Wing Mil

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

5th International Conference-Workshop - CPE 2007

Design and Testing of a Low Cost Peak-Power


Tracking Controller for a Fixed Blade 1.2 kVA
Wind Turbine
Horizon Gitano, Soib Taib, and Mohammad Khdeir

Abstract – Wind turbines are a common source of electrical


power generation in rural sites around the world. The
power generated by a wind turbine at a given wind speed is
a non-linear function of rotor speed. This results in a

DC Electrical Load
specific rotor speed for maximum power production at each
wind speed. A control system incorporating a Pulse Width
Modulated DC-DC converter has been designed to vary the
load on the wind turbine thereby forcing it to operate at its
maximum power point. Another important feature of the
controller is the application of electrical stall-breaking of the
turbine. The circuit was then tested on a turbine test bed Fig. 1. Schematic diagram of a wind turbine system
simulating various wind speeds, and the “peak-power
tracking” capability and electrical breaking capability were
verified. The system gave an overall efficiency of 70 to 80% The rotor speed changes with time, due to changes in the
over a wide range of wind speeds and PWM duty cycles.1 wind speed. This continuous change of wind speed
affects the output power of the system, and thus
Index Terms – Wind Turbine, DC-DC converter, Peak
Power Tracking, Power Controller. efficiency [3].
This paper is dedicated to the design of the wind turbine
I. INTRODUCTION controller. The controller must be able to perform the
following tasks:
Wind turbines are widely used for supplying electrical - Output Power limiting
power in isolated areas of high wind. They are considered
- Track the peak output power of the turbine at
an excellent alternative to diesel generators being nominal wind velocities by controlling the
environment friendly. In addition, they are economically rotational speed of the rotor
competitive in more remote regions requiring no - Provide overspeed protection via an electrical
transportation of fuel. Wind electrical generators are breaking system
composed of a wind turbine, electrical generation unit,
The breaking system consists of a dump load which
controller and the load which may be bank of batteries,
can be directly connected across the generator output via
electrical pumps, or any other direct current loads [1][2].
a relay. Several such dump loads may be employed
The block diagram of a wind turbine system is shown in
allowing varying levels of speed control, or they can be
Fig. 1.
used for turbine breaking only.
The wind turbine coverts the wind power to shaft
The design of the power tracker is the most critical part
power directly spinning a permanent magnet rotor
of the controller. It must be capable of measuring the
following the design of Hugh Piggott [3]. The stator is a
output power of the system and adjusting the load on the
three-phase, Y-configuration rectified to DC inside the
wind turbine. The output power is measured via a voltage
electrical generator unit. The controller takes the DC
divider in parallel with the load, and a current sensing
output of the generator and couples it to the load.
resistor in series with the load. The load on the generator
is varied by adjusting the duty cycle of a DC-DC
converter. It updates the measured values periodically,
This work was supported by University Sains Malaysia (USM). It thereby continually adjusting the duty cycle for maximum
was the result of a cooperative work between School of electrical and
electronics engineering and school of Mechanical Engineering.
power extraction.
H. Gitano is with the Department of Mechanical Engineering,
University Sains Malaysia, Nibong Tebal 14300, Penang, Malaysia (e- II. WIND TURBINE CHARACTERISTICS
mail: [email protected]). The output power of the wind turbine is function of
S. Taib is with the Centre of Education, Training and Renewable
