Resolution of Generic Safety Issues: Issue 138: Deinerting of BWR Mark I and Mark II Containments During Power Operations upon Discovery of RCS Leakage or a Train of a Safety System Inoperable (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
The issue of deinerting upon discovery of RCS leakage was identified1414 by DL/NRR, based on data collected by OIE. The related but separate concern of deinerting with one train of a safety system inoperable was also raised.
BWR MARK I and II containments are inerted with nitrogen during normal operations to protect against the build-up of a potentially explosive H2-O2 mixture, in the event of a LOCA or core damage event. Following a LOCA, H2 is released within the containment from zircaloy-water reactions and H2 and O2 are also produced by radiolysis of the coolant. Core damage or melting would add an additional large quantity of H2 as a result of metal/water reaction with fuel cladding and core structural materials. Plant atmosphere systems are designed to maintain containment O2 concentration to less than 5% by volume, or the H2 concentration to less than 4% by volume, to ensure that a combustible gas mixture does not form.
RCS leakage outside of TS limits requires licensees to identify, isolate, and repair the leak to avoid or mitigate the consequences of a LOCA. These steps require plant personnel entry into the containment. In accordance with plant TS, operators typically reduce power, deinert the containment, and allow personnel entry to identify potential RCS leaks. However, deinerting under leak conditions leaves the containment vulnerable to H2-O2 concentration build-up, if the leak progresses to a LOCA or core damage event. This was the primary concern in this issue. A secondary concern was that TS also allow licensees to operate with a deinerted containment for up to 24 hours with one train of a safety system inoperable.
The OIE data consisted of 13 RCS leak event reports in which the containment was deinerted to allow corrective action by plant personnel; these events occurred between 1981 and 1986. Existing NRC guidance for this issue included GDC 41 (Containment Atmosphere Cleanup), from 10 CFR 50 Appendix A, and Section 22.214.171.124 of the GE STS for BWR/5 designs.
With either of the above concerns, the possibility of early gross containment failure with energetic source term release could be significantly increased, thereby increasing public risk. The issue affected 33 BWRs with MARK I or II containments.
A possible solution to this issue was to revise plant TS to require a reactor to be brought to cold shutdown, prior to deinerting the containment, when either unidentified leakage in the containment building or inoperability of a safety system is noted.
The baseline risk assessment used for this issue was the Millstone-1 PRA which was based on the assumption that the containment would always be inerted unless the reactor is in a cold shutdown condition; the PRA provided an assessment of risks associated with accident sequences at full power. In addition, the Millstone-1 PRA took into account TS allowing operations with one train of the safety system inoperable. The following two cases were assessed for this issue:
Case 1 A deinerted containment during a shutdown for an unidentified RCS leak.
Case 2 A deinerted containment with one train of a safety system inoperable.
For this issue, potential public risk reduction results from the change in containment failure mode afforded by not permitting deinerting during power operations under off-normal conditions. The assumed resolution of the issue will not result in a reduction of core-melt frequency.
To analyze the issue, an average consequence factor (C) was determined using the following equation:
|where||R||= Risk (man-rem/RY)|
|fcm||= Core-melt frequency (event/RY)|
|C||= Average Consequence Factor (man-rem/event)|
|fcmn||= Total core-melt frequency for the nth release category event/RY)|
|Cn||= Consequence Factor for the nth release category (man-rem/event)|
The following values were used from the Millstone-1 PRA and NUREG/CR-2800:64
|fcm = 3.09 x 10-4/RY||fcm1 = 1 x 10-6/RY||C1 = 5.4 x 106 man-rem|
|fcm2 = 8 x 10-6/RY||C2 = 7.1 x 106 man-rem|
|fcm3 = 1 x 10-4/RY||C3 = 5.1 x 106 man-rem|
|fcm4 = 2 x 10-4/RY||C4 = 6.1 x 105 man-rem|
Therefore, the value of C was 2.25 x 106 man-rem/event.
From the Millstone 1 PRA, the only dominant LOCA sequence was found to be the small-break LOCA (SB(B)). The frequency of SB(B)(fSB(B)) was 3 x 10-6 event/RY.
