Sie sind auf Seite 1von 7

ORIGINALARBEITEN

Elektrotechnik & Informationstechnik (2011) 128/5: 174–180. DOI 10.1007/s00502-011-0828-4

Doubly fed induction generators for wind turbines – modeling the steady state

G. Ofner, P. Handgruber, R. Seebacher

Doubly fed induction generators with frequency converters on the rotor side are well established for large wind power plants up to several MW. This technology concept represents some advantages compared to fixed speed generators such as variable reactive power control, higher energy yield and better power quality behaviour. Furthermore, only a fraction of the rated power has to be carried by the frequency converter, which reduces its size and cost and improves the efficiency. For the manufacturer of these generators it is important to predict with high accuracy the rotor current and voltage despite its saturation and non linear effects. Moreover, the power quality of such a grid connected plant is getting more and more important as the number of the installed plants is rapidly increasing and the grid codes are becoming stricter. That is why the prediction of the emitted harmonic current levels is of importance.

Keywords: doubly fed induction generators; induction generators; wind power generation; current harmonics

Doppelt gespeiste Asynchrongeneratoren fu¨ r Windkraftanlagen – Modellierung stationa¨ rer Betriebszusta¨ nde.

Doppelt gespeiste Asynchrongeneratoren erlauben in Verbindung mit Frequenzumrichtern den wirtschaftlichen Betrieb von drehzahlvariablen Windkraftanlagen der MW-Leistungsklasse. Im Vergleich zu drehzahlstarren Systemen bieten diese den Vorteil gro¨ ßerer Leistungsausbeute bei gleichzeitigem Beitrag zur Netzstu¨ tzung. Da der Umrichter nur einen Teil der Anlagenleistung fu¨ hrt, ko¨ nnen dessen Baugro¨ ße und Kosten reduziert werden. Die Vorhersage des Betriebsverhaltens anhand geeigneter mathematischer Modelle ist fu¨ r den Generatorhersteller von entscheidender Bedeutung. Diese Arbeit skizziert anhand eines einfachen Maschinenmodells den grundlegenden Berechnungshergang. Auf ein detailliertes Simulationsmodell fu¨ r stationa¨ re Betriebszusta¨ nde mit besonderem Augenmerk auf Sa¨ ttigungs- und Verlust-Erscheinungen wird ebenso hingewiesen. Die Simulationsergebnisse werden mit Hilfe von Messungen verifi ziert. Ferner werden auch Untersuchungen zur Spannungsqualita¨ t angestellt. Die von der gesamten Windkraftanlage erzeugten Oberschwingungsemissionen finden dabei besondere Beru¨ cksichtigung.

Schlu¨ sselwo¨ rter: doppelt gespeister Asynchrongenerator; Asynchronmaschine; Windkraftanlage; Stromdoerschwingungen

Received October 7, 2010, accepted March 10, 2011 Springer-Verlag 2011

1. Introduction In the last years, wind power acco unted for more newly installed electrical power generating capacity in the EU than any other technology ( Wilkes, 2010 ). Wind turbines with variable speed drivetrains offer the advantages of lower grid disturbances

(flicker) due to varying wind speed s, controllable reactive power, soft synchronisation to the grid and higher energy yields. Variable speed drivetrains can be built with synchronous generators, asyn- chronous generators (squirrel cage) or with doubly fed induction generators (DFIG). Systems with synchronous or squirrel cage generators require a frequency converter that is rated for the full power of the generator. Common systems with doubly fed in- duction generators require a frequency converter that is rated for only about 30% of the full power. This reduces the overall system costs. Although synchronous generators usually have better efficiency than induction generators, the system efficiency is comparable to synchronous generator systems because of the also reduced converter losses ( Mu¨ ller, Deicke, De Doncker,

2002 ).

Due to the annually increasing installed power of wind farms the power quality and the grid stability are of great significance. The grid

operator defines grid codes which regulate on the one side the plant behaviour in a fault case (e.g. low voltage ride through – LVRT) and on the other side the harmonic current emission.

174 heft 5.2011

© Springer-Verlag

2. Modeling of the doubly fed induction generator The use of an induction machine as a doubly fed generator results in a different approach to calculate an operating point of the machine compared to a common motor operation:

" the rotor speed is imposed by the wind turbine and the slip is much higher (range 0.3)

" the active and reactive power are given by the control system and the rotor current and voltage are imposed by a frequency converter.

For the machine designer the electric rotor values in each operating point are of high interest. These operating points cover a wide range of speed, active and reactive power generation.

