1 Department of Wind Energy, Technical University of Denmark2 Fluid Mechanics, Department of Wind Energy, Technical University of Denmark3 unknown
The aerodynamic performance of the MEXICO (Model EXperiments In Controlled cOnditions) rotor at five tunnel wind speeds is predicted by making use of BEM and CFD methods, respectively, using commercial MATLAB and CFD software. Due to the pressure differences on both sides of the blade, the tip-flow will produce secondary flow along the blade, consecutively resulting in decreases of torque. To overcome the above-mentioned issue, a variety of tip-correction models are developed, while most models overestimate the axial and tangential forces. To optimize accuracy, a new correction model summarized from CFD results by Shen is adopted in this paper. In order to accurately simulate the separation point and the separation area which is caused by the adverse pressure gradient, the CFD method using SST turbulence model is used to solve the three-dimensional Reynolds averaged equations. The first order upwind is used for the advection schemes, and the discrete equations are solved with simple algorithms. In addition, uniform velocity and static temperature are given as inlet boundary conditions, and static pressure is given as the circumferential outer boundary condition and the outlet boundary condition. The boundaries of fan-shaped both sides are defined as rotationally periodic connection, and the freeze rotor model is applied at the interface of the rotating and stationary domains, which means the relative position of rotating and stationary domains is fixed when calculating the flow field. Speed no-slip conditions are applied to solid walls such as blades. In this paper, two different meshing methods are used to generate a hexahedral grid for the rotating domain and a tetrahedral grid for stationary domain, between which comparison of the deviation of axial force on 60% blade cross section under the design condition (Vtun=15 m/s) leads to a clear decision of the better mesh method with less deviation. Taking the better mesh method into consideration, the final number of rotating domain grids is calculated according to verification of grid independence, with an amount of 2, 961, 385. The conclusion of this paper will be illustrated from the following points: first, the comparison of the calculated and the experimental angle of attack distribution along the span direction shows that the maximum relative errors of the attack angle calculated by BEM and CFD respectively are -0.402 and 0.099; it further illustrates that the experimental results are substantially between the results obtained by the two methods, and closer to the result of CFD at the blade tip. Meanwhile, the axial force on the blade increases with increasing radius, while the tangential force shows small change. All of the axial and tangential force in each section increases with increasing wind speed. Additionally, the maximum relative errors of axial force calculated by BEM and CFD respectively are -0.139 and -0.096. In a word, the experimental data are in good agreement with the results calculated by BEM and CFD, confirming the reliability of the MEXICO data. Second, the SST turbulence model can better capture the flow separation on the blade and has high aerodynamic performance prediction accuracy for a horizontal axis wind turbine in axial inflow conditions. Finally, the comparisons of the axial and tangential forces as well as the contrast of the angle of attack indicate that the prediction accuracy of BEM method is high when the blade is not in the stall condition. However, the airfoil characteristic becomes unstable in the stall condition, and the maximum relative error of tangential force calculated by BEM is -0.471. As a result, prediction accuracy of the BEM method needs to be further improved.
Nongye Gongcheng Xuebao, 2013, Vol 29, Issue 18, p. 67-74