Resolution of Generic Safety Issues: Issue 143: Availability of Chilled Water Systems and Room Cooling (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
In past years, several nuclear plants experienced problems with safety system components and control systems that resulted from a partial or total loss of HVAC systems. Many of these problems occurred for two reasons: (1) the desire to provide increased fire protection; and (2) the need to avoid severe temperature changes in equipment control circuits. Since the Browns Ferry fire, considerable effort had been expended to improve the fire protection of equipment required for safe shutdown. Generally, this improvement had been accomplished by enclosing the affected equipment in small, isolated rooms. However, the result was a significant increase in the impact of the loss of room cooling. Another problem resulting from loss of room cooling was the technological advances in control circuit design. With the introduction of electronic integrated circuits, plant control and safety improved; however, these circuits were more susceptible to damage from severe changes in temperature caused by the loss of room cooling.
It was believed that failures of air cooling systems for areas housing key components, such as RHR pumps, switchgear, and diesel generators, could contribute significantly to core-melt probability in certain plants. Because corrective measures are often taken at the affected plants once these failures occur, the ACRS believed that the impact of these failures on the proper functioning of air cooling systems had not been reflected in the final PRAs of plants. Thus, plants with similar, inherent deficiencies may not have been aware of these problems.1317
Operability of some safety-related components is dependent upon operation of HVAC and chilled water systems to remove heat from the rooms containing the components. If chilled water and HVAC systems are unavailable to remove heat, the ability of the safety equipment within the rooms to operate as intended cannot be assured.
A possible solution to this issue would require a reevaluation of each plant's room heat load and heat-up rate in order to identify areas in which a reduction in the dependence of equipment operability on HVAC and room cooling may be implemented. While the total elimination of this dependence may not be possible at all plants, this reevaluation would identify areas in which this dependence was critical. The determination of the critical dependencies and the ability to reduce them could be accomplished through the use of a plant-specific PRA such as an IPE. After the critical dependencies were identified, each plant would implement procedural changes (to provide alternate cooling) to eliminate or reduce the dependencies where possible. Hardware modifications may be needed for situations in which a procedure change cannot be implemented to reduce a critical dependency.
The next step in the possible solution would be the issuance of a generic letter that would require licensees to: (1) evaluate the dependencies of plant safety systems and equipment operability on HVAC and room cooling; (2) identify areas in which this dependence was critical; (3) identify appropriate procedure changes and hardware modifications to minimize the effects of the dependencies on plant risk; (4) submit this evaluation to the NRC for review and approval of the proposed modifications; and (5) implement the approved procedural changes and hardware modifications. The generic letter would include guidance on acceptable procedures licensees could use to evaluate the potential dependencies in the design of these systems. The generic letter would also include alternative solutions for improving the independence of systems that are critical to plant risk. It was assumed that a research project would form the basis for a more fully-developed solution and for the guidance in the generic letter.
The Grand Gulf Unit 1 and Oconee 3 PRAs were used as representative of BWRs and PWRs, respectively. A total of 90 PWRs and 44 BWRs were potentially affected by this issue and were expected to continue operating for an average of 28.8 and 27.4 years, respectively.64 For comparative purposes, 71 operating plants were assumed: 47 PWRs and 24 BWRs.
Several systems at Grand Gulf 1 were affected by HVAC or room cooling failures in locations that included: (1) emergency switchgear and battery rooms; (2) standby diesel generator rooms; and (3) pump rooms for RHR, RCIC, high pressure cooling system (HPCS), and low pressure cooling system (LPCS). A loss of cooling was assumed to fail the following: (1) operating diesel generators in 15 minutes; (2) battery, emergency switchgear, and low pressure ECCS (i.e., RHR and LPCS) pumps in 4 hours; and (3) HPCS and RCIC in 12 hours. An examination of the Grand Gulf 1 emergency ventilating system fault trees provided a list of basic events that result in failure of systems due to HVAC failure.64
Next, the core-melt sequences for the dominant core damage categories were examined. When a sequence was found that contained one of the above-mentioned basic events, it was reviewed to determine the effect of eliminating the dependence on HVAC/room cooling. Three effects were possible. First, the sequence could no longer result in core damage. These scenarios were eliminated, resulting in a direct reduction in CDF. Second, the HVAC/room cooling basic event was replaced with another basic event (e.g., replacing diesel generator room cooling failure with a diesel generator hardware failure). The new sequence was compared with the other sequences in the dominant sequence mode and, if an identical sequence was found, the new sequence was absorbed, resulting in a CDF reduction by the amount of the frequency of the absorbed cutset. The third effect was similar to the second except that no other sequence was found to be identical. In this case, this same new sequence was substituted for the original and a new sequence frequency was calculated. The contribution to changing the core-melt frequency was the difference between the original and the new frequency.
