Hierarchical failure simulation for machinery prognostics

摘要

A "prognostics" capability is an essential element of a condition-based maintenance schemem. Prognostics involves monitoring to ascertain the current health of a machine in combination with predictive modeling, and implies the ability to confidently predict remaining life under a range of operating conditions so as to guarantee operation for a minimum time under extreme conditons. Development of efficient unmerical models of material damage accumulation and component faulure are key parts of a prognostics capability. Such models need not approach perfect fidelity, as errors can be detected by monitoring, but must be accurate enough to guarantee safe operation over some time window into the future. A hierarchical modeling approach is being developed to address this need, with the following features: (1) local material failure is modeled approximately using contimuum damage mechanics; (2) component failure is modeled through numerical solution of representative boundary value problems; and (3) system failure is modeled dynamically by representing individual components as discrete elements having associated damage states. Damage mechanics solutions for individual components are used to define characteristic "component damage modes," wich are used to construct low-order component models for use in system simulation. These simulations are used to dientify system failure modes and accompanying observables, as well as to assess remaining life.