People and Projects

Microstructure simulation of metal castings Savithri Sivaraman,
with Roschen Sasikumar and Elizabeth Jacob

Background : Molten metals solidify with a dendritic (tree-like) morphology under most processing conditions. The alloying elements get segregated during solidification and the scale of these inhomogenities are decided by the scale of the dendrite structure. The grain size distribution in a material is an important microstructural parameter. In a casting, the grain size distribution depends on the cooling conditions as well as the distribution of nucleating agents in the melt. It is therefore of interest to study the development of the evolution of dendrites and growth of grains under different processing conditions.

Methods : The main governing equation is the diffusion equation (heat diffusion for thermal dendrites and solute diffusion for alloy dendrites), which is solved on a grid that is fine enough to show up the dendrite morphology. Curvature undercooling and surface tension driven phenomena are included in the simulation. For grain growth we have developed a model that employs a finite difference solution of the governing equations with a local front tracking method. Starting with the simple case of a pure metal solidifying from an undercooled melt, models have been developed for binary alloys solidifying under different conditions, directional solidification of alloys, structures developed during quenching from the mushy region, etc. The nucleation and growth of each grain is followed in this simulation by using cellular automaton techniques. Nuclei, which can potentially be activated at specified undercoolings, are distributed on a fine grid which represents the molten metal. The temperature is lowered and nuclei are activated. They are allowed to grow into the neighboring cells according to a growth law.

Results : Different stages of dendrite growth of a simulated Al-4.5%Cu alloy dendrite are shown in the adjacent figure. The solid is marked as red, liquid as green and the interface as blue. The columnar and random grain growth simulation using CA method is shown.