Developing parametrically upscaled constitutive and crack nucleation models for the [formula omitted] Ti64 alloy

Abstract

This paper develops Parametrically Upscaled Constitutive Model (PUCM) and the Parametrically Upscaled Crack Nucleation Model (PUCNM) for a commercially used α/β-phase Ti64 alloy. These thermodynamically consistent macroscopic constitutive models bridge spatial scales through the explicit representation of Representative Aggregated Microstructural Parameters (RAMPs). The PUCNM is an indicator of the probability of crack nucleation in the local underlying microstructure. A symbolic regression-based machine learning method operates on data-sets generated by image-based micromechanical crystal plasticity simulations to derive constitutive coefficients as functions of the RAMPs. The PUCM/PUCNMs development (both calibration and validation) uses data from microstructural characterization as well as mechanical tests, including constant strain-rate, creep, and strain-controlled dwell tests for the Ti64 alloy. Parametric study is conducted with the experimentally-validated PUCM and PUCNM to investigate the effect of RAMPs on dwell fatigue crack nucleation life. Finally, the combined PUCM/PUCNM tool is used for examining the impact of microstructure on fatigue crack nucleation in an engine blade model under simulated operating conditions. The results from these examples clearly exhibit the promise of the PUCM/PUCNM models in predicting fatigue crack nucleation in the microstructure of real structural applications, as well as demonstrating the microstructural influence.