Resolution of Generic Safety Issues: Issue 150: Overpressurization of Containment Penetrations (Rev. 1) ( NUREG-0933, Main Report with Supplements 1–34 )
This issue was identified1330 by DSIR/RES and addressed the concern for overpressurization of containment piping penetrations following a containment isolation and subsequent heat-up.
Containment isolation at all nuclear power plants ensures that radioactive materials are contained if an accident or inadvertent release of such materials occurs. Isolation is provided for all piping systems that penetrate the containment. Double barriers are provided to ensure that no single failure of an active component can result in a loss of this isolation function. Typically, this double barrier system is provided by isolation valves inside and outside containment. When containment isolation is required because of an accident or inadvertent release of radioactive materials, these valves are closed to prevent leakage of radioactive materials to the environment.
Overpressurization of the containment piping penetrations could potentially occur during an accident involving a significant increase in the containment temperature. This might occur when water that is trapped between the inner and outer containment isolation valves is heated and expands. Theoretically, heating a constant volume of water from 100F and 100 psia to 200F would increase the pressure to 3000 psia. This pressure increase could fail the penetration or the isolation valves and could provide a direct flow path to the environment from the potentially contaminated containment atmosphere. The pressure increase is mitigated somewhat by the penetration itself expanding because of the temperature increase, as well as the possibility that the isolation valves will not be leak-tight and thus will not pressurize fully.
A possible solution to this issue was to provide a mechanism for preventing water from becoming trapped or for relieving the pressure that could build up in the piping systems between the inner and outer containment isolation valves. Licensees would need to perform thermal and structural analyses of the penetration systems to determine which penetrations, if any, were susceptible to such failure. A pressure relief system would be needed to prevent the pressure increase from failing the penetration. This pressure relief system could consist of the following:
(a) Check valves inside the reactor building instead of the inner containment isolation valves. These valves would prevent water from becoming trapped between the two isolation valves but is only viable for penetrations with flow into containment.
(b) A method to provide pressure relief, such as a rupture disk or safety valve. A storage tank might be needed to contain blowdown liquid or vapor that would be forced through the pressure relief equipment when the equipment is operated.
This issue does not directly impact the potential for a core damage accident but addresses a plant's ability to contain radioactive materials that might be released during a core damage accident. Thus, the only concern was the probability of containment failure resulting from the failure of containment isolation. Therefore, to estimate the potential public risk reduction, the effect of the possible solution on the probability of containment failure, assuming a core damage event has occurred, was evaluated.
It was assumed that 90 PWRs and 44 BWRs with average remaining lives of 28.8 and 27.4 years, respectively, were affected by this issue. Oconee 3 and Grand Gulf 1 were used as the reference PWR and BWR, respectively.
The Oconee 3 PRA54 addressed failure of, or leakage through, containment penetrations as a potential failure mode of the containment. For Oconee 3, the probability of containment penetration leakage (), given the occurrence of a core damage accident, was estimated to be 7.3 x 10-3. This probability was used as the base case value for Oconee 3.
The Grand Gulf 1 PRA54 did not explicitly evaluate containment penetration leakage because none of the accident sequences involving containment isolation failure were found to be among the dominant sequences from a public risk perspective. However, the Grand Gulf 1 PRA was based on WASH-140016 which did assess the conditional probability of containment isolation failure (). Therefore, the base case value used in this analysis was based on the WASH-140016 analysis.
The probability of containment isolation failure is dependent upon the specific core damage sequence that occurs before containment failure.16 As a result, the containment isolation failure probability depends on the prior success or failure of the various engineered safeguards functions. Based on information in Appendix V of WASH-1400,16 a value of 4 x 10-2 was selected as a representative probability for containment isolation failure. This value represented the weighted average of the range of possible values based on the number of observations in WASH-1400.16
The release categories associated with containment isolation failure for Oconee 3 are PWR-4 and PWR-5. For Grand Gulf 1, the affected release categories were not explicitly stated in NUREG/CR-280064 or NUREG/CR-1659.54 It was noted in WASH-140016 and a PRA889 of Oconee 3 that relatively high containment leakage, attainable from failure of containment isolation, would prevent failure of the containment building from potential overpressure caused by H2 gas explosions. Consideration of the containment isolation failure mode was therefore incorporated into accident sequences involving containment overpressure events caused by gas generation, as shown in Appendix B of NUREG/CR-2800.64 A new base case risk value was developed for all BWR accident sequences that involve containment overpressure events. To accomplish this, the core damage sequences presented in NUREG/CR-280064 that could result in containment overpressure were modified to incorporate the base case containment isolation failure probability, rather than the containment overpressure probability. This had the effect of creating a set of new accident sequences that included containment isolation failure events.
