Reliability Modeling, Testing and Optimization for Multi-Component Systems under Stochastic Degradation Propagation

Abstract:

Engineering systems are configured with multiple heterogeneous components/parts to serve desired functionality; system functionality is an integration of the performance of all of the components. Throughout the lifetime, components may experience functionality degradation as a stochastic process due to the aging effect, external shocks and environmental factors such as temperature and humidity. In further, degradation may “propagate” from one component to its adjacent components, causing “dependent” degradation and accordingly, failures. Guaranteeing system reliability metrics at desired levels is one of the utmost concern of engineers alike. The key to address this is to develop analytical models to capture components’ degradation process (including degradation of individual component and the degradation propagation mechanism), design optimal testing plans to obtain components’ effective degradation data for model verification and accordingly, determine the optimal preventative maintenance policies. Specifically,

To derive analytical models for components’ (accordingly, the system’s) degradation process, we will (i) develop physics-statistics-based models to capture individual component’s degeneration mechanism and process, (ii) systematically investigate the degradation propagation mechanism and (iii) develop bidirectional/multi-directional branching models to reflect the spatio-temporal interaction among multiple components’ degradation process. The degradation models need to be fitted into real degradation data for the purpose of validation and parameter estimation. Under most circumstances, degradation data cannot be obtained within desirable test duration if tested under working condition; this necessitates the optimal accelerated testing plan, which includes the choices of sample size, stress levels in the test, test duration, etc. Based on the above studies, the optimal preventative maintenance policies including the time-to-replace (repair) specific components and the components to be replaced (repaired) will be designed to guarantee the system functionality at desired level across time horizon.