Failure Forewarning in NPP Equipment NERI2000-109 Final Project Report
The objective of this project is forewarning of machine failures in critical equipment at next-generation nuclear power plants (NPP). Test data were provided by two collaborating institutions: Duke Engineering and Services (first project year), and the Pennsylvania State University (Applied Research Laboratory) during the second and third project years. New nonlinear methods were developed and applied successfully to extract forewarning trends from process-indicative, time-serial data for timely, condition-based maintenance. Anticipation of failures in critical equipment at next-generation NPP will improve the scheduling of maintenance activities to minimize safety concerns, unscheduled non-productive downtime, and collateral damage due to unexpected failures. This approach provides significant economic benefit, and is expected to improve public acceptance of nuclear power. The approach is a multi-tiered, model-independent, and data-driven analysis that uses ORNL's novel nonlinear method to extract forewarning of machine failures from appropriate data. The first tier of the analysis provides a robust choice for the process-indicative data. The second tier rejects data of inadequate quality. The third tier removes signal artifacts that would otherwise confound the analysis, while retaining the relevant nonlinear dynamics. The fourth tier converts the artifact-filtered time-serial data into a geometric representation, that is then transformed to a discrete distribution function (DF). This method allows for noisy, finite-length datasets. The fifth tier obtains dissimilarity measures (DM) between the nominal-state DF and subsequent test-state DFs. Forewarning of a machine failure is indicated by several successive occurrences of the DM above a threshold, or by a statistically significant trend in the DM. This paradigm yields robust nonlinear signatures of degradation and its progression, allowing earlier and more accurate detection of the machine failure.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- DE-AC05-00OR22725
- OSTI ID:
- 885692
- Report Number(s):
- ORNL/TM-2003/222; TRN: US0604079
- Country of Publication:
- United States
- Language:
- English
Similar Records
Annual Report for NERI Proposal No.2000-0109 on Forewarning of Failure in Critical Equipment at Next-Generation Nuclear Power Plants
Methods for improved forewarning of critical events across multiple data channels