Component Analysis With Competing Risks: How Re-Use of Individual Components of an ESP Impacts Reliability

Abstract

Abstract A rigorous statistical study using a semi-parametric survival analysis approach was applied to an analysis of individual components, with competing risks, of an ESP system. Since its proposal in 1972, Cox's Proportional Hazards (CPH) semi-parametric model has become a primary tool for regression analysis of censored data in medical studies and in the aerospace industry. However, Cox's model assumes that observations are time-independent. This is not the case with ESP components as observations involving these, in the same installation, are naturally related and the multiple observations of an individual component through its re-use are also related and time-dependent. In this paper, we utilize some of recent extensions of standard Cox formulation to analyze components of ESP systems where multiple competing risks from individual components exist (intra-component correlations, intra-installation correlations and time-dependent explanatory variables). These model extensions utilize a more involved analysis undertaking and demand some careful preliminary data treatment. The work illustrates the concept of time-dependent competing risks with an example of an individual component, in our case a specific motor, from its first deployment and subsequent re-deployments in different ESP systems in a systematic and unbiased manner.