Energy Efficiency (CETREE), University Sains Malaysia, Penang
wind velocity cubed. It can be described mathematically
11800, Penang, Malaysia (e-mail: [email protected]). by (1).
M. Y. Khdeir is with the Department of Electrical and Electronics
Engineering, University Sains Malaysia, Nibong Tebal 14300, Penang,
PTurbine = 0.5C P ρAVwind
3
(1)
Malaysia (e-mail: [email protected]).
5th International Conference-Workshop - CPE 2007
where PTurbine, CP, ρ, A, and Vwind are turbine power input voltage (Vi) by a specific factor. This factor is the
(Watts), power coefficient (dimensionless), air density duty cycle of the circuit (D), which is the ratio of the
(kg/m3), blade swept area (m2), and wind speed (m/s), switching element conduction time to the full period time.
respectively [1][2] and [5]. The function of the inductor and the capacitor in the
Both swept area and air density are constants while the circuit is to smooth the output current and voltage,
controlling of the wind speed is beyond the scope of this respectively[7].
paper. Therefore, to maximize the output power of the Vo = DVi (3)
wind turbine we have to optimize the value of the power
coefficient. The power coefficient is a strong, non- linear As switching time plays a significant role in the
efficiency (η) of buck circuit, it must be considered in the
function of the “Tip Speed Ratio”; as shown in Fig. 2 [5]
and [6]. The Tip Speed Ratio is the speed of the tip of the design. According to (4), the higher frequency of the
turbine divided by the wind speed. For a turbine of radius PWM (f) results in a higher efficiency at constant load
R, and rotor speed ωrotor the tip-speed-ratio, λ, is thus: resistance RL, and constant duty cycle.
ωrotor R η=
(tON f ) Vi
2
× 100%
λ= (2) (4)
Vwind RL I i
Where Vwind is the speed of the wind at turbine hub An Insulated Gate Bipolar Transistor (IGBT) is used as
height. In order to maximize Cp we vary the load on the the switching element of the circuit. It was chosen
turbine to achieve a rotor speed ωopt of: because of its ability to handle a higher current as well as
λoptVwind its fast switching speed[8]. The IGBT gate is controlled
ωopt = (3) by an IGBT driver. The duty cycle of the driver is
R determined by the microcontroller, according to the
The optimum tip-speed ratio λopt is a constant values of the sensing elements in the circuit. To enhance
dependent on turbine blade design for fixed blade the switching time of the circuit, the freewheeling diode
turbines. Thus for every wind speed there will be a was replaced by a schottky diode[9]. All power
different rotor speed for maximum power production. components were sized appropriately for 1.2 kVA
Fig. 3 shows the ideal power produced as a function of operation.
rotor speed for several different wind speeds. The goal of The sensing resistors of the buck circuit are used to
our controller is to adjust the rotor speed to operate at its measure the output power of the system. A very small
maximum power point for each wind speed [6]. resistor in series with the load is used to measure the
output current of the circuit. In addition, a voltage divider
III. THE DESIGN OF THE BUCK CIRCUIT is used to normalize the value of the output voltage into a
The basic elements of the buck circuit are shown in value between 0 & 5 V for reading by the controller. The
Fig. 4. It is composed of single switching element. This final buck circuit is shown in Fig. 5.
makes it simple and more efficient compared to other Varying the duty cycle of the switching element
DC-step-down choppers. The buck circuit provides an changes the wind turbine loading level. Zero duty cycle
average output voltage (Vo) that is less than the average means that the switching element is always OFF and no
load is connected to the wind turbine. Thus the maximum
rotational speed will be achieved when the duty cycle is
zero. Increasing the duty cycle decreases the output
impedance of the system, which adds more load to

