Masakazu Shimooka, Makoto Iida, and Chuichi Arakawa The University of Tokyo

1. Basic Study of Winglet Effects On Aerodynamics and Aeroacoustics Using Large-Eddy Simulation Masakazu Shimooka, Makoto Iida, and Chuichi Arakawa The University of Tokyo European Wind Energy Conference & Exhibition Athens, Greece, 27 February – 2 March 2006

2. To optimize the tip shape for increasing public acceptance of wind energy. To clarify winglet effects on aerodynamic performance, loads, noise. To investigate a possibility of application to the blade design tools. Purpose of this work

3. Simulate the whole blade including the tip shape effects, using LES (Large-Eddy simulation) with 300 million grid points. Investigate effects of differences of tip shapes on aerodynamics and aeroacoustics (Direct Noise Simulation). ・ 2 types of winglets whose installation angle is 0, 50 degree. Introduce our current work (Detached-Eddy simulation) based on our knowledge of LES. Outline of this work

4. Simulation results Related research (WINDMELⅢ） Actual tip shape Oliver Fleig, Chuichi Arakawa 23rd ASME Wind Energy Symposium January 5 – 8, 2004, Reno, Nevada Ogee tip shape

5. Developed by Whitcomb Diffuse tip vortices Reduce induced drag Increase thrust and lift force Examples of winglets for blades of rotation ・Tip vane by van Holten　（Wind turbine） ・Mie vane by Shimizu　（Wind turbine） ・Bladelet by Ito　（Marine propeller） Increase of rotor output as results of experiments and numerical analysis such as ・BEM (Blade Element Momentum method) ・VLM (Vortex Lattice Method) What is winglet ? In this work, We use Navier-Stokes simulationto resolve complex structure of tip vortices in detail.

6. SGS Smagorinsky Model (Cs = 0.15) Van Driest Wall damping function Numerical method(1) - Flow field ・ Governing equation: Compressible Navier-Stokes equation ・ Turbulence model： LES Smagorinsky model ・ 3rd order Upwind Finite Difference scheme in space ・ 1st order Implicit Euler scheme in time

7. Near field: Direct noise simulation By compressible LES × Far field: Modeled By Ffowcs Williams-Hawkings (FW-H) equation Numerical method(2) - Acoustic field • Near field 　　（1 to 2 chord lengths） 　・ Direct noise simulation 　・ sufficiently fine grids 　・ Accurate modeling of non-linear effects and wall reflection, refraction, scattering in the near field • Far field 　・ Ffowcs Williams-Hawkings equation 　・ permeable integration surface which does not need to correspond with the body surface

8. Boundary condition inflow • Uniform flow at inlet • Convective boundary conditions at outlet • Wall: No-slip conditions; pressure and density extrapolated • Outer boundaries are very coarse to prevent reflection of high frequency acoustic waves: • Large rate of grid stretching and extreme distance between blade and outer boundaries 　・Half-sphere ・Periodic plane a-b ・Radius of sphere is twice the blade span a b y x z Rotation axis Computational domain

9. ξ ζ η y z x Direct noise simulation 25-30 grid points per wavelength Computational grid 765 points，along the surface (ξ) 193 points，perpendicular to the surface (η) 2209 points，along the span direction (ζ) Total number of grid points,300million Use 14 nodes (112 CPU) on Earth Simulator ξ ・ Single O-grid ・ Minimum wall distance is 2×10-5 corresponding to y+=1 (wall resolved) ・ High concentration of grid points in the blade tip region Grid spacing of airfoil section (ζplane)

10. 50deg. 0deg. Simulation parameters and tip shapes • Re = 1.0x106 Reference is the chord length at tip c = 0.23(m)， and the effective flow velocity at tip Ueff = 61.74(m/s) • Mach = 0.18 at tip • Δt = 3.6x10-5c/Ueff = 1.3x10-7(s) Ueff Ueff Tip shape (top: 50deg., bottom: 0deg.)

11. Flow field - Tip vortex 0deg. 50deg. Vorticity magnitude iso-surfaces

12. Smaller but more complex structure Pressure contours at the trailing edge 0deg. 50deg. ・trailing edge at the very tip (y/c=1.0) ・Winglet diffuses tip vortices.

13. Vorticity magnitude contours at the near wake y/c =2.0 y/c =1.8 y/c =1.6 50deg. y/c =1.4 y/c =1.2 y/c =1.0 y z 0deg. x

14. 0deg. 50deg. 50deg Vorticity magnitude contours and iso-surface (|ω|=4.0) ・Winglet reduces the strength of tip vortices .

15. z z x x Spanwise velocity components contours 0deg. 50deg. ・Spanwise velocity (w) component contours at y/c=0.7 ・Reduced downwash effect, and Spread of wake in spanwise direction.

16. Rotational torque and Flap moment Main blade Winglet Main blade Winglet Hub side Hub side Tip side Tip side ・ Increase of rotational torque at the winglet and the main blade near the winglet. ・ Reduction of flap moment at the winglet.

17. Larger suction peak at the leading edge More sufficient recovery of pressure at the trailing edge Pressure distribution 50deg. 0 deg. Suction side

18. 0deg. 0deg. 50deg. 50deg. Point A Point B Acoustic field – Near field SPL (dB), ref: 2×10-5(Pa) SPL (dB), ref: 2×10-5(Pa) Frequency (Hz) Frequency (Hz) ・ Point A is where the tip vortex is developed. ・ Point B is slightly downstream from the trailing edge of main blade near the winglet.

19. Blade (dB) (dB) Smaller but more complex vortices caused by winglet emit strong noise In high frequency. Acoustic field – Far field Integration surface for FW-H equation (yellow surface) 0deg. 50deg. Far field overall sound pressure level (OASPL) Integration from 1kHz to 12.5kHz (2.3m downstream from rotor)

20. Current Work— Detached-Eddy Simulation — for NREL Phase VI

21. Pressure distribution (U∞=7.0m/s) α=11.8° α=10.1° α=12.2° α=7.4° α=8.3°

22. Flow field (U∞=25.1m/s) U∞ Vorticity magnitude iso-surface (|ω|=0.2) and contours Streamlines

23. Conclusions • We succeeded in capturing winglet effects in detail, using 300 million grid points in Earth Simulator. - Diffuse and reduce tip vortices. - Reduce downwash effect, and Spread wakein spanwise direction. • This simulation will be very useful for designing optimal tip shapes. • We have performed Detached-Eddy simulation as the first step for less computational costs • This simulation is based on our knowledge of grid dependence in LES.