Application of Bayesian Methods in Reliability Data Analyses

Abstract

The development of the theory and application of Monte Carlo Markov Chain methods, vast improvements in computational capabilities and emerging software alternatives have made it possible for more frequent use of Bayesian methods in reliability applications. Bayesian methods, however, remain controversial in Reliability (and some other applications) because of the concern about where the needed prior distributions should come from. On the other hand, there are many applications where engineers have solid prior information on certain aspects of their reliability problems based on physics of failure or previous experience with the same failure mechanism. For example, engineers often have useful but imprecise knowledge about the effective activation energy in a temperature-accelerated life test or about the Weibull shape parameter in the analysis of fatigue failure data. In such applications, the use of Bayesian methods is compelling as it offers an appropriate compromise between assuming that such quantities are known and assuming that nothing is known. In this paper we compare the use of Bayesian methods with the traditional maximum likelihood methods for a group of examples including the analysis of field data with multiple censoring, accelerated life test data, and accelerated degradation test data. The development of the theory and application of Monte Carlo Markov Chain methods, vast improvements in computational capabilities and emerging software alternatives have made it possible for more frequent use of Bayesian methods in reliability applications. Bayesian methods, however, remain controversial in Reliability (and some other applications) because of the concern about where the needed prior distributions should come from. On the other hand, there are many applications where engineers have solid prior information on certain aspects of their reliability problems based on physics of failure or previous experience with the same failure mechanism. For example, engineers often have useful but imprecise knowledge about the effective activation energy in a temperature-accelerated life test or about the Weibull shape parameter in the analysis of fatigue failure data. In such applications, the use of Bayesian methods is compelling as it offers an appropriate compromise between assuming that such quantities are known and assuming that nothing is known. In this paper we compare the use of Bayesian methods with the traditional maximum likelihood methods for a group of examples including the analysis of field data with multiple censoring, accelerated life test data, and accelerated degradation test data.

Different Bayesian Methods for Updating the Fatigue Crack Size Distribution in a Tubular Joint

“…In reliability problems, informative prior distributions can be constructed using physical theory, computational analysis, and expert opinions (e.g. see [19] and [21]). In assessing probability distributions based on expert opinion, there are many potential biases that must be minimized [19].…”

The Application of Downhole Vibration Factor in Drilling Tool Reliability Big Data Analytics—A Review

“…Vibration combined with other downhole factors such as temperature, shock, and resonance have a damaging impact on drilling tool failure and lifetime [4,21,47,[66][67][68]. Table 2 lists research efforts on applying vibration factor in reliability modeling and predictions based on FRBD by statistical modeling, Bayesian method, data mining, Monte Carlo simulation, advanced computing, and analytics methods [17,69,70].…”

Reliability Meets Big Data: Opportunities and Challenges

“…Yan (2012) shows how to use a Bayesian approach to solve a complicated inverse problem to identify crack properties. Li and Meeker (2013) provide an introduction to the basic ideas of using Bayesian methods for reliability data analysis and illustrate the methods with four basic kinds of reliability data (single distribution analysis of field failure data, accelerated life testing, accelerated destructive degradation data analysis, and accelerated repeated measures degradation data analysis).…”