Resolution of Generic Safety Issues: Task I.A.4: Simulator Use and Development (Rev. 6) ( NUREG-0933, Main Report with Supplements 1–34 )
The objectives of this task were to: (1) establish and sustain a high level of realism in the training and retraining of operators, including dealing with complex transients involving multiple permutations and combinations of failures and errors; and (2) improve operators' diagnostic capability and general knowledge of nuclear power plant systems.
ITEM I.A.4.1: INITIAL SIMULATOR IMPROVEMENT
ITEM I.A.4.1(1): SHORT-TERM STUDY OF TRAINING SIMULATORS DESCRIPTION
This TMI Action Plan48 item called for a short-term study of training simulators to collect and develop corrections for identified weaknesses.
A study of training simulators was undertaken and NUREG/CR-1482299 was published in June 1980. Thus, this item was RESOLVED and no new requirements were established.
ITEM I.A.4.1(2): INTERIM CHANGES IN TRAINING SIMULATORS
This TMI Action Plan48 item called for the development of requirements to correct specific training simulator weaknesses, based on the short-term study results from Item I.A.4.1(1).
This item was RESOLVED with the issuance of Regulatory Guide 1.149439 in April 1981 and new requirements were established.
ITEM I.A.4.2: LONG-TERM TRAINING SIMULATOR UPGRADE
The four parts of this item were combined and evaluated together.
Nuclear power plant simulators were recognized as an important part of reactor operator training and this TMI Action Plan48 item called for a number of actions to improve simulators and their use.
There was significant interaction among the simulator-related action items and clear separation of this item was difficult. Item I.A.4.2 had a number of components dealing with long-term upgrades. The NUREG-066048 description called for research to: (1) improve the use of simulators in training operators; (2) develop guidance on the need for and nature of operator action during accidents; and (3) gather data on operator performance. Specific research items mentioned included simulator capabilities, safety-related operator action, and simulator experiments. The item also called for the upgrading of training simulator standards, specifically the updating of ANSI/ANS 3.5-1979. A regulatory guide endorsing this standard and giving the criteria for acceptability was also mentioned. The final portion of Item I.A.4.2 called for a review of simulators to ensure their conformance to the criteria.
At the time the issue was initially evaluated, a significant portion of the activities to be conducted had been completed. For example, ANSI/ANS 3.5 was revised and issued in 1981 and Regulatory Guide 1.149,439 which endorsed this standard, had been published along with numerous research reports. It was clear that the regulations, the ANS standard, and the regulatory guide did not require a site-specific simulator. 10 CFR 55 states that, if a simulator is used in training, it "shall accurately reproduce the operating characteristics of the facility involved and the arrangement of the instrumentation and controls of the simulator shall closely parallel that of the facility involved." ANSI/ANS 3.5-1981 called for a high degree of fidelity between the simulator and the "reference plant." However, there was no requirement that the reference plant be the same facility that the personnel in training would operate. Regulatory Guide 1.149439 explicitly made the distinction stating "the similarity that must exist between a simulator and the facility that the operators are being trained to operate is not addressed in the guide and should not be confused with the guidance provided that specifies the similarity that should exist between a simulator and its reference plant."
The work that had been completed for Item I.A.4.2(1) included the issuance of NUREG/CR-2353300 (Volumes I and II), NUREG/CR-1908,416 NUREG/CR-2598,417 NUREG/CR-2534,418 NUREG/CR-3092,419 and NUREG/CR-3123.653 This item, however, had long-range requirements calling for: (1) the review of operating experience to provide data on operator responses; and (2) the design and conduct of experiments to determine operator error rates under controlled conditions. Items I.A.4.2(2) and I.A.4.2(3) were completed with the issuance of Regulatory Guide 1.149.439 Item I.A.4.2(4) addressed the long-term training simulator improvement criteria which were established in Regulatory Guide 1.149439 and initiated in FY 1982. However, the staff review of submittals from simulator owners for conformance with the criteria was an ongoing task in 1983. Therefore, the outstanding portions of this issue (the continuation of simulator research and the review for conformance to acceptability criteria) were evaluated.
Use of simulators with high fidelity to the reference plant would significantly improve operator training in dealing with abnormal conditions thereby reducing operator error. Operators' performance under accident conditions was expected to be enhanced. Thus, a potential core-melt would be avoided and overall core-melt frequency reduced.
A possible solution was to establish a high level of realism in the training and retraining of plant operators by developing simulators with a high degree of fidelity to the reference plant.
