Simulation of grain refinement of Al-8Si-0.2 Mg alloy inoculated with Al-Nb-B via an improved cellular automaton model

In this paper, an improved cellular automaton (CA) model with low grid anisotropy has been implemented using the zigzag capture rule and growth anisotropy reduction with diffusion method. The improved CA model can describe the evolution of the spherical growth, dendritic growth, and undercooling field, thus achieving a more accurate estimation of grain size than previous models. The model was used to simulate nucleation behavior and grain size of Al-8Si-0.2 Mg (wt.%) alloy inoculated with Al-Nb-B refiner, quantitatively revealing the factors that suppressed nucleation. Results show that when the inoculant particles were uniformly distributed, latent heat was the main factor restricting nucleation. Latent heat inhibited nucleation by reducing the available undercooling and terminating nucleation at the recalescence. When considering the agglomeration of particles, the effects of latent heat and solute suppressed nucleation (SSN) on nucleation inhibition accounted for 37.57 % and 58.58 %, respectively. Agglomeration caused the particle spacing to be smaller than that of a uniform distribution, and the SSN effect significantly increased as the separation distance decreased, resulting in a large portion of particles losing nucleation potency. In addition, it was found that the refinement by high cooling rate was attributed to not only providing more undercooling but also reducing SSN zone thickness.