Resolution of Generic Safety Issues: Issue 35: Degradation of Internal Appurtenances in LWRs (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
This issue was identified493 after AEOD completed a study on internal appurtenances in LWRs. This study, AEOD/E101,47 was initiated because of the relatively high number of LERs that described events in which internal appurtenances (flow straighteners, orifices, diffusers, etc.) in the secondary system piping became loose or dislodged.368
Originally, the safety concern was that, if a steam line break were to occur in a PWR, any loose objects in the secondary piping could become missiles during the steam generator blowdown and rupture one or more steam generator tubes.429 A combined SGTR and MSLB was not a design basis accident. However, the issue was broadened to include loose objects in all LWRs, presumably in all locations. With such a broad definition, it automatically followed that there was a relatively large number of safety aspects. In general, a loose object causes problems either by causing impact damage or by blocking flow. In addition, the presence of a loose object of internal origin automatically implies that the system or component from which the loose object originated becomes degraded.
Concern with loose objects was by no means new since Issue B-60, "Loose Parts Monitoring System," extensively studied the occurrence and safety significance of loose objects within the primary system. Therefore, the evaluation of Issue 35 did not include the primary system. In addition, degradation of the ESFs was not considered since the general issue of ESF reliability was addressed in other issues, e.g., Issues A-45, B-4, and II.E.3.2. Moreover, since ESFs are not in operation during normal operation of a plant, degradation of internal appurtenances in service should not be a problem. Thus, the only concern that remained to be evaluated was the secondary system, the focus of the AEOD study.47
In addition to the MSLB/SGTR scenario described above, a loose or disengaged object in the secondary system can have additional safety significance such as:
(1) A loose object in the feedwater system can cause a loss-of-feedwater transient.
(2) A sufficiently massive object could cause a feedwater line break.
(3) A loose object in the main steam system could cause a transient. In many plants, plugging or isolating the main steam lines will also cause loss of main feedwater. A massive object could cause a steam line break.
(4) A loose object in a PWR steam generator could cause an SGTR.
(5) A loose object could prevent containment isolation valves from closing in the event of an accident.
(6) A loose object could conceivably cause a small LOCA in non-safety systems connected to the primary system (e.g., RWCU System). These are primary rather than secondary systems which were not covered in Issue B-60 and thus remained within the scope of Issue 35.
The systems considered were comprised primarily of piping rather than plenums and thus were not amenable to the loose parts detection program required by Issue B-60. About all that could be done was more frequent inspection and/or greater care in design and assembly, each where appropriate.
Each of the scenarios described under Safety Significance above was examined.
(1) Steam Line Break with SGTR
The frequency of a steam line break was estimated to be 10-3/RY, about 10% of which was expected to occur within containment (See Issue A-22). Assuming this frequency, the probability of a loose part being present at the time of the break was needed. AEOD studies352, 433 listed 12 such events as of June 15, 1982, which corresponded to 360 PWR-years. It was assumed that these reported events constituted 20% of the total events. In addition, it was assumed that a loose object would go unnoticed for 2 years, i.e., at least a complete reload cycle. The probability of a loose object being present during such a two-year period was estimated using the Poisson formula:
It was further assumed that, given an MSLB event and a loose part present somewhere in the steam generator feedwater lines, there was a 10% chance of the loose object rupturing a steam generator tube. Using these numbers, an event tree was constructed. Branches accounted for whether the break was inside or outside containment, whether the MSIV closed or not, whether feedwater to the affected steam generator was shut off, and whether the HPSI system operated of failed. The result was:
Core-melt/PWR-year = 8.7 x 10-8
Man-rem/PWR-year = 0.32
The dominant sequence was as follows:
|Steam line break occurs||10-3/PWR-year|
|Loose object is present||0.28|
|Steam generator tube(s) rupture||0.10|
|Break is inside containment||0.10|
|MSIV is closed||0.90|
|Feedwater to affected steam generator continues
(containment fails due to overpressure)
|HPSI successfully cools core, but noble gases and some iodine are released||0.98|
|Net probability||2.5 x 10-7/PWR-year|
|Release is ~ PWR-8||7.5 x 104 man-rem|
|Net risk for this sequence||0.20 man-rem/PWR-year|
(2) Loss of Feedwater Transients and Transients Induced by Loose Objects in the Steam Lines
AEOD studies352, 433 listed 17 events of this nature in a period covering about 600 RY. Thus, the frequency of reported events was 17/600 per reactor-year, or 2.8 x 10-2/RY. Again, there was probably a sizable number of loose objects that were not reported. However, not all loose objects cause transients and those that do probably were expected to be reported. Thus, the frequency of transients was estimated to be the same as the frequency of reported events, 2.8 x 10-2/RY. Such transients normally have no safety consequences. However, they can initiate an accident if other equipment fails. To estimate risk, the frequencies of the transient-initiated sequences in Tables V 3-14 and V 3-16 of WASH-140016 were scaled to match the frequency estimated above. The results are shown below in Table 3.35-1.
