Resolution of Generic Safety Issues: Issue 61: SRV Line Break Inside the BWR Wetwell Airspace of Mark I and II Containments (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
The SRVs of a BWR plant provide protection against overpressurization of the reactor primary system. During normal operation, the SRVs which are mounted in the main steam lines open on high pressure permitting steam to escape from the reactor vessel. SRV discharge lines carry the steam through the drywell, into the wetwell, and discharge into the suppression pool thus condensing the steam. Failure to condense the steam would eventually lead to rupture of the containment boundary and possibly loss of reactor coolant inventory.
This issue postulates a break in the SRV discharge line in the wetwell airspace above the suppression pool of Mark I and II plants. Coupled with the line break is a failure of the relief valve to close after its actuation in response to the transient. The relief valve must be postulated to remain open for a significant amount of steam to escape, bypass the pool, and threaten overpressurization of the containment vessel with rupture in approximately ten minutes. This issue was identified as a potential generic safety issue at the April 29, 1982 meeting of the ACRS Subcommittee on Hydrodynamics and was formally proposed as a generic safety issue by GIB on August 20, 1982.
The scenario described above would result in a direct release of reactor coolant and effluents to the environment. If major core damage or core-melt were to occur, either as a result of the above event or as an independent event, large off-site releases of radioactivity would be experienced.
The three possible solutions postulated are as follows:
1. Reduce the probability of containment failure* for the stuck open SRV with discharge line failure in the wetwell air space by the automation of the Containment Spray System (CSS).
2. Reduce the probability of containment failure* for the stuck open SRV with discharge line failure in the wetwell air space by increased inspection of the discharge lines in the wetwell air space.
* For the purpose of analysis of this issue, overpressure failure of the BWR Mark I and II containments was assumed to result in a core-melt event through loss of the suppression pool. Containment failure for normal SRV discharges to the pool are not assumed as these loads provide only one component of the combined loads (SRV, LOCA, and Seismic) for which the containment (wetwell) is designed.
3. Reduce the probability of containment failure* for the stuck open SRV with discharge line failure in the wetwell air space by installation of guard pipes around the SRV discharge lines in the wetwell air space.
Technical analysis of the first prospective solution for this issue was performed by PNL and is documented in NUREG/CR-2800.64 Elements of the PNL analysis were used by the NRC staff to include the other two prospective solutions.
The issue is a generic concern but is limited to BWR reactors using the Mark I or Mark II containments. Using the data base in Appendix C of NUREG/CR-2800,64 the issue was assumed to be applicable to 24 plants with Mark I containments (22 operating and 2 yet to be licensed for full power) and 10 plants with Mark II containments (2 operating and 8 yet to be licensed for full power).
Solution 1 - Automate Containment Spray
The PNL analysis of the issue made use of a risk assessment performed by BNL.295 The BNL risk assessment identified a series of new accident sequences which had not previously been considered in BWR risk assessments. The series of events were described as T•P•D•Z, T•P•D•(1-Z), and T•P•D•(1-Z)•W where:
T is the frequency of an anticipated transient with relief valve actuation. T = 4/RY.
P is the probability that a relief valve sticks open. P = (1.5n) (4.5 x 10-3)/Transient, where n = the number of valves actuated by the transient.
D is the probability of the failure of a relief valve discharge pipe when subjected to relief valve discharge flow. D = (1/n)(7.4 x 10-5) /event.
Z is the probability that containment spray is not manually activated soon enough after the initiation of the transient (10 min.) to prevent containment overpressure failure. Z = 5 x 10-1 /demand.
W is the probability that containment failure occurs even when containment spray is actuated within the 10 minute period following the initiation of the transient. W = 1.5 x 10-3/demand.
