ABSTRACT
As one of the main causes of failure of turbine engine components, low-cycle fatigue (LCF) damage seriously affects structural performance and reliability. This chapter aims to develop an effective approach for the probability analysis of turbine blade LCF damage. It proposes combining the distributed collaborative concept, moving least squares, and the substructure method, the substructure-based distributed collaborative moving least squares (SDCMLS). The chapter reconstructs the probabilistic strain-life relationships considering the influence of confidence level, in which fatigue parameters are regarded as functions of a standard normal variable. MLS regression method improves the classical least squares method, reconstructing continuous functions through compact support concept, which enables establishing multi-variable non-linear surrogate model. With the involvement of multi-physical domains, multi-output responses and multi-components, the probabilistic analysis of gas turbine blades is a typical multi-level high-non-linear probability analysis problem. The probabilistic strain-life relationships are integrated with the proposed SDCMLS to promote the LCF life prediction in accordance with the distributed collaborative concept.
