Multi-Level Accelerated Sub-Iterations for Fluid-Structure Interaction

Abstract

Computational fluid-structure interaction is most commonly performed using a partitioned approach. For strongly coupled problems sub-iterations are required, increasing computational time as flow and structure have to be resolved multiple times every time step. Many sub-iteration techniques exist that improve robustness and convergence, although still flow and structure problems have to be solved a number of times every time step. In this paper we apply a multilevel acceleration technique, which is based on the presumed existing multigrid solver for the flow domain, to a two-dimensional strongly coupled laminar and turbulent problem and investigate the combination of multilevel acceleration with the Aitken underrelaxtion technique. It is found that the value for the under-relaxation parameter is not significantly different when performing sub-iterations purely on the coarse level or purely on the fine level. Therefore coarse and fine level sub-iterations are used alternately, where it is found that performing 3 or 4 coarse level sub-iterations followed by 1 fine level sub-iteration resulted in the highest gain in efficiency. Although the total number of sub-iterations increases slightly by 30%, the number of fine grid iterations can be decreased by as much as 65–70%.