|PWR-1||8.1 x 10-10||5.4 x 106|
|PWR-2||8.1 x 10-9||4.8 x 106|
|PWR-3||1.1 x 10-9||5.4 x 106|
|PWR-4||1.9 x 10-10||2.7 x 106|
|PWR-5||5.4 x 10-10||1.0 x 106|
|PWR-6||5.4 x 10-9||1.5 x 105|
|PWR-7||2.7 x 10-8||2.3 x 103|
|BWR-1||2.7 x 10-9||5.4 x 106|
|BWR-2||1.6 x 10-8||7.1 x 106|
|BWR-3||5.4 x 10-8||5.1 x 106|
|BWR-4||5.4 x 10-9||6.1 x 105|
|Core-melt Frequency:||4.3 x 10-8/PWR-year|
|7.8 x 10-8/BWR-year|
|Public Risk:||0.051 man-rem/PWR-year|
(3) Feedwater Line Break and Steam Line Break
The frequency of line breaks due to loose objects was very small, since these lines are quite massive. Nevertheless, steam flows through steam lines at roughly 300 mph and a loose object traveling with the flow would have considerable impact when it encounters a 90 bend. The result, if there is damage, would likely be a hole punched in the piping rather than a complete circumferential break.
It was assumed that the frequency of line breaks was 1% of the frequency of transients estimated above. This resulted in an estimated break frequency of 2.8 x 10-4/RY. Although this number was judgmental in nature, if the actual frequency were an order of magnitude higher, two steam line breaks would have occurred in the 630 RY accumulated at the time the issue was initially evaluated.
In PWRs, steam line breaks outside of containment are not particularly significant from the point of view of the reactor. The licensing analysis of such an event concentrates on the cooldown reactivity transient (and associated radiological consequences), assuming a complete circumferential break, end-of-cycle moderator temperature coefficients, and failure of the highest worth control rod to insert. It is most unlikely that a loose object would cause a complete circumferential break, regardless of the probabilities of the other assumptions of the licensing basis. It should also be noted that opening an ADV is equivalent to a steam line break passing 10% of rated steam generator flow. Feedwater line breaks outside containment are still more innocuous, since check valves will prevent blowdown of the steam generator.
Inside containment, breaking a steam line will dump the mass and energy content of a steam generator to the containment. If feedwater to this steam generator is not cut off, continued steam production could endanger the containment. (Feedwater line breaks are less troublesome, since breaking the feedwater line is guaranteed to shut off feedwater to the steam generator which is blowing down.)
This was exactly the safety concern of Issue A-22, "PWR Main Steam Line Break - Core, Reactor Vessel, and Containment Response." If the priority parameters calculated for Issue A-22 were scaled to the frequency of 2.8 x 10-5/PWR-year estimated for a break inside containment, the result would be:
|Public Risk:||0.00038 man-rem/PWR-year|
In BWRs, steam line breaks and feedwater line breaks are small LOCAs. In view of the presence of two MSIVs in each steam line and two check valves in each feedwater line in a BWR, only breaks inside containment were considered. Again, the estimated frequency was 2.8 x 10-5/BWR-year, just as in the PWR case. In addition, it was assumed that the break was an "S1" LOCA (equivalent diameter of 2 to 6 inches). The following values were obtained by scaling Table V 3-16 of WASH-140016 to 2.8 x 10-5 "S1" event/BWR-year.
|Release Category||Frequency/RY||Consequences (man-rem)|
|BWR-1||9.3 x 10-10||5.4 x 106|
|BWR-2||8.4 x 10-9||7.1 x 106|
|BWR-3||1.9 x 10-8||5.1 x 106|
|BWR-4||1.9 x 10-9||6.1 x 105|
|Core-melt Frequency:||3 x 10-8/BWR-year|
|Public Risk:||0.16 man-rem/BWR-year|
This particular scenario was addressed in the steam generator tube integrity Issues A-3, A-4, A-5, 66, and 67 and was not considered further in this issue.
