Skip to main content

Optimization and parametric analysis of a novel design of Savonius hydrokinetic turbine using artificial neural network

Research Abstract

This study focuses on enhancing the efficiency of vertical axis Savonius Hydrokinetic turbines designed for
marine applications, historically characterized by a power coefficient below 0.1. Prior efforts aimed at improving
rotor performance have primarily involved modifications to blade designs. In this article, a new approach is
introduced, incorporating twisted blades inspired by the Archimedes screw turbine. Utilizing a 3D incompressible
flow analysis based on the Navier-Stokes equation, this research explores and compares the turbine’s
effectiveness with varying screw pitches (0.5, 0.75, 1). The system of equations is solved numerically using
ANSYS 2020 R2 fluid fluent. The performance assessment involves contrasting each proposed rotor against a
pitchless semi-circle rotor. An innovative aspect of this work involves investigating the impact of asymmetry
using two different ratios (2:1 and 3:1). Specifically, the lower half of the optimal pitch screw remains constant,
while the upper half varies based on these ratios. To understand performance trends, the study employs visualizations
of pressure, velocity contours, and streamlines to grasp the flow field and its underlying principles.
Turbulent kinetic energy and eddy viscosity are also visualized. The results reveal an 18.25 % improvement in
performance with the proposed rotor featuring a pitch screw of 0.5. Notably, the asymmetric rotor with a 2:1
ratio demonstrates the highest performance. According to the ANN, the optimum pitch screw value is determined
to be 0.6, achieving a power coefficient of 0.1938. This investigation employs novel design modifications and
asymmetrical configurations, offering valuable insights into significantly enhancing the performance of Savonius
turbines for marine applications.

Research Authors
Shehab Osama, Hamdy Hassan, Mohamed Emam
Research Date
Research Journal
Applied Energy
Research Pages
124921
Research Publisher
Elsevier
Research Vol
378
Research Year
2025