Incipient Transient Detection in Reactor Systems: Experimental and Theoretical Investigation

Tsoukalas, Lefteri H. ; Revankar, S.T. ; Wang, X ; Sattuluri, R.

Purdue University

关键词: Probability;    Training;    Testing;    Transients;    Progress Report;   

DOI  :  10.2172/850582
RP-ID  :  DOE/ID/14339
RP-ID  :  FG07-02ID14339
RP-ID  :  850582

美国|英语

来源: UNT Digital Library

【 摘 要 】

The main goal of this research was to develop a method for detecting reactor system transients at the earliest possible time through a comprehensive experimental, testing and benchmarking program. This approach holds strong promise for developing new diagnostic technologies that are non-intrusive, generic and highly portable across different systems. It will help in the design of new generation nuclear power reactors, which utilize passive safety systems with a reliable and non-intrusive multiphase flow diagnostic system to monitor the function of the passive safety systems. The main objective of this research was to develop an improved fuzzy logic based detection method based on a comprehensive experimental testing program to detect reactor transients at the earliest possible time, practically at their birth moment. A fuzzy logic and neural network based transient identification methodology and implemented in a computer code called PROTREN was considered in this research and was compared with SPRT (Sequentially Probability Ratio Testing) decision and Bayesian inference. The project involved experiment, theoretical modeling and a thermal-hydraulic code assessment. It involved graduate and undergraduate students participation providing them with exposure and training in advanced reactor concepts and safety systems. In this final report, main tasks performed during the project period are summarized and the selected results are presented. Detailed descriptions for the tasks and the results are presented in previous yearly reports (Revankar et al 2003 and Revankar et al 2004).