(5) Loss of Containment Isolation Capability
Loose objects can interfere with valve operation, particularly since valve seats are natural collection points for debris. Most valves in the secondary system are not safety-related; interference with these valves will at most cause a transient, as discussed earlier. The safety-related valves include steam safety valves and isolation valves. Of these, the steam safeties are not susceptible to damage by loose objects under normal circumstances, since there is no flow to carry objects into them nor are most loose objects likely to float upwards in steam. Even if a loose object were carried into a safety valve during a safety valve actuation, the overpressurization analysis assumed one failed valve. Thus, a loose object plugging one safety valve would not result in overpressurization of the secondary side.
Interference with isolation valves (preventing complete closure) is more plausible, since these valves are usually passing flow during normal operation. AEOD studies352, 433 listed only one event of this nature in a period of 600 RY. Since isolation valves are tested periodically and problems are reportable, it was unlikely that the actual number of events was significantly larger than the number of reported events. Thus, it was expected that the frequency of occurrence of an inoperable isolation valve due to a loose object would be on the order of 1.7 x 10-3/RY.
The longest interval between isolation valve tests is a full 18-month fuel cycle. A simple application of the Poisson formula produced a probability estimate of 2.5 x 10-3 for the failure of an isolation valve somewhere in the plant to close on demand. Failure of an isolation valve does not automatically mean that the containment fails to isolate, since there are double isolation valves on primary systems and even secondary system isolation valves are usually backed up by check valves, turbine stop valves, etc. However, the potential for a common mode failure is quite high for loose part events. It was assumed that failure of one isolation valve would result in failure of the containment to isolate.
The effect of a containment isolation failure of this nature was to change accident scenarios, which otherwise would have resulted in Release Category PWR-7 or PWR-9, into Release Categories PWR-5 or PWR-8, respectively. Using the WASH-140016 frequencies for PWR-7 and PWR-9, a multiplication by the estimated containment isolation failure probability gave the change in the PWR-5 and PWR-8 frequencies:
|PWR-7, -9 Frequency/Year||Containment Failure Probability||PWR-5, -8 Frequency/Year||PWR-5, -9 Consequences (man-rem)||Public Risk (man-rem/year)|
|4 x 10-5||2.5 x 10-3||1.0 x 10-7||1.0 x 106||0.100|
|4 x 10-4||2.5 x 10-3||1.0 x 10-6||7.5 x 104||0.075|
|Public Risk:||0.175 man-rem/PWR-year|
The corresponding calculation for BWRs was more difficult. The analog of the PWR-7 to PWR-5 case, which corresponded to a jammed isolation valve releasing core-melt activity which would otherwise be trapped within containment, did not exist for a BWR, since BWR core-melts were expected to cause containment overpressurization and failure anyway. The analog of the PWR-9 to PWR-8 case, which corresponded to a jammed isolation valve releasing contaminated containment atmosphere during and after a successfully mitigated LOCA, existed in theory, but no appropriate BWR release category had been calculated. The pragmatic assumption was made that the BWR analog of the PWR-9 to PWR-8 case will result in roughly the same radiological release.
|Public Risk:||0.075 man-rem/BWR-year|
(6) Small LOCA in System Connected to Primary
AEOD studies352, 433 listed one loose object in a system connected to a PWR primary loop. In 360 PWR-years, the frequency of reported loose object events was estimated to be 2.8 x 10-3 /PWR-year. If these represented 10% of the actual events, the actual event frequency was 2.8 x 10-2/PWR-year.
