A Novel Immersed-Boundary Method for Multiple Moving and Interacting Bodies

Licentiate thesis, 2007

This thesis describes the development, implementation and validation of an implicit, second order accurate, finite-volume and instationary immersed-boundary method for simulating the detailed flow around multiple arbitrary moving and interacting bodies. The potential for flows including moving bodies or boundaries, such as multiphase flows still has to be fully realized. The knowledge of the small scale flows around immersed interacting bodies are poor and to fill this gap in knowledge detailed simulations are needed. The method is implemented in a segregated finite-volume solver with a staggered variable configuration. The body surfaces (immersed boundaries) are triangulated by employing the GNU Triangulated Surface (GTS) library. In the method the fluid velocity is constrained by an implicit immersed boundary condition, closing the equations governing the fluid flow. The immersed boundary condition mirrors the velocity field over the triangulated immersed boundary such that it exactly follows the immersed boundary and the resulting reversed fictitious velocity field inside the immersed boundary is excluded from the continuity equation to preserve the mass in the immersed boundary cells. The interactions between bodies are modeled by a proposed and validated novel triangle-triangle based collision force. The forces acting upon the bodies are determined by integrating the pressure and the viscous forces over body. The method has been validated by simulating the drag force on a stationary sphere up to Reynolds number 100. To validate the instationary part of the model one and two sedimenting spherical bodies have been simulated at low Reynolds numbers. The proposed collision force has been validated by simulating a head on collision of two spherical bodies immersed in a fluid. From the simulations it is shown that the model accurately simulates the detailed flow around arbitrary moving and interacting bodies at low Reynolds numbers. In the future the method will be implemented into a fully coupled flow solver with RANS and LES models such that flow with higher Reynolds numbers can be simulated.