Furling Control

Wind Energy Program Wind Energy Program
Modeling and Control of a

Bergey-Type furling Wind
Turbine
Dr. Marwan Bikdash
Department of Electrical Engineering
North Carolina A&T State University
Greensboro, NC 27410
7/15/00 M. Bikdash 1
Objectives
• To study the overspeed-protection furling mechanism
of the Bergey Wind Turbine and
– Model the furling mechanism (for real-time)
– Effect of aerodynamics, generator, electrical side
• Build a simulation package
• To design an active yaw mechanism for larger
turbines (and improve performance)
– Control Objectives
– Control Laws and Algorithms
– Actuation mechanism
July 14-15 M. Bikdash 2

Schedule And Status:

• Completed the Lagrangian derivation of EOM of
furling mechanism
• Analyzed YawDyn Output
• Approximated YawDyn data using fuzzy models
(real-time; derivatives for linearization)
• Analyzed steady-state behavior of turbine
• Studying actuation choices for active mechanism
• Concurrently writing software in MATLAB and
Simulink
• Designing control laws

• Bergey Excel 10KW Turbine Bergey Wind Turbine

• Operates in the boundary layer of
the Earth (turbulence)
• Tail always “points” in the average
direction of the wind
• Uses a passive auto-furling
mechanism to protect
turbine/generator combination from
overspeeding in high winds
• Low maintenance/fail safe
• New 40KW turbine envisaged: Is a
(still fail-safe) autofurling
mechanism desired?
• Use furling to achieve protection
AND optimum performance.

Start-up Wind Speed: 3.4 m/s (7.5 mph)

Cut-in Wind Speed: 3.1 m/s (7 mph)
Specifications of the
Rated Wind Speed: 13 m/s (29 mph) 10 KW Bergey Wind
Rated Power: 10,000 Watts
Cut-Out Wind Speed: None Turbine
Furling Wind Speed: 15.6 m/s (35 mph)
Max. Design Wind Speed: 53.6 m/s (120
mph)
Type: 3 Blade Upwind
Rotor Diameter: 7 m (23 ft.)
Blade Pitch Control: POWERFLEX®
Overspeed Protection: AUTOFURL
Drive: Direct
Temperature Range: -40 to +140 Deg. F
Generator: Permanent Magnet Alternator
Output Form: 3 Phase AC, Variable
Frequency (Regulated 48 - 240 VDC after
VCS-10 or 240 VAC, 1Ø, 60 Hz with
Powersync® inverter)

θ
C is c.o.g
North Furling angle
v
between vertical planes
positive CCW
Tail Angle between
L3 r OT
vertical planes α
L4 blades v
α B
b2
L1 Rotor
A v Axis
Yaw axis offset L2 θ b1
v
View for figure 2

a2
A-Frame is inertial and points
in the average direction of wind θw
v
a1
Top View of Turbine
Before Tilting Wind makes angle θ w
with A-Frame

v ure 3
v
c3 fig
Hinge Axis about which c2 w for
Vie
the tail furls
v
β b3
v
(furled) b2
ed) Yaw Axis

furl
(un
l
Tai Side view

Other Views
v
v
Vie
d3
wf
d3 Tai
v v
or f
l
igu
d1 d2
re 4
γ
ψ
ψ
rotor
Hinge Axis about
which the tail furls
v Padded stops
c3
ω
v Yaw axis Top View Along
c1
Hinge Axis
Front View Along Rotor

Notation

Notation (Continue)

Notation (Continue)

Coordinate Frames

Drag and Lift on Tail
assumption
assumption

Potential and Kinetic Energies
From yaw
axis to
aerodynamic
center of tail
Rotational speed of tail

with respect to inertial A
Frame
Kinetic Energy

Lagrange’s Equation of
Motion

Generalized Force for Furling Angle
assumption

Tail Aerodynamic Furling Moment
assumption
Depends on
the angle
between tail
and wind

assumption

Tail Aerodynamic Yaw Moment
as
su
mp
tio
n Depends on
the angle
between tail
and wind as
well as yaw
angle

Tail Aerodynamic Moments

Equations of Motion For

Furling Mechanism
assumption

General Wind Turbine Model

Wind Speed V Relative wind
direction −θ w
Aerodynamics
∆θ
(YawDyn ⇒ Fuzzy)
Mechanical rpm
Torque
Moments, Generator Electrical

