Crystal plasticity modeling and data-driven approach for fatigue life estimation of additively manufactured Ti-6Al-4V alloy

Abstract

The limited use of additively manufactured Ti-6Al-4V (AM Ti64) alloy in critical load–bearing applications stems from an incomplete understanding of its fatigue behavior, the underlying causes and mechanisms, and the absence of reliable predictive modeling. This study aims to bridge this gap by attempting to aid a microstructure–sensitive modeling with the number of cycles to failure. Low cycle fatigue (LCF) tests are performed to failure at room temperature with five different strain amplitudes, with cyclic softening noted in all tests. A crystal plasticity model is developed and used for analyzing the fatigue indicator parameters (FIPs). Synthetic microstructures that statistically resemble the experimentally observed microstructure obtained using Electron Backscatter Diffraction (EBSD), are used. Grain-averaged and Band-averaged Fatemi–Socie FIPs are employed to evaluate the likelihood of crack initiation. These FIPs are derived from the output of CPFE model and volume-averaged for each strain amplitude. Following the elastic–plastic shakedown, the highest 5% of volume-averaged FIPs are analyzed using a Gumbel extreme value distribution. A Bayesian inference approach is used to associate the Gumbel distribution’s characteristics of FIPs with fatigue life, demonstrating a strong correlation with the experimental data on fatigue life. This work shows that a consistent correlation between FIPs and the number of cycles to failure can be established, offering a predictive tool for fatigue life assessment.