Probabilistic modeling of uncertainties in fatigue reliability analysis of turbine bladed disks

Abstract

Turbine bladed disks normally operate under complex loadings coupling with uncertainties originate from multiple sources, including material variability, load variation and geometrical uncertainty. The influence of these uncertainties on mechanical response of engineering components are critical for their fatigue assessment and reliability evaluation. In this work, a general framework for fatigue reliability analysis is developed by coupling the Latin hypercube sampling with FE analysis to describe the combined effects of multi-source uncertainties. Fatigue reliability analysis of a full-scale bladed disk under multi-source uncertainties was performed as well as sensitivity analysis for fatigue design. In order to describe the manufacturing errors or tolerances, random dimensions are inputted. Comparing the predicted fatigue lifetime distributions with/without geometrical uncertainty, it shows that geometrical uncertainty matters in structural fatigue reliability. Particularly, sensitivity analysis indicates that the geometrical uncertainty exerts more critical influences on the fatigue lifetime and reliability of the turbine bladed disk than others. The sensitivity factors of three typical dimensions emerges the influence of designed sizes and dimensional tolerances on the failure probability, which provides a reference for engineering design.