Quantitative assessment of human contribution to risk in nuclear power plants

It is known from operation history of nuclear power plants (NPPs) that human factors considerations and human reliability analysis (HRA) are necessary for complete safety analysis of NPPs. HRA is usually performed to estimate the influence of human errors on the unavailability of various safety systems of NPPs;The objective of this work is to assess quantitatively the human contribution to the unavailability of those safety systems involved in the S(,2)C accident sequence. S(,2)C sequence is defined as a small loss of coolant accident (LOCA) and core meltdown after containment failure. This sequence is selected for this study because it makes a significant contribution to risk. In order to do that, gross operator error rates were estimated using past experience in the operation of pressurized water reactors (PWR) for the failure mode involving valves in those systems in the S(,2)C sequence. These systems are the containment spray injection system (CSIS), the containment spray recirculation system (CSRS), the containment heat removal system (CHRS) and the sodium hydroxide addition system (SHAS). The information was extracted from the licensee event reports (LERs). Appropriate statistical methods were used to estimate the error rates and the unavailabilities of those systems. A comparison was made with other estimates reported in other sources. Sources used were the Handbook of Human Reliability Analysis with Emphasis on NPP Applications (NUREG/CR-1278) and the Reactor Safety Study (WASH-1400). The estimated error rates from these sources were used to analyze the modified fault trees for the CSIS and CHRS systems. Fault tree analysis (FTA) is commonly used in reliability analysis and probabilistic risk assessment (PRA). The reliability analysis of the CSIS and CHRS was carried out by means of a fault tree methodology using the PREP and KITT-1 codes. It was found that the unavailability, Q, estimated using LERs for the two systems is comparable to that obtained using the other two sources of data