Development of a fatigue life prediction concept in the very high cycle fatigue range based on covariate microstructural features

Abstract

In the present work the dependence between the size and space distribution of defects relating to the material quality as well as the size and location of failure initiation defects in fatigue specimens correlating with their fatigue lives in the VHCF-range was investigated. The investigations were made for two reference materials, a nickel-based superalloy Nimonic 80A and a metastable austenitic stainless steel AISI 304 (1.4301) with a high deformation-induced martensite volume fraction. The effect of typical damage-relevant defects for the investigated materials was modeled by corresponding failure-relevant parameters. The stress concentration at crack initiating twin boundaries as well as regular grain boundaries in Nimonic 80A was quantified using a misorientation factor by Blochwitz et al. [1] and a developed crack initiation parameter. The effect of size and location of extrinsic defects in 1.4301 representing the type-II-materials was estimated by means of a stress intensity factor with consideration of the local stress at defects. The investigation of the distribution of failure-relevant parameters in the single specimens showed that crack initiation predominately takes place at defects with the maximum values of the defined parameter. Applying the findings of the fatigue test results generated in the framework of this project, the observed dependence between the failure-relevant parameters and corresponding number of cycles until failure or crack initiation was modeled.

The analysis and statistical modeling of the defined damage-relevant defects was carried out on the basis of metallographic investigations of the reference materials in the as-received condition. Using the extreme value statistics (EVS) the size and (if necessary) space distributions of the larger values of defined damage-relevant defects were modeled for metallographic samples. These models were used in order to evaluate the value of failure-relevant parameters in fatigue specimens and corresponding fatigue lives. The good agreement of experimental and modeling results as well as the likely application of the method on other alloys are discussed in the paper.