Resolution of Generic Safety Issues: Issue 137: Refueling Cavity Seal Failure ( NUREG-0933, Main Report with Supplements 1–34 )
On August 21, 1984, the Haddam Neck plant experienced failure of a refueling cavity seal during preparations for refueling. The failure of the seal caused 200,000 gallons of water to drain from the refueling cavity into the lower levels of the containment building in 20 minutes. No fuel was being transferred at the time. If a similar seal failure were to occur at a plant during fuel transfer, fuel elements could be uncovered and could result in high radiation exposure to plant personnel, possible fuel cladding failure, and release of radioactive material. Also, because the refueling cavity is connected to the spent fuel storage pool, the potential exists for this seal failure to initiate drainage of the spent fuel pool, if the fuel transfer canal were open at the time.
Refueling cavity seal failure could lead to an event sequence not previously considered by the NRC, i.e., uncovering spent fuel being transferred and spent fuel in storage in the spent fuel pool. These sequences are not considered explicitly in the SRP11 nor in the NRC guidance pertaining to acceptability of facility designs, technical specifications, operating procedures, and emergency procedures. The SRP11 Sections that may be affected by this issue are 9.1.2 "Spent Fuel Storage," 9.1.3, "Spent Fuel Pool Cooling and Cleanup System," and 15.7.4, "Radiological Consequences of Fuel Handling Accidents." Regulatory Guide 1.251154 may also be affected.
Following the event at Haddam Neck, IE Bulletin No. 84-031158 was issued requiring licensees to investigate the potential for refueling cavity seal failures at their plants. Of the total responses received from OLs (72 plants with 112 units), 26 plants (40 units) use an inflatable reactor cavity seal. Most of the plants use a pressurized bladder of similar configuration to that used at Haddam Neck. Some significant differences between the Haddam Neck seal bladder design and those used at other plants were noted as well as other plant design features which might provide some capability to cope with the consequences of reactor cavity seal failure.
This issue is related to Issue 82, "Beyond Design Basis Accidents in Spent Fuel Pools," but was determined not to be encompassed within the scope of Issue 82.1155 In Issue 82, a refueling cavity seal failure was not included as an initiating event for accidental draining of a spent fuel pool. Therefore, Issue 137 was established as a new generic issue rather than expand the scope of Issue 82 to include reactor cavity seal failures as an additional initiator of possible spent fuel pool accidents.1156,1160
A refueling cavity seal failure is itself considered to be an initiating event for an accident sequence. The immediate result of a refueling cavity seal failure during fuel transfer is the loss of water from the refueling cavity. The possible safety consequences are as follows: (1) high radiation levels in the containment due to uncovering of spent fuel in transfer; (2) radioactive material release in the containment building due to rupture of fuel pins (by self-heating after uncovering); (3) high radiation levels in the spent fuel building due to uncovering of stored spent fuel; and (4) radioactive material release outside the containment building due to rupture of fuel pins in the storage pool. The consequences involving the spent fuel pool are based on the assumption that the fuel transfer canal connecting the refueling cavity to the spent fuel pool is open at the time of the initiating seal failure and that the canal cannot be closed.
Pneumatic rubber seals similar to the one at Haddam Neck (used mainly at PWRs) are most vulnerable to failure. They are susceptible to misalignment, improper inflation, puncture, and rupture. Other types of seals, such as the permanent steel bellows on most BWRs, have been more reliable than pneumatic seals.1157 At least 45 plants are equipped with pneumatic seals: 38 operating PWRs, 4 operating BWRs, and 3 PWRs under construction. Thirty-six of the affected plants use a seal with a single inflatable gland. Nine plants (including Haddam Neck at the time of the seal failure there) use a seal with two inflatable glands separated by a metal spacer ring. Only five plants indicated in their response to IE Bulletin No. 84-031158 that their spent fuel pools could possibly drain through a failed refueling cavity seal.
The proposed resolution to this issue is actually made up of several resolutions which apply to different plants according to each individual plant's refueling cavity configuration, seal design, and operating procedures. The various aspects of the proposed resolution are based on assessments by the Haddam Neck staff1159 and the NRC.1156 All of the proposed actions are aimed at bringing the affected plants into conformance with the features employed at the nuclear plants that are least vulnerable to a refueling cavity seal failure and its consequences. Several of the proposed actions have already been implemented at Haddam Neck.
The overall proposed resolution includes both mitigative and preventive measures.
Proposed mitigative measures include: (1) temporary reinforcement of existing seals until permanent corrective measures are implemented; and (2) implementation of procedures to assure prompt operator response to a leak and to gross seal failure (completed at Haddam Neck). Preventive measures include: (1) installation of improved-design seals at plants with single inflatable seals; (2) replacement of double inflatable seals with permanent steel seals (completed at Haddam Neck); and (3) installation of a coffer dam to prevent spent fuel pool draining through the refueling cavity at plants where this is possible (completed at Haddam Neck).
