Investigation of parallel efficiency of an adaptive flow solver

AbstractParallel efficiency, in a domain decomposition based approach, strongly depends on a partitioning quality. For an adaptive simulation partitioning quality is lost due to the dynamic modification of the computational mesh. Maintaining high efficiency of parallelization requires rebalancing of the numerical load. This paper presents numerical experiment with adaptive and dynamically load balanced flow application. It is shown that through a relatively inexpensive process of repartitioning high parallel efficiency is maintained