DIAGNOSTIC MODELING AND DIAGNOSABILITY EVALUATION OF MECHANICAL SYSTEMS

Consideration of diagnosability in product design promises to increase product quality by reducing maintenance time without increasing cost or decreasing reliability. Methods for investigating the diagnosabilit y of mechanical and electro-mechanical systems are described and are a pplied to the Bleed Air Control System (BACS) on the Boeing 747-400. T he BACS is described and a diagnostic model is developed using informa tion from the system Failure Modes and Effects Analysis. Emphasis is p laced on the relationships between the system's functions and its comp onents. Two metrics for the evaluation of system diagnosability and tw o metrics for the evaluation of component diagnosability are defined. These metrics emphasize diagnostic ambiguity and are combined with the probability of different system failures to weight the effects of eac h failure. Three modified systems are produced by reassigning function s from one component to another. The resulting effects on the system a nd component diagnosability are evaluated. We show that by changing th ese relationships system diagnosability can be improved without adding sensors or other components.