Abstract

Over the past decade, nuclear power plants have been transitioning from periodic to predictive maintenance strategies in order to reduce operational costs. This type of transition requires that new data be retrieved and different decision processes employed. Advanced monitoring and data analysis technologies are essential for supporting predictive strategies, as they can provide precise information on the health of a given component, track its degradation trends, and estimate its expected failure time. With such information, maintenance operations can be performed on a component shortly before its expected failure time. Classical reliability modeling is not suited for this dynamic context of maintenance operations, thus a new reliability mindset is required — one in which component health information (and not failure rates/probabilities) is propagated from the component to the system level in order to inform and optimize plant operations. The present paper addresses this situation by providing an alternative reliability modeling approach based on component margins rather than failure rates/probabilities. In this approach, component condition data and diagnostic/prognostic information can be directly employed to quantify component health in terms of margin. Lastly, we show how margin values can be propagated from the component to the system level in order to assess system health and identify those components most critical to system operation.

This content is only available via PDF.
You do not currently have access to this content.