Fig. 2. The variation of the power factor with tip-speed-ratio

Fig. 4. The basic elements of a buck circuit


Vwind 1 > Vwind 2 > Vwind 3

{
{

Fig. 3. Wind turbine characteristics


Fig. 5. The final schematic of the buck circuit
5th International Conference-Workshop - CPE 2007
the wind turbine; thereby slowing the rotor. The rotor IV. BRAKING SYSTEM
reaches the minimum speed when duty cycle is one.
Some environmental conditions, for instance
A simple model suffices to determine the relationship hurricanes, include very high wind speeds. This may
between the duty cycle of the dc-dc converter and the result in runaway and destruction of the wind turbine.
expected shaft speed. Assuming a constant power input
and efficiency the output electrical power I²R should also Additionally, it may be a necessary to stop the wind
be constant. The resistance is essentially an average turbine spinning, if there is a mechanical or electrical
between the connected load resistance R (weighed by the fault or even during preventive maintenance. Thus some
switch ‘on’ time) and an infinite resistance (weighed by kind of braking system for the wind turbine is required.
the switch ‘off’ time), thus the effective average resistance Many turbines include a mechanical braking design
is R/D where D is the duty cycle. For a Permanent Magnet (i.e. furling or tilting the axis of the turbine away from
(PM) DC motor or generator such as ours, the relationship incoming wind). Alternatively, frictional beaks may be
between the generator current and rotor speed is linear incorporated into the wind turbine. The disadvantage of
decreasing from the stall current Istall at rotor speed ω = 0, such system is the short life expectancy and additional
to zero at the unloaded speed ωNL. Mathematically we system cost and complexity [4] and [10]. Another option
would write: is to provide electrical braking. Under high wind
 ω  conditions heavy dump loads may be connected directly
I = I stall − I stall   (5) to the generator to slow the turbine enough causing the
 ω NL  blades to “stall” aerodynamically. Once stalled, the
Knowing that I² times the effective resistance R/D is torque produced by the turbine is greatly reduced, and it
constant allows us to write: spins at a very slow speed [4].
2 The braking system design is shown in Fig. 7. It
 ω  R consists of a relay that will connect the voltage of the
2
I stall  1 −  =C (6)
generator to a dump load upon the triggering of its coil.
 ω NL  D In sufficiently strong winds the generator output will
exceed a pre-set trigger level (i.e. breaking threshold).
Where C is a constant. Rearranging to solve for the rotor Even while charging the nominal load, the controller
speed ω we have: measures the generator voltage via a voltage divider into
0.5 analogue input channel. When the generators voltage
 CD 
ω = ω NL − ω NL  2 
 (7) exceeds the breaking threshold, the breaking relay is
 I stall R  activated, connecting a high power, low resistance dump
load directly to the generator output.
The purpose of adding anti-shock inductor in series
This shows that the rotor speed will decrease as the square with the dump load and the generator voltage is to avoid
root of the duty cycle. The factor ωNL (C / I2stall R)0.5 arcing at the contacts.
represents the speed decrease from ωNL at a duty cycle of
unity. If we assume that the load can completely stall the Stalling the wind turbine to avoid run away is the
rotor, i.e. ω = 0 at D = 1, then we can rewrite the primary purpose of the breaking system. It can also be
relationship as: used to avoid the over charging the load (i.e. batteries)

ω = ω NL (1 − D 0.5 )
when the output voltage of the buck circuit is too high.
(8) The presence of different dump loads is helpful in the
This relationship is shown graphically in Fig. 6. partial reducing the turbine power to the load by
converting it to heat in the dump loads. Alternatively
some useful loads (such as pumps) can be used as dump
1
loads, only being actuated at high wind speeds. The
0.9
response of the braking circuit is shown in Fig. 8.
0.8
Breaking is initially overridden manually, and the
Normalized Rotor Speed ω

0.7

0.6
generator is allowed to run over the breaking threshold
0.5
limit, which for this test was set at 11.8V. At
0.4
approximately 7 seconds breaking is re-enabled, and the
0.3 controller immediately goes into breaking mode, reducing
0.2 the rotor speed to less than 30 rpm. At intervals of
0.1 approximately 10 seconds the controller releases the
0 breaks allowing the turbine to spin up until the threshold
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Duty Cycle
is again exceeded, and breaks are reapplied. This
continues until the wind speed is reduced below the
breaking threshold, and normal operation ensues.
Fig. 6. Rotor speed versus duty cycle for an ideal DC-DC converter and
load
5th International Conference-Workshop - CPE 2007

Fig. 7. The schematic of the braking circuit

14

12
Generator Voltage (Volt)

10

0
0 10 20 30 40 50
Tim e (s )

