Washington, DC 20555-0001 NRC Job Code L1120
DISCLAIMERPortions of this document may be illegible in electronic image products. Images are produced from the best available original document.
AbstractThis report presents a limited assessment of the conservatism of the Accident Sequence Evaluation Program (ASEP) human reliability analysis (HRA) procedure described in NuREG/CR-4772. In particular, the ASEP post-accident, post-diagnosis, nominal HRA procedure is assessed within the context of an individual's performance of critical tasks on the simulator portion of requalification examinations administered to nuclear power plant operators. An assessment of the degree to which operator performance during simulator examinations is an accurate reflection of operator performance during actual accident conditions was outside the scope of work for this project; therefore, no direct inference can be made from this report about such performance. The data for this study are derived from simulator examination reports from the NRC requalification examination cycle. A total of 4071 critical tasks were identified, of which 45 had been failed. The ASEP procedure was used to estimate human error probability (HEP) values for critical tasks, and the HEP results were compared with the failure rates observed in the examinations. The ASEP procedure was applied by PNL operator license examiners who supplemented the limited information in the examination reports with expert judgment based upon their extensive simulator examination experience. ASEP analyses were performed for a sample of 162 critical tasks selected randomly from the 4071, and the results were used to characterize the entire population. ASEP analyses were also performed for all of the 45 failed critical tasks. Two tests were performed to assess the bias of the ASEP HEPs compared with the data from the requalification examinations. The first compared the average of the ASEP HEP values with the fraction of the population actually failed and it found a statistically significant factor of two bias on the average. The second test partitioned the critical tasks into subgroups based on the ASEP HEP values and compared the subgroup average ASEP HEP values with the observed subgroup failure rates. It found little or no bias for small ASEP HEP values, but a considerable bias for larger ASEP HEP values.