Derivation of values for the probabilities of the individual events (T, P, D, Z, and W) is documented in the PNL risk assessment.64
Of the three new event sequences, BNL determined that the T•P•D•Z sequence is by far the dominant sequence. The PNL review affirms this conclusion. The base case probability of core-melt for the T•P•D•Z sequence was found to be 10-6/RY, the probability of the T•P•D•(1-Z) sequence (a non-core-melt event) was also found to be 10-6/RY, and the T•P•D•(1-Z)•W event (a core-melt event) was found to be 1.5 x 10-6/RY. The T•P•D•Z and T•P•D•(1-Z)•W events are assumed to result in early containment failure and, as a result, BWR Category 2 or 3 releases would be expected. The T•P•D•(1-Z) event is expected to result in containment pressures in excess of design but not containment failure or core-melt and, as a result, a BWR Category 5 release (a dose consequence of about five orders or magnitude less than the above core-melt events). Therefore, the dominance of the T•P•D•Z sequence is established.
The consequences of the T•P•D•Z event are taken to correspond approximately to the WASH-140016 BWR release Categories 2 and 3 because the sequence is expected to result in both an early containment failure and early core-melt. Consequences for these release categories are expressed in man-rem. The total whole-body man-rem dose is obtained by using the CRAC Code64 for the particular release category. The calculations assume a uniform population density of 340 people per square mile (which is average for U.S. domestic sites) and a typical (midwest plain) meteorology. The results of the dose equivalence calculations are expressed in Table D.1 of NUREG/CR-2800.64 The total integrated dose equivalents for BWR Release Categories 2 and 3 are found to be 7.1 x 106 and 5.1 x 106 man-rem, respectively. Using these dose consequences and the frequency of the T•P•D•Z event, a base case risk of 6.6 man-rem/RY was determined.
Separation of the CSS from the ECCS and automatic actuation of the CSS are assumed to reduce the probability of Z (containment failure) by two orders of magnitude (i.e., Z* = 5 x 10-3/demand). This results in an adjusted case probability of core-melt for the T•P•D•Z sequence of 10-8/RY and an adjusted case risk of 6.6 x 10-2 man-rem/RY.
Subtracting the adjusted case values from the base case values, a core-melt frequency reduction of 9.9 x 10-7/RY and public risk reduction of 6.5 man-rem/RY are obtained for the resolution of this issue. Since the issue is assumed to affect 36 BWRs with Mark I or II containments and these affected plants have an average remaining life of 26.8 yrs, the total public risk reduction is 6,000 man-rem and the total yearly core-melt frequency reduction is 3.4 x 10-5/RY.
Solution 2 - Inspect SRV Discharge Lines in Wetwell Air Space
For the purpose of these evaluations, it is assumed that the inspection of the SRV discharge lines in the wetwell airspace at a much increased frequency (at about 3-year intervals) will result in a risk reduction equivalent to that for Solution 1. This may be an overly optimistic assumption but, as shown later, is a sufficient assumption for the comparison of potential solutions.
Solution 3 - Guardpipes Installed Around SRV Discharge Lines in Wetwell Airspace
This solution was also assumed to result in a risk reduction equivalent to that of Solution 1.
Industry Cost: Industry costs were estimated for each solution separately.
Solution 1 - Automate Containment Spray
Implementation of the CSS modifications would require a new containment penetration for a suction line, pump and motor installation, routing pipe for suction and discharge lines and connection to the existing containment spray header, and installation of circuitry for its automatic actuation. Assurance of ECCS integrity is assumed to be best achieved through a separate suction line for the CSS.
Manpower for design, engineering analysis, scheduling, purchasing, planning and QA is estimated to be 285 man-weeks/plant. Labor for installation of the CSS modifications is estimated to be 108 man-weeks/plant for the 24 backfit plants. The 10 forward-fit plants are assumed to be able to achieve the same modifications for a lower labor expenditure, since the modifications can be scheduled into the original construction schedule and the labor expenditure for forward-fit plants is thus estimated to be 86 man-weeks/plant. This results in a total estimated industry cost of $34.1M for the CSS modifications.
The additional CSS equipment and actuation circuitry is assumed to increase plant maintenance and surveillance manpower requirements by 1 man-week/year per plant. This is estimated to be an increase of $2.1M for the industry over the lifetime of the affected plants.
