EVALUATION OF RELIABILITY . • USING RESPONSE SPECTRA FOR PREDICTIVE MAINTENANCE APPLICATIONS \

Abstract

Reliability evaluation using response spectra is developed through this work as an efficient technique of predictive maintenance applications. The objective is to quantitatively expect machine failure based on the spectral variations. Both approaches of spectral and reliability analyses have the same bases of statistical aspects. Hence, reliability analysis and spectral analysis can be interacted to achieve the developed technique. This work is concerned with the reliability evaluation of a rolling element using its response spectra based on investigation of the rolling contact fatigue phenomena. Rolling contact fatigue life is obtained in terms of spall size by applying crack propagation theory. Spectral analysis provides the peak value of response spectra in terms of spall size. Thus, the rolling contact fatigue life equation is modified as a function of the peak value of response spectra of the rolling element due to failure. Weibull spectral reliability function is then evaluated against the rolling element operating time from this response spectra to achieve its remaining life based on conditional reliability. An interactive computer program for Fault Diagnosis System (FDS) is developed to implement the developed technique in this work. FDS for windows is provided as a software to be used for failure detection and predictive maintenance of rolling element bearings. The spectral reliability evaluation has been implemented for detection of serious deterioration of a ball bearing installed on bullgear shaft of a centrifugal air compressor.