Abstract

In this article, authors have studied genetic algorithm-based optimization technique to optimize rotor profile for elliptic shaped Savonius-style wind turbine with an aim to maximize the coefficient of performance. Genetic algorithm has been used to optimize design variables having distinct values and discontinuous and nondifferentiable objective functions. Optimization procedure using genetic algorithm uses the following steps: initialization, assessment, assortment, crossover and lastly alteration. Once the genetic algorithm is initialized, then the evaluation process trails, where each parametric value is evaluated based on the fitness function stated as objective function. Then the GA operators i.e assortment, cross over and alteration are applied. At the end of GA operation procedure, a new set of values of design parameter is generated. This procedure is endlessly iterated until the convergence criteria is met. Then the optimized and non-optimized profiles are studied using numerical simulation. Initially a two-dimensional numerical model is developed and validated against experimental results. The two-dimensional analysis is conducted using k-ω shear stress transport model. Unsteady Reynold’s Averaged Navier Stoke’s equations have been solved to simulate the flow field of a Savonius-style rotor. This analysis has been executed using finite volume approach in Fluent 17.2 version. Grid independence study is performed to curtail the effect of grid size on the flow field portrayals. The optimization technique implemented on the Savonius-style wind turbine, generated design parameters that were able to yield a coefficient of performance value of 0.398. The coefficient of torque and coefficient of performance values are studied for both optimized and non-optimized profile as a function of tip speed ratio. Numerical simulation predicted a maximum gain of 41% for coefficient of performance at TSR = 1.0 over for optimized profile over the non-optimized profile.

This content is only available via PDF.
You do not currently have access to this content.