The evaluation was performed assuming that all dependence on HVAC and room cooling could be eliminated. The elimination of this dependence was expected to be accomplished by a reevaluation of room heat capacity and heat-up, by the implementation of procedures that provide alternate cooling, or by the implementation of hardware modifications that would eliminate the dependence. The result of the evaluation was a CDF reduction of 8 x 10-6 event/RY at Grand Gulf 1. It should be noted that this reduction was a maximum using the above assumptions and might not have been achievable in all plants. However, every nuclear plant should realize some reduction in the probability of core damage.
The HVAC systems at Oconee 3 consisted of various flow-through systems as well as individual room coolers. Two flow-through systems, one for the control room and one for the auxiliary building, and the room coolers for the low pressure injection (LPI) and reactor building spray system (RBSS) were examined in detail. The PRA determined that only the room coolers for LPI and RBSS had a potential contribution to core damage; this contribution was examined below. The unavailability values were obtained from Appendix A of NSAC-60.889
|Room Cooler Failure||= (Fan 1)(0perator + Fan 2 )|
|= (1.9 x 10-4)[10-3 + (1.9 x 10-4)]|
|= 2.4 x 10-7|
|LPI Pump Fails (Start and Run)||= 1.3 x 10-3|
|Room Cooler Contribution||= 0.02%|
|RBSS Pump Fails (Start and Run)||= 8.3 x 10-3|
|Room Cooler Contribution||= 0.003%|
As a final part of the evaluation, an examination of the leading core damage sequences from internal events was made. Failures of the LPI and RBSS pumps were not contributors to any of these sequences.
A second potential contribution from HVAC or room cooling was through the loss of service water which is the heat sink. Service water, however, also provides direct cooling of the high pressure injection (HPI) pumps. An examination of the event trees for Oconee 3 shows that core damage mitigation using the low pressure systems could only be accomplished if HPI was successful first. Therefore, a failure of service water would result in a loss of HPI and core damage and no contribution from loss of room cooling would be found.
Based on the preceding discussion, it was concluded that no further analysis using NSAC-60889 would be performed because of a lack of dependency on HVAC and room cooling. Therefore, to model the effects of the solution on a generic PWR, the scaling technique described in NUREG/CR-280064 was used. In this technique, it was assumed that the core-melt frequency for a BWR was 0.45 times that for a PWR. In addition, it was assumed that the public risk for a BWR plant was approximately 1.2 times that for a PWR plant. This resulted in a core-melt frequency reduction estimate of 1.8 x 10-5/RY.
The reduction in core-melt frequency for BWRs was translated into a reduction in public risk of about 57 man-rem/RY.64 For the 44 BWRs, this amounted to a risk reduction of 6.9 x 104 man-rem. The public risk reduction for PWRs was calculated to be 47 man-rem/RY.64 For 90 PWRs, this amounted to a risk reduction of 1.2 x 105 man-rem.
Industry Cost: The proposed solution called for licensees to evaluate the dependencies of HVAC and chilled water systems on room cooling systems and the subsequent effects on plant risk. This engineering analysis was assumed to require approximately 1 man-year of effort ($100,000/plant) and would include the identification and evaluation of dependencies as well as identification of appropriate procedure changes and hardware modifications to minimize the effects of the dependencies on plant risk. Licensees would submit their evaluations to the NRC for review.
The next step would be for licensees to implement the proposed procedural changes and hardware modifications at their plants. Procedural changes would be proposed where possible because of the expense and occupational doses associated with implementing hardware modifications in radiation areas. However, it was believed that procedural changes would not resolve the dependencies in some cases, particularly where failure of room cooling was a critical failure mode for an important safety-related component. In these cases, it may be necessary to replace a particular piece of equipment with one that can be operated independently of HVAC and chilled water systems. Alternatively, a component of the HVAC or chilled water system, such as one or more of the room coolers, may be replaced with one of a different design that can be operated independently of the HVAC or chilled water systems.
