Abstract
‘Damage informatics’ means here a total life-cycle management methodology including event scenario making, statistical damage analyses and risk-based maintenance decision-making. Event scenarios are expressed in terms of event trees for various components in steam turbines and gas turbines. The field damage data of specific machine types are analysed statistically and the probability of damage or failure can be expressed through the bivariate distribution function of total start-up cycles and operation time. For gas turbine nozzle cracking data, a damage evolution law for low cycle fatigue cracking is applied and the nozzle position dependence is clarified as the major cause of data dispersion. The risks are obtained by the product of probability and consequence of damage/failure and then shown to provide the basis of maintenance decision-making.
Acknowledgements
The author should like to acknowledge Mr. Yudai Iida and Mr. Takumi Yoshida, the post graduate students, who conducted a part of the data analysis.