Original Paper(Vol.63 No.8 pp.608-613)

A Predictive Method of Optimum Repair Intervals and Repair Amounts for Steam Turbine Casings Based on the Statistical Analysis of Crack Length Data

Kazunari FUJIYAMA and Kazuhiro SAITO

Abstract:Statistical analyses of the field inspection database were conducted for steam turbine high pressure inner casings. The apparent reducing trends of crack amounts at nozzle box fit radius were exhibited as repair numbers increased. This trend may be caused by the blunting of the corner root radius due to the grind-off actions for cracks. The average and upper trends of maximum crack length with repeated inspection/repair numbers were expressed by Green function successfully, indicating initial increasing and subsequent decreasing trends. To estimate total crack length, log-normal distribution was applied to predict the individual crack length from the maximum crack length and the total crack length was obtained from the summation of distributed crack length over the total number of cracks. The average growth rates of maximum crack length against start-up cycles showed similar reducing trends as maximum crack length. Using the average crack growth rates of maximum crack length and total crack length as the function of the numbers of repeated repairs, optimum repair intervals were demonstrated as the combination of initial frequent repair followed by long intervals after 3rd or 4th repair. The expression of crack growth rate as the function of the number of repeated repairs could be also very useful for maintenance scheduling.

Key Words:Steam turbine, Casing, Crack length, Damage, Repair, Stress concentration, Creep-Fatigue, Statistical distribution