Reliability modeling that combines Markov analysis and Weibull distributions

Abstract

When systems possess components with wearout failure characteristics or non-constant hazard rates and in standby redundancy configurations, the most common method currently utilized in industry for handling the reliability predictions of such systems is based on Monte Carlo Simulations. Monte Carlo Simulations are relatively easy to develop, but accuracy of the approximations that are produced dependents on the number of simulation trials selected. To obtain high accuracy for moderately complex system reliability models, Monte Carlo Simulation based system reliability models need to be run for large numbers of trials, in many cases greater than 10,000 trials, to achieve accuracy to the 4th or 5th decimal, which is sometimes required for Department of Defense (DoD) contracts in the aerospace industry. Markov Analysis is an alternate approach for modeling system reliability, which produces higher accuracy results than Monte Carlo Simulation based modeling, and requires fewer iterations.