Report

Technologies for Sustainable Built Environments Centre Click to edit Master title style Dynamic Stall in Vertical Axis Wind Turbines Rosario Nobile | Dr Maria Vahdati | Dr Janet Barlow | Dr Anthony Mewburn-Crook In Figure 4, strong instability is observed for large angle of Mesh Overview Dynamic stall is an intrinsic phenomenon of Vertical Axis Wind Turbines (VAWTs) at low tip speed ratios (TSRs) and its nature can affect fatigue life and energy output of a wind turbine. A two-dimensional Vertical Axis Wind Turbine (VAWT) is explored. The analysis is conducted by using Computational Fluid Dynamics (CFD) tools. The attacks and low TSRs due to deep dynamic stall. In addition, The mesh, as shown in Figure 2, is mainly composed of the development of several peaks, especially for negative angle of attacks and low TSRs can be associated with the three sub-domains: one fixed sub-domain outside the rotor, one dynamic sub-domain around the blades of the development of upstream wakes that interact with the rotor and one fixed sub-domain for the remaining part of downstream blades. the rotor. Conclusions numerical results are compared with experimental data. Fixed Sub-domain The key conclusion of this numerical study is that a CFD tool will allow the visualisation of the flow aerodynamics involved during the operation of a VAWT that is not Dynamic Sub-domain Introduction The last few years have proved that Vertical Axis Wind Figure 3. An example of dynamic stall for the 2-D simulation at low TSR and different positions of the blades. Turbines (VAWTs) are more suitable for urban areas than Horizontal Axis Wind Turbines (HAWTs) 1, 2, 3. However, the aerodynamic analysis of a VAWT is very complicated than Recently, to study a full scale wind turbine in the wind tunnel is an infeasible task due to size limitations and costs involved. Therefore , a Computational Fluid Dynamics (CFD) Software, ANSYS 12.0, is selected for this study and code adopted is able to show dynamic stall that is typical found in VAWTs at low TSRs. Also, from this numerical Results analysis appears that in order to achieve a good agreement The numerical simulations obtained during the present selection of the turbulent method is fundamental. However, study are mainly compared with an experimental study it is strongly suggested to develop a more sophisticated 3-D In Figure 2, the local mesh around the blades of the rotor is carried out in 2010 6. The SST method shows a good model that is more realistic than 2-D. refined for accurate and efficient resolution of the boundary layer and wakes. agreement with the experimental data than the k- and k-ε conventional wind turbines and at low tip speed ratios (TSRs<5), VAWTs are subjected to a phenomenon called dynamic stall. This can really affect the fatigue life of a VAWT if it is not well understood. possible with ordinary wind tunnel tests. In general the CFD Fixed Sub-domain Figure 2. Mesh and sub-domains for the two-dimensional .VAWT Boundary conditions and turbulence method Symmetrical boundaries were used for the top and bottom methods. Figure 4 shows how the lift and the drag between numerical and experimental data , the right References coefficients, Cl and Cd, are affected by different angles of 1. S. Mertens, Wind energy in the built environment: concentrator effects of buildings. TU Delft, 2006, pp. 3-14. attack and TSRs. The curve shapes are in good agreement 2. S. Stankovic, N. Campbell, and A. Harries, Urban Wind Energy. Earthscan, 2009 with the experimental data, which is the red line on the right side. 3. C. J. Ferreira, G. van Bussel, and G. van Kuik, 2D CFD simulation of dynamic stall on a Vertical Axis Wind Turbine: verification and validation with PIV measurements, presented at the 45th AIAA Aerospace Sciences Meeting and Exhibit, 2007, pp. 1-11. in order to reduce time and memory costs only a 2-D case is explored. parts of the 2-D model with no-slip boundary conditions at the two sides. An opening boundary was chosen for the Method output and a constant wind was defined for the inlet.The three Reynolds-Averaged Navier-Stokes (RANS) turbulence 5. J. Larsen, S. Nielsen, and S. Krenk, Dynamic stall model for wind turbine airfoils, Journal of Fluids and Structures, vol. 23, no. 7, 2007, pp. 959-982. Geometry methods are: the standard k-ω model, the standard k-ε model and the SST model 4. 6. S. Wang, D. B. Ingham, L. Ma, M. Pourkashanian, and Z. Tao, Numerical investigations on dynamic stall of low Reynolds number flow around oscillating airfoils, Computers & Fluids, vol. 39, no. 9, 2010, pp. 1529-1541. As shown in Figure 1, the 3-D solid model of the rotor was generated with ProEngineer 4.0. And the 2-D model of the Dynamic stall The author would like to thank my academic supervisors Dr M. Vahdati and Dr J. Barlow and my industrial supervisor Dr A. Mewburn-Crook for their supports for this work. As shown in Figure 3, dynamic stall is mainly characterised • by flow separations at the suction side of the airfoil 5. This The author is also grateful to the EPRSC and MatildasPlanet for funding this project. can be summarised in four crucial stages: 2-D • Leading edge separation starts, • Vortex build-up at the leading edge, • • Detachment of the vortex from leading edge and build-up of trailing edge vortex, Department of Technologies for Sustainable Built Environments, University of Reading, Whiteknights, RG6 6AF • Email: [email protected] • www.reading.ac.uk/tsbe • Figure 1. Three dimensional rotor of a straight-bladed Darrieus wind turbine obtained with ProEngineer 4.0 and two dimensional rotor extrapolated from middle plane. Acknowledgements • VAWT was generated from the middle plane and imported into ANSYS CFX 12.0. 3-D 4. D. C. Wilcox, Turbulence Modeling for CFD. DCW industries La Canada, 2006. Detachment of trailing edge vortex and breakdown of leading edge vortex Contact information Figure 4. Lift and Drag coefficient (Cl and CD ) from numerical studies and experimental data.