Resolution of Generic Safety Issues: Issue 149: Adequacy of Fire Barriers (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
This issue was raised in SECY-89-1701320 and addressed the effect of premature failure of fire barriers on CDF. Concern for fire barriers arose as a result of an NRC-sponsored Fire Risk Scoping Study1211 which focused on existing fire protection practices for control rooms, remote shutdown areas, control transfer areas, and local control areas.
10 CFR 50 Appendix R and the SRP11 contain requirements for the protection of safety-related equipment from fires and related fire-suppression activities. Tests on fire barrier components are conducted in accordance with industry standards ASTM-E119 (fire barriers in general), NFPA-251 (walls and columns), and NFPA-252 (doors and dampers). The tests assure a certain degree of fire barrier reliability; however, they do not simulate actual fire conditions. Test procedures do not include the requirement for component testing under a pressure gradient.
Fire barriers, typically walls or doors, are designed to ensure that a fire will be contained within its enclosure to the extent possible. Tests at SNL1211 suggested that fire barriers in operating nuclear plants might not properly contain the spread of fires because: (1) barrier penetration seals, doors, and dampers may not be sufficiently reliable to prevent fire spread; and (2) deficiencies may exist in the methods used to rate fire barriers.
A rapid energy release from a fire could generate a pressure differential across a fire barrier which could result in a premature failure of the barrier. Many variables such as compartment air tightness, materials, and geometry will affect the magnitude of the pressure gradient. Studies1211 have shown that core-melt frequency is sensitive to fire barrier reliability. This issue affected all LWRs.
A possible solution to this issue involves licensee searches for plant-specific vulnerabilities, an assessment of the reliability of fire barriers based on site-specific experience, and modifications, if necessary.
It was assumed that 90 PWRs and 44 BWRs with an average remaining operating life of 28.3 years were affected by this issue. It was also assumed that 25% of all plants would be required to make some modifications. Cost estimates were based on 1982 dollars.
An estimate of the core-melt frequency due to fire barrier failure was calculated based on the operating history of nuclear power plants. At the time of this evaluation, there were approximately 1700 RY of nuclear power operation. Over this entire period, the only nuclear power plant fire that approached inducing a core-melt was the Browns Ferry fire of March 22, 1975. Thus, a crude but reasonable point estimate for the frequency of a fire with at least the potential core-melt impact of Browns Ferry is 1/(1,700 RY) or 6 x 10-4/RY.
However, the Browns Ferry fire did not result in a core-melt. It was estimated64 that the maximum probability for core damage, given the Browns Ferry fire, was 0.03, occurring 2.5 hours from the start of the fire and assuming all relief valves failed after 30 minutes and remained so for at least 12 hours. Applying this conservative estimate to the point estimate for a Browns-Ferry-type fire, a conservative point estimate for the frequency of core damage from fire was calculated to be (0.03)(6 x 10-4/RY) = 1.8 x 10-5/RY.
Since the Browns Ferry fire represented one in which a fire barrier "failed" (the barrier had previously been breached by non-fire-related activities), this estimate would be a conservative one for core-melt frequency due to fire barrier failure. It lies within the range 10-5/RY to 10-4/RY estimated in NUREG/CR-5088,1211 comparing well with the geometric mean (3 x 10-5/RY) over that range.
Nuclear power plants have made numerous improvements since the Browns Ferry fire to reduce the likelihood of a fire-induced core-melt. Moreover, increased pressures leading to barrier failure resulting from a fire within a compartment in a U.S. nuclear power plant is highly unlikely and is not considered credible. This is based on the fact that compartments are sufficiently ventilated (not fully air-tight), have low and not easily ignitable combustible contents, are provided with early warning fire/smoke detection capability, and, in areas where fire hazards exist that could affect safe shutdown capability, automatic suppression capability is provided.1365 Therefore, it would be reasonable to lower the above conditional probability for core damage from 0.03 to 0.0003 (i.e., a factor of 100 reduction). This resulted in a point estimate for the core-melt frequency due to fire barrier failure of 1.8 x 10-7/RY. Assuming that a 90% reduction in the failure probability of a fire barrier is attainable,1211 the adjusted case core-melt frequency was estimated to be 1.8 x 10-8/RY.
Based on PNL calculations,64 the average LWR dose factor is 3.3 x 106 man-rem. Using this value, the base case risk was estimated to be (1.8 x 10-7/RY)(3.3 x 106 man-rem) or 0.594 man-rem/RY; the adjusted case risk was (1.8 x 10-8/RY)(3.3 x 106 man-rem) or 0.0594 man-rem/RY. Thus, the potential risk reduction is (0.59 - 0.0594) man-rem/RY or 0.53 man-rem/RY. For 25% of the 134 affected plants with an average remaining life of 28.3 years, the total public risk reduction was estimated to be (0.53)(0.25)(134)(28.3) man-rem or approximately 500 man-rem.
Industry Cost: It was estimated64 that 6 man-months will be required to search all 112 operating plants for the potential vulnerabilities identified in this issue. This effort would include a review of plant drawings and existing fire hazards analyses and a walk-through inspection of potentially susceptible areas. Assessment of the reliability of fire barriers, based on site-specific experience, was estimated to require 6 man-months. This effort will involve reviewing the existing fire hazards analyses, estimating the conditions likely to be experienced in an actual fire, and performing sensitivity studies. Based on a cost of $100,000/man-year, the industry cost for the above 12 months of effort will be $100,000/plant. For the 112 operating plants, this cost will be $11.2M.
Following licensee review and assessment of existing fire barriers, some modifications at 25% of the 112 operating plants was conservatively estimated to require $50,000/plant for a total cost of $1.4M. Operation and maintenance of improvements would not incur any additional costs. The cost for future plants was considered to be negligible in comparison to that for OLs.
NRC Cost: It was assumed that the staff will have to issue an information notice or a generic letter with a possible TS change. Based on NUREG/CR-4627,961 this cost was estimated to be $11,0OO/plant. For 112 plants, this cost will be $1.23M. Review and approval of licensee proposals was estimated to require 2 man-weeks/plant for the 28 operating plants that will require modifications. At $2,270/man-week, this cost was estimated to be approximately $127,000.
Total Cost: The total industry and NRC cost associated with the possible solution was approximately $14M.
Based on an estimated public risk reduction of 500 man-rem and a cost of $14M for a possible solution, the value/impact score was given by:
Plant TS require compensatory measures to be taken in the event that a fire barrier is made inoperable. Generally, a fire watch is established for the affected barrier and for the areas which it provides fire separation. In addition, since the Browns Ferry fire, there has not been a case where a fire in a nuclear power plant has challenged the integrity of a fire barrier or has contributed to its failure as a result of differential pressure. In response to the IPEEE program,1354 licensees will have to justify their fire barrier maintenance programs in support of justifying the adequacy of their fire barriers.1365
Based on the potential public risk reduction, this issue was given a low priority ranking (see Appendix C) in October 1992. 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 $27,777/man-rem which placed the issue in the DROP category.