Creep-fatigue reliability assessment for high-temperature components fusing on-line monitoring data and physics-of-failure by engineering damage mechanics approach

Abstract

A new framework for creep-fatigue reliability assessment of high-temperature components is proposed in this paper, during which the monitoring data and physics-of-failure are integrated into the natural degradation process on the basis of engineering damage mechanics approach. Firstly, the surrogate model is constructed to establish the mapping from on-line monitoring data to elastic loads spectrum. Then, the plastic stress is corrected by Neuber rule, and thus the creep-fatigue life prediction models are combined with the linear damage summation rule to calculate the damage accumulations of components at dangerous positions in service. Secondly, uncertainty quantification is performed to reveal the probabilistic damage accumulations and modeled by stochastic process. Finally, the component-level reliability assessment of steam turbine rotor considering the correlation of damage accumulations at dangerous positions is explored by using the copula function. It shows that the combination of inverse Gaussian process model and Gumbel copula are appropriate to describe the damage accumulations and to analyze the dependencies at dangerous positions. Furthermore, the comparative research is also performed with respect to the independent assumption and hotspot method. The independent treatment tends to be conservative and hotspot method may be non-conservative in reliability assessment, which demonstrates the applicability of the proposed framework and is expected to be promoted in other high-temperature components.