Prediction of distortions and pattern allowances in steel sand castings

Modeling the thermomechanical behavior of the bonded sands used for steel sand casting is of great importance for the prediction of distortions and pattern allowances. In this study, distortions created by mechanical interactions between the casting and sand mold are measured from two experimental setups and then predicted by finite element stress analyses. The casting geometries involve a hollow cylinder for the first experiment and U-shaped bracket for the second. The temporal evolutions of 1) the cylinder’s inner diameter and 2) the gap opening between the bracket legs are measured in situ utilizing LVDTs (Linear Variable Differential Transformers) connected to quartz rods. The considerable distortions measured during the cylinder and bracket experiments are mainly caused by core expansion and core restraint, respectively. For the simulations, a one-way temperature-displacement coupling is adopted, in which temperatures are predicted using commercial casting simulation software and then used as inputs for the finite element stress analyses. The steel is modeled as an elasto-visco-plastic material, whereas the Drucker Prager Cap model is employed for the bonded sand. It is found that sand dilation (i.e., the volumetric expansion of a granular media due to a shear force) must be considered for the cylinder experiments. Otherwise, the inner diameter expansion observed during solidification is far under-predicted. For the bracket, a crack plane must be included in the stress simulation model. If not, the outer mold restrains the bracket legs from being pushed outward and distortions are under-predicted. By matching the predicted displacements with the measurements, a constitutive dataset for bonded sands is developed, whose predictive capability is then demonstrated through a case study.