Resolution of Generic Safety Issues: Task HF8: Maintenance and Surveillance Program (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
The purpose of the Maintenance and Surveillance Program (MSP) effort is to provide direction for the NRC's efforts to assure effective nuclear power plant maintenance. The program will be based on the current NRC regulatory approach to maintenance and an evaluation of the effectiveness of current industry efforts in the maintenance area.
The NRC's current regulatory approach to nuclear power plant maintenance is concentrated on: (1) QA during design, construction, and operation for structures, systems and components important to safety (10 CFR 50, Appendix B); and (2) surveillance requirements to assure that the necessary availability and quality of such systems and components is maintained (10 CFR 50.36). Despite extensive surveillance testing requirements, the NRC's rules and regulations provide no clear programmatic treatment of maintenance. NRC additionally requires reporting of significant events (10 CFR 50.72), including personnel errors and procedural inadequacies which could prevent fulfillment of safety functions and exceeding of TS limits. The NRC does not stipulate maintenance requirements for systems which are not safety-related. Many challenges to safety systems originate from systems/components which are classified as not safety-related. The relationship between non-safety grade control systems and safety systems is being addressed in USI A-47.
The MSP is intended to integrate the NRC's efforts to assure effective nuclear power plant maintenance and to do so in a manner that is consistent with and responsive to the Commission's 1984 Policy and Planning Guidance.745 The program addresses the problems and issues which exist and proposes development of alternative NRC approaches to regulating nuclear utility maintenance activities consistent with the Policy and Planning Guidance. The scope of the program includes all aspects of maintenance required to carry out a systematic maintenance and surveillance program. It includes conventional maintenance and repair plus such things as surveillance and test activities, equipment isolation, post-maintenance testing, independent verification, maintenance management, administrative control, personnel selection and training, procedures, and technical documentation.
Since the TMI-2 accident in 1979, it has been evident that faulty maintenance practice is a principal contributing factor to operating abnormalities. Preliminary estimates indicate that, aside from design deficiencies, more than 35% of the abnormal nuclear power plant occurrences reported to Congress since 1975 may be directly attributed to maintenance error, with the trend towards a worsening maintenance situation as plants age.740 Reviews of operating experience show a high frequency of degraded system performance due to both the lack of maintenance (especially preventive maintenance) and improperly performed maintenance, including human error during repair and surveillance testing.740
The proposed solution to this issue is to implement a systematic maintenance program as addressed in the NRC's preliminary MSP with the following five objectives:
(1) To assure that needed maintenance is being accomplished, especially in counteracting system and equipment aging effects, by taking appropriate preventive and corrective action to minimize equipment failures.
(2) To reduce failures from improper maintenance to an acceptable level and to assure safety through effective maintenance management, personnel selection and training, procedures, administrative control, and design for maintainability.
(3) To assure proper integration of maintenance operations and other organizational interfaces for maintenance activities which can affect plant safety.
(4) To improve the effectiveness of nuclear power plant maintenance programs in reducing the number of challenges to safety systems (e.g., reactor scrams).
(5) To optimize surveillance requirements to assure equipment availability when required without excessive equipment out-of-service intervals for testing and to eliminate the unnecessary exposure for transient trips due to excessive test frequencies of logic and initiation systems.
The following paragraphs describe the background and approach to quantifying the base and adjusted cases for this issue. The background description relates to the subjects of aging and maintenance in an overall sense. The approach makes assumptions based on the background information. The subjects of aging and of overall effect of maintenance are considered below.
Aging: The effects of system and equipment aging is considered as part of the MSP because adequate maintenance and surveillance can counteract aging effects. NUREG/CR-249776 (p. 5-4) discusses the variation of significant precursors with plant age. This can be assumed to reflect general equipment deterioration and the subsequent impact on plant safety in general. Trends for a number of initiating events or demand failures were presented for data up through 1979. For PWRs, failure rate trends for long-term core cooling were given as constant and perhaps increasing. For BWRs, only the ADS demand failure showed an increasing failure trend, based on a small number of observed events. The emergency power system failure trend was given as constant, perhaps increasing. The general conclusion was that no clear variation in the number of significant events with plant age has been demonstrated.
