CFD optimizations are characterized by long and highly variable objective function evaluation times. The aim of this study is to improve the parallel performance using a new load balancing master-worker parallelization strategy. First the new method is introduced and compared with the standard parallelization strategy. Finally two illustrative examples are provided.
Aizpurua, I., Lehmkuhl, O., Borrell, R., Perez, C., Oliva, A. A load balancing method for parallel genetic CFD optimizations. A: International Conference on Parallel Computational Fluid Dynamics. "ParCFD 2015: 27th International Conference on Parallel CFD". Montreal: 2015.