When dealing with design process of compressor blades, predominantly deterministic models are used for High Cycle Fatigue (HCF) investigations. The existing scatter in factors such as material inhomogeneity of the blade material and loading condition is accounted for by safety factors that often end up in conservative designs. An alternative way to account for these uncertainties is the application of probabilistic models. More information about the scatter in different sources together with probabilistic models can lead to a more robust design process.
In order to compute the stresses acting in a compressor blade, the Finite Element (FE) method is widely used as standard tool. This method may show mesh dependence. Therefore, mesh requirements always exist in FE computations.
In this work, a probabilistic HCF investigation is carried out for a transonic compressor rotor blade. The sensitivity of the volume based weakest-link probabilistic model (WL) due to different mesh properties of the blade is investigated. The goal is to provide advice for better finite element meshing of the blades based on linear type solid elements for the computation of stress history. The mesh types of the blade are the input parameters for the probabilistic HCF investigation.
A stress invariant based HCF local criterion, Sines, and a critical plane criterion, Findley, are used in weakest-link to describe the failure probability for the 12% Cr-steel material used for the compressor blade. The estimation of the weakest-link and the local HCF criteria material parameters are performed using HCF experimental data based on 2 million load cycles obtained for smooth and notched specimens.
The study shows that the choice of the mesh property through the thickness of the compressor blade has much more effect on the failure probability predictions compared to the in-plane mesh property of the blade.