Two accident sequences were considered: (1) refueling cavity seal failure resulting in serious transfer canal drainage; and (2) seal failure resulting in spent fuel pool drainage. The probabilities of these accidents were developed from the frequencies of the steps in the accident sequences. The frequency of the initiating event, an inflatable refueling cavity seal failure, was estimated to be 10-2/RY based on historical information compiled in NUREG/CR-49821157 and corroborated by PNL calculations.64 In addition, in NUREG/CR-4982, it was estimated that the seal failure rate decreased by a factor of 10 to 10-3/RY, due to improvements in design and increased awareness of the problem following the Haddam Neck incident.
The frequency of transfer canal drainage accidents (Y) was estimated by taking the product of the following: (a) the frequency of refueling cavity seal failure; (b) the probability that spent fuel is being transferred at the time of seal failure; and (c) the probability that reactor operators do not recover from the seal failure incident in time to prevent drainage of the refueling cavity. The probability that fuel is being transferred is estimated based on the assumption that, during a refueling outage at an average plant (i.e., with seal in place and canal flooded), fuel is actually in transit through the canal only a portion of the time. We have, therefore, assumed that the frequency of spent fuel being in transit concurrent with seal failure to be 0.5/RY. The frequency of non-recovery is taken from NUREG/CR-49821157 in which it is estimated to be 5 x 10-2/demand. Thus, the frequency of refueling cavity seal failure resulting in serious transfer canal drainage (Y) is estimated to be:
Y = (10-3/RY)(0.5)(5 x 10-2) = 2.5 x 10-5/RY
The parameter Z, the frequency of refueling cavity seal failure resulting in serious spent fuel pool drainage, is estimated to be the product of the frequency of spent fuel pool drainage and the probability of no recovery. The frequency of spent fuel pool drainage resulting from inflatable seal failure has been estimated at 10-5/RY.1157 This value incorporates the estimated seal failure rate of 10-3/RY. The frequency of nonrecovery from this accident is assumed to be the same value (5 x 10-2/demand) as that used above to estimate the value for parameter Y. The resulting frequency for parameter Z is:
Z = (10-5/RY)(5 x 10-2/demand) = 5 x 10-7/RY
The next step in calculating radiation release frequencies is to multiply the accident frequencies by their respective containment failure probabilities. The applicable containment failure mode for Case Y is assumed to be represented by containment penetration leakage. The containment failure probability used in this issue was taken from Appendix A of NUREG/CR-2800,64 which is based on the Oconee 3 PWR. Typical Technical Specifications for PWRs require that containment doors be closed and that penetrations be either closed or capable of being closed by automatic containment purge and exhaust isolation valves. The containment failure probability associated with penetration leakage is 7.3 x 10-3. However, spent fuel pool building integrity is not required during fuel transfer and we have, therefore, assumed a containment integrity failure probability of 1. The release frequencies for parameters Y and Z are calculated as follows:
Spent Fuel Pool
|Fy||= (Y)(7.3 x 10-3)||Fz||= Z x 1|
|= (2.5 x 10-5/RY)(7.3 x 10-3)||= (5 x 10-7/RY) x 1|
|= 1.8 x 10-7/RY||= 5 x 10-7/RY|
For the fuel transfer canal drainage scenario (Case Y), it was assumed that the events resulted in the uncovery of one fuel assembly, which is assumed to be in the transfer process at the time. It was also assumed that the exposed assembly undergoes overheating and melting and thus the site release was based on the NUREG/CR-280064 consequence calculated for PWR Category 4 and BWR Category 4 events. However, since the damaged fuel is limited to a single fuel assembly, the PWR and BWR consequence factors were reduced by a factor of 300 and 600, respectively. (The typical PWR core has about 300 fuel assemblies and the older BWR-2 reactors have nearly 600 fuel assemblies.) The dose consequences for the fuel transfer canal drainage scenario were thus estimated to be 9000 man-rem/ event and 1100 man-rem/event for PWR and BWR plants, respectively.
