Abstract:
The difficulty in choosing the right combination of time step and time increment in simulating a flow-driven rotating rotor is largely attributed to the wind speed dependent time constant which is defined as the elapsed time for the rotor to reach its stable rotational speed. Therefore, a combination of time step and time increment for one wind speed may not work for other wind speeds. A CFD sensitivity analysis of a Savonius rotor is conducted in this study in order to improve its simulation accuracy in predicting the rotor rotational speed by manipulating the time step for a wind speed range of 2–6 m/s. An optimum combination of time step and time increment for minimum error for first are obtained by CFD analysis. Based on a reference parameter of 6 m/s wind speed, other time step for 2, 3, 4 and 5 m/s wind speeds were calculated using the number of revolutions required to reach a steady state rotational speed. The resulted rotational speeds generated were then compared with the actual experiment in an open circuit wind tunnel. It is observed that the rotor only starts to rotate just after zero-time step and gradually accelerates until it reaches a constant rotational speed. Consequently, higher time step is required for low speeds and the converse is true for the high speeds. Lower time step can result in non-convergence solution. The result shows that the predicted RPM of Savonius rotor can be made close to the experimental data at the expense of higher computing time. Accuracy of predicted RPM is within 5% as compared to 20% without sensitivity analysis.
