Digital sand core physics: Predicting physical properties of sand cores by simulations on digital microstructures

In this work, we compare measurements of strength, permeability and thermal conductivity of blown anorganically bound sand cores to simulation results on digitally generated unit cell models resolving the fine sand grain-binder aggregate on the micrometer scale. The quality and operationality of sand cores used in foundry applications is strongly affected by different characteristics of the sand and binder used, for instance chemical composition, porosity, grain size distributions and binder volume fraction. On the one hand, sand cores should be strong enough to survive the casting process. On the other hand, after casting extracting the sand should be easy. Currently, in industry, finding cost-efficient recipes for mixing sand and binder types requires tedious calibration and expert's knowledge. This originates from an incomplete understanding of the sensitivity of the physical properties of the resulting sand core w.r.t. the sand-binder characteristics. Furthermore, even with a controlled sand shooting process, the final sand core exhibits a heterogeneity, for instance in density, which needs to be accounted for during the design process. In a previous work, we have introduced a method to generate stochastic microstructure models based on X-ray micro-computed tomography scans and simulated compactification, which serves as the starting point for further simulation studies. In this work, we improve the previously introduced method towards better applicability to industrial problems: Firstly, we dispense with the time-consuming morphological dilation and use background CAD models instead. Secondly, we include a broader variety of real sand shapes by utilizing modern image processing tools. With this tool at hand, we simulate strength, permeability and thermal conductivity on these microstructures, and subsequently compare those to dedicated experiments, laying the foundation for “Digital Sand Core Physics”.