An Efficient Method for the Determination of the Probability of Failure on the Basis of LCF Data: Application to Turbogenerator Design

Turbogenerator rotors and coil retaining rings are highly loaded components, under low cycle fatigue at every machine switch on and switch off transient. Their failure during service may lead to very serious consequences. The present paper utilizes several experimental data, concerning the static, cyclic and fatigue behaviour of the rotor material, in order to perform a probabilistic analysis. All the main methodological issues are tackled, consisting in the experimental data processing, in the choice of the random variables that are involved in the fatigue process, in the determination of their distributions, in the practical computation of the probability of failure and of the safety index in the machine life range. The here implemented numerical method has remarkable properties to be emphasized: low computational times, generality of application, and accuracy. In order to test the method for accuracy, the obtained results were validated by a comparison with those of a Monte Carlo simulation and with reference data reported in the current literature.