This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Validation of different modeling approaches: Standard wake models Hybrid model using IBL approach Linear CFD-RANS model Non-linear CFD-RANS model Large-Eddy Simulations
Results Nysted wind farm Horns Rev wind farm
Conclusions
Acknowledgments J. Garza and colleagues at DONG Energy R. Barthelmie at Indiana University DONG Energy, Vattenfall and E.On P.-E. Réthoré at the Risø /DTU
BackgroundGOAL : To validate and improve our estimated wake losses and energy
productions.
QUESTIONS : How to capture the interaction between multiple turbine wakes
(deep array effect) as well as between wakes and the PBL?
METHODOLOGY : Validate different modeling approaches Standard engineering wake models Hybrid model using IBL approach Linear CFD-RANS model Non-linear CFD-RANS model Large-Eddy Simulations
• Based on a balance of momentum to model single wakes (Jensen 1983, Katic 1986)
• Assumes an initial velocity deficit immediately behind the turbine rotor, calculated from the turbine’s thrust coefficient (Ct) and an empirically determined wake-decay constant (k)
• The wake-decay constant sets the linear rate of expansion of the wake with distance downstream
• Based on Navier-Stokes equations with simplifying assumptions (Ainslie 1988)– No pressure gradient term;– Beyond 5 rotor diameter downstream the wake profile is roughly
Gaussian and the centerline deficit decays monotonically;– Etc.
• Valid only at distances farther than ~ 2-3 rotor diameters downstream of a turbine.
• The model runs fast on any PC → suitable for turbine layout optimization.• An industry standard for calculating wake losses
• Hybrid model based on internal boundary layer (IBL) growth and Eddy Viscosity (Brower and Robinson 2009)
• Assign a roughness to each turbine and assume that an internal boundary layer develops at the bottom and top of the turbine rotor (based on Frandsen 2007).
• Couple the IBL growth model with Eddy Viscosity• The model runs fast on any PC → suitable for turbine layout optimization• An industry standard for calculating wake losses (openWind).
• Linear RANS model + actuator disk developed by Risø/DTU• Designed for sites with homogeneous terrain and roughness• Fully integrated within WAsP
Garza, J. et al. (2011). “Evaluation of two novel wake models in offshore wind farms ". Proceedings from the EWEA Offshore conference, 29 Nov. - 1 Dec 2011. 10 p.
• RANS model using a k- turbulence closure. • Based on the commercial RANS software Ansys CFX. • Added of an actuator disk to model wakes. • Does not take atmospheric stability into account (at the moment)
Garza, J. et al. (2011). “Evaluation of two novel wake models in offshore wind farms ". Proceedings from the EWEA Offshore conference, 29 Nov. - 1 Dec 2011. 10 p.
Based on the actuator disk theory (Adams et al. 2007, Réthoré et al. 2008) , a wind turbine is modeled as :
Drag force due to the thrust force that a turbine exert on the upwind flow.
Source of turbulent kinetic energy representing the sub-grid scale turbulence due to the turbine-induced wakes. It includes the effects of the blade tip, blade shed and root vortices. Ct(|u|) = thrust coefficient,
Cp(|u|) = power coefficient, u = wind speed vectors, = air density,A = area swept by the blades.
These results were kindly provided by Garza et al. (2011). “Evaluation of two novel wake models in offshore wind farms”. Proceedings from the EWEA Offshore conference, 29 Nov. - 1 Dec 2011. 10 p.
• The Fuga and WindModeller models align very well with the observed normalized power production within the third column of turbines .
• The Park model performs better at Horns Rev than it did at Nysted but, as with the EV model, the profile remains relatively flat after the fourth column
• The Jensen and Fuga models show better accuracy over 30° wide sector than 5°.
• The Park and Eddy Viscosity models works well within the first 3 columns. However, they are typically not able to capture the wake losses beyond the 3rd column from the front (→ deep array effect).
• DAWM captures the wake losses in large array much better than either the EV or Park model.
• The Fuga and Windmodeller models also showed promise, though a full comparison was not possible.
• ARPS performed reasonably well for these initial tests and merits further research
• Need more detailed power production data with concurrent meteorological conditions (e.g. IEA WakeBench experiment)
Acknowledgments J. Garza and colleagues at DONG Energy R. Barthelmie at Indiana University DONG Energy, Vattenfall and E.On P.-E. Réthoré at the Risø /DTU