The assessment of this issue was conducted by PNL staff64 with experience in reactor operator licensing, reactor operation, and general reactor safety, in consultation with General Physics Corporation. General Physics Corporation provided utility training services and had experience in reactor simulators, providing procurement and startup assistance, operation and maintenance services, and simulator modifications.
In the assessment of this issue, it was necessary to acknowledge that many of the TMI items associated with operator training were interrelated and that ranking problems surface when an attempt is made to assess these independently. For example, this item was related to Items I.A.2.6(1,2,3, and 5), which dealt with training improvements, including the enhanced use of existing simulators. I.A.4.1, dealt with initial simulator improvement, including short-term and interim changes in training simulators. However, the final safety ranking of this issue was relatively insensitive to changes in the basic assumptions used to distinguish these interrelated issues by the very nature of the ranking matrix. Therefore, it was possible to establish a priority ranking for this issue, despite the possible overlapping of potential benefits and costs with the other interrelated issues.
It was assumed that the major effect of these issues, both in terms of safety benefit and cost incurred, would be the enhancement of the level of realism imparted by simulators. The modeling capabilities given under Item I.A.4.1(2) and in ANSI/ANS 3.5-1981 reflected this feature.
It was assumed that, in order to provide the intended level of realism, site-specific simulators would be acquired. Such simulators would be significantly more realistic when compared to the specific facilities, both in layout and operation, than existing generic simulators. In addition, they were assumed to enhance transient and accident modeling capabilities.
It was clear that provision of site-specific simulators, while not explicitly required, would meet the requirements of Item I.A.4.1(2), the fidelity requirements of ANSI/ANS 3.5-1981, and the accurate reproduction requirements of 10 CFR 55. Less sweeping simulator enhancements might also fulfill these requirements but would have to be decided on a case-by-case basis. Therefore, it was assumed that the enhancement would be effected by the introduction of site-specific simulators.
The public risk reduction (and occupational dose reduction due to accident avoidance) were associated with the reduction in operator error expected to result from the training and requalification of operators on improved simulators. Inasmuch as any studies relating human error rates to the realism of simulator training were not available, this assessment was based primarily on PNL engineering judgment. Therefore, it was estimated that a reduction in operator error rate of 30% would result from the resolution of this issue. This estimate implied that, for specific instances, the improvement could be much greater, but the 30% reduction was used as an estimate of the average improvement.
There were 90 PWRs and 44 BWRs affected by this issue with average remaining lives of 28.8 years and 27.4 years, respectively. The representative plants selected for analysis were Oconee-3 and Grand Gulf-1 for PWRs and BWRs, respectively. (It was assumed that the fractional risk and core-melt frequency reductions for Grand Gulf would be equivalent to those for a PWR which was calculated directly.)
All release categories were affected by the resolution of the issue. The calculated core-melt frequencies were 8.2 x 10-5/RY for PWRs and 3.7 x 10-5/RY for BWRs. The reduction in these frequencies, based on the 30% reduction estimated for operator error, was 1.3 x 10-5/RY for PWRs and 5.9 x 10-6/RY for BWRs.
The dose calculations were based on a reactor site population density of 340 people/square-mile and a typical midwest meteorology. The resulting total reduction in public risk was 150,000 man-rem.64
Industry Cost: The major effect of the resolution of these issues was assumed to be the acquisition and use of site-specific simulators. The cost of such an undertaking would be substantial. If improved modeling changes were possible on existing simulators, the cost to industry would be substantially smaller. However, this was not clear at the time of the evaluation and it was assumed that new simulators would be required. (The impact of this assumption could be weighed subsequently in the final safety priority ranking. The assumption could be reevaluated at that time for any appropriate modifications.)
Assuming that new simulators would be required, the principal implementation cost would be the purchase of the simulators and provision of the new training materials. The capital cost of a simulator was estimated to be $7M. The provision of training materials was estimated to be equivalent to a 7 man-year effort.
It was assumed that all reactors, both operating and planned, would be affected. However, not every reactor would require a simulator. Many reactor sites have two or more reactors located together. If these reactors were sufficiently similar, a single simulator could serve them. Examining the list of 134 operating and planned power reactors, it was estimated that 62 additional site-specific simulators would be adequate. This assumed that 20% of the potential simulators were not required because either a site-specific simulator already existed or the plant in question was an older facility with limited remaining life.
