Remove from My Interests
We'll discuss the multi-physics flow problems like Fluid-Structure Interaction (FSI) involve complex interaction physics and require solution of non-linear partial difference equations. Efficient numerical solvers are extremely useful tools for researchers to study the multi-physics interaction behavior. The advent of parallel algorithms and high performance computing have further revolutionized the field of computational engineering. It is therefore important to accelerate the legacy solvers using state of the art parallelization techniques. Currently, optimization of a discrete finite difference based Immersed boundary solver (IB) is undertaken to efficiently study the external or internal flow behavior around complex geometries at low Reynolds number. The performance enhancement is required in the computationally heaviest components of the solver, i.e., tagging of the intercepted cells and solving continuity and momentum equations. The computational efficiency is improved by utilizing OpenACC programming standards for parallel computing on Graphical Process Units (GPU) and using different iterative solvers for solving velocity-pressure correction equation.
Do Not Sell My Personal Information