Abstract
Bivariate log-normal distribution analyses were conducted on the specific output class of steam turbine components such as rotors, moving blades, nozzles, casings. The damage phenomena were erosion, creep deformation, fatigue cracking, fouling, scoring and so on excluding the repeated events after repairmant. Time and cycles to the occurrence of each event were regressed as log-normal type distribution function respectively but either contribution should not be ignored. Then bivariate log-normal type distribution functions were applied to the event data and proved to be almost successfully identified to represent the distribution characteristics. The quadratic exponent Q of the bivariate log-normal distributions was introduced to show the equi-probability contours of time and cycles and used to evaluate equi-risk functions where the consequence was attributed to the cost of repairment. There after total bivariate risk functions was established and the total optimization of repair cost regarding to time and cycles pattern of steam turbine plant. This tool was proved to be useful for realizing notional maintenance planning in actual components.
