Reliability prediction techniques.

Abstract

This thesis considers methods for predicting reliability and mean time to first failure of complex systems. The need for these methods, is a result of recent advances in pure materials technology and fabrication procedures, that potentially lead to more reliable systems. In this thesis, two types of complex systems are considered. Chapter I, treats systems, with k independent parameters, each exhibiting degradation failure. A model is developed for predicting, the reliability and mean life of such systems, from data taken early in life tests. Therefore saving many hours of test time; and eliminating the need of testing to failure. Chapter II, treats complex systems composed of independent systems, each exhibiting constant failure and repair rates. From the verbal description of a complex system, a Markov model is developed, from which predictions of reliability, availability and mean time to first failure, are made. This model also allows for the inclusion of a random waiting time between failure and beginning of repair.