The cost of the 62 new simulators spread over 134 reactors yielded $3.2M/reactor in capital cost and 3.2 man-year/reactor to provide new training materials. The operation and maintenance of the new simulators was estimated to require 3 man-years/simulator. Again, sharing the expense for 62 simulators over 134 reactors yielded 1.4 man-years/reactor. Industry may also experience costs stemming from participation in simulator experiments and research; however, these costs would be small in comparison to the costs related to new simulators. Based on these assumptions, the total industry cost was obtained as follows:
|Thus, the total industry implementation cost was (134 plants)$(0.32 + 32)M/plant or $470M.|
(2) Operation and Maintenance
Therefore, the total industry cost was $(470 + 530)M or $1,000M.
NRC Cost: There was no cost for development of a solution since all work was essentially complete and a solution had been identified. The principal costs were the continuation of research and the conduct of the confirmatory reviews. No additional development costs were foreseen as ANSI/ANS 3.5 was being revised and necessitated a revision to Regulatory Guide 1.149.439
The continuing research was treated as an implementation cost. It was estimated to require one NRC man-year and $1M in contractor support. (This included the remaining costs associated with Item I.E.8.) The confirmatory reviews were also treated as an implementation cost and were estimated to require 4 man-weeks/ simulator, or a total of 248 man-weeks for the assumed 62 new simulators.
The operational review cost to the NRC was minimal. It was assumed that annually each simulator would be audited to ensure that reference plant updates had been adequately represented on the simulator. Such an annual review was estimated to require 2 man-weeks/simulator or 124 man-weeks/year for all 62 new simulators assumed. NRC costs were estimated as follows:
|Continuing Research:||1 man-year =||0.33 man-week|
|Initial Simulator Reviews:||248 man-weeks =||1.9 man-weeks|
Based on a total NRC manpower of 2.23 man-weeks/plant, the implementation cost was given by:
With contractor support estimated to be $1M, the total implementation cost was $(0.6783 + 1)M or $1.7M.
(2) Review of Operation and Maintenance
The cost for review of operation and maintenance for all affected plants was [(90 PWRs)(28.8 years) + (44 BWRs)(27.4 years)]($2,100/RY) or $8M.
Thus, the total implementation, operation, and maintenance cost was $(1.7 + 8)M or $9.7M.
Total Cost: The total industry and NRC cost associated with the possible solution was $(1,000 + 9.7)M or $1,010M.
Based on an estimated public risk reduction of 150,000 man-rem and a cost of $1,010M for a possible solution, the value/impact score was given by:
The estimated reduction in occupational dose was 820 man-rem, based on accident avoidance only, since there were no implementation or maintenance dose reductions associated with resolution.
Based on the estimated risk reduction of 150,000 man-rem and the value/impact score of approximately 150 man-rem/$M, this issue was given a high priority ranking (see Appendix C). In view of the large estimated risk reduction, this ranking was essentially unaffected by any reasonable uncertainties in the cost estimates.
ITEM I.A.4.2(1): RESEARCH ON TRAINING SIMULATORS
This item was evaluated in Item I.A.4.2 above and was given a high priority ranking (see Appendix C). In April 1987, the issue was RESOLVED with the publication of Revision 1 to Regulatory Guide 1.149439 and new requirements were established.1045
ITEM I.A.4.2(2): UPGRADE TRAINING SIMULATOR STANDARDS
This item was and RESOLVED with the issuance of Regulatory Guide 1.149439 in April 1981 and new requirements were established.
ITEM I.A.4.2(3): REGULATORY GUIDE ON TRAINING SIMULATORS
This item was RESOLVED with the issuance of Regulatory Guide 1.149439 in April 1981 and new requirements were established.
ITEM I.A.4.2(4): REVIEW SIMULATORS FOR CONFORMANCE TO CRITERIA
This item was evaluated in Item I.A.4.2 above and was given a high priority ranking (see Appendix C). Staff efforts in resolving the issue resulted in the publication of a rule and a simulation facility evaluation procedure.
When this item was originally identified in 1980, the staff's approach was to require a submittal from each licensee in compliance with a regulatory guide (which later was issued as Regulatory Guide 1.149439) and to conduct a review of each simulator; there was no simulator regulation in effect at that time. However, in 1983, Section 306 of the Nuclear Waste Policy Act (Public Law 97-425) directed the NRC, in part, to establish "requirements for operating tests at civilian nuclear power plant simulators." This Congressional mandate had the effect of superseding the original intent of Item I.A.4.2(4) and required the staff to develop regulations for simulators. As a result, the approach taken by the staff for the resolution of Item I.A.4.2(4) was modified to comply with the Congressional mandate. The work scope was changed to reflect the fact that licensees, under the proposed regulation, would be required to certify their plant-referenced simulators to the NRC, and that NRC would perform an audit only when a need was identified, or upon request. Only in the case of those few licensees (estimated to be six), which were expected to seek NRC approval for a simulation facility that did not include a plant-referenced simulator, would the staff be obligated to review simulator documentation.