It is assumed that the CSS modifications could be accomplished during a refueling outage for the backfit plants and accommodated within construction schedules for the foward-fit plants. Therefore, no replacement power costs are estimated.
The total industry cost is thus estimated to be $36.2M.
Solution 2 - Inspect SRV Discharge Lines in Wetwell Airspace
It is assumed that an adequate reduction in the probability of SRV discharge line failure can be achieved by requiring an ISI using visual and/or radiographic techniques at a frequent interval. We have assumed that the lines would be inspected three times in every 10-year operating cycle.
Because of the physical location and the lack of permanent structures in the proximity of the SRV discharge lines in the BWR Mark I and II containment wetwells, we have assumed that a 2-man team would require one shift to erect portable scaffolding, perform the inspection, and remove the scaffolding for each discharge line.
The number of discharge lines varies from about 9 in some Mark I plants to about 18 in the Mark II plants. Using the above estimated manpower per SRV discharge line (16 man-hours), the average number of discharge lines for the Mark I and Mark II plants, the assumed inspection frequency, and the previously identified affected group of plants and their respective projected lifetimes, we estimate about 18,000 man-weeks of inspection to be required. In addition, we have assumed that for every hour of physical inspection time it is expected the licensees will be required to expend a like amount of time in support of the inspector. We have, therefore, estimated that this solution would require an additional 18,000 man-weeks of supportive services (planning, report writing, film reading, repairs, logistics, etc.) over the lifetime of the affected plants.
At a cost of $100,000/man-year, we estimate the licensees total cost of this solution to be $72M.
Solution 3 - Guardpipes Installed Around SRV Discharge Lines in Wetwell Airspace
The installation of guardpipes around SRV discharge lines in the BWR wetwell airspace has been done at some European installations. The task is complicated by restricted access (in the Mark I containment especially) and the necessity to provide adequate support for the impingment loads and pool swell loads. In operating plants, the job is further complicated by the fact that the work must be performed in a low level radiation environment.
We believe that the complete installation of guardpipes in the hostile environment will require at least 3 months. Therefore, we believe that it would be extremely optimistic to assume that this solution could be pursued without an interruption of plant power generation. We estimated that it would require 1 month of replacement power at each of the 24 operating plants for a total cost of $216M for replacement power only.
Because the replacement power costs are envisioned to be so large, we did not perform a more detailed estimate of total licensee costs for this solution. However, we feel confident that total licensee costs for this solution would exceed $250M.
NRC Cost: For all three solutions, it was estimated that 1 man-year ($100,000) would be required to complete resolution of the issue, review and approve new requirements, and issue implementation orders. One man-week/plant was assumed for the review of licensees' plant modifications and/or operational changes. In addition, 0.1 man-week/RY was estimated for long term inspection of the licensee surveillance activities associated with the plant modifications and/or operational changes. This results in an estimated cost of $300,000 for the above review and inspection activities. The total NRC cost for resolution of the issue is thus estimated to be $400,000.
Assuming that all 3 solutions will result in the same public risk reduction (6,300 man-rem), Solution 1 appears to be the most effective solution. It is the least costly and is expected to result in the least ORE. The value/impact score for Solution 1 is given by:
All 3 potential solutions to this issue will entail entries to the containment wetwell and activities in a low level radiation field. In Solution 1, only a portion of the expended manpower would be performed in the radiation field. In Solution 3, almost all of the physical manpower would be expended in the radiation environment. In Solution 2, all the inspection effort would be in the radiation environment.
We have assumed that all efforts performed in the containment wetwell and the ECCS pump room are performed in a 15 millirem/hr field. Using the previous labor manpower estimates, we have estimated an ORE of 2,100 man-rem for Solution 1 (automatic CSS) and 11,000 man-rem for Solution 2. Solution 3 (guardpipes) would be expected to have a greater ORE than Solution 1.
