抄録
Probabilistic Risk Analyses (PRAs) are increasingly being used as a tool for supporting the acceptability of design, procurement, construction, operation, and maintenance activities at nuclear power plants. Since the issuance of Generic Letter 88-20 and subsequent Individual Plant Examinations (IPEs) and Individual Plant Examination of External Events (IPEEEs), the US Nuclear Regulatory Commission has issued several Regulatory Guides such as RG 1.174 and RG 1.177 to describe the use of PRAs in risk-informed regulatory activities. The PRA models developed for the IPEs were typically based on common cause models that are static relative to changing plant configurations. The IPE models contain common cause models that limit the ability to realistically assess current issues. For example, IPE modeling of common cause failures limit the ability to perform online maintenance during "at-power" operation. The limitations of such models result in skewed risk measures and minimal risk insight information. IPE common cause models related initiating events have resulted in non-conservative estimates of risk impacts when equipment is removed from service. The IPE common cause models do not explicitly address the differences between corrective and preventative maintenance. The intent of this paper is to introduce the concept of a dynamic common cause model. A dynamic common cause model must have sufficient fidelity and level of detail to properly capture the risk impact of changing plant configurations. Many current PRA applications, such as Maintenance Rule, require risk evaluations that involve the removing equipment from service. This paper summarizes the characteristics of a dynamic common cause model in terms of the required enhancements to the IPE model. Specific examples of PRA model enhancements will be presented to illustrate the level of detail required to properly address current issues. The examples are based on actual implementation of modeling upgrades to a nuclear power plant PRA. The paper will also discuss additional modeling enhancements required to address emerging risk issues.
