Microstructure-sensitive notch root analysis for dwell fatigue in Ni-base superalloys

Abstract

Macroscopic viscoplastic constitutive models for γ– γ′ Ni-base superalloys typically do not contain an explicit dependence on the underlying microstructure. Microstructure-sensitive models are of interest in many applications since microstructure can vary in components, whether intentional or not. In such cases, the use of experiments from one microstructure condition to fit macroscopic models may be too limiting. The principal microstructure attributes that can significantly affect the cyclic stress–strain response of γ– γ′ Ni-base superalloys are the grain size and γ′ precipitate volume fraction and size distributions. An artificial neural network (ANN) is used to correlate the material parameters of a macroscale internal state variable cyclic viscoplasticity model with these microstructure attributes using a combination of limited experiments augmented by polycrystal plasticity calculations performed on other (virtual) microstructures within the range characterized experimentally. The trained model is applied to an example of a component fatigue notch root analysis with dwell periods at peak load to demonstrate the methodology and explore the potential impact of microstructure-sensitive constitutive models on life prediction for notched structures subjected to realistic load histories.