Thrusts
Load
Controller
Yaw/Furling Dynamics
θ
Building the Simulation Software

1. Use the Lagrangian to derive the basic equations of
motion of furling mechanism
2. Run YawDyn to characterize the turbine power
output and the corresponding aerodynamic forces
on turbine (as a function of wind speed, rpm, and
angle between wind and rotor
3. Obtain fuzzy approximation
4. Add friction and stop terms
5. Add generator (Simulink) and electrical load/power
electronics model (Simulink? Sabre?)
6. Add/modify dynamics for actuator
7. Add model for disturbances (turbulence?)
Generator Model
• Intersection of turbine Torque-Speed characteristics
with that of generator yield equilibrium point at onset
of furling
• This curve depends on the electrical load seen by
the generator
• Resistive load (interesting?)
• Battery charger (needs work)
• General electrical load with power electronics
control
• through power electronics
• Assume utility grid interface controlled to always
draws rated power at rated speed
Reduced Efficiency After Furling

Output power(KW)
The generator output power vs. wind speed (Bergey data)

15 After furling,
Power limiting with synchronous Turbine operates at
10 inverter
significantly less
5
efficiency.
0
0 5 10 15 20 25 30 35 40 45
wind speed(mph)
In this range of wind speeds, Turbine

furls and unfurls constantly. This
“hunting” behavior, while consistent with
the over-speed protection function, is
undesirable.

Hunting Behavior
Tower thrust VS. wind speed
Tower thrust in LBS
2500
Design thrust load
2000
1500
1000
500
0
0 20 40 60 80 100 120 140
Wind speed in mph
Hunting
We can use tower thrust

experimental data to approximate
turbine thrust as a function of wind
speed or rpm.

YawDyn Software
• Complexity of HAWT aerodynamics mathematical
models
• Once the geometry of the turbine and blades are
specified, the aerodynamic forces and moments of the
turbine can be computed using YawDyn.
• YawDyn is a software package developed by C.
Hansen and his colleagues at the University of Utah.
• Numerical Computations of Aerodynamic forces
expensive numerically.
• Example YawDyn calculation for a 40 kW Bergey-type
turbine.

YawDyn Power Curves

0.6
• 40 kW wind
0.5
turbine
Power Coefficient
0.4 • at 10 degrees
40 wind angle
0.3
• different rpms
0.2 60
0.1
80
0
0 2 4 6 8 10 12 14 16 18 20
Tip Speed Ratio

Different (Yaw -wind) angles

0.6
0.5 10 • 40 KW
Power Coefficient
20 Turbine
0.4 • At 60 rpm
30
0.3 • Dtheta =
angle
0.2 between
0.1
wind and
rotor
0 0 5 10 15
Tip Speed Ratio

Power Curves at different rpms
250
• Power levels
RPM = 100
off for every
Power generated in kW
200
rpm as wind
150
RPM = 80 speed
100
increases
RPM = 60 • Power
50 increases
RPM = 40
with rpm
0 RPM = 20
• Maximum
-50
0 5 10 15 20 25 30 35 40
allowable
Wind speed m/s rpm

Fuzzy Approximation of YawDyn
• YawDyn is computationally expensive for real-time

simulations
• Alternative: To generate (learn) a fuzzy inference
system (FIS) that approximates YawDyn.
• Gathering a sufficient amount of input/output data to
use for the learning process.
• Gathering made possible by a MATLAB interface
that allows running several iterations of YawDyn
with out user interaction.

MATLAB Interface
Wind Velocity
Wind Angle OUTPUT

YawDynVB.m FixOut.m yawdyn.plt
RPM
Wind.m rmp.m YawDyn
yawdyn.ipt
yawdyn.wnd

Architecture of Proposed FIS

• Several FIS architectures can be used to
approximate YawDyn.
• Direct approach: Use wind speed, relative wind
angle, and rpm as inputs.
• Simplest Approach: Use power coefficient—CP
versus the tip speed ratio—TSR.
• Selecting a specific approach includes a trade
off between learning time and complexity and
approximation quality.
• The next slides will show the implemented FISs.

