11 th International Conference on Nuclear Engineering Tokyo, JAPAN, April 20-23, 2003 ICONE11-36354 1 Copyright © 2003 by JSME DIESEL GENERATORS TESTING INTERVAL OPTIMIZATION AND RELIABILITY DATA UNCERTAINTY MINIMIZATION Robertas Alzbutas Lithuanian Energy Institute Breslaujos 3, LT-3035 Kaunas, Lithuania Phone: +370 37 401945 Fax: +370 37 351271 e-mail: robertas@mail.lei.lt Juozas Augutis Lithuanian Energy Institute Breslaujos 3, LT-3035 Kaunas, Lithuania Phone: +370 37 451349 Fax: +370 37 351271 e-mail: j.augutis@if.vdu.lt Eugenijus Uspuras Lithuanian Energy Institute Breslaujos 3, LT-3035 Kaunas, Lithuania Phone: +370 37 401926, Fax: +370 37 351271 e-mail: egeniju@isag.lei.lt Keywords: Unavailability, Diesel Generators, Testing Interval Optimization, Reliability Data, Uncertainty Minimization. ABSTRACT The object of the considered work was to investigate the possibility to change and optimize the testing intervals in such way, that the safety level of Ignalina NPP would not be decreased. The investigation of Emergency Diesel Generators System (EDGS) at Ignalina Nuclear Power Plant is performed applying the reliability related measures. In order to estimate and minimize the reliability data uncertainty the Bayesian updating approach is also investigated. As the considered measure and failure data are mainly related to unavailability of EDGS, the unavailability model is considered and used for the testing interval optimization. The considered approach, models, and received results can be used in the framework of the living probabilistic safety analysis and reliability data updating in order to minimize data uncertainty. The analysis results showed that the interval of testing EDGS at the Ignalina NPP could be increased up to two month. Due to change of testing interval from one month up to two month the average unavailability is changing insignificantly and doesn’t exceed the acceptable limit of unavailability. 1. INTRODUCTION The Diesel Generators (DGs) are usually tested periodically in order to check their conditions. In view of their importance, the reliability of DGs should be very high. Since these DGs are standby equipment and operate only during demand or during surveillance tests, their demand failure probability should be very low and once they operate their operational unavailability should be also very low. However, the testing frequency is chosen mainly by engineering judgment, and according to general practices. Having the reliability data, the mathematical modeling can be used to support the