It is further suggested that the operating time on the majority of the plant safety systems is very small. In many cases, the operating time is only that experienced during testing intervals. While aging effects cannot be ruled out at this time, the likelihood of their showing any significant role in safety systems is small. As a result, it is proposed that aging effects, if any, would best be modeled by failure rates increasing in the balance-of-plant. This would manifest itself as an increase in plant transients requiring shutdown.
Maintenance: A central aspect of the maintenance and surveillance program is the increased efficiency of maintenance operations and the assumed resultant reduction in errors committed during maintenance. It is believed that, if an integrated maintenance program were implemented, increases in preventive maintenance would reduce the need for corrective maintenance during plant operation. The MSP would provide a decrease in improper maintenance due to better training, procedures, human factors engineering, etc. The maintenance program is also seen as improving maintenance such that fewer transients will occur because of better maintained equipment.
In order to evaluate this issue, it is necessary to estimate and bound the likely magnitude of these effects and the degree to which current maintenance and surveillance approaches can deal with the program. Existing information is reviewed below.
Utility Maintenance Experience: The idea of this program is to increase the role of preventive maintenance and thus decrease corrective maintenance required for failures during operation. An examination of current experience indicates that corrective actions now represent the smaller fraction of recorded maintenance actions.
To characterize existing utility maintenance programs, NUREG/CR-3543741 (p. 20) indicates that between 64% and 80% of the age-related LER failures examined were detected by routine testing and surveillance performed in accordance with the plant TS or maintenance program. Detection after failure during plant operation could then be assumed to occur in 20% to 36% of the failures.
This indicates that the present unsystematic preventive maintenance programs at the majority of nuclear power plants is still detecting a substantial number of the events related to equipment degradation before actual failure in operation occurs. The 36% figure could be assumed to bound the category of failures during operation.
This result could also be expected to follow for a large part of the safety-related systems since the operating time on these systems is only during periodic test. Some examples of exceptions are the AFW systems in some designs and instrumentation channels for systems such as the reactor protection system and the electrical power systems.
An examination of transients in an EPRI study307 indicates that almost every transient category can also be considered as involving equipment failures, although these would be in the BOP. Although one could argue that preventive maintenance may not be as strict in this portion of the plant as opposed to safety systems, the utilities obviously have an economic incentive to maintain this portion of the plant as well. As a result, it is assumed that BOP failures and hence transients are subject to the same detection percentages mentioned above.
Base Case: The base case is the same as for the original Oconee and Grand Gulf risk assessments with the following exception. To model the effects of plant aging, it is proposed that the BOP transient frequencies be increased by 10%. This would reflect increased failures due to plant aging. This 10% value is felt to be an appropriate "trip level" beyond which the present surveillance programs would detect the increased failures. Thus T2, T3, and T23 frequencies of 3/RY, 4/RY, and 7/RY, respectively, are increased by 10% for the base case. The base case parameters are T2 and T3 for Oconee and T23 for Grand Gulf.
The resolution of USI A-47 (which is expected to be resolved prior to this issue) may reduce the base and adjusted case transient frequencies and also result in a decrease in the predicted reduction in transient frequency for parameters T2 and T3. However, it is anticipated that the resulting changes will not significantly impact the results and conclusion contained herein.