Two Fuzzy Inference Systems

Wind Speed Rotor Power
Relative Wind Angle GeneralYaw Nacelle Yaw Moment

6 MFs / Input Lateral Hub Force
864 Coefficients
RPM 9388 Training Points Rotor Thrust
Relative Wind Angle

CpvsTSRTheta
6 MFs / Input Coefficient of Power
Tip Speed Ratio 108 Coefficients
9388 Training Points

YawDyn vs. Sugeno GernalYaw

Using random samples

YawDyn FIS
Vwind ∆θ rpm Protor P̂rotor Run Time
(sec)
Run Time
(sec)
35 31 35 17.95 18.06 4.306 0.110
29 50 21 6.744 6.732 3.185 0.130
17 11 34 9.825 9.848 4.427 0.100
29 18 61 57.33 57.30 7.361 0.110
12 41 48 21.80 22.04 7.691 0.110
23 53 82 83.08 83.37 11.28 0.130
18 17 45 22.88 23.11 6.209 0.130

Performance of FIS2
) YawDyn FIS
∆θ TSR PC PC Run Time Run Time (sec)
(sec)
13 1.2732 0.0123 0.0129 7.400 0.0710
1 2.7322 0.1196 0.1209 16.35 0.0100
56 3.2620 0.0924 0.0929 11.71 0.0200
48 0.7716 0.0051 0.0053 3.485 0.0100
16 0.9502 0.0062 0.0059 3.905 0.0200
44 4.8564 0.2233 0.2155 7.351 0.0100
27 1.0104 0.0076 0.0076 5.398 0.0100
for random test input samples

Fuzzy Linearization
• Use the Interpretable Sugeno Approximator (ISA)
• Make its consequent polynomials rule-centered.
• A typical ISA rule has the form:
Rk: If x1 is A1 and x 2 is A2 and… xn is An Then
u k = b0k + b1k ( x1 − r1k ) + b2k ( x2 − r2k ) + ... + bnk ( xn − rnk )
• rk is the rule center (centers of the membership

functions of all inputs tested by the kth rule.)
• If the membership functions are chosen properly
(local and differentiable everywhere),
• the b coefficients can be interpreted as Taylor
series coefficients.

Interpretation of Coefficients
Some coefficients
of the Sugeno
engine can be
interpreted as the
derivative of the
function at rule
centers. In this
case, these are
aerodynamic
sensitivity
derivatives.

Conclusions ANNIE 99
• Fuzzy approximation techniques were used to

derive FISs that approximate YawDyn.
• Excellent approximation
• FIS with fewer inputs learns faster
• More general FIS approximates better
• 1 to 2 orders of magnitude speed up in
computation
• Using ISA yields estimates of derivatives as
well to obtain linearized models

Moment Curves for 40KW

5
Dtheta = 10 • Nacelle
4
degrees and Nacelle Yaw Moment is
Moments
3 RPM = 60 dominant!
• Hence
2 Rotor Thrust specifying
1 furling wind
speed is not
0 Lateral Hub Force easy
-1
0 5 10 15 20 25 30 35 40
Wind Speed

Vwind = [7 10 13 16 19] m/sec, 80 rpm

Power in KW Due to Nacelle
100 20
V 10 Yaw
50 Moment of
0
Starts negative Nacelle
0 -10 becomes
0 20 40 60 80 0 20 40 60 80
dominant
Total Aerodynamic Yaw Due to thrust as the rotor
20 moves
4 in KNm
10 away from
0 2 wind.
-10 0
0 20 40 60 80 0 20 40 60 80
Angle between wind and rotor (degrees)

Equilibrium Conditions
Generator Torque-Speed Characteristics

(dependent on electrical load and power electronics)
3 Equations
Unknowns

Onset of Furling

psi=0 L1=0.2032
Torque in kNm
16 15
14
10
Onset
wind speed in m/s
12
5
of
10
8 0
Furling.
6 8 10 12 14 16 6 8 10 12 14 16
Dtheta in deg No
Omega in rad/s
12 12
10
Omega in rad/s
10 control.
8 8
6 6
4 4
6 8 10 12 14 16 2 4 6 8 10 12
Torque in kNm
120 150
Power in kW
100
80 Power in kW 100 Any
60 50 generator
40
torque-
20 0
6 8 10 12
Dtheta in deg
14 16 4 6 8 10
Omega in rad/s
12 speed
Curve
16 10
14 Torque in kNm
8
wind
12 6
speed
in m/s 10 4
8 2 8 Dtheta
4 6 8 10 12 14 16 10 12 14 16 in deg
12 12
Omega 10 10
in 8 8
rad/s 6 6
4 Torque
6 8 10 12 14 16 4 2 4 6 8 10 12 in kNm
120 120
100 100 Power in kW
Power 80 80
in kW 60 60
40 40 Omega
20 20 4
6 8 10 12 14 16 6 8 10 12 in rad/s
Dtheta
In deg Utility grid interface drawing rated
power at any wind speed

