Detecting component failure potential using proportional hazard model

The proportional hazard model provides a framework for incorporating measurable parameters (covariates) into the description of failure rates. In this report a methodology is presented for calculating the failure probability based on the parameters measured at a routine surveillance test of the component. These parameters are expected to vary in an identifiable way before a failure indicating an impending failure, and are termed symptomatic emitters. A decision framework for a repair/replacement policy based on such information is presented. The use of symptomatic emitter information in determining a component failure potential has great utility, particularly in highly reliable systems like those in nuclear power plants. Further work on application of this methodology is recommended. If actual data are not readily available, simulated data can be developed using the simulation approach presented in this report. 4 refs