The cost estimates for the procedural changes and hardware modifications were taken from NUREG/CR-4627.961 It was assumed that each plant would implement 2 procedural changes and replace 8 room coolers. The procedural changes were anticipated to be relatively complex and, therefore, would require approximately $3,600/change or $7,200/plant. The proposed solution was somewhat similar to the removal and replacement of the containment spray heat exchangers because work would be performed in a radiation environment and in a relatively congested area.961 It was also stated in NUREG/CR-4627961 that these cost estimates should be applied only to plants that were operating or at the 70% or higher completion state of construction. The removal and replacement cost was therefore applied only to the assumed 71 operating plants. The proposed resolution was assumed to result in no incremental costs for plants that were under construction. The removal and replacement costs were summarized as follows for each operating plant:
|(1) Purchase Cost (8 Room Coolers)||= $ 968,000|
|(2) Labor Cost (12,400 man-hrs @ $21.74/hr times labor productivity adjustment, removal, engineering, and QA factors)||= 356,000|
Therefore, the industry cost for implementation was estimated to be $100,000/ plant for the engineering study, $1.324M/operating plant for the hardware upgrade, and $7,200/plant for the procedure changes. These upgrades and changes were anticipated to be one-time activities. No additional resources were expected for operation and maintenance of the upgraded equipment. Thus, the total industry cost for all affected plants was approximately $108.4M.
NRC Cost: The first step in the proposed solution would involve issuance of a generic letter to all licensees to evaluate the dependencies on adequate room cooling of the operation of HVAC and chilled water systems. Issuance of the generic letter was assumed to cost approximately the same as a simple TS change, as described in NUREG/CR-4550,1318 including the necessary reviews and approvals; this cost was $13,000. The generic letter would include a description of the proposed resolution and guidance on acceptable procedures for licensees to use to evaluate the potential dependencies in the design of these systems and alternative solutions to improving the independence of systems that are critical to plant risk. It was assumed that a research project would form the basis for a more fully-developed solution and for the guidance in the generic letter. This research project was assumed to be performed by a contractor and was estimated to require 6 man-months at a cost of $50,000.
Review and approval of the proposed modifications were estimated to require approximately 1 man-week/plant or $2,270/plant. No incremental resources were foreseen for review of the operation and maintenance of the upgraded equipment. Thus, the total estimated NRC cost for all plants was approximately $370,000.
Total Cost: The total industry and NRC cost for implementing the possible solution was estimated to be $(108.4 + 0.37)M or approximately $109M for the 71 affected plants.
Based on an estimated public risk reduction of 1.9 x 105 man-rem and a cost of $109M for the possible solution, the value/impact score was given by:
(1) The analysis for estimating core-melt frequency reduction was performed for BWRs and scaled for PWRs using the scaling relationships given in NUREG/CR-2800.64 This assumption may have introduced considerable additional uncertainty to the estimated core-melt frequency reduction for PWRs. Nevertheless, the contribution due to BWRs alone approached the high priority threshold value, even if the contribution from PWRs were smaller.
(2) Labor for removal and replacement of 8 heat exchangers within containment was estimated in NUREG/CR-4627961 to be approximately 4,000 man-hours. A productivity factor of 3.1 was used to adjust the labor estimate to account for inefficiencies caused by work in radiation zones, congested areas, and difficult access areas. The total labor in radiation zones was estimated to be 12,400 man-hours/operating plant. Assuming a radiation field of 25 millirem/hour inside containment where the room coolers are located, implementation of the possible solution was estimated to result in an ORE increase of 310 man-rem/plant.
Because the estimated risk reduction from eliminating (or decreasing) the dependence of safety systems on HVAC and room cooling could be quite significant, the issue was given a high priority ranking (See Appendix C). In resolving the issue, the staff found that, although there might be some plant-specific accident sequences involving loss of HVAC/room cooling that could contribute significantly to core damage, any vulnerabilities that might exist were expected to be identified by licensees as part of the IPE process; a study of the issue was published in NUREG/CR-6084.1550 The staff's regulatory analysis was documented in NUREG-14271549 and the issue was RESOLVED with no new requirements.1551 In an RES evaluation,1564 it was concluded that consideration of a 20-year license renewal period did not affect the resolution.