Quantification of Input Uncertainty Propagation Through Models of Aluminum Alloy Direct Chill Casting

Insight into transport phenomena in complex solidification processes, such as direct chill (DC) casting, that cannot be found from experimental observation can be gained from numerical simulations. These predictions depend on material, process, and numerical parameters which contain inherit uncertainties due to experimental measurements or model assumptions. A fully transient numerical model of the direct chill casting process of Al-4.5 wt pct Cu was used to examine the propagation of input uncertainty to outputs of interest. The effect of microstructural model parameters, thermal boundary conditions, and material property input uncertainties were examined. Probability density functions were calculated based on these input uncertainties for metrics that characterize the ingot macrosegregation and sump depth. The macrosegregation-level predictions depend strongly on parameters that control the formation of the rigid mushy zone and shrinkage-driven flow. The heat release and transfer in the mushy zone are the dominant factors for determining the sump depth.