CFD optimization of a transonic compressor using multiobjective GA and metamodels

Paper in proceeding, 2012

This paper present an automatic design process where a non-deterministic, global search optimization is utilized to optimize a first stage rotor of a highly loaded transonic compressor. The first part is focused on finding the best suited metamodel that can be used to accelerate the design process. The second part presents the results of using the meta model within the design process for an industry relevant case. Using the radial basis functions as acceleration technique for the optimization was seen to be very successful. The meta model assisted optimization reduced the total design time from approximately 2 weeks to 3.5 days given that 8 designs could run in parallel on a cluster. The 3D optimization produced a pareto front where it was possible to select blades having either high efficiency or high stability.