Issue 85 focused on failures of the VB valves attached to steam lines which discharge to the pressure suppression pool in BWR containment buildings. During the course of the analysis of Issue 85, it was determined that the failure of a VB valve on an SRV discharge line in the closed or near-closed position poses a potential for increased hydrodynamic loads on the SRV discharge line and the containment wetwell and, as such, the risk associated with that scenario should be considered in this issue (Issue 61). Likewise, the closed failure mode of VB valves on the HPCI and RCIC turbine exhaust lines poses a similar threat to the containment wetwell integrity and must also be considered in this generic issue. Failure of the HPCI or RCIC turbine exhaust line VB valves in the open position could result in bypass of the suppression pool and overpressurization of the wetwell and thus also must be considered in the analysis of this issue.
The analysis of the SRV discharge line and HPCI/RCIC turbine exhaust line VB valve failure is described as follows:
SRV Discharge Line VB Valve: Failure of the SRV discharge line VB valve in the open position is considered in Issue 85. Our review of the Grand Gulf dominant risk sequences found that none of the identified sequences were appropriate for the case in which the VB fails such that the valve disc is fixed in the closed or near-closed position. Failure of the VB in the closed or near-closed position, when combined with a second actuation of its SRV ("second pop"), could result in increased hydrodynamic loads. If the increased hydrodynamic loads are severe enough, the following failures could occur: (1) failure of the wetwell structure, (2) failure of the SRV discharge line, or (3) failure of the SRV.
We developed new sequences which are appropriate for the failure of the VB in the closed or near-closed position and subsequent events. The initiating events are common to all the sequences and are defined and assigned the indicated probabilities as follows:
|(T1)||= frequency of a transient initiated by loss of power||=0.2/RY|
|(T23)||= frequency of all other transients resulting in reactor shutdown||=7.0/RY|
|(SRV)1||= probability of SRV actuation for a T1 transient||=1.0/event|
|(SRV)1||= probability of SRV actuation for a T23 transient||=0.8/event|
|(SRV-2)1||= probability that an SRV, once actuated in response to a T* transient, will undergo a second opening||= 1.0/event|
|(SRV-2)23||= probability that an SRV, once actuated in response to a T23 transient, will undergo a second opening||= 0.8/event|
|(X)||= probability of VB valve failure||= 0.0093/demand|
|(Y)||= probability that, given a VB valve failure, the disc remains in a static closed or near-closed position||= 0.01/VB failure|
Bases for the values assigned to the above specific events are explained below. The frequencies for the T1 and T23 events were taken from the Grand Gulf PRA. Since loss of offsite power will result in closing of the MSIVs, it was assumed that, for the T1 transient, the probability of both the initial opening of an SRV, (SRV)1, and a subsequent second opening of the SRV, (SRV-2)1, would be 1/event.
For the T23 transient, a probability of SRV actuation (SRV)23 and second SRV actuation (SRV-2)23 of 0.8/event was assumed because most BWRs can accommodate a trip from about 50% power with adequate heat rejection through the turbine bypass and many trips occur during start-up or at low power. The probability of VB valve failure (X) was derived by PNL64 from LER data. The probability of a VB failure occurring such that the disc is firmly stuck in the closed or near-closed position was assumed to be 0.01/event because, although it is theoretically possible for this failure to occur, it is not the expected failure and no instance of this type of failure has been observed.
