Variable-Speed Wind Turbine Controller

Nolan D. Caliao Supervisor: Dr. A. Zahedi

Abstract
The Simulation Model
Uncontrolled wind turbine configuration such as stall-regulation captures energy relative Reference Wind
Pitch (ref) Speed
to the amount of wind speed. This configuration requires constant turbine speed Rotor
Wind speed
because the generator that is being directly coupled is also connected to a fixed- Ref. Rotor Controller + Pitch Angle Actuator
Turbine
Speed (PI/PID) +
frequency utility grid. In extremely strong wind, only a fraction of available energy is +   Limit

captured. Plants designed in such configuration are not economically feasible to run at (error)

this occasion. Thus, wind turbines operating at variable speed are better alternatives.
Figure 5: Block diagram of the simulation model

A controller design methodology applied to a variable-speed, horizontal axis wind

turbine was developed. A simple but rigid wind turbine model was used and linearised
to some operating points to meet the desired objectives. From a reference value; by
using blade pitch control the deviation of the actual rotor speed is minimised. The
performances of PI and PID controllers were compared relative to a step wind
disturbance. Results show comparative responses between these two controllers. With
the present methodology, despite the erratic wind data, the wind turbine still manages
to operate at the stable region 88% most of the time.

The Physical Model

Figure 6: Wind speed data at 1 Hz. Figure 7: Wind turbine disturbance response
IT ngear (13 m/s – 14 m/s wind speed step change)

K
• high wind, mean = 15.7 m/s; standard
deviation = 6.9 m/s • Step disturbances are the simplest to model
TA TE
• low wind, mean = 9.5 m/s; standard and analyze yet they represent the most
deviation = 4.2 m/s) severe disturbances a wind turbine is likely
to encounter.
Aero Low High
dynamic speed Gear
speed
box
shaft shaft
The Performance Outputs
Figure 2: Physical model of drive train,
I = turbine and generator inertias, K =
Figure 1: Significant parts of a wind spring stiffness
turbine

The Mathematical Model

ITT = TA − TE  1st order model

TA = C  Cq  ( ,  )(vw ) 2
Figure 8: Performance Under High Figure 9: Comparison of the rotor speed and
Where:
Wind of the linear model A for PI blade pitch angle traces for the PID and the
• IT = moment of inertia of turbine rotor.
controller PI controllers of the linear model A.
• T = angular shaft speed.
• TE = mechanical torque necessary to turn the generator.
• TA = aerodynamic torque, is expressed in terms of the variable torque coefficient Cq(,) which is The Performance Indicators
dependent on the tip speed ratio () and blade pitch angle () and wind speed v.
• C = 0.5AR;  is air density, A is rotor swept area, R is rotor radius. Two metrics determines the performance of the Table 1: RMS and ADC values for each model
Performanc High wind
controller.
ITT = 1  T +  2  vw + 3  
Model
 linearised model e check PID PI
• The root mean square (RMS) of the error Reference RMS 0.0325 0.0464

T  Cq  between the actual rotational speed and the Point ADC 0.6599 0.7000
 2 = IT  = C  vw op 2 Cq − op  desired fluctuations is minimised
 T op  op  op  Linear A
RMS 0.0308 0.0457
  • The actuator duty cycle (ADC) was used to ADC 0.5042 0.5638
2 Cq measure the actuator motion during the
T  
Linearised coefficients RMS 0.0314 0.0467
 2 = IT  = C   vw  simulation. ADC is the total number of
Linear B
ADC 0.4258 0.4859
  op   degrees pitched over the period of the
op op RMS 0.0304 0.0451
simulation. Linear C
T Cq ADC 0.3741 0.4452
1 = IT  = C  R vw
T op 
op
Table 2: Average tip speed ratio and wind turbine power under high wind.
Ave. tip speed ratio Std. deviation of tip speed
Ave. power (W)
() ratio ()
The Operating and Linearised Points Model
PID PI PID PI PID PI
Reference
6.87 6.76 2.62 2.83 8.9580 104 9.4669 104
Point
Linear A 7.07 6.86 2.71 2.83 1.7798 106 1.7886 106
Linear B 6.79 6.79 2.61 2.82 1.5864 105 1.6157 105
Linear C 6.80 6.72 2.53 2.79 1.3495 105 1.3972 105

Conclusion
Since life of most of the wind turbine components is determined by its capacity to
withstand high wind speed, a highly turbulent wind speed data was used as an input
to the first-order nonlinear wind turbine model. RMS and ADC of linear model C are
Figure 4: Reference and linearised considered the preffered values however linear model A has better power output.
Figure 3: Cq under the normal and stall
regions. Regions belonging to below operating points.
crit and above crit represent the Generally, the PID controller has better performance in terms of the RMS and ADC
• operating tip speed ratio (op) → maximum Cp
normal and the stall regions measures. If the power generation output is however the important criteria in the
respectively. • the objective of the controller is to set the blade design, the PI controller should be the preferrence controller.
pitch (op) at a certain operating value so as to
attain and maintain maximum Cp as possible
during operation.

Electrical and Computer Systems Engineering

Postgraduate Student Research Forum 2001