The adjusted case values of the affected parameters were estimated by adding to the base case values the probability of failure of the penetration system (failure of one or more penetrations) that would arise from overpressurization. A conservative approach was taken to develop a new containment leakage sequence of events that incorporated the potential for overpressurization. This sequence consisted of the following events: (1) containment isolation is successful; (2) water becomes trapped between inner and outer isolation valves; (3) containment heating causes heating and expansion of the water between the isolation valves; and (4) the water expansion causes the penetration to fail, such that a leakage path occurs between the containment atmosphere and the environment. Based on this sequence of events and using the rare event approximation, the additional probability of containment system failure was obtained from the product of the following terms:
|N||= number of penetrations that are susceptible to overpressurization|
|P||= probability water becomes trapped between isolation valves|
|P||= probability inboard and outboard isolation valves are leak-tight|
|P||= probability penetration overpressurizes to rupture, given that the penetration is leak-tight and full of water.|
|P[N]||= probability that the penetration fails in a manner that results in a leakage path from the containment atmosphere to the environment, given that the penetration ruptures due to overpressurization.|
N depends on the types of penetrations and isolation valves at each plant. Only liquid penetrations are susceptible to this type of failure and penetrations provided with check valves are not. To determine the value for N, the description of containment penetrations given in NSAC-60889 was assumed to represent both PWRs and BWRs. A total of 62 penetrations were listed, of which, 36 were provided with check valves or were not liquid-carrying lines and were not susceptible to this containment failure sequence. Therefore, the value of N was 26.
No information was available to calculate P. Containment isolation valve closures are timed such that one valve closes slightly sooner than the other to prevent water from becoming trapped between the valves. Therefore, this event could be caused by failure of the containment isolation system control logic or circuitry to function as intended, or by failure of the valve to close when intended. P was assigned a value of 0.5.
In this analysis, P was set equal to unity. Thus, no credit was taken for the fact that penetration overpressurization would not occur if one of the isolation valves were not leak-tight. It was estimated that there was approximately a 30% chance that one of these valves would not be leak-tight.1331
To assist in estimating P, a simplified engineering analysis was performed by DSIR/RES to determine the stress and strain that a penetration would experience, assuming that it is leak-tight and full of water. For the purposes of the analysis, a typical penetration was approximated as a 2" diameter, 12" long cylinder fabricated of steel and having a yield point of 30 ksi. Assuming that the water was initially at 100F and its temperature increased to 200F as a result of an accident, it was calculated that the hoop stress would exceed the yield point. However, the volume of the cylinder would only have to increase 2.6% to accommodate the expansion of the water. This corresponded to a plastic strain of 1.3% in the diameter of the penetration, conservatively assuming no plastic strain in the axial direction. If the water were heated to 300F, the diametric plastic strain needed to accommodate the expansion of the water would be 3.8%, again conservatively assuming no plastic strain in the axial direction. These values of strain were far below the values that would be expected to cause rupture of the penetration. Also, as expansion of the penetration volume occurs due to plastic deformation, the pressure of the trapped water decreases, further decreasing the likelihood of rupture. Using the above information, but considering that there is some probability that the material used to fabricate the penetration has an undetected flaw, the value of P was estimated to be 10-4.
For there to be a leak path that satisfies the definition of P, there must be a failure inside and outside the containment. One possibility is that the penetration rupture "runs" past the containment vessel. The other possibilities involve failures of both containment isolation valves, or the containment penetration and one isolation valve. Such failures must be simultaneous since the failure of one component relieves the pressure and eliminates the possibility of sequential failures. It was estimated that the value of P was 0.1.
Based on the above estimates, the additional probability of containment system isolation failure was the product of N, P, P, P, and P, and was approximately (26)(0.5)(1)(10-4)(0.1) = 1.3 x 10-4.
Incorporating the values into the Oconee 3 and Grand Gulf 1 PRAs resulted in a potential public risk reduction of 1.3 x 10-2 man-rem/RY and 3 x 10-3 man-rem/RY, respectively. Thus, the total potential public risk reduction was about 40 man-rem for all 134 affected plants.
