Uncertainty analysis of tenon structure fretting fatigue life

DOI:	10.11809/bqzbgcxb2023.09.025
Keywords:	Monte Carlo; beta distribution; bimodal normal distribution; failure probability; tenon structure
Abstract:	Considering the uncertainty analysis of fretting fatigue life of tenon structure, the modified Monte Carlo sampling method is proposed, which utilizes beta distribution sampling and bimodal normal distribution sampling methods to sample variables. Based on the theory of importance sampling, the failure probability of corresponding variable distribution characteristics is derived, and the corresponding failure probability evaluation processes are built. The effects of distribution, scaling, and translation parameters on the performance of failure probability estimation of the proposed method are investigated by using the nonlinear functional function. Results indicate that the distribution of samples in central and edge region are effectively balanced by beta distribution method. However, too small scaling parameter can only cover partial failure regions, affecting the evaluation of failure probability. There exists an optimal translational value for bimodal normal distribution sampling methods, since too small will make little change and too big leads to less distribution in failure zone with high probabilities. Compared with simple random sampling method under the same convergence criteria, one magnitude less sample size is achieved for bimodal normal distribution, and it also performs better than beta distribution sampling method. Based on this method, 105 times of Monte Carlo simulation is conducted on the fan blade tenon, whose failure probability of fretting fatigue life is calculated to be 0.046 4%. The global sensitivity index analysis shows that the tensile pressure has the most significant impact on the failure probability of fretting fatigue life of the tenon structure.
Published:	2023-09-28
PDF HTML