Fig. 8 Braking system response at a threshold of 11.8 V


Fig. 9. Program flowchart

V. THE MICRO-CONTROLLER Some of the variables are initialized in the beginning of


The micro-controller is responsible for measuring the the program as following:
input voltage and the output power of the wind turbine, - Duty Cycle: 7%
and then making decisions based on the readings. The - Step size: +7, positive sign means increase
controller used for this project is PIC16F818. - Read initial value of the power on the first run of
The important features of this controller are the built-in the controller
PWM generator and five analog-to-digital converter - Set the breaking threshold voltage.
channels. The frequency of PWM is held constant, but the
duty cycle is controllable either automatically or manually VI. TEST BED
depending on the running mode of the program. Our controller is being designed for implementation on
The sequence of the program operations is shown in a fixed-blade horizontal-axis wind turbine in Mozda, India
Fig. 9. The controller was programmed to operate in two [3]. Lacking direct access to the target wind turbine we
different modes: Test Mode and Auto-Run Mode. The built a test bed which has a similar characteristic to that of
user can select between them via a switch. Under the test the wind turbine, though at lower power levels.
mode, the duty cycle is varied manually through two push The test bed is composed of DC motor that drives two
buttons. One button is used to decrease the duty cycle and phase AC generator. The output AC voltage is rectified
another is used to increase it. The purpose of this mode is and filtered using a diode bridge and a capacitor. The
to allow flexibility for taking data. output DC voltage of the test bed is simulating the output
The auto-run mode is used in normal operation. In this DC voltage of the wind turbine. DC power supply which
mode the duty cycle is continuously incremented by an feeds the DC motor simulates the wind speed. The
amount Step, which may be either positive or negative. differences between our test bed and the wind turbine are
Initially the duty cycle is set low with a positive Step. At shown in schematically in Fig. 10.
each cycle of the program execution the output power is
measured and compared to the previous cycle’s power. If
the current power value is less than the power at the
previous cycle, then the sign of Step is reversed. In all
cases, the duty cycle is then incremented by the value of
Step, which may actually be negative, thereby decreasing
the value of the duty cycle.
This “gradient” method will result in operation around
the peak of the power curve, and will also allow quick
adaptation to varying wind speeds without measuring
either wind speed or turbine RPM directly.
Fig. 10. Differences between the test bed and the turbine
5th International Conference-Workshop - CPE 2007
Several tests were performed on the test bed. First, it 2500
was necessary to ensure that the test bed has a single
maximum in the power curve at a constant supply voltage.
Secondly, we wanted to verify that the peak in the power 2000

curve changed as a function of input power. Fig. 11 and

Rotor Speed (RPM)


Fig. 12 show the difference in behavior at different power 1500
supply conditions and they prove the presence of distinct
peaks at different simulated wind speeds. 1000
The initial testing of the DC-DC controller was aimed
at establishing the relationship between duty cycle and 500
rotor speeds at a constant load resistance and supplied
current to the DC motor, i.e. simulated wind speed. As can
0
be seen in figure 13, varying the duty cycle has a strong 0 0.2 0.4 0.6 0.8 1
influence on the rotor speed of the test bed. At the tested Duty Cycle
load and supply configuration the rotor speed was reduced
from 2250 rpm at 5% duty to 515 RPM at a duty cycle of
85%. Fig. 13. The effect of the Duty cycle on the rotor speed in the test bed

direction is continued for several program cycles until the


VII POWER TRACKING TESTING peak power is passed at around 1000 RPM. The controller
The controller was tested on the test bed to prove its then wanders back and fourth about this peak power
ability to track the maximum power. Two tests were done output.
initially using the auto-run mode to prove the ability of the Fig. 15 shows another run beginning at very low duty
controller to track the power starting from different values cycle. As the test bed is lightly loaded it operators at high
of the duty cycle. The duty cycle in the first test, shown in speed (2300 RPM), because the duty cycle is low the
Fig. 14, starts from the highest duty cycle. At 100% duty output voltage is also low (around 7V). As the duty cycle
cycle with the given load and simulated wind speed the is increased the test bed is loaded, slowing down, and the
rotor is turning at less than 500 RPM. The circuit output voltage increases. Again this continues until the
decreases the duty cycle, allowing rotor to speed up and system is operating near the peak power output at
increasing the output voltage, and thus power. This approximately 1250 RPM, 13.5V. Some scatter in the
output data can be seen. This is due to the mechanism
0.14 Peak Power 0.7 system dynamics. Once-the duty cycle is changed the
Point electrical output increases nearly instantly. The increased
0.12 0.6 load does not immediately change the rotor speed,
however, due to rotor inertia. Thus each step up in duty
Output Power (Watt)

