Uncertainty assessment in system performability analysis

Uncertainty quantification remains a critical area in the analysis and simulation of system reliability, availability and performance. Uncertainty could either be data based, or system-based and hence driven by the underlying stochastic nature of the process. In a previous paper, the authors had provided three algorithms to handling data uncertainty in system availability analysis. In this paper, these ideas are extended into the analysis of inherent system uncertainty for both availability and performance analysis. A new Monte Carlo method for analyzing the variability in both system performance and availability is presented. A new method for calculating maximum likelihood estimates for scenarios of components including multiple types of repairs and maintenance is presented. In addition, a new method for obtaining the distribution of the Weibull parameters under complex scenario conditions is presented. In this paper, these new algorithms are illustrated with representative case studies drawn from the power generation industry.