A meta-model based cross-sectional shape of a Savonius hydrokinetic turbine for sustainable power generation in remote rural areas

Abstract

Energy accessibility and transition converge on exploring non-conventional renewable energy sources and the technology to harness them. An interesting and abundant resource is hydrokinetics. This work presents a Savonius cross-sectional blade shape modification that enhances the turbine performance in low-flow speed applications through a metamodel-based process. The blade profile is described by a Bézier curve control point as the parametrization strategy for generating a set of geometries to evaluate with COMSOL CFD. The obtained performance parameter of each geometry is defined as the output, and their control points parameters as inputs. This data set is utilized to train an Artificial Neural Network (ANN) to describe the interaction of blade shape and performance. The ANN is subsequently used as the target function in a Genetic Algorithm, to get the blade shape that best fits the model. A geometry with a power coefficient of 0.2405 results in an operational condition of 0.8 m/s flow speed at 1.1 Tip-Speed-Ratio. It means a performance increase of 8.3% compared with a standard turbine in the same conditions. This achievement leads to the implementation of this technology to supply the base load of rural households with a riverine resource of around 1 m/s flow speed.