; and (5) use other mitigation paths with larger decontamination factors.
It was concluded that Alternatives (2) and (3) were not cost-effective. Alternative (4) was not considered to be warranted on a generic basis because of the small magnitude and uncertainties in the associated risk. Alternative (5) was estimated to result in a very small change in risk. As a result, the staff adopted Alternative (1). The regulatory analysis for the resolution of this issue was published in NUREG-13721307 and related studies were documented in NUREG-11691308 and NUREG/CR-5397.1309 Thus, this issue was RESOLVED and no new requirements were established.1310