Wind Turbine/Farm Noise : Propagation Modelling
- 1. Development of an advanced noise propagation model for noise optimization in wind farm Emre Barlas Supervisors: Wen Z. Shen, Wei J. Zhu, Jens N. Sørensen PhD Defense DTU Wind Energy Department of Wind Energy Fluid Mechanics 26.01.2018 1
- 2. 1. Introduction & Motivation 2. Atmospheric Acoustics & Modelling 3. Sound Propagation & Wind Turbine Wake 4. Wind Turbine Noise Generation & Propagation Model 5. Preliminary Wind Farm Study 6. Conclusions & Future Work Content
- 3. 1. Introduction (1/3) Issue of Wind Turbine Noise
- 4. 1. Introduction (1/3) Issue of Wind Turbine Noise Generation (Design, Operation Conditions, Inflow etc.)
- 5. 1. Introduction (1/3) Issue of Wind Turbine Noise Propagation (atmospheric conditions, terrain, ground characteristics etc. )
- 6. 1. Introduction (1/3) Issue of Wind Turbine Noise Perception (social background, age, landscape etc.
- 7. Motivation & Objectives Motivation Inaccurate wind farm noise assessment would result in: Under-prediction: causes downregulation of wind turbines under certain atmospheric conditions. Over-prediction: causes turbines to be located at less resourceful sites. 1. Introduction (2/3)
- 8. Objectives are; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) SECTION 2 & 3 • to develop a suitable source model that can handle the variability of wind turbine noise generation. - SECTION 4 • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) - SECTION 4 • to prepare a code to be applied for wind farm noise mapping and/or optimization. - SECTION 5 Motivation & Objectives 1. Introduction (3/3)
- 9. 1. Introduction & Motivation 2. Atmospheric Acoustics & Modelling 3. Sound Propagation & Wind Turbine Wake 4. Wind Turbine Noise Generation & Propagation Model 5. Preliminary Wind Farm Study 6. Conclusions & Future Work Content
- 10. 2. Atmospheric Acoustics & Modelling (1/6)
- 11. Engineering approach – Ray model (very fast, limited accurate). Accurate numerical approach: Time domain (expensive, propagation in all directions) – DNS, LES/CAA: compute source + propagation, based on solving Navier- Stokes equations. – FDTD: given source + propagation, based on solving Euler equations Accurate numerical approach: Frequency domain (less expensive, one way propagation) – Parabolic equation (PE) – Fast Field Program (FFP) (layered atmosphere and homogeneous ground) 2. Atmospheric Acoustics & Modelling (2/6)
- 12. 2. Atmospheric Acoustics & Modelling (2/6) Engineering approach – Ray model (very fast, the least accurate). Accurate numerical approach: Time domain (expensive, propagation in all directions) – DNS, LES/CAA: compute source + propagation, based on solving Navier- Stokes equations. – FDTD: given source + propagation, based on solving Euler equations Accurate numerical approach: Frequency domain (less expensive, one way propagation) – Parabolic equation (PE) – Fast Field Program (FFP) (layered atmosphere and homogeneous ground)
- 13. Propagation Model : Parabolic Equation Method 2. Atmospheric Acoustics & Modelling (3/6)
- 14. Propagation Model : Parabolic Equation Method Solve the wave equation with; General assumptions; Implies; • Harmonic wave - Frequency Domain • Axisymmetric - 2D • Far field - Long range propagation • One way propagation - Waves that are traveling from the source to the receiver (no backscattering • Effective Speed of Sound (optional) - Moving atmosphere is replaced by a hypothetical motionless medium with the effective sound speed. 2. Atmospheric Acoustics & Modelling (3/6)
- 15. Propagation Model : Parabolic Equation Method Solve the wave equation with; General assumptions; Implies; • Harmonic wave - Frequency Domain • Axisymmetric - 2D • Far field - Long range propagation • One way propagation - Waves that are traveling from the source to the receiver (no backscattering • Effective Speed of Sound (optional) - Moving atmosphere is replaced by a hypothetical motionless medium with the effective sound speed. OR 2. Atmospheric Acoustics & Modelling (3/6)
