Optimization of a diffuser augmented kinetic turbine with the island model

Abstract An augmented kinetic turbine with a sectioned diffuser parameterized with thirty degrees of freedom is optimized using computational flow field simulations combined with the island model for parallelizing the optimization. The geometry and mesh generation is performed with an in-house design framework. The parallel optimization and the necessary database for migration are implemented in Python. For all runs an evolutionary optimization algorithm is used. Three different groups of runs with different initializations and migration strategies are investigated. The study shows good speedup values for a diverse initialization in combination with the database migration. Moreover, for several runs, a superior speedup value is observed. Additionally, the database migration handles homogeneous initializations better than commonly used strategies.