Uncertainty quantification in ICME workflows for fatigue critical computational modeling

Abstract

Fatigue critical applications of digital twins require an emphasis on computing the driving forces to form and grow fatigue cracks in materials. The concept of a digital twin is extended to incorporate microstructure-sensitive fatigue in the context of ICME. Uncertainty is ubiquitous in ICME workflows, and robust uncertainty quantification strategies can facilitate reliable prognosis support. This work addresses epistemic uncertainty from model form and parameters for microstructure-sensitive simulation of fatigue. Model form uncertainty is studied through statistical volume element sampling strategies. A strategy to reduce this uncertainty is applied and a second study of model parameter uncertainty stemming from macroscopic calibration of a Ti-6Al-4V constitutive model is discussed.