In analyzing the first concern, i.e., deinerting the containment with an unknown leakage into the containment building, two different schools of thought concerning the effect of detected leakage upon LOCA expected frequency were addressed. Employing the leak-before-break theory, detection of leakage within the containment building would not be associated with an increase in LOCA frequency. Conversely, it could be assumed that leakage from a through-wall pipe crack would increase the probability that the pipe would break causing a LOCA. As a result, the unknown leakage concern was analyzed from both perspectives as Case 1.
It was assumed that, whenever the containment building is deinerted, any core-melt event will result in containment failure due to H2 burn with a probability of 1. TS permit licensees to deinert the containment building of BWR MARK I and II designs, 24 hours prior to a scheduled shutdown. LERs gathered over the 5-year period revealed 9 events in which the plants were deinerted before reaching a hot shutdown condition and one event in which the plant was deinerted before shutdown with one train of ECCS inoperable. Of the 9 leakage events, 8 were found to be valve stem, body packing, or seal failures and one was determined to be a through-wall crack in a RCS line. Assuming that the containment was deinerted for 24 hours during plant operation for each of these events and that during the five-year period 29 plants operated with an assumed power production factor (i.e., historical fraction of the calendar-year that plants operate at or near full power) of 0.7, the following fractions of plant operating history were calculated. Each represents the fraction of the operating history that a plant experiences a deinerted condition, as a result of the particular set of circumstances:
|Va||- deinerted due to unknown leakage (9 events)||= 2.4 x 10-4|
|Vb||- deinerted due to unknown leakage actually due to through-wall pipe leak (1 event)||= 2.7 x 10-5|
|Vc||- deinerted due to unavailability of one train of a safety system (1 event)||= 2.7 x 10-5|
Assuming that, if the containment is deinerted and a core damage event occurs, the conditional probability of containment failure (Fcon) due to H2 explosion is 1, the conditional probability of containment failure becomes the probability that the containment is deinerted (Vn). The consequence of containment failure due to H2 explosion in the Millstone-1 PRA is best represented by the consequence factor (C2) for a Category 2 release, i.e., early large containment failure. Therefore, the risk during periods while the plant is deinerted in preparation for a shutdown with an unknown leakage, assuming that detection of leakage does not increase the probability of a pipe break (LOCA), was calculated as follows:
|= (3.1 x 10-4)(2.4 x 10-4)(1)(7.1 x 106) man-rem/RY|
|= 0.528 man-rem/RY|
This was the base case risk. If plants are not permitted to deinert during periods of unknown leakage in the containment building, the adjusted case risk would be:
|= (3.1 x 10-4)(2.4 x 10-4)(2.25 x 106) man-rem/RY|
|= 0.167 man-rem/RY|
The change in risk (r) is (r - r) or 0.361 man-rem/RY. When applied to the 33 affected plants with an average remaining life of 18.6 years, the total risk reduction attainable by not permitting plants to deinert when an unidentified source of leakage in the containment building exists was estimated to be 222 man-rem.
If it is assumed that the probability of a LOCA (in this case SB(B)) is increased by two orders of magnitude if a through-wall leak exists, the potential risk reduction afforded by not permitting plants to deinert when an unidentified source of leakage in the containment building exists is calculated as follows:
|r = fcm(Va - Vb)FconC2 + (fcm)'VbFconC2|
|where||[fSB(B)]' = fSB(B) x 100 = 3.10 x 10-4/RY|
|[fcm]' = fcm - fSB(B) + [fSB(B)]'= 6.08 x 10-4/RY|
Thus, the base case risk (r) was 0.585 man-rem/RY.
If deinerting is not allowed when there is an unknown leakage in the containment building, the adjusted case risk is calculated as follows:
r' = fcm(Va - Vb)C + (fcm)'VbC = 0.186 man-rem/RY
Thus, the change in risk due to resolution of this aspect of the issue is (0.585 - 0.186) man-rem/RY = 0.399 man-rem/RY. When applied to the 33 affected plants with an average remaining life of 18.6 years, the potential risk reduction due to resolution, if it is assumed that through-wall leakage increases the probability of LOCA by two orders of magnitude, is 245 man-rem.
