Experimental and Analytical Investigation of Cyclic Crack Initiation in Nickel Based Super Alloy With Stress Concentration Features

Abstract

Accurate prediction of cycles to crack initiation in critical turbine components is a major issue in turbomachinery design, especially in components with highly concentrated stress such as turbine blades with cooling holes. Several viscoplastic and lifing methods have been used successfully to predict shakedown and cycles to failure, however complicating factors still exist that produce challenges for traditional methods. Therefore newer methods utilizing constitutive modeling with consideration for isotropic and polytropic hardening have been developed to better capture evolution of cyclic behavior of the material. Presence of mean stress and stress concentration factors are some of the complications that can be better accounted for using constitutive models. The present paper evaluates experimental and theoretical life of specimen made from nickel based super alloy with high stress concentration features under cyclic conditions with mean stress. The specimen geometry and loading were designed to mimic trailing edge holes in an F class IGT turbine blade. Experiments were conducted at an elevated temperature at two peak stress values to determine sensitivity to applied load at operating temperature similar to engine. Cycles to crack initiation are analytically evaluated using the well-known Manson-Coffin method with Morrow mean stress correction and two distinct methods for strain range evaluation. First method is the traditional Ramberg Osgood shakedown that has been extensively used in the industry. Second method is constitutive Chaboche based model run with linearized FEM results. Constants for Chaboche model are determined from Ramberg-Osgood constants with a method that takes into account yield surface evolution and hardening constants, in addition to rate dependent stress relaxation factor that can be used to model dwell time effects. Methods to decrease computational time with constitutive model are discussed. Analytical results are compared with the experimental data, and advantages and disadvantages of both methods including computational times are discussed.