- 16. 2. Atmospheric Acoustics & Modelling (6/6) With turbulence Without turbulence Transmission Loss for 800 Hz Wind direction dB
- 17. 1. Introduction & Motivation 2. Atmospheric Acoustics & Modelling 3. Sound Propagation & Wind Turbine Wake 4. Wind Turbine Noise Generation & Propagation Model 5. Preliminary Wind Farm Study 6. Conclusions & Future Work Content
- 18. Receiver 3. Sound Propagation & Wake (1/6) SOFAR Channel in the ocean (sound fixing and ranging channel) Noise Behind a Barrier in the atmosphere SPL with Non Varying Wind SPL with Varying Wind
- 19. 3. Sound Propagation & Wake (2/6) Turbine : NM 80 - Incoming Turbulence Int. : 3% Snapshot of the flow field obtained from unsteady simulations Time Averaged Wind Profile
- 20. 3. Sound Propagation & Wake (3/6)
- 21. 3. Sound Propagation & Wake skipped
- 22. 1. Introduction & Motivation 2. Atmospheric Acoustics & Modelling 3. Sound Propagation & Wind Turbine Wake 4. Wind Turbine Noise Generation & Propagation Model 5. Preliminary Wind Farm Study 6. Conclusions & Future Work Content
- 23. 4. Combined Model (1/19) Generation (Design, Operation Conditions, Inflow etc.) Propagation (atmospheric conditions, terrain, ground characteristics etc. )
- 24. COMBINED MODEL OVERVIEW UNSTEADY FLOW MODEL (Large Eddy Simulation) Aeroelastics tool (FAST v8) + Semi Empirical Airfoil Noise (BPM) Propagation (Parabolic Equation) 1 2 3
- 26. 4. Combined Model (3/19) Source : Aeroelastics + Aeroacoustics Turbulent Boundary Layer Trailing Edge noise Separation-Stall noise Turbulent Inflow Noise
- 27. Store the highest SPL contributor airfoil location along each blade (freq, blade, receiver, time) 4. Combined Model (4/19) Source : Aeroelastics + Aeroacoustics Source Locations For 2 frequencies
- 28. Run PE (freq,blade,receiver,time) 4. Combined Model (5/19)
- 29. 4. Combined Model (6/19)
- 30. Source Only Simulations OSPL @ 2 m receiver height 20-1000 Hz Interpolated in between receivers Variability is caused by • Blade Movement • Angle of attack and TI change due to turbulent atmosphere Lacking propagation physics • Atmosphere (refraction) • Ground (reflection) 4. Combined Model (8/19)
- 31. Coupled Simulations OSPL @ 2 m receiver height 20-1000 Hz Interpolated in between receivers Z0 – Roughness Value : 0.6 m Neutrally stratified atmosphere Grassland 4. Combined Model (9/19)
- 32. Amplitude Modulation 4. Combined Model (10/19)
- 33. Amplitude Modulation Source Only Simulations Wind Direction High AM at crosswind 4. Combined Model (10/19) TOP VIEW
- 34. Wind Direction High AM at crosswind & Downwind & Upwind Amplitude Modulation Coupled Simulations 4. Combined Model (11/19) TOP VIEW
- 35. Amplitude Modulation Coupled Simulations Wind Direction High AM at crosswind & Downwind & Upwind 4. Combined Model (12/19) TOP VIEW
- 36. LES Flow Field Output • Horizontal slice at hub height • Vertical Slice at the rotor 4. Combined Model (13/19)
- 37. Unsteady nature of wake + unsteady nature of the source results in ‘’unexpected’’ far field modulation Flow @ Hub height With WakeOSPL @ 2m height With Wake Flow @ Hub height Without WakeOSPL @ 2m height Without Wake
- 38. Amplitude Modulation - Coupled Simulations Wake Effect 4. Combined Model (15/19)
- 39. 4. Combined Model (16/19) Variability Flow variability During a day for the same hub height wind speed
- 40. 4. Combined Model (17/19) Variability - SNAPSHOT FLOW SOUND PRESSURE LEVELS
- 41. 4. Combined Model (18/19) Variability – TIME AVERAGED SPL OSPL @ 2 m height
- 42. 4. Combined Model (18/19) Variability – TIME AVERAGED SPL OSPL @ 2 m height Spectra at different distances
- 43. 4. Combined Model (19/19) Variability - AMPLITUDE MODULATION Ampl. Modul.