2.1 Machine model The basis for this simple approach is the well-known equivalent circuit of the induction machine. Assuming a lossless machine, the active power flow of the machine can be expressed with

Ofner, Georg, Dipl.-Ing. Dr., ELIN Motoren GmbH, Elinmotorenstraße 1, 8160 Preding/Weiz, Austria; Handgruber, Paul, Dipl.-Ing., Christian Doppler Laboratory for Multiphysical Simulation, Analysis and Design of Electrical Machines, Graz University of Technology, Kopernikusgasse 24/III, 8010 Graz, Austria; Seebacher, Roland, Ass.-Prof. Dipl.-Ing. Dr., Institute of Electrical Drives and Machines, Graz University of Technology, Inffeldgasse 18, 8010 Graz, Austria (E-mail: georg.ofner@elinmotoren.at)

e &i elektrotechnik und informationstechnik

G. Ofner et al. DFIG for wind turbines – modeling the steady state

P mech ¼ P G ¼ P 1 þ P 2

ð1Þ

whereas the stator power P 1 and the rotor power P 2 depend on the slip s and can be expressed as a function of the grid power P G .

ð2Þ

ð3Þ

P 2 ¼ s P 1

s ¼ 1

n

n syn

P 1 ¼

P G

1 s

ð4Þ

P 2 ¼ P G

s

1 s

ð5Þ

The goal is to calculate from a given grid power, rotor speed n and grid power factor cosðG Þ the unknown rotor voltage U 2 and current I 2 . Starting from the required grid power and grid power factor, the grid current can be determined

q

S

G

I G ¼

S 1 ¼

p

ffiffiffi

3

2

1

U

G

þ Q 2

G

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

P

Q G ¼ S G sinðG Þ

ð6Þ

ð7Þ

ð8Þ

The real and imaginary parts are given by

I G; real ¼

I G; imag ¼

P G

p

ffiffiffi

3

Q

U

G

G

p

ffiffiffi

3

U

G

ð9a Þ

ð9bÞ

The stator side current can be calculated in the same manner.

I 1; real ¼

I 1; imag ¼

P 1

p

ffiffiffi

3

Q

U

G

G

p

ffiffiffi

3

U

G

ð10aÞ

ð10bÞ

At last, the rotor side current is calculated from the rotor power.

I 2; real ¼

I 2; imag ¼

P 2

p

ffiffiffi

3

U

Q

RG

G

p

ffiffiffi

3

U

G

ð11aÞ

ð11bÞ

To assess the machine performance over the whole speed range, the rotor side current and voltage are essential. This is because the rotor side values vary strongly with the rotor speed. From an electric point of view the possible operating area of the wind

point of view the possible operating area of the wind F i g . 1 .
point of view the possible operating area of the wind F i g . 1 .

Fig. 1. Extended model with dependencies

Mai 2011 | 128. Jahrgang

ORIGINALARBEITEN

turbine is limited by the maximum rotor voltage and current. The induced rotor voltage delimits the minimum and maximum rotor speed, whereas the maximum fre quency converters current deli- mits the possible rotor active power resp. apparent power. The machine equations presented above give a good overview of the operating performance of the doubly fed asynchronous machine, but with restrictions. To predict the machine behavior more exactly some non linear effects have to be considered. These nonlinearities and dependencies as well as the extended machine

model are shown in Fig. 1 and discussed in more detail in ( Ko¨ nig, 2009; Ofner et al., 2010 ). Figure 2 shows the influence and the necessity to take into account these effects.

850 800 750 U G = 594 V 700 U G = 660 V 650
850
800
750
U G = 594 V
700
U G = 660 V
650
U G = 726 V
600
550
−1000
−500
0
500
1000
1500
Rotor current I 2 (A)

Reactive power Q G (kvar)

Fig. 2. Rotor current against reactive power. Solid lines indicate simulation results with main field saturation. Dashed lines indicate simulation results without saturation. Marks indicate measured values

3. Power quality The current strong growth in the area of renewable energy plants

makes it necessary to regulate more strictly the grid connection

conditions. These regulations are focusing on two main fields, name-

ly the behaviour in a grid fault case and the harmonic distortion and emission. To consider a stable grid in a case of a temporary voltage or power variation the connected energy suppliers have to stay operational and not disconnected from the grid. Such a com- mon grid transient event is a low voltage ride through (LVRT). Here the grid voltage is temporarily reduced and returns back to the rated level after several grid periods. In this fault period the wind energy plants have to stay connected and support the grid with reactive power respectively to provide short circuit power. On the other side, with the increasing number of decentral energy suppliers the num- ber of harmonic distortion sources are increasing, too. The conse- quence of this fact is that the influence on the voltage quality is intensified. To limit the grid distortions the grid operators regulate – as required by law – the maximum distortion levels. Depending on the grid configuration and other guidelines there are different national standards. Table 1 gives a short overview of some stan- dards. Most standards cover basically the topic in the same way, differences can be found by the individual emission levels for the harmonics. A short example of a grid connected single wind plant should demonstrate the differences of the presented standards and policies in Table 1. Figure 3 shows the simple plant configuration. The

medium voltage grid is modeled as voltage source with impedance and is characterized with the short circuit apparent power. Between the low voltage wind plant and the grid a transformer is installed. Figures 4 and 5 show for the given example the resulting limit

