Reliability Prediction Methods: A Survey and Selection for Mechanical Design-For-Reliability Available to Purchase

There are many reliability prediction methods applicable to mechanical systems design. This paper specifically discusses the methods of probability or failure rate generation for reliability prediction. The paper consists of three parts. The first part is to survey the methods. We survey five most frequently used failure probability generation methods: statistical frequency and modeling method, similarity and comparative assessment method, physics based failure modeling, Monte Carlo simulation, and expert elicitation. We discuss the technical rationale and scientific foundation of each method and illustrate them with application examples. The second part is to evaluate and compare the methods. We identify the following attributes for evaluating and ranking the methods: the closeness of the method to the design; the validity and fidelity of the prediction results; the extensiveness of the analysis effort involved and data needs; the applicability of the methods at different product life cycle stages; and the limitations and cautions of using the prediction results to assist design-for-reliability. The third part is to establish a selection framework from applicable methods based on the ranking result of the second part, to assist practical use of the methods for mechanical design-for-reliability.