Abstract:
Historical experience has documented the limitations of traditional component removal, based upon "statistical safe life" estimates. With the aid of emergent diagnostic and prognostic techniques, the obvious choice to improve the accuracy of component retirement is through its condition monitoring and automated health management.
Benefits of this approach include: improved safety, heightened system readiness, and reduced life cycle costs resulting from better understanding of the timing of critical system failure modes. A critical step towards accomplishing some of these economic and logistic goals is through the development of efficient condition monitoring systems. For different process components from existing data one can get knowledge base of acceptable parameter responses over the operational regime. Replacing conservative threshold limits of measurement with component that are updated on regular intervals can produce much more accurate predictions of a' component's health and failure progression. This approach also benefits the maintenance decision process by improving fault classification/diagnosis. Predictions of the component future health status will provide the lead-time necessary to schedule maintenance proactively and avoid catastrophic failures.