For the scenario in which wetwell failure is postulated, the sequence of events leading to severe core damage is given by:
fem = [(T1)•(SRV)•(SRV-2)1+(T23)•(SRV)23•(SRV-2)23•(SRV-2)23]•(X)•(Y)•(FCON)•(CM)
In this scenario, (FCON) and (CM) are defined and assigned the indicated probabilities as follows:
|(FCON)||= probability that the wetwell fails due to increased hydrodynamic loads||= 10-4 to 10-5 /demand|
|(CM)||= probability that transient escalates into a severe core damage event because of wetwell failure||= 0.1/event|
FCON was assumed to be in the range of 10-4 to 10-5/demand after considerable discussions with CSB, GIB, MEB, and SGEB which revealed that: (1) results from tests of the Monticello and Caorso (Italy) SRV discharge lines, when extrapolated through engineering judgment, would indicate a maximum hydrodynamic load on the wetwell of less than twice the load used for design of the wetwell, and (2) when this increased hydrodynamic load is considered in a mechanistic combination of applicable loads (as opposed to the non-mechanistic load combination used for design), failure of the containment wetwell would not be expected. The probability of severe core damage (CM) caused by containment wetwell failure (i.e., loss of recirculation water inventory) is the same value that was calculated previously. This value was obtained by adjusting the probability of core-melt derived for a PWR with loss of recirculation coolant (0.25/demand) to account for the significantly greater volume of water available for injection from the Condensate Storage System in the BWR design, as well as the availability of more "avenues" for getting that water to the reactor. Using the above accident sequence and the indicated event probabilities results in a calculated core-melt probability of 4.35 x 10-9/RY. Since the failure of the wetwell due to hydrodynamic loading would occur early in the transient, it was assumed that the consequences of this event would be best approximated by the BWR Category 2 Release Category (i.e., 7.1 x 10-6 man-rem/event). When applied to the population of 44 BWRs with an average remaining life time of 27.4 years, a total potential public risk of 37.2 man-rem was calculated for this scenario.
The second scenario (failure of an SRV VB valve in the closed or near-closed position when combined with a second SRV opening, "2nd POP") causes an increased dynamic load on that SRV discharge line. The analysis of this scenario, using the previously identified common initiating events and the containment failure and core-melt probabilities, indicates a total estimated public risk potential of 144 man-rem. Since this scenario represents just another "avenue" for SRV discharge line failure in the wetwell air space, the public risk potential was calculated previously.
The failure of an SRV VB valve in the closed or near-closed position when combined with a "2nd POP" of its SRV might also result in damage to the SRV (scenario three) because of increased dynamic loads or water hammer in the SRV discharge piping. This might cause the SRV to either not open under its next opening demand or fail in the open or leaking position. We have not attempted to evaluate the fail-to-open-on-demand event because of the extensive degree of redundancy in the design of the BWR SRVs and safety valves and the fact that failure of an SRV to open is a DBA. (PRAs for BWRs have consistently shown that DBAs are not a significant contributor to public risk.) For the case of induced failure of the SRV in the open position, examination of the Grand Gulf PRA reveals that the probability of SRV failure in the open or leaking position (P) used in all PRA scenarios is 0.1/demand. To include the effects of VB failure in the closed position in those scenarios resulting in an SRV failure in the open or leaking position, one would add the probability of that event to the 0.1/demand assumed in the PRA and recalculate the risk involved with all accident sequences in which P is one of the events in the sequence. We have calculated the additional probability of the SRV open or leaking failure due to VB failure in the closed position and SRV "2nd POP," using the conservative assumption that this event will always result in an SRV open (or leaking) failure. We calculated this additional probability of P to be 1.5 x 10-4/demand. This is insignificant compared to the 0.1/demand assumed in the Grand Gulf PRA calculations and can therefore be neglected.
It should be noted that two design features of some BWR plants were not factored into the above calculations, and that consideration of these features would result in a reduction of the calculated potential core-melt frequency and public risk. Specifically, these design features are parallel VB valves and SRV low-low level reset logic. Nearly half of the BWRs have two VB valves per SRV discharge line in a parallel flow path arrangement (i.e., a redundant VB valve). For the plants with this design arrangement, the potential public risk associated with a VB valve failure would be one to two orders of magnitude less than we have calculated for the failed-closed VB event. Sixteen of the BWRs have adopted a low-low reactor coolant level SRV reset logic as a means of reducing the number of second SRV openings. If this factor were to be included in a more rigorous analysis, the public risk associated with the SRV VB valve failed-closed event would be reduced, but not by as large a factor as for the parallel VB valve design.
Adding the potential public risk estimates for the 3 failure scenarios associated with the SRV VB valve fail-closed (or near-closed) mode results in a total maximum potential public risk of about 180 man-rem.
