Reliability Prediction of Systems with Recurrent Failures and a Specified Set of Covariates

• Activity Number: 488
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Risk Analysis
• Author(s): Nasser Fard*+ and Alexandre Mendes
• Companies: Northeastern University and Northeastern University
• Keywords: Reliability ; Hazard Rate ; Proportional ; Covariates ; DOE

Abstract

A non-repairable system could be subject to a repeated number of one or more types of failures during its useful life. Systems' failed components are replaced to bring them to operating conditions. This study considers multiple non-terminal failures leading to a tolerance limit, at which a system is defined as non-operational. The proportional hazard model (PHM) which considers a baseline hazard function and linear combination of the risk factors is used. The predictive model is obtained by assessing the significance for a set of covariates, identifying whether they change the baseline survivor function. After assessing the proportionality and linearity assumptions, the method of partial maximum likelihood is used to estimate the regression coefficients, and then survival functions are estimated. An Experimental design is developed to test units with a fixed number of covariates for a specified time. Experimental results show that the PHM method with correction for lack of heterogeneity identifies the significance of covariates and interactions that are used for reliability assessment. A statistical software (SAS) is applied to perform the required calculations.