1. Wind Energy Program School of Aerospace Engineering Georgia Institute of Technology Computational Studies of Horizontal Axis Wind Turbines PRINCIPAL INVESTIGATOR: Lakshmi N. Sankar NREL/SNL TECHNICAL MONITORS: Alan Laxson (NREL), Scott Schreck (NREL), Walter Wolfe (SNL) OBJECTIVE(S): –Develop a first-principles based methodology for the prediction of horizontal axis wind turbine performance. –Use the methodology to study the effects of tower shadow, atmospheric turbulence and yaw angle on rotor blade loads. –Reduce the computational cost of modeling the 3-D viscous flow field, through the use of phenomenological models.

2. Wind Energy Program SCHEDULE AND STATUS: This is a three year effort, spanning the period May 1997 - May 2000. Year 1 Goal: –Develop and validate a first-principles based method for the prediction of horizontal axis wind turbine aerodynamics. –Status: Completed. Year 2 Goal: –Incorporate Atmospheric boundary layer effects, boundary layer transition models, 1-equation turbulence models, and validate against available data. –Status: Completed. Results to be presented today. Year 3 Goal: –Incorporate tower effects; Examine existing stall models and tip loss models in light of computed data; Make computer codes available to interested researchers and industries.

3. Wind Energy Program BUDGET: Year 1: $ 56, 805 –Covers 200 hours of P. I. Time and a graduate student. Year 2: $ 59,267 Year 3: $ 61,883

4. Wind Energy Program TECHNICAL RESULTS: Outline of the present methodology Recap of Results for Phase II and Phase III rotors Transition Model Studies Effects of Yaw on Rotor Loads and Power Generation

5. Wind Energy Program PRESENT HYBRID METHODLOGY The flow field is made of –a viscous region near the blade(s) –A potential flow region that propagates the blade lift and thickness effects to the far field –A Lagrangean representation of the tip vortex, and concentrated vorticity shed from nearby bluff bodies such as the tower. –Method is unsteady, compressible, and does not have singularities near separation lines. –Method described in AIAA Journal of Aircraft, Vol. 34, No.5, 1997, pp. 635-640. N-S zone Potential Flow Zone Tip Vortex

6. Wind Energy Program SAMPLE GRID A fully automated grid algebraic generation procedure has been developed. User only needs to specify the airfoil shape and twist distribution at a few radial locations. The grid generator automatically divides the zones into Navier-Stokes and Viscous Zones, based on user input.

7. Wind Energy Program SAMPLE RESULTS - Phase III Rotor

8. Wind Energy Program Sample results - Phase II Rotor

9. Wind Energy Program The hybrid code rapidly converges to steady state when one exists (19 seconds/iteration on a HP Model 750 Workstation)

10. Wind Energy Program Transition Models Transition to turbulent boundary layers can have a dramatic effect on the flow over the rotor, and power generation. A number of “Engineering” models are available in literature. These models were developed from 2-D steady flow applications, and may be applied in 3-D flows using a “strip theory” approximation. Two transition models were incorporated into the hybrid code: –Eppler Model, Michel’s Model

11. Wind Energy Program Eppler’s Model This model is in wide use in many of the airfoil analyses and design codes used by the wind turbine industry. Transition is said to occur when laminar flow separates, or when: Reynolds number based on momentum thickness Ratio of energy thickness over momentum thickness Roughness Factor

12. Wind Energy Program Michel’s Model This model is in wide use in fixed wing aircraft industry. Reynolds No based on momentum thickness Reynolds Number based on distance from leading edge=u  x/

13. Wind Energy Program Transition Line on the Rotor Upper Surface Phase III Rotor, 6 m/s wind

14. Wind Energy Program Transition Line on the Rotor Lower Surface Phase III Rotor, 6 m/s wind

15. Wind Energy Program Transition line on the Upper Surface Phase III Rotor, 8m/s Wind

16. Wind Energy Program Conclusions on Transition Model Study On both the upper and the lower surface, Eppler’s model predicts a transition location that is either comparable to, or upstream of Michel’s predictions. On the lower surface, the pressure gradients are favorable. This leads to a thinner boundary layer. Both these criteria predict that transition will occur aft of the corresponding upper surface locations. Transition line location appears insensitive to the turbulence model used, except at inboard stations.

17. Wind Energy Program Modeling Inflow Turbulence and Yaw Effects Present Methodology has been modified to account for inflow turbulence and yaw on the rotor loads and power generation. A steady cross flow, a boundary layer profile, or an unsteady freestream condition may be prescribed, with minor change to the present code. This information impacts the outer flow (potential) field far away from the rotor as follows:

18. Wind Energy Program Typical Wind Conditions for the Phase IV Rotor (NREL Database)

19. Wind Energy Program

20. Variation of Computed Power over an Entire Revolution 10 m/s, 20 Degree Yaw A 3 per rev variation was dominant. Caused by the 120 degree phase difference between the three blades.

21. Wind Energy Program CONCLUSIONS: A first-principles based methodology for Predicting Power Generation by HAWT rotors has been proposed and validated. Two turbulence models (Baldwin-Lomax, Spalart- Allmaras) and two transition models have been implemented. A formulation is in place for modeling yaw effects and inflow turbulence. The code correctly predicts the expected 3/rev variation in power. The computed power generation in the presence of yaw effects is in overall agreement with measured data.

22. Wind Energy Program FUTURE PLANS: Tower effects will be modeled using an overset grid methodology, where the tower, nacelle and the rotating blades are modeled on separate grids. The existing theories for static and dynamic stall delay, and for tip loss effects will be examined in light of computed data. Additional simulations for the NREL Phase IV Rotor under yaw conditions will be done and compared with NREL data. Results for these calculations will be presented this time next year.