Design Yaw Control for constant 40 KW at

any wind speed

Assuming control moment at the yaw
Wind Energy Program axis only Wind Energy Program
80
In degrees Yaw = angle between rotor and wind
60
40 Furling = angle between tail and rotor
L1=0.2032, Lac=4.728
20
0 8
Yaw-Furling ≈angle between wind and tail
10 12 14 16 18 20
40 Furling Curves obtained
Torque (KNm)
30 under the assumption of
P(KW); Om(rad/s)
20 constant rated (40KW) power
10 delivered to electrical
0 8 10 12 14 16 18 20
generator under rated speed.
15
YawDyn -Tail -Control Yaw Moments in KNm.
10 • Tail Moment essentially constant!
• YawDyn: Rotor thrust + Nacelle +
5 Lateral
• Difference must be picked up by Yaw
0 8 10 12 14 16 18 20 Control Torque (Acting on nacelle only)
V wind in m/s

Design Yaw Control for any desired

schedule of Power and rpm

80
Yaw
60
40 Furling
10%
20
Yaw-Furling
allowable
0 8 10 12 14 16 18 20
Overspeed
1.4
In per unit
1.2 Speed
1 P
Maintaining 40KW output but
Torque with 10% allowable
0.8 8 10 12 14 16 18 20 overspeed requires 27% less
10 Yaw control
-Tail
5
Yaw Moments
YawDyn -Control in KNm
0
8 10 12 14 16 18 20
V wind in m/s
Class of Actuators

Linear Motor
Geometry
F C
Q
P
B: the tail hinge
A: the yaw axis

Conclusions • Aerodynamics of tail?

• Simulation software in • Generator/Load/Power
MATLAB and Simulink Electronics? (d-q
• Equations of motion for Simulink model)
furling mechanism • Analysis of equilibrium
derived • onset of furling
• Reliance on YawDyn • scheduled
• Hybrid fuzzy crisp power/rpm as
modeling for real-time function of Vwind
simulations
• linear motor
• Nacelle moment
dominant in furling • Linearization? Stability?
behavior. Reshape? Hopf bifurcation?

Furling Control

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Furling Control

Diunggah oleh

Hak Cipta:

Format Tersedia

Wind Energy Program Wind Energy Program

Modeling and Control of a

July 14-15 M. Bikdash 2

Schedule And Status:

July 14-15 M. Bikdash 3

• Bergey Excel 10KW Turbine Bergey Wind Turbine

July 14-15 M. Bikdash 4

Start-up Wind Speed: 3.4 m/s (7.5 mph)

July 14-15 M. Bikdash 5

View for figure 2

July 14-15 M. Bikdash 6

ed) Yaw Axis

July 14-15 M. Bikdash 7

July 14-15 M. Bikdash 8

July 14-15 M. Bikdash 9

July 14-15 M. Bikdash 10

July 14-15 M. Bikdash 11

July 14-15 M. Bikdash 12

Drag and Lift on Tail

July 14-15 M. Bikdash 13

Potential and Kinetic Energies

Rotational speed of tail

July 14-15 M. Bikdash 15

July 14-15 M. Bikdash 16

Generalized Force for Furling Angle

July 14-15 M. Bikdash 17

Tail Aerodynamic Furling Moment

July 14-15 M. Bikdash 18

July 14-15 M. Bikdash 19

Tail Aerodynamic Yaw Moment

July 14-15 M. Bikdash 20

Tail Aerodynamic Moments

July 14-15 M. Bikdash 21

Equations of Motion For

July 14-15 M. Bikdash 22

General Wind Turbine Model

Moments, Generator Electrical

Building the Simulation Software

Reduced Efficiency After Furling

The generator output power vs. wind speed (Bergey data)

In this range of wind speeds, Turbine

July 14-15 M. Bikdash 26

We can use tower thrust

July 14-15 M. Bikdash 27

July 14-15 M. Bikdash 28

YawDyn Power Curves

July 14-15 M. Bikdash 29

Different (Yaw -wind) angles

July 14-15 M. Bikdash 30

Power Curves at different rpms

July 14-15 M. Bikdash 31

Fuzzy Approximation of YawDyn

• YawDyn is computationally expensive for real-time

July 14-15 M. Bikdash 32

Wind Angle OUTPUT

Wind.m rmp.m YawDyn

July 14-15 M. Bikdash 33

Architecture of Proposed FIS

July 14-15 M. Bikdash 34

Two Fuzzy Inference Systems

Relative Wind Angle GeneralYaw Nacelle Yaw Moment

Relative Wind Angle

July 14-15 M. Bikdash 35