Adjusted Case: The adjusted case involves transients affected in the base case by reducing them because of the implementation of a systematic maintenance program. A comparison of U.S. and Japanese data on automatic scrams for 1981 and 1982 provide the basis for an adjusted case reduction in frequency of transients. U.S. data for 1981 and 1982 automatic scrams indicate a frequency of 5.3/RY whereas the comparable Japanese data indicate a frequency of 0.4/RY. After discussions with PNL researchers in the human factors maintenance area, it is assumed for this analysis that, if an integrated maintenance program were implemented in the U.S., the U.S. automatic scram frequency could be reduced to 2/RY. This factor of 2.65 reduction from the 1981 and 1982 U.S. average of 5.3/RY is assumed to be applied to the base case transient frequencies T2 and T3 for Oconee and T23 for Grand Gulf. Thus, applying the factor of 2.65 related to improved maintenance results in the adjusted case transient frequencies.
Also, for the adjusted case, it is proposed here that integrated maintenance and avoidance of errors can impact unscheduled maintenance outages during power operation, reducing the duration (t) and the outage frequency (f). The model that is used in this analysis to represent maintenance outages is expressed as the following equation for unavailability Q(TM) of systems due to test and maintenance where H1 and H2 are contributions from human performance, D1 and D2 are contributions for design, t is expressed in hours/act, and f is expressed in acts/month.
Q(TM) = [(H1 + D1)t)][(H2 + D2)f/720]
Factors H1 and D1 initially add to one as do factors H2 and D2. The model initially assumes H1 and D1 as 50% each and H2 and D2 as 25%/75% split, respectively. It is assumed for this analysis that improved maintenance from an implemented maintenance program results in a 10% improvement in human performance related to outage duration (t) and a 25% improvement in human performance related to outage frequency (f). Thus, a test and maintenance term of 0.0021 in the base case become 0.0019 in the adjusted case.
This issue affects all 134 BWRs and PWRs operating or under construction. For this analysis, Oconee 3 was selected as the representative PWR and Grand Gulf 1 was selected as the representative BWR.
The improvement affects all categories of PWR and BWR releases as defined in WASH-1400.16 The total whole body man-rem dose is obtained by using the CRAC Code64 assuming an average population density of 340 persons per square mile (which is the mean for U.S. domestic sites) from an exclusion area of a half-mile radius about the reactor out to a 50-mile radius about the reactor. A typical midwest plain meteorology is also assumed. Based upon these assumptions and the preceding discussions, the base case core-melt frequencies are 4.95 x 10-5/RY and 3.81 x 10-5/RY for PWRs and BWRs, respectively. With the maintenance improvements as described, the adjusted case core-melt frequencies become 3.08 x 10-5/RY and 2.08 x 10-5/RY for PWRs and BWRs, respectively.
The base case and adjusted case core-melt frequencies are distributed over the following release categories:
|Base Case (RY)-1||Adjusted Case (RY)-1|
|PWR-1 = 2.7 x 10-8||PWR-1 = 2.6 x 10-8|
|PWR-2 = 3.1 x 10-6||PWR-2 = 1.2 x 10-6|
|PWR-3 = 2.3 x 10-5||PWR-3 = 1.3 x 10-5|
|PWR-4 = 5.0 x 10-8||PWR-4 = 1.9 x 10-8|
|PWR-5 = 3.3 x 10-7||PWR-5 = 2.2 x 10-7|
|PWR-6 = 3.4 x 10-6||PWR-6 = 3.1 x 10-6|
|PWR-7 = 2.3 x 10-5||PWR-7 = 1.5 x 10-5|
|BWR-1 = 1.1 x 10-7||BWR-1 = 8.4 x 10-8|
|BWR-2 = 3.5 x 10-5||BWR-2 = 1.8 x 10-5|
|BWR-3 = 1.4 x 10-6||BWR-3 = 1.0 x 10-6|
|BWR-4 = 1.6 x 10-6||BWR-4 = 1.2 x 10-6|
The reduction in core-melt frequency per release category results in a per-plant reduction in public risk of 64 man-rem/RY and 123 man-rem/RY for PWRs and BWRs, respectively. Based upon an average remaining life of 28.8 years for the 90 PWRs and 27.4 years for 44 BWRs, the total best estimate public risk is reduced by 3.1 x 105 man-rem.
