Use of corrective action programs at nuclear plants for knowledge management

Abstract

Due to the uncertainty of many of the factors that influence the performance of the humans in nuclear power plant maintenance activities, we propose using Bayesian networks to model this kind of system. In this study, several models are built from the information contained in the Condition Reports from the Corrective Action Program at a nuclear power plant. This first study, using actual nuclear power plant data, includes a method for data processing and highlights some potential uses of Bayesian networks for improving organizational effectiveness in the nuclear power industry. The tool described in this paper is designed to provide a systematic approach to assist in managing an organization's knowledge base and support improvements in organizational performance. This paper describes the utilization of cause codes recorded in the Corrective Action Program for determining their effect on consequential events.