Integrating Deformable Trailing Edge Geometries in Modern MegaWatt Wind Turbine Controllers

1
Integrating Deformable Trailing Edge Geometries
in Modern Mega-Watt Wind Turbine Controllers

• Peter Bjørn Andersen Ph.D. Student Risø-DTU
• Lars Henriksen Ph.D. Student Risø-DTU
• Mac Gaunaa Senior Scientist Risø-DTU
• Christian Bak Senior Scientist Risø-DTU
• Thomas Buhl Senior Scientist Risø-DTU
• European Wind Energy Conference
  ExhibitionBrussels Expo, Belgium, 31 March - 3
  April 2008

2
Agenda
Introduction Controller Results Conclusion
3
Introduction (17)
4
Introduction (27)
Deformable Trailing Edge Geometry (DTEG)
Aileron Flap
5
Introduction (37)
HAWC2() simulation based on the 5MW reference
turbine. () HAWC2 is the aero-elastic
tool used for wind turbine simulations at Risø.
6
Introduction (47)
A profile section with pitot and DTEG
7
Introduction (57)

• DTEG (structure) part
• 10 of chord
• /- 8 degree deflection possible
• from /-8 to -/8 in simulated dt (0.01s)
• no effects of hysteresis
• no overshoot or other dynamics
• DTEG (aerodynamic) part
• max ?CL(a,ß8deg) 0.29
• min ?CL(a,ß-8deg) -0.29
• Dynamic stall model by Gaunaa and Andersen

8
Introduction (67)
Dynamic stall model by Gaunaa and Andersen
Downwash geometric AOA
Trailing Edge (TE) separation (Kirchhoff)
Wake effect effective AOA
Fully separated lift
Fully attached lift
Dynamic TE separation based on dynamic boundary
layer
Project forces on camber (Gaunaa)
Gaunaa / (Theodorsen)
Dynamic lift based on mix between fully attached
and fully separated lift, mixing rule given by
dynamic trailing edge separation
9
Introduction (77)
Turbulent time series (Mann), wind shear (power
law), tower shadow (potential)
10
Agenda
Introduction Controller Results Conclusion
11
Controller (18)
aerodynamic controller part
elastic controller part
pitch controller part
power controller part
12
Controller (28)
An inverse Theodorsen/Gaunaa model Possibility
to use running averages or reference AOA values
(The Ka factor) Objective level out
stochastic signal AOA / Vrel
13
Controller (38)
The Ka factor (low wind speeds) Seeking to
optimize for axial induction CL/CD peek of
Cp-surface Ka0 at rated rotor speed(!) (based
on blade planform design)
Residual for Ka1
Residual for Ka0
14
Controller (48)
An elastic response model Objective level out
deterministic signal Mj (blade root moment)
15
Controller (58)
Elastic response
RESPONSE FROM DTEG (1)
(2)
Response reversal (function of torsional
stiffness)
(5)
(3)
(4)
DTEG Step response at t0s
16
Controller (68)
Pitch communication model Objective DTEG and
blade pitch work together
17
Controller (78)
Pitch servo modeled as a 2nd order system Max
pitch rate 8 degree/sec Power controller
model (gains provided by NREL part of the UpWind
project) Omega filter to remove free-free
oscillations
18
Controller (88)
Controller tuning DTEG Simplex
algorithm with various objectives a 10 min.
turbulent series takes about 1-2 weeks to
tune/optimize POWER Simple relation ?power(?pi
tch) Scheduling Mixing rule
19
Agenda
Introduction Controller Results Conclusion
20
Results (15)
21
Results (25)
22
Results (35)
23
Results (45)
24
Results (55) different turbulent seeds
25
Agenda
Introduction Controller Results Conclusion
26
Conclusion

• Comments on the suggested DTEG/PITCH/GENERATOR
  controller
• Positive
• The DTEG controller can reduce 10 min ultimate
  and fatigue loads for blade root moment and
  tower root moment.
• Reduced pitch activity
• Possibility to tune turbine power output at low
  wind speeds. (DTEG assists in staying close to
  Cp,max on the Cp surface)
• Negative
• Long gain tuning time.
• Future work

27
Future work
Acoustic noise reduction
A real turbine
Position of DTEG
Extreme directional change in wind direction
Dimension of DTEG
Main shaft (fatigue)
Extreme wind conditions (gusts)
Yaw misalignment
Tilting moment
Blade flapwise, extreme (bending, buckling)
Hardware in the loop
Sensor dynamics/hysteresis
Blade edgewise (fatigue)
Floating turbines
Yaw system (extreme)
Tower welding (fatigue)
Offshore
Power consumption
Main bearing (fatigue)
Negative wind shears
Lightning
CFD
Stand still
Gear (fatigue)
Monte Carlo simulations
Wind farm issues
Pitch regulation
IEC Load case
Two bladed turbine
Sensor delay
Stability
Model based controller
Emergency shut down
Foundation (extreme)
Signal noise
Gray box style controller