The final rule was published1077 as 10 CFR 55.45 and states, in part: "The operating test will be administered in a plant walkthrough and in either (i) a simulation facility which the Commission has approved for use after application has been made by the facility licensee, or (ii) a simulation facility consisting solely of a plant-referenced simulator which has been certified to the Commission by the facility licensee." In support of these regulations, the staff initiated a program to develop a procedure for its evaluation of selected certified simulation facilities. This procedure was subjected to a pilot test prior to being issued in draft form for comment. As a result of comments received, the procedure was revised and issued in final form as NUREG-12581084 in December 1987. Thus, the item was RESOLVED and new requirements were established.1098
ITEM I.A.4.3: FEASIBILITY STUDY OF PROCUREMENT OF NRC TRAINING SIMULATOR
The description of this issue in NUREG-066048 was as follows:
"In addition to the increased use of industry simulators for training of NRC staff (notably, the work by OIE with the TVA training center simulators), a feasibility study of the lease or procurement of one or more simulators to be located in the NRC headquarters area will be performed. These simulators would be used in familiarizing the NRC staff with reactor operations, in assessing the effectiveness of operating and emergency procedures and in gathering data on operator performance. The study will include development of specifications, development of procurement and commissioning schedules, estimation of costs, and comparison with other methods of providing such training for NRC personnel."
The intent of this issue was to improve the NRC staff's familiarization with reactor operations. The study was an effort to establish the feasibility of procuring an NRC training simulator. The issue had no direct bearing on public risk reduction and, therefore, was considered to be a Licensing Issue.
Technical studies262,263,264 of the issue performed by BNL indicated that existing simulators had significant modeling limitations. It was established that the capability of existing simulators was not acceptable at any but near-normal operating conditions, and that the lack of technical capability during two-phase conditions was significant. These results had an adverse effect on the feasibility of a training simulator for the NRC staff. Thus, this Licensing Issue was resolved.
ITEM I.A.4.4: FEASIBILITY STUDY OF NRC ENGINEERING COMPUTER
The purpose48 of this study was to fully evaluate the potential value of and, if warranted, propose development of an engineering computer that realistically modeled PWR and BWR plant behavior for small-break LOCA and other non-LOCA accidents and transients that may call for operator actions. Final development of the proposed engineering computer would depend on a number of research efforts.
Risk assessment tasks (interim reliability evaluation program, or IREP, for example) to define accident sequences covering severe core damage would also provide the guidelines for the experimental and analytical research programs needed to improve the diagnostics and general knowledge of nuclear power plant systems. The programs would assist the development and testing of fast running computer codes used to predict realistic system behavior for these multiple accident studies. These codes would provide the basic models for use in the improved engineering computer as well as the capability for NRC audit of NSSS analyses. This issue had no direct effect on public risk reduction and, therefore, was considered a Licensing Issue.
Reports262,263 on the review of PWR and BWR simulators were completed by BNL while work on Plant Analyzers continued at BNL, INEL, and LASL. RES believed that Plant Analyzers (Engineering Computer) would be helpful in uncovering potential operational safety problems in LWRs, caused by operator errors or equipment malfunctions, which would lead to risk reductions through increased operator awareness, improved procedures, and equipment redundancy.
The Plant Analyzer is not a design tool but rather an aid to the NRC staff in performing an audit function in the licensing process. After the second year of research on the Engineering Computer (Nuclear Plant Analyzer), it was concluded that it was not feasible to develop a device that would be sufficiently accurate and function with sufficient speed (i.e., faster than real accident progression time) to give a plant operator information adequate to guide action he or she should take during an accident. It was found, however, that a Nuclear Plant Analyzer, which takes output from an NRC safety analysis code such as TRAC or RELAP and displays plant accident conditions in schematic form on a video screen, would considerably ease the burden of understanding the results of complex safety analysis calculations. The Plant Analyzer also would allow the safety analyst to interpose simulated operator actions into an accident calculation underway. Based on these findings, the objectives of the development program were reoriented toward assistance for plant safety analysis and away from operator accident assistance.
A Management Plan968 for the Nuclear Plant Analyzer was prepared by the staff and included a listing of products expected to enter the regulatory arena in fiscal years 1985 through 1989. The staff concluded that it was not feasible to develop an Engineering Computer to provide input for operator actions during plant accidents; it was feasible to develop a device to give NRC an improved capability to audit NSSS analyses and this was being done in accordance with the Management Plan. Thus, this Licensing Issue was resolved.