SPH simulation for 3D non-isothermal injection molding filling process using GPU acceleration

The nature of high viscosity and pressure in the injection molding process poses a great challenge for numerical simulation in terms of numerical stability, especially when using particle-based meshless methods. In the present work, 3D filling stage of injection molding is simulated using smoothed particle hydrodynamics (SPH) method. To counter the instability caused by high viscosity and pressure, various methods including a new non-penetration boundary treatment, modified low-dissipation Riemann solver, kernel gradient correction and particle shift technique are applied. GPU parallel computing is achieved by using Taichi language to boost computing efficiency. 3D non-isothermal injection molding process is performed for rectangular cavity, tensile test specimen and a customized transparent injection mold which we intend to perform visual injection experiment to verify our simulation in future work. The properties of flow field such as pressure and velocity are shown and compared with Moldflow simulation. The results of our simulation show good agreement with Moldflow.