In BWRs, all systems are connected to the primary side and the definition of systems needed some modification. Here, it meant systems connected to the reactor, exclusive of the main steam, condensate and feedwater, and normally idle safety systems. No such loose objects had been reported. Therefore, it was assumed that the rate of occurrence would be the same as for PWRs: 2.8 x 10-2/RY.
Flow rates for such systems are low. Moreover, they are equipped with various types of leak detection coupled with automatic isolation. Therefore, it was very unlikely that a loose object would cause an unisolated leak. It was assumed that the probability of such a leak was on the order of 10-3, given the presence of a loose object. Thus, the overall frequency of the leak was estimated to be on the order of 2.8 x 10-5/RY. The size of such a LOCA would be in the "S2" class. Scaling WASH-140016 Tables V 3-14 and V 3-16 to this frequency, the results are shown below in Table 3.35-2.
|Release Category||Frequency/RY||Consequences (man-rem)|
|PWR-1||2.8 x 10-9||5.4 x 106|
|PWR-2||8.4 x 10-9||4.8 x 106|
|PWR-3||8.4 x 10-8||5.4 x 106|
|PWR-4||8.4 x 10-9||2.7 x 106|
|PWR-5||8.4 x 10-9||1.0 x 106|
|PWR-6||5.6 x 10-8||1.5 x 105|
|PWR-7||5.6 x 10-7||2.3 x 103|
|BWR-1||5.6 x 10-10||5.4 x 106|
|BWR-2||2.8 x 10-9||7.1 x 106|
|BWR-3||1.1 x 10-8||5.1 x 106|
|BWR-4||1.1 x 10-9||6.1 x 105|
|Core-melt Frequency:||7.3 x 10-7/PWR-year|
|1.5 x 10-8/BWR-year|
|Public Risk:||0.63 man-rem/PWR-year|
|The scenarios above added up to the following:|
|Core-melt Frequency:||8.6 x 10-7/PWR-year|
|1.2 x 10-7/BWR-year|
|Public Risk:||1.18 man-rem/PWR-year|
At the time of the initial evaluation of this issue in February 1984, there were 95 PWRs and 47 BWRs operating, planned, or under construction, with an estimated, aggregate remaining life of 3,400 PWR-years and 1,600 BWR-years, respectively. This allowed the following estimates to be made:
|Man-rem, Total||= 5,000|
|Core-melt/RY||= 6 x 10-7|
|Core-melt/Year||= 9 x 10-5|
Industry Cost: It was postulated that 10 man-weeks spent on inspections every refueling outage would be about 90% effective in revealing degraded appurtenances. If refuelings occurred every 18 months, this cost would be $13,300/RY. The cost of actual repair, replacement, or upgrading was not included since this would eventually have to be done anyway. It should also be noted that avoiding even one unnecessary reactor scram would pay for 20 years of inspections. Thus, there was some actual financial advantage to licensees. Based on a remaining lifetime of 5,000 RY for all reactors, the total industry cost for the solution to this issue was $66.5M.
NRC Cost: NRC costs were estimated to be about $500,000 (5 man-years). This estimate was higher than usual since, at the time of the initial evaluation of the issue, licensing effort did not emphasize BOP systems and opposition and delay were likely.
Total Cost: The total industry and NRC cost associated with the possible solution was estimated to be $(66.5 + 0.5)M or $70M.
Based on an estimated public risk reduction of 5,000 man-rem and a cost of $70M for a possible solution, the value/impact score was given by:
None of the scenarios described above were dominant; however, all followed much the same pattern. The estimated frequency of loose part occurrences was unlikely to be more than a factor of 10 too low or too high. The probability of a loose part causing an accident (purely judgmental estimates) might be uncertain by as much as a factor of 20 in some cases. The estimates of consequences should be within a factor of five. Finally, the estimates of cost could have been off by a factor of 10. If log normal distributions were assumed, the man-rem and core-melt figures would have been within a factor of about 60 and the priority score within a factor of 100.
Based on Appendix C, the core-melt/year estimate was barely in the medium priority range (because of the large number of plants affected) and all others factors were in the low priority range. Therefore, the issue was given a LOW priority ranking in February 1984. In NUREG/CR-5382,1563 it was concluded that consideration of a 20-year license renewal period could change the ranking of the issue to medium priority. 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 $14,285/man-rem, which did not affect the low priority ranking.