0.1 0.5
Output Current (A)

cycle results in an ‘arc’ up (as voltage output increasing)


0.08 0.4 and to the left (as the system gradually slows to a new
steady state operating point). Similarly decreases in duty
0.06 0.3
cycle will result in arcs down and to the right. As data was
0.04 Linear – Output Current 0.2 taken continuously during operation, these ‘arcs’ then
Poly – Output Power
deviate from the actual steady-state operating curve.
0.02 0.1
Fig. 16 shows the system response to varying simulated
0 0 wind speeds. Initially the system is started at low duty
0 5 10 15 20 25 cycle (the nominal start condition) and it quickly stabilizes
Output Voltage (V)
at the optimum power point near 1200 RPM and 13.5V
output. The DC motor is then given less power simulating
Fig. 11. Test Bed response for variable resistive load at low simulated a second wind speed and the system again stabilizes as the
wind speed
new peak power, now 900 RPM and 12V. Finally, a
lower wind speed is tested and the system finds a peak
Peak Power power at 500 RPM and 10.2 V.
0.12 Point 2
1.8 14
0.1 1.6
O utput Pow er (Watt)
O utput Current (A)

12
1.4
0.08
Output Voltage (V)

1.2 10
0.06 1
8
0.8
0.04 6
Linear – Output Current 0.6
Poly – Output Power 0.4
0.02 4
0.2
2
0 0
0 10 20 30 40 50 60 0
Output Voltage (V) 0 500 1000 1500 2000
Rotor Spe ed (RPM)

Fig. 12. Test Bed response for variable resistive load at higher simulated Fig. 14. Auto Power tracking starting from high duty cycle at constant
wind speed. load
5th International Conference-Workshop - CPE 2007
16 100

14 90
12 80
Output Voltage (V)

10 70
8

Efficiency %
60
6
50
4
40
2
30
0
20
0 500 1000 1500 2000 2500
Rotor Speed (RPM) 10

0
Fig. 15. Auto Power tracking starting from low duty cycle at constant 0 20 40 60 80 100
load Duty Cycle

14 Fig. 17 The change in the efficiency versus duty cycle at constant load
and constant frequency
V wind1
12 should have parts cost of less than $30 an output
efficiency of 70-80% and be capable of power up to 1.2
10
V wind3
V wind2 kW.
Output Voltage (V)

