Modeling Multiphase Flows with Incompressible SPH on GPU

Incompressible multiphase flows with complex interface geometries are involved in many industrial and environmental applications. In contrast to the widely-used Eulerian approaches, as a Lagrangian method, SPH handles the interface representation naturally without the need for computationally expensive interface tracking techniques. Incompressible SPH, as opposed to its more established and easily parallelizable counterpart, the Weakly Compressible SPH, enforces the incompressibility by solving a Pressure Poisson equation at each time step and is known for its accuracy. However, incompressible SPH has higher computational costs and is rarely applied in 3D problems. This project explores the accuracy and efficiency of incompressible SPH on GPU with second-order consistent discretization operators on multiphase phenomena revealed by benchmark numerical experiments such as Rayleigh Taylor Instability, 3D cubic droplet deformation, and 3D bubble rising.

Contributors: Lijing Yang, Milad Rakhsha, Wei Hu