An efficient global reliability analysis method for turbine disc’s multi-failure modes considering importance distribution technology

Abstract

Abstract This paper presents an improved efficient global reliability analysis (EGRA) method for a turbine disc considering the importance distribution of multi-failure modes, such as low cycle fatigue (LCF) and creep-fatigue (CF). Failure mode effects and criticality analysis (FMECA) technology is firstly used to analyse the criticality and occurrence probability of each failure mode. Importance factor S is then proposed to quantify the importance of the corresponding failure mode. In order to improve the reliability analysis’s efficiency, expected feasibility function (EFF) criterion in sequential sampling method is selected through numerical experiments to generate the samples of random variables. Secondly, EGRA method is established by combining sampling method with Kriging model, in which the performance function is fitted using the key sampling points near both the failure boundaries and high uncertainty region. On the basis of EFF criterion, the composite EFF criterion applied to EGRA method is developed. This method avoids repeated analysis on non-significant failure modes, which further improves the reliability analysis’s efficiency for the multi-failure modes system. Through a binary nonlinear example, it is proved that the prediction accuracy is 99.8% compared with Monte Carlo method. Moreover, the computational time of reliability is reduced by half compared with EFF criterion, which proves the prediction efficiency of composite EFF criterion. Finally, the reliability analysis on a turbine disc is conducted using the improved EGRA method combining with the failure mode importance distribution technology. Results show that the fatigue life at 99.9% reliability is decreased by 7.4% after considering the importance of failure modes, indicating that the ignoration of failure modes’ importance would lead to a more dangerous prediction result.