Vacuum Breakers on Other Steam Lines: We found that only the HPCI and RCIC turbine discharge lines can additionally discharge steam to the wetwell pool and are equipped with VBs. Our review of the HPCI and RCIC system drawings from the OIE Training Center BWR Systems Manual and conversations with one of the Training Center BWR instructors indicate that the vacuum relief lines for both turbine exhausts are 2 in. lines and have multiple VBs and a motor-operated valve in series/parallel arrangements which are designed to provide redundacy for both the VB fail-closed and fail-open scenarios. However, the sensitivity to the fail-open (leaking VB) event is greatly heightened because the HPCI and RCIC turbine discharge line vacuum is relieved by the vacuum line to the wetwell air space instead of the drywell as is done with the SRV discharge lines.
We have analyzed VB failure events for these lines in the same manner as the analysis for the SRV line VBs, using the VB failure frequency determined by PNL.64 As in the case of the SRV VBs, we found that, for the assumption of VB failure in the closed or near-closed positions, there were 3 possible severe core damage scenarios: (1) failure of the containment wetwell structure due to increased hydrodynamic loads, (2) failure of the HPCI or RCIC turbine exhaust line in the wetwell air space, or (3) induced failure of the HPCI or RCIC turbine system.
Considering the specific HPCI and RCIC turbine exhaust line and vacuum relief line configurations, we calculated the following public risk potential for each of the three scenarios: (1) 0.18 man-rem, (2) 24 man-rem, and (3) 1.6 man-rem. Thus, the public risk potential for the HPCI/RCIC VB fail-closed scenarios is about 26 man-rem, a relatively small value.
Such is not the case for the HPCI and RCIC VB fail-open scenario because the leakage would bypass the containment suppression pool resulting in pressurization of the wetwell airspace (similar to SRV and HPCI or RCIC turbine discharge line failures in the wetwell air space). Our analysis of this scenario indicates a potential public risk of 820 man-rem for this scenario. Since the end result of this scenario is severe core damage as a result of wetwell air space overpressurization, the potential public risk from this scenario is considered in this issue.
We regard the above calculated public risk associated with failures of the HPCI and RCIC turbine exhaust VBs to be a closer approximation to an upper bound estimate than to an average estimate which would normally be used in a prioritization analysis. Available LER data indicate that there has only been one confirmed HPCI or RCIC VB failure (a leakage failure). A more rigorous determination of the failure rate for HPCI/RCIC VBs would probably result in a calculated failure rate of one or two orders of magnitude less than that calculated by PNL64 for SRV and HPCI/RCIC VBs from the combined failure data. However, the use of such failure data (one confirmed failure) would introduce large uncertainty in the estimate of HPCI/RCIC VB failure rate.
As calculated above, the maximum potential public risk associated with HPCI and RCIC turbine exhaust line VB failures is about 350 man-rem. Therefore, the total maximum public risk that can be attributed to the failure of SRV discharge line VB valves in the closed (or near-closed) position and the failure of HPCI and RCIC turbine exhaust line VBs is about 1,030 man-rem. If it is assumed that the resolution of this issue will result in an order of magnitude reduction in public risk from these VB failures, an additional maximum public risk reduction of about 930 man-rem can be allocated to the resolution of Issue 61. Factoring this into the priority score calculation and the total public risk reduction results in a revised priority score of about 190-man rem/$M and a revised potential public risk reduction of about 7,000 man-rem. Therefore, addition of the VB failure concerns will not alter the priority recommendation.
Staff efforts on the resolution of Issue 61 have been initiated. The approved Task Action Plan for the issue includes a sub-task devoted to the development of an accurate estimate of the failure rate of HPCI/RCIC turbine exhaust line VB valves. This is essential in determining whether improvement in HPCI and RCIC exhaust line VB valves is warranted, should a resolution for Issue 61 be adopted which does not result in automation of containment sprays.
The calculated value/impact score was indicative of a medium priority ranking and it was believed that, if ORE estimates were considered, it would tend to lower the priority of the issue. In August 1986, a regulatory analysis998 performed by the staff showed that the contribution of the postulated event to core-melt frequency and public risk was lower than previously estimated. The regulatory analysis998 was based on information published in NUREG/CR-4594.999 Thus, this issue was RESOLVED and no new requirements were established.