Inverse Uncertainty Quantification of a MOOSE-based Melt Pool Model for Additive Manufacturing

In this paper, we employ Bayesian inverse Uncertainty Quantification (UQ) to quantify the input parameter uncertainties in a MOOSE-based melt pool model for the additive manufacturing (AM) of nuclear fuels. Inverse UQ is the process to inversely quantify the input uncertainties while keeping models consistent with measurement data. The inverse UQ process takes into account uncertainties from models, codes and measurements while simultaneously characterizing the uncertain distributions in the input parameters, instead of only providing best-fit point estimates. We employ measurement data on melt pool sizes (lengths and depths) to quantify the uncertainties in several melt pool model parameters such as power absorption coefficient, emissivity and convection coefficient, as well as the specific heat, thermal conductivity, and viscosity of the melt pool liquid. The posterior uncertainties from inverse UQ can be used to replace ad-hoc expert judgment in future forward UQ and validation studies of the melt pool model. Melt pool simulations based on the posterior distributions produced results closer to the measurement data.