Multi-fidelity shape optimization of hydraulic turbine runner blades using a multi-objective mesh adaptive direct search algorithm

A robust multi-fidelity design optimization methodology has been developed to integrate advantages of high- and low-fidelity analyses, aiming to help designers reach more efficient turbine runners within reasonable computational time and cost. An inexpensive low-fidelity inviscid flow solver handles most of the computational burden by providing data to the optimizer by evaluating objective functions and constraint values in the low-fidelity phase. An open-source derivative-free optimizer, NOMAD, explores the search space, using the multi-objective mesh adaptive direct search optimization algorithm. A versatile filtering algorithm is in charge of connecting low- and high-fidelity phases by selecting among all feasible solutions a few promising solutions which are transferred to the high-fidelity phase. In the high-fidelity phase, a viscous flow solver is used outside the optimization loop to accurately evaluate filtered candidates. High-fidelity analyses results are used to recalibrate the low-fidelity optimization problem. The developed methodology has demonstrated its ability to efficiently redesign a Francis turbine blade for new operating conditions.