Multi-Objective Optimization of an Axial Flow Turbine Design Using Surrogate Modeling and Genetic Algorithm

Abstract

Abstract A surrogate model-based multi-objective design optimization methodology using a nondominated sorting genetic algorithm is used to maximize the power output and efficiency of an axial flow turbine. A set of high-fidelity data obtained from ANSYS-CFD simulations on space-filling samples is used for developing a surrogate model. The methodology uses (i) a Latin hypercube experimental design for selecting space-filling samples, (ii) a genetic algorithm for determining parameters of a Kriging model, and (iii) axial gap, tip clearance, and rotation angle of nozzle profile of turbine as predictor variables. Flow in the turbine is characterized by the presence of jet and wake flow, shock in the rotor blade flow passage, and tip clearance vortex. Study shows that (i) the axial gap controls the mixing of the jet and the wake flows and provides proper blade incidence which reduces the shock strength in the rotor blade flow passages and (ii) an optimum sized tip clearance vortex controls tip clearance leakage. The optimization provided a set of Pareto-optimal solutions that are nondominated in terms of power and efficiency. Verification of a selected design configuration from the Pareto-optimal solution using computational fluid dynamics (CFD) analysis showed that the process of optimization has been able to fine-tune the axial gap and the tip clearance.