For the spent fuel pool drainage scenario (Case Z), core-melt was not assumed. Radiation exposure to the public was assumed to occur only in the event of a coincident refueling cavity seal failure and an open fuel transfer canal i.e., a drainage path for the spent fuel pool. Radiation exposures to the public were calculated using non-core-melt accident Release Categories PWR 9 and BWR 5. The dose consequences for these release categories were estimated to be 120 man-rem/event and 20 man-rem/event, respectively.64
The public risk is obtained by multiplying release probabilities (Fy and Fz) by their corresponding public dose consequences. The per-plant public risk estimates are:
|Spent Fuel Pool Drainage|
|PWR: (5 x 10-7/RY)(120 man-rem/event)||= 6 x 10-5 man-rem/RY|
|BWR: (5 x 10-7/RY)(20 man-rem/event)||= 1 x 10-5 man-rem/RY|
|Transfer Canal Drainage|
|PWR: (1.8 x 10-7/RY)(9.0 x 103 man-rem/event)||= 1.6 x 10-3 man-rem/RY|
|BWR: (1.8 x 10-7/RY)(1.1 x 10* man-rem/event)||= 2.0 x 10-4 man-rem/RY|
Thus, the total public risk estimates are as follows:
|PWR: (6 x 10-5 + 1.6 x 10-3) man-rem/RY||= 1.7 x 10-3 man-rem/RY|
|BWR: (1 x 10-5 + 2.0 x 10-4) man-rem/RY||= 2.1 x 10-4 man-rem/RY|
The resolution (SIR) was conservatively assumed to eliminate the problem of refueling cavity seal failure completely. Thus, the accident sequence frequencies and public risk were effectively reduced to zero as a result of the implementation of the SIR.
There are 45 affected plants (41 PWRs and 4 BWRs) with an average remaining lifetime of 28.8 years. Thus, the total risk associated with this issue is [(41)(28.8)(1.7 x 10-3) + (4)(28.8)(2.1 x 10-4)] man-rem or 2.25 man-rem.
Industry Cost: Industry cost related to implementation of the proposed resolution, and to operation and maintenance after implementation, were estimated to be at least $4.1M.
NRC Cost: NRC costs in support of the industry's efforts were expected to be, at a minimum, $2.4M.64
Based on an estimated risk reduction of 2.25 man-rem and a minimum total cost of $6.5M for the proposed solution, the value/impact score is given by:
ORE for post-accident cleanup and repair were considered both for drainage of the fuel transfer canal and for drainage of the spent fuel pool. For the spent fuel pool event, an ORE estimate of 1880 man-rem/event was taken from Appendix D of NUREG/CR-2800.64 It was based on a small LOCA in which the ECCS functions as intended and no fuel melting occurs, but some fuel cladding ruptures. For the refueling canal event, melting of the fuel assembly in transit was assumed. Therefore, assuming an event that results in a portion of the core melting, an ORE of 7640 man-rem/event was estimated.64 The per-plant occupational dose reduction due to accident avoidance was estimated as follows:
|Spent Fuel Pool Drainage|
|(5.0 x 10-7/RY)(1880 man-rem/event) = 9.4 x 10-4 man-rem/RY|
|Transfer Canal Drainage|
|(2.5 x 10-5/RY)(7640 man-rem/event) = 1.9 x 10-1 man-rem/RY|
For the 45 affected plants over the 28.8 years average remaining lifetime, the maximum expected reduction in ORE for cleanup and repair efforts necessitated by reactor cavity seal failure is (9.4 x 10-4 + 1.9 x 10-1) man-rem/RY x 45 plants x 28.8 yrs = 247 man-rem.
Normally, ORE is incurred in making physical modifications or inspections in a high radiation field to implement the resolution of an issue and to clean up a facility during recovery or decommissioning efforts following an event; these aspects are covered above. However, for this issue, the scenario associated with a rapid drainage of the refueling canal with the transfer of a spent fuel assembly in process could subject plant personnel in the containment building to radiation exposure prior to their evacuation of the building. Analysis performed by PNL64 indicates that, for the case of a single PWR fuel assembly lying in the bottom of a dry refueling canal, exposure levels on the order of 10,000 R/hr on the operating deck at the extreme edge of the canal are possible.
Extrapolation of the PNL analysis64 indicates a potential general radiation level on the order of 100 to 500 R/hr on the operating deck due to reflection from the
containment dome. Considering the possible canal drainage time, alarm levels, and evacuation times, it appears unlikely that any operator would receive a lethal dose prior to evacuation; however, it is likely that a few people could receive a whole-body dose on the order of 50 to 100 man-rem. In terms of offsite public exposure, this dose is about comparable to some of the higher probability/lower consequence core-melt scenarios. Since the frequency of this event is estimated to be 2.5 x 10-5/RY, we believe the likelihood of the event is sufficiently low and, therefore, do not perceive a need to alter the priority of this issue on the basis of potential accidental exposure of utility staff.
Based on the low estimated public risk posed by refueling cavity seal failures, the low value/impact ratio of the suggested resolution, and the fact that IE Bulletin No. 84-031158 was issued to address the concern, this issue was placed in the DROP category.