Paper presented at the International Society of Logistics (SOLE) 1999 Symposium, Las Vegas, Nevada, August 30 – September 2, 1999. 1 Development of a Framework for Predicting Life of Mechanical Systems: Life Extension Analysis and Prognostics (LEAP) Frank L. Greitzer, Edward J. Stahlman, Thomas A. Ferryman, Bary W. Wilson, Lars J. Kangas, Daniel R. Sisk Pacific Northwest National Laboratory Richland, Washington 99352 Abstract—The focus of this paper is on health monitoring of complex mechanical systems for diagnostics, prognostics, and maintenance scheduling. We discuss challenges faced in developing and implementing a general methodology for predicting the remaining useful life of mechanical systems, and challenges to institutional and logistical processes for exploiting prognostics. Introduction Prognostics is the process of predicting the future state of a system. Prognostics systems comprise sensors, a data acquisition system, and microprocessor-based software to perform sensor fusion, analysis, and reporting/interpreting of results with little or no human intervention in real-time or near real-time. It offers the promise of minimizing failures (especially failures “in the field”), extending the time between maintenance overhauls, and reducing life-cycle costs. But prognostics is still in a research and development phase, and implementing prognostics is a monumental task on several levels—the technical challenges involving hardware and sensor technologies, the analytical challenges involving predictive methods, and the logistical challenges centering on how to make use of prognostic information. Advances in Data Collection Advancements in electronics, sensors, computer processing speed and memory, and commu- nications are enabling more reliable and less expensive field data collection to support diagnos- tics and prognostics. Some examples of such advancements are smart microsensors, ultrasonic sensors, acoustic emission sensors, smart memory cards, radio-frequency tags/multi-sensor modules, and cellular data links. With control microprocessors, these sensors and instrument packages may be fabricated within a size, cost, weight, and power requirement that will allow deployment directly onboard host equipment. Data Analysis Methods As hardware and sensor technologies make it more feasible to collect critically needed field data, interest has grown in improving analysis techniques. A hardware and software architecture for diagnostic and prognostic analysis of sensor data for a turbine engine application has a system-level architecture that is sufficiently general to apply to a variety of mechanical systems (Greitzer et al., 1999). Analysis proceeds through a series of stages or components, beginning with sensor validation and progressing through diagnostics and prognostics analyses. This