| Multiphase flows associated with interfacial dynamics, steep jump in fluid properties and moving boundaries between different phases pose substantial computational challenges in terms of both modeling as well as computational cost. The present work uses an immersed boundary technique to model the interfacial dynamics, Lagrangian markers to track the moving phase boundaries, and a stationary Cartesian grid to solve all the flow governing equations. Time dependent triangulated surface meshes are employed to represent the time dependent interfaces shape and location. Based on the solution characteristics, multi-level, three-dimensional adaptive grid techniques are incorporated into the computational framework to help meet the resolution requirements. Furthermore, a conservative marker redistribution technique is developed to maintain a desired marker-spacing, and a connectivity preserving level contour-based reconstruction technique is devised to handle topological changes associated with the interfacial dynamics. The flow equations were solved using the projection method with a finite volume staggered grid formulation on adaptive grids. For phase change problems, accuracy of the mass transfer computation critically affects the overall computational outcome. Efforts have been made to address this via a sharp interface-based mass transfer mode combined with the immersed boundary method. The capabilities and accuracy of the individual components and overall computational system are tested with a range of computations including demonstration of improvements with conservative interface restructuring, reconstruction and its effect on immersed boundary solution accuracy, estimation of computational cost saving with adaptive grids, rising bubble coalescence, binary drop collision, and stationary bubble growth in a superheated liquid pool. The method has demonstrated its capability for handling high density ratio, O(1000), multiphase fluid dynamics. |
PDF Full Text Request