- 44. 5. Preliminary Wind Farm Study (1/9) A wind farm in Shanxi region of China over complex terrain 25 turbines : 93 m diameter, 70 m hub height
- 45. 5. Preliminary Wind Farm Study (2/9) RANS Flow Field - EllipSys3D Stream-wise velocity different 2d slices Wind Direction Wind Direction
- 46. 5. Preliminary Wind Farm Study (4/9) Time dependent source locations and 2D PE slices
- 47. 5. Preliminary Wind Farm Study (4/9) Time dependent source locations and 2D PE slices
- 48. 5. Preliminary Wind Farm Study (5/9) Receiver distribution around the wind farm
- 49. 5. Preliminary Wind Farm Study (6/9) OSPL around the wind farm
- 50. 5. Preliminary Wind Farm Study (7/9) Parameter study with 4 cases. CASE NUMBER TERRAIN WIND 1 FLAT NO WIND (homogenous atmosphere) 2 FLAT LOG WIND 3 COMPLEX LOG WIND 4 COMPLEX RANS WIND
- 51. 5. Preliminary Wind Farm Study (8/9)
- 52. 5. Preliminary Wind Farm Study (9/9) Difference between flat and complex terrain Difference between log wind and RANS flow (both complex terrain)
- 53. 6. Conclusions and Future Work Objectives were; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) • to develop a suitable source model that can handle the variability of wind turbine noise generation. • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) • to prepare a code to be applied for wind farm noise mapping and/or optimization. 53
- 54. 6. Conclusions and Future Work Objectives were; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) Fortran/MPI implementation for Parabolic Wave Equation (vector or scalar PE). • to develop a suitable source model that can handle the variability of wind turbine noise generation. • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) • to prepare a code to be applied for wind farm noise mapping and/or optimization. 54
- 55. 6. Conclusions and Future Work Objectives were; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) Fortran/MPI implementation for Parabolic Wave Equation (vector or scalar PE). • to develop a suitable source model that can handle the variability of wind turbine noise generation. Point sources represented within PE model, in which the source power levels are obtained from airfoil aeroacoustics combined with aeroelastic simulations and high fidelity solvers LES. • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) • to prepare a code to be applied for wind farm noise mapping and/or optimization. 55
- 56. 6. Conclusions and Future Work Objectives were; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) Fortran/MPI implementation for Parabolic Wave Equation (vector or scalar PE). • to develop a suitable source model that can handle the variability of wind turbine noise generation. Point sources represented within PE model, in which the source power levels are obtained from airfoil aeroacoustics combined with aeroelastic simulations and high fidelity solvers LES. • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) Day & Night / Ground & Terrain effects (next slide). • to prepare a code to be applied for wind farm noise mapping and/or optimization. 56
- 57. 6. Conclusions and Future Work Objectives were; • to develop a high-fidelity sound propagation model (accuracy / comp. demand) Fortran/MPI implementation for Parabolic Wave Equation (vector or scalar PE). • to develop a suitable source model that can handle the variability of wind turbine noise generation. Point sources represented within PE model, in which the source power levels are obtained from airfoil aeroacoustics combined with aeroelastic simulations and high fidelity solvers LES. • to investigate the various effects (i.e. wind and temperature gradient, ground cover, atmospheric and wake turbulence, turbine operation conditions) Day & Night / Ground & Terrain effects (next slide). • to prepare a code to be applied for wind farm noise mapping and/or optimization. Wind Farm Study 57
- 58. 6. Conclusions and Future Work • Time Averaged SPL can be captured within reasonable accuracy via moving 2D source + mean flow • Near field daytime levels are higher than night time levels due to enhanced turbulence under convective conditions. • The atmospheric conditions affect propagation significantly and in the far field levels at night time are higher than daytime. (contrary to near field) • Wake is an important factor for downwind propagation. Wake deficit entrapment • Unsteady Wake + Unsteady Sources lead to enhanced far field AM. 58
- 59. 6. Conclusions and Future Work Future Work • Unsteady Wind Farm Noise • Wind Farm Control • Further validation 59