Berlin 2008 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

MM: Fachverband Metall- und Materialphysik

MM 50: Materials Design I

MM 50.1: Vortrag

Freitag, 29. Februar 2008, 10:15–10:30, H 0111

An adaptive grid algorithm for solving multi-scale solidification processes on large domains — •Sebastian Waniorek, Michael Selzer, and Britta Nestler — Institute of Computational Engineering, Karlsruhe University of Applied Sciences

To gain insight into interdiffusion microstructures occurring between two contact alloys is critical for the design of new materials and surface coatings. In ternary and more generally multicomponent systems, the so-called Kirkendall effect is accompanied by a phase interface migration, both influencing the material properties. The simulation of interdendritic eutectic growth in one domain is, due to the multiple scales involved, a major challenge in computational engineering. We use a phase-field model coupled with multicomponent diffusion equations to simulate the outlined solidification processes. To numerically solve the phase-field equations with high accuracy, a small physical distance between computation points in the interface is needed. Thus a large computational domain is required to resolve the whole effect. These problems are hardly solvable on a uniform grid due to the vast memory and computation time requirements. However most approaches solve the mentioned equations this way. Since a computation with high resolution is only required in areas of steep gradients, we adapt the computation grid to the accuracy affordances. By this approach, the computation effort in the bulk solid phase and in liquid regions is minimized. We apply the method of adaptive mesh refinement to a finite difference solver and enable computations of multicomponent diffusion processes even on personal computers.