Application of uncertainty quantification techniques to ductile damage predictions in the third Sandia Fracture Challenge

Abstract

This report documents predictions in response to the third Sandia Fracture Challenge (Kramer et al. in Int J fract, 2019) that employs 3D component built with additive manufacturing processes. These predictions employ a modeling and simulation framework that supports uncertainty quantification by rigorous statistical methods and fast-running surrogate models. This work introduces a novel ductile damage formulation supported by recent experimental observations. This report details a calibration process that updates prior material property distributions to posterior distribution by aligning distributions of computational experiments to measurements. Finally, we discuss in detail the assumptions (with rationales) and techniques employed to set-up the challenge analyses and to process these results. Blind predictions of the challenge geometry align with global measurements in the elastic and plastic regimes. There is less agreement after the onset of ductile failure, and this lack of agreement is discussed at the end of the report. Local measurements show additional variability not captured by the statistical processes. Results in this study support the use of existing continuum-scale material models for additively manufactured components.