The second case, i.e., containment deinerting prior to shutdown with a train of a safety system inoperable, was analyzed as follows. The LER data from the 5-year study period revealed only one instance of deinerting prior to shutdown with a train of a safety system (LPS in this case) inoperable. This results in a fraction of operating history for this condition Vc = 2.7 x 10-5 as shown earlier. It was conservatively assumed that unavailability of one train of any safety system will increase the core-melt frequency by one order of magnitude, i.e., (fcm)z = 3.1 x 10-3/RY.
The base case risk (rz) was calculated by the following relationship:
|= (3.1 x 10-3)(2.7 x 10-5)(1)(7.1 x 106) man-rem/RY|
|= 0.594 man-rem/RY|
If resolution of the issue does not permit deinerting with one train of a safety system inoperable, the adjusted case risk rz is calculated from the following relationship:
|= (3.1 x 10-3)(2.7 x 10-5)(2.25 x 106) man-rem/RY|
|= 0.188 man-rem/RY|
Thus, the change in plant risk for this case due to resolution of the issue is (rz - rz) = (0.594 - 0.188) man-rem/RY = 0.406 man-rem/RY. When applied to the affected population of 33 plants over the average remaining life of 18.6 years, the potential risk reduction attainable by not permitting deinerting with one train of any safety system inoperable is 249 man-rem.
Industry Cost: Industry costs included preparation and implementation of TS changes for the affected plants and replacement power costs that would be incurred because TS eliminating deinerting prior to shutdown under off-normal conditions will lengthen plant outages by up to 24 hours per shutdown. At $18,000/ plant, the total industry cost for simple TS changes was estimated to be $594,000.961
Assuming that each time a plant must shut down for an unidentified leak in the containment building and/or a safety system train inoperable not allowing the 24-hour deinerting period will add one day to the plant outage, the average replacement power cost was estimated to be $300,000/day for each instance. Ratioing the 9 containment leakage events and the 1 loss of a safety system train event from the 5-year survey of LERs, it was determined that for the 33 affected plants over their remaining life of 18.6 years, 38 leakage events and 4 safety system train events would be expected. This equates to an industry replacement power cost of $12.6M ($11.4M for leakage events and $1.2M for loss of a safety system train events).
Thus, the total estimated industry costs were $12M and $1.8M for the leakage and safety system inoperability aspects of the issue, respectively.
NRC Cost: Resolution of either or both parts of this issue will require the issuance of a backfit order and the development, review, and approval of a revised TS for each of the affected plants. Development and approval of the resolution of the issue was estimated to require a staff effort of $100,000 and a technical assistance contractual effort of $250,000, for a total of $350,000. Imposition and implementation of the resolution of the issue, i.e., review and approval of a simple TS change, were estimated to be $11,000/plant, and $363,000 for the 33 affected plants. Thus, the total NRC cost was estimated to be $713,000.
Total Cost: For Case 1, the total industry and NRC cost associated with the possible solution is $(12 + 0.713)M for leakage events and loss of a safety system train events. For Case 2, the total industry cost is $(1.8 + 0.713)M.
Separate value/impact scores were calculated for each case.
(1) Case 1 - Leakage Only
(a) Based on no increased probability of LOCA,
(b) Based on two orders of magnitude increase in LOCA probability,
(2) Case 2 - Safety System Train Inoperable
License Renewal: The remaining life of the plants used to calculate the risk, cost, and value/impact scores was based on the assumption that the total operating life of the existing operating plants was limited to 40 years. The potential for license extension was also considered with the assumption that 75% of existing operating plants would apply for license extensions of 20 years.
The additional risk reduction increment from license extensions would not result in a total potential risk reduction of more than 500 man-rem for either aspect of this issue. Since both the risk reduction estimates and the licensee costs estimates were a direct function of remaining plant operating life, the value/impact scores remain essentially unchanged by consideration of license extension.
Based on the value/impact scores and potential risk reduction for both the leakage and safety system train inoperability aspects, the issue was given a low priority ranking (see Appendix C) in October 1991. Consideration of a 20-year license renewal period did not change the priority of the issue.1564 Further prioritization, using the conversion factor of $2,000/man-rem approved1689 by the Commission in September 1995, resulted in an impact/value ratio (R) of $10,040/man-rem which placed the issue in the DROP category.