| As a contribution to develop a foundry simulation software, the algorithm and numerical results of different filling solvers are presented and compared in this thesis. The predicted flow pattern of PlicIf solver basically matches well with the actual flow in simple geometry castings. But in complex geometry problems, the liquid front face tends to follow the momentum of the air. As an improvement to the PlicIf solver, a droplet solver was developed. This is a Lagrangian solver. The metal liquid is modeled as numerous droplets. They are initialized with proper velocity and diameter at the inlet of the filling domain. The surface layer of elements on the inner side of the die wall are treated as bouncing elements. The normals for bouncing elements are obtained from the STL files. While traveling in the cavity, the droplets bounce between die walls. After travelling for a certain time interval, the droplet solver is halted and droplet data (mass and momentum) are mapped to the element data (vof and velocity). DFP (Divergence Free Projection) solver is then solved to get a better velocity field to help the droplets travel around corners. However, even with the updated velocity field from the DFP solver, the piling up of droplets at corners could not be avoided completely. As a result, after each step of filling, extra liquid in an element is mapped to their neighbouring elements to enforce incompressibility condition. A stress solver will be included later to deal with this problem more efficiently. |
