The Risk Based Maintenance (RBM) procedure was applied to steam turbine casing as a typical example of components suffering from combined damage modes such as creep-fatigue cracking. Risk analysis was conducted through the field inspection database for cracks at the portion under creep-fatigue conditions. The primary stage of RBM is semi-quantitative risk analysis for risk prioritization of events using risk matrixes coupled with parts breakdown trees and event trees. The secondary stage is quantitative probabilistic risk assessment (PRA) to optimize maintenance intervals for the prioritized issues. The unreliability functions were expressed in two-dimensional log-normal type probability functions of operation time and start-up cycles. As the field data include censored or sustained ones, the precise correlation factors are not always obtained from the data sets. To estimate the most likely correlation factor, the Bayesian inference was introduced to the analysis of two-dimensional probability functions. The risk functions of operation periods were obtained using the assumed operation pattern, that is, the ratio of start-up cycles to operation time by substituting this relation into the two-dimensional probability distribution functions. Total expected cost function was defined as the sum of periodical repair cost rate and the total risk cost of subject event and component. The cost function has usually the optimum cost point and it can be used as the basis of the decision making of maintenance intervals.

This content is only available via PDF.
You do not currently have access to this content.