8 IX FURTHER WORK
6
The circuit will continue to be tested and characterized
V wind1 >V wind2 >V wind3 on a high-power test bed. A final weatherized version will
4
be instrumented for performance verification and
implemented on the target turbine in Mozad, India.
2
APPENDIX A COST OF THE COMPONENTS
0
Table 1 Approximate prices of electrical components
0 500 1000 1500 2000
Rotor Speed (RPM)
Item Function Price (USD)
Fig. 16. Auto power tracking at different simulated wind speeds and
High current N-
constant load type IGBT. The
IRG4PC50U 7.9
switching element
Fig. 17 shows the efficiency as a function of duty cycle. of the buck circuit.
The efficiency varies non-linearly with duty cycle at NPN transistor to
constant load, and constant PWM frequency. The inductor transfer the braking
has a significant effect on the efficiency, because of its BU810 pulse from the 2.09
influence on the output current [9], and it will suffer I2R micro-controller to
losses as well. The change in the efficiency can be divided the relay.
into three stages. At very low duty cycles, less than 20%,
the discharge time of the inductor is less than the OFF IR2117 IGBT gate driver 2.06
time of the pulse width modulation signal. This results in a PIC16F818 PIC microcontroller 1.72
discontinuity of the output current and lost efficiency. PNP transistor. To
The discontinuity time decreases with the increase of transfer PWM
the duty cycle, because the OFF time of the pulse width BC108 0.82
signal to the IGBT
modulation gets smaller. Above 20% duty cycle the driver
inductor current never reaches zero because the discharge To filter the output
time of the inductor becomes greater than the IGBT OFF 470uF, 200V
voltage of the Buck 2.35
time. From about 20% to 80% the buck circuit is fully capacitor
circuit
performing as step-down-converter with an average
efficiency of approximately 76%. Above 80% duty cycle To filter the output
the efficiency jumps to 90% because the OFF time of Inductor current of the buck 5.5
PWM is insufficient to drive the IGBT fully OFF. In this circuit
case the DC-DC converter begins to act as low impedance Schottky diode for
MBR1535CT 2.5
resistive link between the generator and the load. the buck circuit
L7800 Voltage regulator 0.47
VIII CONCLUSION Miscellaneous Resistors, switches,
3.5
A simple, low cost wind turbine controller has been Components etc…
developed. Its peak power tracking capability and
electrical braking function have been verified on a test bed
simulating various wind speeds. The resulting controller
5th International Conference-Workshop - CPE 2007
APPENDIX B THE PARAMETERS OF MOZDA HAWT
- 48 V, 1.2 kVA, Y connected three phase AC
generator
- The rotor has 12 poles
- Stator has 9 coils, each coil made from 1.1 mm
diameter enameled copper.

ACKNOLEDEMENTS
We would like to thank the Engines and Energy
Conversion Laboratory at Colorado State University for
their support of this project.

REFERENCES
[1] F. Wien, (2006, Nov.). Distribution Generation and
Renewable Wind Power. KEMA Nederland BV. [Online].
Available:
http://www.cda.org.uk/megab2/elecapps/PQ_Guide/832-
wind-power.pdf
[2] G. E. van Baars and Peter M. M. Bongers, "Wind turbine
control design and implementation based on experimental
models" in Proc. 1992 Decision and Control Conf., pp.
2454-2459.
[3] R. Werther and M. Pitterle, "Field Installation of a
Horizontal Axis Wind Generation in India". [Online].
Available: http://carbon.cudenver.edu
[4] K. Rajambal, B. Umamaheswari, and C. Chellamuthu.
(2005, Jun.). Electrical Braking of Large Wind Turbines.
[Online]. Available: http://www.sciencedirect.com
[5] G. L. Johnson, Wind Energy Systems. Prentice Hall,
Englewood Cliffs, 1985.
[6] M. Simones, B. K. Bose, and R. J. Spiegel, "Fuzzy Logic
Based Intelligent Control of a Variable Speed Cage Machine
Wind Generation System," IEEE Trans. on Power
Electronics, vol. 3, pp. 87-95, Jan. 1997.
[7] M. H. Rashid, Power Electronics Circuits, Devices, and
Applications, Prentice hall, Upper Saddle river, New Jersey
1993.
[8] Application Note AN-990: Application Characterization of
IGBTs. International Rectifier. [Online]. Available:
www.irf.com/technical-info/appnotes/an-990.pdf
[9] D. Schelle and J. Castorena (2006, Jun.). Buck-Converter
Design Demystified. Maximum Integrated Products,
Sunnyvale, Cliff. [Online]. Available:
http://www.powerelectronics.com
[10] G. Sterzinger, M. Svrcek. (2004, Sep.). Wind Turbine
development: Location and Manufacturing Activity.
Renewable Energy Policy Project (REPP). [Online].
Available: http://www.crest.org
[11] J. P. Agrawal, Power Electronic Systems Theory and Design,
Prentice Hall, Upper Saddle River, New Jersey 2001.

You might also like