Multi-objective optimization of a bidirectional impulse turbine

Real-life engineering problems have multiple objectives, which mostly are conflicting in nature, and these problems can be solved through multi-objective optimization (MOO) procedure. In the present problem, a high-fidelity computational fluid dynamics model coupled with multiple-surrogate assisted genetic algorithm based MOO has been solved for performance enhancement of a wave energy extracting axial impulse turbine. Response surface approximation, Kriging, neural network and a weighted-average surrogate (WAS) were used to generate population for the MOO procedure and Pareto optimal fronts (PoF) of the objectives were produced. The design variables were number of rotor blade and guide vanes and the objectives were minimization of pressure drop and maximization of shaft power of the turbine. It was found that a cross-validation error analysis is inevitable to find the degree of fitness of a surrogate. The WAS-produced PoF shows better performance as compared to that of the other surrogates. The surrogates based on minimum cross-validation errors produce slightly lesser performance than the WAS. The efficiency, which is a function of both the objectives, was relatively increased by 1/411% through the current investigation.