Boris Fischer, Fraunhofer IWES

© Fraunhofer IWES

Reducing rotor speed variations of floating wind turbines by compensation of non-minimum phase zeros (NMPZ)Boris Fischer, Fraunhofer IWES

PI

P

10-2

10-1

100

101

10

20

30

40

50

60

ma

gn

itu

de

(d

B)

10-2

10-1

100

101

-540

-360

-180

0

180

frequency (rad/s)

ph

as

e (

de

g)

lin. model, w/o complin. model, with comp

FAST, w/o compFAST, with comp

© Fraunhofer IWES, B. Fischer 2

Strategy of the pitch control system

increasedwind speed

pitch-to-feather

nacellemovesupwind

In the region above rated wind speed controller might introduce negative

damping instability strategy: “hide” nacelle motion from the

controller by proper closed-loop bandwidth

reducedrotor thrust


Onshore vs. floating

In the region above rated wind speed closed-loop bandwidth < lowest relevant

structural frequency onshore: first tower bending mode floating: first tower pitching mode bandwidth significantly lower

Consequently, increased variation of the rotor speed in (Larsen and Hanson, 2007):

±10% onshore ±30% floating

Many DFIG-designs max. rotation speed variation = ±30%

bending mode (~0.3 Hz)

rigid body mode (~0.03 Hz)


Transfer function from blade-pitch to generator speed

ngeneratorbblade

ngenerator(s) = G(s) · bblade(s)

PI

ndesired

G(s)

s

jw

: zero

: pole

natural frequency of the platformpitch mode


ngenerator(jw) = G(jw) · bblade(jw)

10-2

10-1

100

101

10

20

30

40

50

60

ma

gn

itu

de

(d

B)

10-2

10-1

100

101

-540

-360

-180

0

180

frequency (rad/s)

ph

ase

(d

eg

)

NMPZ limit the closed-loop bandwidth

ngeneratorbblade

G(jw)

PI

ndesired


10-2

10-1

100

101

10

20

30

40

50

60

ma

gn

itu

de

(d

B)

10-2

10-1

100

101

-540

-360

-180

0

180

frequency (rad/s)

ph

as

e (

de

g)

10-2

10-1

100

101

10

20

30

40

50

60

ma

gn

itu

de

(d

B)

10-2

10-1

100

101

-540

-360

-180

0

180

frequency (rad/s)

ph

ase

(d

eg

)



Compensation of the NMPZ

ngenerator

Tgenerator vnacelle

bblade

PI

ndesired

P

Approach introduced by:[Leithead and Dominguez, EWEC 2006]

ngenerator(jw) = G(jw) · bblade(jw)


Discussion on the control method

multivariate control design methods “include” NMPZ compensation

approach is in line with common SISO practice

readily available sensor signal: nacelle acceleration filtering

constrain demand of generator torque

integration with the supervisory control system

careful assessment of loads, especially drive train

PI

ngenerator

Tgenerator vnacelle

bblade

ndesired

P


Simulation study:land-based designed pitch controller on a floating turbine

Benchmark system: NREL 5MW baseline wind turbine OC3-Hywind spar buoy Simulations with FAST &

Matlab/Simulink

Source: NREL

set-up Pitch controller

Generator controller

A onshore high bandwidth

const. power

B floating low bandwidth

const. torque

C floating + comp.

highbandwidth

const. torque + comp.


Example time series, 14 m/s mean wind speed

300 400 500 600 700 8005

10

15

20

25

time (s)

win

d s

pee

d(m

/s)

300 400 500 600 700 800900

1000

1100

1200

1300

1400

time (s)

gen

erat

or

spee

d(r

pm

)

w/o comp.with comp.

300 400 500 600 700 8000

5

10

15

time (s)

bla

de

pit

ch(d

eg)

300 400 500 600 700 80020

30

40

50

60

time (s)

gen

erat

or

torq

ue

(kN

m)

300 400 500 600 700 8000

2

4

6

8

time (s)

pla

tfo

rm p

itch

(deg

)

300 400 500 600 700 8002000

3000

4000

5000

6000

time (s)

shaf

t to

rqu

e(k

Nm

)

reduced rotorspeed variations

30% 10%

increasedshaft torque

300 400 500 600 700 8005

10

15

20

25

time (s)

win

d s

pee

d(m

/s)

300 400 500 600 700 800900

1000

1100

1200

1300

1400

time (s)g

ener

ato

r sp

eed

(rp

m)

w/o comp.with comp.

300 400 500 600 700 8000

5

10

15

time (s)

bla

de

pit

ch(d

eg)

300 400 500 600 700 80020

30

40

50

60

time (s)

gen

erat

or

torq

ue

(kN

m)

300 400 500 600 700 8000

2

4

6

8

time (s)

pla

tfo

rm p

itch

(deg

)

300 400 500 600 700 8002000

3000

4000

5000

6000

time (s)

shaf

t to

rqu

e(k

Nm

)


Fatigue load calculation

DLC1.2 (normal operation) 4, 6, 8, …, 24 m/s mean wind speed metocean data according to

[Jonkman, 2007] 5 random seeds 55 runs/config.

Results DELs from rainflow count are

similar, except for main shaft torsion (+50%)

drive train DELs from load duration distribution similar

shaft blade ip blade oop tower ss tower fa0

0.5

1

1.5

2

2.5

DE

L r

atio

flo

atin

g t

o la

nd

-bas

ed

General DELs from cumulative rainflow count

1.0

5

0.99

3

0.9

8

1.4

1

2.4

7

1.5

6

0.98

9

1.0

3 1.3

6

2.4

5

B to AC to A

torque bend. moment shear force0

0.2

0.4

0.6

0.8

1

1.2

1.4

DE

L r

atio

flo

atin

g t

o la

nd

-bas

ed

Drive train DELs from load duration distribution

0.994 1.02 0.991 1.01 1.02 0.991

B to AC to A


Conclusion

NMPZ compensation nacelle velocity & generator

torque increased bandwidth of the blade-

pitch control loop

Simulation study land-based designed pitch

controller on a floating turbine fatigue loads are similar,

exception: main shaft (DEL +50%) rotor speed variations reduced to

onshore values (30%10%)

PI

ngenerator

Tgenerator vnacelle

bblade

ndesired

P

10-2

10-1

100

101

10

20

30

40

50

60

ma

gn

itu

de

(d

B)

10-2

10-1

100

101

-540

-360

-180

0

180

frequency (rad/s)

ph

as

e (

de

g)




Thank you for your attention!

The HiPRwind project receives funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant

agreement n°256812

Boris Fischer, Fraunhofer IWES

Documents

Transcript of Boris Fischer, Fraunhofer IWES