The main task of the work was to conduct a numerical analysis of the influence of various types of Savonnius wind turbines geometries on the turbine performance. The initial design of the turbine was modified in three-ways: blade twisting, stator blades shielding and rounding of the end plates. In each of the three cases, simulations were carried out to identify the best performing model. In order to prepare the models and conduct the simulations the Ansys software was used. The work have been divided into 2 stages. The first stage was the preparation of the reference model. The reference geometry is a standard Savonius turbine with the overlapping rotor blades. The dimensions of the rotor blades and their overlap ratio have been selected for the optimal turbine efficiency with a tip speed ratio equal to 0.8. The second stage involved the implementation of the turbine modifications and compare the subsequent results with the reference model. This stage was divided into 3 sub-stages. Firstly, a blade twisting modification was carried out. The purpose of increasing the twisting angle of a rotor was to increase the area of a rotor blade and reduce aerodynamic drag in case of the returning blade, thus reducing the oscillatory character of the power produced by the Savonius wind turbine. Secondly, the stator blades were implemented into the turbine design. This type of solution was prepared in order to increase the air stream acting on the turbine blades without changing the turbine design. In this case, in order to check the wind direction dependence on turbine performance, for each model two different wind directions were analysed. The last sub-stage involved rounding the end plates of the turbine. The purpose of implementing such an idea was to narrow the section between the end plates, and thus increasing the speed of air flowing through the gap between the rotor arms. In this way, the resultant resistance force acting in the opposite direction to the turbine rotation should be reduced. The analysis showed that rotor geometry changes should be carried out with great care and it needs reliable numerical computations to verify the effect. Most of the proposed changes caused the turbine performance to decrease but few of them give promising results but it needs further analysis.