Industry Cost: The implementation of all tasks, identified in the Draft Maintenance Program Plan,740 involve principally costs associated with labor. Some of the labor intensive costs for each plant are:
|Task 2.2.4||Evaluation of Regulatory Alternative (Principally the establishment of the preventive maintenance program by the licensee)||=10 man-years|
|Task 2.3||Assess role of Safety System Monitoring(Principally to code and label piping, valves, etc.)||=1 man-year|
|Task 2.5.2||Plant Maintainability (Job performance aids and task analysis)||=2 man-years|
|Task 2.5.3.d||Upgrade Maintenance Procedures (3.5 man-years for rewriting maintenance procedures and 1.5 man-years for improving document control)||=5 man-years|
|Task 2.5.4||Maintenance Personnel Qualifications and Training (Qualification and Training Program)||=1.25 man-years|
Equipment and other labor costs are expected to cost $0.2M/plant. Thus, the total industry cost for implementation is given by:
(134 plants)[(19.25 man-years/plant)($100,000/man-year)] + $0.2M = $280M
Labor and engineering costs are estimated to be 3.7 man-years/RY. In addition, 16 days will be added to the annual plant outage time to permit additional maintenance. At an estimated cost of $300,000/day for replacement power costs, the additional maintenance costs will be $4.8M/RY. However, the added maintenance is expected to reduce the automatic scram frequency by 3.3 events/RY. Based on 1981 data, each scram results on the average in a two-day outage time. Hence, about 7 days replacement power costs are saved each plant-year, or (7 x $300,000) = $2.1M saved per plant-year. Thus, the annual cost for operation and maintenance is $(4.8 - 2.1)M/RY + (3.7 man-yrs/RY)($0.1M/man-yr) = $3.1M/RY.
Total industry cost for maintenance and operation
|=[(90 PWRs)(28.8 years) + (44 BWRs)(27.4 years)]($3.1M/RY)|
Total industry cost for the implementation and maintenance and operation is $(12,000 + 280)M = $12,280M.
NRC Cost: Estimated NRC costs for implementation are $1.2M and $10,000/RY for operation and maintenance review. Thus, total NRC costs are estimated to be $1.2M + ($10,000)[(90 x 28.8) + (44 x 27.4)] = $38M.
Based on a risk reduction of 3.1 x 105 man-rem, the value/impact score is given by:
The total industry cost benefit resulting from accident avoidance costs is calculated to be $100M. Occupational dose calculations predict a total industry implementation dose of 2.7 x 104 man-rem. This dose results principally from a 1% increase in occupational dose to provide better identification (labels, etc.) for piping, valves, and other control devices. The operation and maintenance dose is believed to be nil. It is felt that the increase in preventive maintenance requirements will be more than offset by improved maintenance training, maintainability, and less frequent unplanned maintenance. Some have estimated as much as a 50% dose reduction to maintenance personnel.
Although the value/impact score was low, the total potential risk reduction justified a high priority ranking. It was recommended that a value/impact analysis be accomplished before individual requirements were finalized to ensure that the most cost-beneficial solutions were are implemented. The objective of this issue was to initiate the NRC's efforts to assure effective nuclear power plant maintenance in accordance with the 1984 Policy and Planning Guidance which included a description of the problems and issues to be addressed and an evaluation of the alternative NRC approaches to regulate nuclear utility maintenance activities.
In 1985, a NRC Maintenance and Surveillance Program Plan was developed. Activities included a survey of ongoing maintenance practices and an evaluation of their effectiveness. These activities were completed in 1986 and documented in NUREG-1212.1013 Subsequently, the Commission approved a Policy Statement1116 in March 1988 and directed the staff to develop a proposed notice of rulemaking for Commission review in 1988. Thus, this issue was RESOLVED with the issuance of the Policy Statement and no new requirements were established.1117