Resolution of Generic Safety Issues: Issue 7: Failures Due to Flow-Induced Vibrations (Rev. 1) ( NUREG-0933, Main Report with Supplements 1–34 )
Flow-induced vibrations caused by vortex shedding resulting from rapid area change, buffeting due to random flow turbulence, fluid structures interaction instability, leakage excitation, steady operation of positive displacement pumps, and cavitating valves can cause failure of equipment, electrical wiring or components, pumps, valves, and piping systems. The three major failure mechanisms are high cycle fatigue, impact, and fretting (wear).
Vibration problems inside the reactor vessel manifest themselves as worn guide tubes, loose guide thimbles, cracked shrouds, and cracked nozzles and spargers. Charging pumps have been damaged by cavitation as well as turbulent buffeting vibrations that show up as cracked casings and welds. Vibrating valve internals (in closed and open positions) result in cracked and worn valve seats as well as cracked welds. Other failures that can result from vibration include loosened bolts, broken fittings, leaking snubbers, damaged pipe hangers, broken wires, thrown switches, loosened relays, damaged printed circuit boards, loosened instrument terminals, radiation monitor failures, false instrumentation activation, and open breakers.
The problem of excessive vibration is important because it can often lead to damage of multiple components. These events are frequently precursors to more serious events, inasmuch as continued occurrences can result in pipe cracks, failures of valves and snubbers, and damage to electrical and mechanical equipment. Other aspects of this problem include the effects of vibration (as well as water hammer) on engineered safety features following severe transients.
The report in NUREG-05724 was geared toward encouraging research in the area of flow-induced vibration and toward encouraging more attention to such phenomena when plants are still in the design stage. Given the present state of the art, it was doubtful that anything more than the practice of correcting vibration problems as they occur could be done. However, the ACRS noted that in 54 out of the 171 LERs (32 percent) studied no action was taken to prevent recurrence. Thus, a partial solution would be to encourage more positive measures in a greater percentage of vibration event followup actions.
The following five aspects of the problem were evaluated:
|(1)||Cracking and leaks caused by vibration can lead to small LOCAs.|
|(2)||Excessive vibration and minor equipment failures can cause transients. An increased frequency of transients implies some safety significance, since plant transients occasionally exercise some safety equipment.|
|(3)||Excessive vibration during operation of an engineered safety feature could cause failure of a safety system just when it is needed.|
|(4)||Vibration can damage nearby electrical and mechanical equipment.|
Of the five safety aspects listed above, only the first two were explicitly considered. The third aspect, which involves the possibility of vibration-induced damage to ESFs, has long been recognized. It is for this very reason that ECC systems are provided with keep-fill systems and high-point vents, BWR standby liquid control pumps have vibration-damping accumulators, etc. In addition, regular testing of these systems should uncover any remaining vibration problems.
The fourth aspect, which involves damage to nearby electrical or mechanical equipment, was judged to be insignificant compared to the first two aspects, since the redundancy in safety system design makes it extremely unlikely that a critical function will be lost. Moreover, spatial interactions such as these were investigated in Issue A-17, "Systems Interaction." Water hammer was addressed in Issue A-1, "Water Hammer."
The leaks caused by vibration have generally been very small, "weeping" through small cracks or leakage past stem packing. Such leaks are of little direct concern since the rate of coolant loss is well within the normal makeup capacity. Nevertheless, the potential for a larger leak exists. Because it is expected that large leaks will be far less likely than small leaks (particularly since most vibration-caused leaks occur in pipes of 1-inch diameter or less), it will be assumed that these LOCAs are of magnitude "S2" in WASH-140016 nomenclature.
The ACRS4 study found 171 vibration-initiated events in a 2-year period. Some (e.g., "piping cracks") were definitely leaks or leak related, but some of the ACRS listings (e.g., "welds") are ambiguous. Thus, a leak frequency cannot be obtained from the ACRS document alone.
Although not identified in the ACRS report, flow-induced vibrations can and have caused damage to reactor vessel internals and core components, including fuel pins and control rods. To a degree, the loose parts and vibration monitoring issues (B-60 and B-73), discussed in Section 2 of this report, address this aspect. Furthermore, the safety significance of these vibration-induced events should be reasonably approximated by this evaluation of vibration-induced LOCAs and transients.
A search of the NPRDS files produced 258 vibration-related events. (Not all events are reported to NPRDS.) Ninety-nine of the 258 involved water systems, 59 of which involved primary coolant, and 26 of the 59 resulted in a leak. None of the 26 leaks caused a reactor shutdown, however.
The ARCS study implied that flow-induced vibration events occur about once per reactor-year. The NPRDS search suggested that about one quarter of these events result in at least a minor primary leak. We will assume, based on judgment, that about 0.1 percent of the leaks will be serious enough to be considered an S2 LOCA. Thus, the estimated frequency of S2 LOCAs due to flow-induced vibrations is 2.5 x 10-4 event/RY.
