1 Fluidic Load Control for Wind Turbine Blades C.S. Boeije, H. de Vries, I. Cleine, E. van Emden, G.G.M Zwart, H. Stobbe, A. Hirschberg, H.W.M. Hoeijmakers Engineering Fluid Dynamics
Dec 13, 2015
1
Fluidic Load Control for Wind Turbine BladesC.S. Boeije, H. de Vries, I. Cleine, E. van Emden, G.G.M Zwart, H. Stobbe, A. Hirschberg, H.W.M. Hoeijmakers
Engineering Fluid Dynamics
2
Overview
• Introduction
• Experimental Setup
• Numerical Setup
• Results
• Conclusions
3
Introduction
• Fatigue loads affect energy production costs– Amount of required materials for structure– Maintenance– Reliability during life span
• As wind turbines become larger, fatigue loads become more important
• Aim: Reduction of Fatigue Loads
4
Introduction (cont’d)
• Fatigue loads due to:– Variable wind, Turbulent inflow, Wind shear
– Gravity
– Tower shadow
– Yaw misalignment
– Wake interaction (in wind farms)
• State of the art load control:– Passive: through aero-elastic response of structure, e.g. tension-
torsion coupling
– Active: pitching of (individual) blades
• Pitching of blades not fast enough to control rapidly changing loads
5
Introduction (cont’d)
• Concepts for fast control:– Conventional trailing edge flaps
• Chow et al., Journal of Physics:Conference Series 75 (2007) 012027
– Flexible trailing edge flaps• Barlas & Van Kuik, Journal of Physics:
Conference Series 75 (2007) 012080
– MEM tabs• Chow & Van Dam, J. of Aircraft, Vol. 43,
No. 5, 2006, pp. 1458-1469
6
Introduction (cont’d)
• Present study: load control concept using fluidic jets– Boundary layer separation control– Pitch control
• Concept: injection of air from multiple orifices or slits controlled individually– Slits located nearby trailing edge of blade– Continuous injection of air directed normally to blade surface
7
Introduction (cont’d)
• Initial steps in investigation:– Study of flow around non-rotating blade section– 2D flow:
• Long slits: length L=O(c), width w=O(0.01c)• Simultaneous operation, no control system
– Uinj/U∞=O(1)
• Important issues:– Is continuous injection from long slits located near trailing edge
effective?– Are jet velocities of Uinj/U∞=O(1) sufficient?– How fast can aerodynamic performance be changed, i.e. what is
response time?
8
Experimental Setup
• Twente University’s closed loop wind tunnel:– Test section 0.7 m x 0.9 m
– Maximum flow velocity 65 m/s
• Investigated blade section: – NACA 0018, chord length c = 0.165 m
– 4 slits (L=0.15 m, w=0.001 m),located at x/c=0.9
• Investigated cases:– Rec=6.6x105, M∞=0.176
– Uinj/U∞=1.2
– Five angles of attack between -12° and +12°
9
Experimental Setup (cont’d)
• Static pressure measurements at 4 locations: x/c = 0.042, x/c = 0.221 (on both sides of airfoil)
• Injection of air at x/c = 0.9– Compressed air (6.5 bar) fed to Piccolo tube in aft compartment
– Choked flow in holes Piccolo tube (constant mass flow)
– Injection velocity determined from static pressure measurement in aft compartment
10
Numerical Setup
• Commercial flow solver ANSYS CFX 11.0• Reynolds Averaged Navier-Stokes equations for compressible
steady and unsteady flow• Shear Stress Transport eddy viscosity turbulence model• Spatial discretization: blend of 2nd- and 1st-order accurate
schemes• Time discretization (unsteady flow cases): 2nd-order accurate• C-type structured grids,466x89x3 cells, y+<1
11
Numerical Setup (cont’d)
• In experiments sharp inner edges of slit:– Vena Contracta will form inside slit
– Hot Wire Anemometry measurements show effective slit width of 70% of actual slit width
• Jet modeled as boundary condition at airfoil surface: constant normal velocity of 1.2U∞ and slit width of 0.7W
• Quasi 2D flow: full span slit
12
Results
NACA 0018 Rec=6.6x105 Uinj/U∞=0.0 or 1.2
• Flow on lower surface separates
• cl increases (less negative) due to jet by ≈0.4 (comp)
13
Results (cont’d)
NACA 0018 Rec=6.6x105 Uinj/U∞=0.0 or 1.2
• Flow on lower surface separates
• cl increases due to jet by ≈0.4 (comp)
14
Results (cont’d)
NACA 0018 Rec=6.6x105 Uinj/U∞=0.0 or 1.2
• Flow on lower surface separates
• cl increases due to jet by ≈0.4 (comp)
15
Results (cont’d)
• Comparison of calculated pressure distributions with
experimental data:
– In general, experiments and computations show the same trend:
increase of lift
– Calculations show more pronounced effect of injection on pressure
on side where slit is located, clearly visible for positive angles of
attack
• Possible causes: 3-dimensionality of flow, or applied turbulence model
does not accurately predict transition
16
Results (cont’d)
• Instantaneous streamlines superposed on contour plot of Mach number field for α=+8°, Rec=6.6x105
– Left: Uinj/U∞=0; Right: Uinj/U∞=1.2
At TE: flow tangential L.S. At TE: flow tangential to U.S. due to entrainment jet formation of recirculation zone
17
Results (cont’d)
• Predicted lift curves, NACA 0018 Rec=6.6x105 Uinj/U∞=0.0 or 1.2
• Δcl0.4 independent of α • dcl/dα not affected
18
Results (cont’d)
• Transient response of lift for α=+8°, Rec=6.6x105
• Computation started from free-stream conditions, injection starts at tU∞/c=15
• 95% of Δcl obtained in ΔtU∞/c4
• 50% of Δcl obtained in ΔtU∞/c1
19
Conclusions
• Studied is load control for NACA 0018 airfoil for α[-12°, +12°]• Load control by continuous injection of air from slits:
– slits located near trailing edge of lower side airfoil x/c=0.9– jet directed normal to airfoil surface, Uinj/U∞=O(1)– slit width O(0.01c)
• Computations and experiments show same trend: increase of lift – Computational results show that an increase of Δcl0.4 can be
obtained, half of this in a dimensionless time of ΔtU∞/c1– Computational results predict more pronounced effect of jet than
experimental results
• Promising option for load control