Evaluation of Improved Methods for Plant Data Validation in Automated Fatigue Monitoring

Abstract

This paper presents the results of an engineering review that examined the accuracy and effectiveness of specific enhancements to an automated data validation algorithm used as part of an on-line fatigue monitoring system. One of the major challenges in developing an effective fatigue monitoring system that processes plant computer data is the quality of the original data stream. In these systems, the goal is to minimize, and eventually eliminate, spurious conservative fatigue monitoring results caused by bad input data passed through the plant computer. Reaching this goal frees engineering users from having to correct the faulty data in the original input stream and reprocess it. The WESTEMS™ stress and fatigue monitoring program at the Beznau Nuclear Power Plant Units 1 and 2 was used to develop and test the enhancements to the data validation algorithm designed to reach this goal. The engineering review included both quantitative and qualitative assessments of data-related errors and warnings, as well as a detailed comparison of the automated data validation and correction algorithm against a parallel user-assisted (manual) process. Effectiveness of the method was quantified by comparing the overall fatigue accumulation rates resulting from the manual data checking process with those produced by the fully automated approach. The positive results of the review demonstrate the effectiveness of the enhanced automated approach presented.