A Meta-Analysis of Miner's Rule

Abstract

Since its introduction in 1945, Miner's Rule has enjoyed a ubiquitous role in providing remaining useful life (RUL) estimates despite its inaccuracy under various conditions. This paper examines the use of Miner's Rule, especially over the past 25 years, in order to explore potential pathways for improving its use. Special attention is given to approaches researchers have taken towards the damage limit value, model accuracy, and uncertainty characterization. This meta-analysis revealed that an alternative model which simply provides more accuracy than Miner's Rule will not retire Miner's Rule. Wide adoption of an alternative model requires balancing increased accuracy with model simplicity. The authors propose that such balance will best be achieved with a model that possesses nonlinear and probabilistic elements. They further hypothesize a probabilistic version of the Marco-Starkey model as a potential candidate. Recommendations for future work include efforts towards a common definition of failure, probabilistic characterizations of the damage limit, an extensive comparison of alternative models to identify candidate models offering optimal balance, and an increased partnership among academia, industry, and government in verification efforts of alternative models.