To estimate the frequency of plant transients, we need an estimate of the fraction of vibration events which result in a reactor scram. Here, the ACRS table is more useful. Of the categories listed in NUREG-05724 Appendix D-IV, the events involving damaged valves, air systems, pump problems, and valve opening or closing are likely to cause a plant transient. These categories total 25 of the 171 events, or about 15 percent of the total. Therefore, since vibration events occur about once per reactor-year, the frequency of transients is estimated to be 0.15 event/RY.
An S2 LOCA or a transient can have a wide range of consequences, ranging from none to a major release, depending on the availability of various safety systems. Therefore, it is necessary to consider the entire probability-weighted spectrum. To do this, the S2 and T event sequences in WASH-140016 were re-normalized to the frequencies estimated above.
One modification is necessary. The only available fix, even if carried out with 100 percent efficiency, can only correct 32 percent of the problem by preventing recurrence. If we make the more realistic assumption that the fix is carried out with 95 percent efficiency, the F to be used is 30 percent of the estimated S2 and T frequencies. The results of the calculation are shown in Table 3.7-1. The total risk reduction for 43 PWRs and 27 BWRs, over an average remaining plant life of 32 years, is 2,980 person-rem.
Industry Cost: Costs were difficult to estimate since vibration damage can affect a wide spectrum of components. The types of corrective action listed in NUREG-05724 were as follows: add vibration dampers, redesign, strengthen, modify flow system, stud, and preventive maintenance. All of these actions except the last one are plant modifications that involve considerable paperwork (safety review, etc.). For these actions, an average cost of $50,000 (one-half staff-year) was estimated. If one event/RY occurs during the remaining average plant life of 32 years for operating plants, and bearing in mind that the "fix" applies to 30 percent of these events, the average cost was estimated to be $480,000/plant.
NRC Cost: NRC costs were estimated to be 6 staff-months of preparatory generic work plus 1 staff-week/plant to get letters out at a cost of about $150,000.
Total Cost: The total cost was estimated to be $[0.15 + (70)(0.48)]M or approximately $34M.
|Release Category||S2 Sequence F||T Sequence F||Consequences (person-rem)||FR|
|PWR-1||7.5 x 10-9||1.4 x 10-9||4.9 x 106||54.4 x 10-2|
|PWR-2||2.3 x 10-8||1.4 x 10-8||4.8 x 106||1.8 x 10-1|
|PWR-3||2.3 x 10-7||1.8 x 10-9||5.4 x 106||1.3 x 100|
|PWR-4||2.3 x 10-8||3.2 x 10-10||2.7 x 106||6.3 x 10-2|
|PWR-5||2.3 x 10-8||9.0 x 10-10||1.0 x 106||2.4 x 10-2|
|PWR-6||1.5 x 10-7||9.0 x 10-9||1.4 x 105||2.2 x 10-2|
|PWR-7||1.5 x 10-6||4.5 x 10-8||2.3 x 103||3.6 x 10-3|
|BWR-1||1.5 x 10-9||4.5 x 10-9||5.4 x 106||3.2 x 10-2|
|BWR-2||7.5 x 10-9||2.7 x 10-8||7.1 x 106||2.4 x 10-1|
|BWR-3||3.0 x 10-8||9.0 x 10-8||5.1 x 106||6.1 x 10-1|
|BWR-4||3.0 x 10-9||9.0 x 10-9||6.1 x 105||7.3 x 10-3|
Based on an estimated public risk reduction of 2,980 person-rem, the value/impact score is given by:
(1) Frequency estimates are uncertain up to a factor of 10. Although this was based primarily on judgment, it should be noted that if frequencies were more than a factor of 10 higher than estimated, some successfully mitigated vibration-induced S2 LOCAs should begin to show up in operating experience.
(2) It was assumed that the public dose estimates were uncertain to no more than a factor of 5. Similarly, the cost estimates were uncertain to a factor of 5, since it is unlikely that the average cost of vibration event follow-up will be less than 1 staff-month, nor is it likely to be over 2.5 staff-years. However, it should be noted that prevention of vibration-initiated plant transients will save some plant downtime. If credit for this averted cost were included, the net cost to the licensee would be less, and might well be zero or negative. Assuming a log normal distribution, if S = 9 x 101, the estimated range of S is 4 x 100 to 2 x 103.
Based on the value/impact score above, this issue fell in the medium priority category with small risk. As noted, the economic costs, if considered, would increase the priority ranking and should be an incentive without NRC action for the industry to take the actions necessary to prevent failures from flow-induced vibration. However, a generic solution to this issue, other than those discussed in Issue C-12, was not apparent and each specific flow-induced vibration event needs to be separately evaluated for safety significance and value/impact. This evaluation served to show that flow-induced vibration was not an issue associated with large risks or easily solved with relatively little expense. Therefore, the solution to each specific flow-induced vibration event should be pursued separately. Thus, this issue was DROPPED from further pursuit.