Investigating the impact of genetic algorithm operators (selection, crossover, and mutation) on the simple genetic algorithm performance for blade shape optimization and defining the best suited to maximize the power

Abstract

Since a reduction in greenhouse gases is required, the use of renewable energy systems, which offer cleaner energy, needs to be advanced. The horizontal axis tidal turbine (HATT) is considered as a renewable energy extraction device. However, HATT has a high levelized cost of energy (LCOE), which could be reduced by blade shape optimization to improve the hydrodynamic efficiency. This will increase the extracted power and reduce the resultant thrust of the turbine. One promising method for blade shape optimization is the use of genetic algorithm (GA) with the implementation of blade element momentum theory (BEMT). Furthermore, the GA performance is influenced by the GA parameters including the GA operators as the main parameters. Additionally, each optimization problem has its own optimum GA parameters. The lack of studies on the impact of GA operators on the simple GA performance for blade shape optimization is the reason for this study. This study investigates the impact, where the chord and twist angle distributions are optimized to maximize the power. This is applied for fixed rotor parameters, operating conditions, and GA parameters. One study outcome is that random, steady-state and tournament selections are acceptable to use for blade shape optimization with the used crossover and mutation types except for swap mutation, while roulette wheel and stochastic universal selection are not recommended except for random mutation. At the same time, rank selection is recommended when combined with the used crossover and mutation types, except for swap mutation and some other combinations. Furthermore, the power improvement achieved among all the combinations ranged from 9.87 % to 21.14%. Finally, the best combination that achieved the best power is (rank selection, uniform crossover, and random mutation).