Industry Cost: To implement the possible solution, licensees would be required to perform analyses to determine if certain penetrations were vulnerable to overpressurization following containment heat-up. These analyses were estimated to require 4 man-weeks/plant. In addition, safety analyses and QA-related activities would be needed because of the installation of hardware inside containment that requires considerable attention to QA in designing the penetrations' pressure relief systems. Two man-weeks of labor per penetration were estimated for the design and safety analyses. The number of vulnerable penetrations was assumed to be 20% of the total penetrations without check valves, or 7 penetrations. Therefore, at 2 man-weeks/penetration, 14 man-weeks/plant would be required.
Installing new hardware within containment at operating plants (i.e., backfit) would require about 1 man-week of labor in radiation zones; plants under construction (i.e., forward-fit) would not require labor in radiation zones. The material costs were estimated to be $900/penetration and $6,300/plant, including labor for pipefitters, welders, radiation monitoring staff, and instrument technicians. For forward-fit plants, the hardware costs remained the same, but labor costs would be reduced because personnel would not be working in radiation zones. Therefore, a 50% reduction in labor requirements was estimated, i.e., 0.5 man-week/plant.
Based on the above estimates, the total labor required was 19 man-weeks/plant for backfit plants and 18.5 man-weeks/plant for forward-fit plants. Therefore, at $2,270/man-week, the estimated industry labor cost was $43,000/backfit-plant and $42,000/forward-fit plant. With a total of 71 backfit plants and 63 forward-fit plants, the total estimated industry implementation cost was $6.5M.
It was estimated that 28 man-hours/RY would be required to conduct periodic (monthly) testing of the pressure relief system. At $2,270/man-week, this cost was $1,589/RY. For the 134 plants, the total estimated cost for operation and maintenance was $6M. Using a 5% discount rate, the present worth of the recurring costs associated with plant maintenance and operation was $3.3M.
NRC Cost: It was estimated that 6 man-months would be required for the staff to develop acceptable methods, data, and acceptance criteria for licensees to use when evaluating the vulnerability of penetrations to the overpressure phenomenon analyzed in this issue. At $2,270/man-week, the total cost for this development was $54,000.
About 2 man-weeks/plant were estimated for reviewing and evaluating licensee calculations of the stresses within the penetrations and for reviewing the design, safety analyses, and QA documentation for the penetration pressure relief systems. At a cost of $4,500/plant, the total cost for this effort was estimated to be $600,000.
After implementation, the NRC would have to inspect the operation and maintenance of the penetration isolation systems; 1 man-hour/RY was estimated as sufficient for NRC review of each system. Therefore, the annual labor requirement was 0.18 man-week/RY for seven such systems. At $2,270/man-week, the total cost for the inspection of the 134 affected plants was $1.5M. Assuming a 5% discount rate, this cost was $880,000.
Total Cost: The total estimated industry and NRC cost associated with the possible solution to this issue was estimated to be $11.3M.
Based on an estimated public risk reduction of 40 man-rem and a cost of $11.3M for the possible solution, the value/impact score was given by:
(1) For backfit plants, an estimated 5 man-hours/penetration of labor in a radiation zone would be required and, assuming 7 penetrations/plant, the total ORE would be 35 man-hours/plant. The dose rate was assumed to be 25 millirem/hour, which was representative of the dose rate inside containment during reactor shutdowns. The implementation dose was therefore estimated to be about 0.9 man-rem/plant. For the 71 backfit plants, the ORE was greater than the total averted public dose.
(2) Routine testing and inspection of the penetration pressure relief systems were assumed to occur once every 30 days, similar to testing of the containment isolation valves. The testing was assumed to be performed by a 2-man team and to last about 10 minutes. Assuming 7 penetrations to be tested, the total operation and maintenance dose was estimated to be about 0.7 man-rem/RY.
(3) The public risk reduction estimated for this issue was overestimated for several reasons. First, no credit was taken for the protection from overpressure that would be provided if one of the isolation valves were not leak-tight. Investigation of a previous containment issue indicated that there was approximately a 30% chance that one of the valves would not be leak-tight.
(4) The costs were estimated assuming that the containment isolation systems were located away from the containment building wall. This is not the case for many of the isolation valves which are located adjacent to the containment walls. This assumption tended to minimize the costs because they would be clearly higher with modification of the containment structure to accommodate the pressure relief system. As a result, the cost estimates provided above were believed to be low.
The estimated public risk associated with overpressurization of containment penetrations was not significant. Based on the value/impact assessment and the staff's simplified engineering analysis, this issue was placed in the DROP category (See Appendix C). In an RES evaluation,1564 it was concluded that consideration of a 20-year license renewal period did not change the priority of the issue.