© Springer-Verlag

heft 5.2011

175

ORIGINALARBEITEN

G. Ofner et al. DFIG for wind turbines – modeling the steady state

Table 1. Examples of national and international standards and policies for emission levels

Title

Validity

Year

BDEW-Richtlinie 2008; Technische Richtlinie Erzeugungsanlagen am Mittelspannungsnetz, Richtlinie fu¨ r Anschluss und Parallelbetrieb von Erzeugungsanlagen am Mittelspannungsnetz Technisch Organisatorische Regeln fu¨ r Betreiber und Benutzer von Netzen, Hauptabschnitt D2:

Germany

2008

Austria

2006

Richtlinie zur Beurteilung von Netzru¨ ckwirkungen Technische Regeln zur Beurteilung von Netzru¨ ckwirkungen TF 3.2.6; Wind turbines connected to grids with voltages below 100 kV, Technical regulations for the properties and the control of wind turbines IEEE Std 519-1992; IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems GB/T 14549-93; Quality of electric energy supply, Harmonics in public supply networks

Germany, Austria, Switzerland Denmark

2008

2004

USA, Canada

1992

China

1993

690 V 30k V ∼ Transformer DFIG S A =3,1MVA Grid =50 MVA S kV
690 V
30k V
Transformer
DFIG
S A =3,1MVA
Grid
=50 MVA
S kV

j kV =45°

Fig. 3. Example of a grid connected single wind power plant

7 BDEW-Richtlinie TOR D2 6 TF 3.2.6 IEEE 519-1992 GB/T 14549-93 5 4 3 2
7
BDEW-Richtlinie
TOR D2
6
TF 3.2.6
IEEE 519-1992
GB/T 14549-93
5
4
3
2
1
0
0
5
10
15
20
25
30
35
40
45
50
I
in % of fundamental current
n,permitted

Harmonic order

Fig. 4. Current limit levels for odd harmonics

6 BDEW-Richtlinie TF 3.2.6 5 IEEE 519-1992 GB/T 14549-93 4 3 2 1 0 0
6
BDEW-Richtlinie
TF 3.2.6
5
IEEE 519-1992
GB/T 14549-93
4
3
2
1
0
0
5
10
15
20
25
30
35
40
45
50
I n,permitted
in % of fundamental current

Harmonic order

Fig. 5. Current limit levels for even harmonics

harmonic current levels for each standard. Figure 4 demonstrates that not all standards consider even harmonics. In addition to the harmonics with multiple of the fundamental frequency, interharmo-

176 heft 5.2011

© Springer-Verlag

nics have to be considered, too. Speed variable drives with induction machines emit a series of these interharmonics and the harmonic distortion analysis has to consider these. Only the BDEW-Richtlinie 2008 comments on this fact and defines limits and refer to the IEC 61000-4-7 for measurement of the harmonics and interharmonics.

4. Polyharmonic machine model Beside the Finite Element Method (FEM) different approaches for a polyharmonic machine model can be found in literature:

" The so-called harmonic wave theory describes the individual har- monic fields analytically. To determine the resulting air gap field the interaction between the individual fields has to be taken into account (Seinsch, 1992; Oberretl, 2007).

" The multiple coupled circuit approach models the windings as electromagnetically coupled circuits. The magnetic coupling of the individual windings with each other is expressed by means of inductances based on the winding function theory (Luo et al.,1995; Kral, 1999).

" Another possible method provides the magnetic equivalent circuit theory. Thereby an electric machine is modeled as a network of magnetic resistors or conductances (Ostovic´ , 1989).

For the proposed model the multiple coupled circuit theory comes to implementation.

4.1 Calculation of inductances The inductance between two single windings can be written as (Luo et al.,1995):

L i ; j ¼ w i w j m 0 rl

d

00

2 p

Z n i N j d

0

ð12Þ

while w i and w j are the numbers of turns per winding, r is the average radius of the air gap, l is the effective armature length and is the particular angular position along the stator inner surface. d 00 denotes a constant air gap length increased by the Carter-factor and saturation effects. The winding functions n i and N j are defined as:

and

with

n i ¼

for i 1 G i 2 0 ; else

1

;

N j ¼ n j hn j i

hf i ¼

2p

1

2p Z

0

f d

ð13Þ

ð14Þ

ð15Þ

e &i elektrotechnik und informationstechnik

G. Ofner et al. DFIG for wind turbines – modeling the steady state

i 1 and i 2 are the start and end angle of the winding i , f is an auxiliary function. Equation (12) allows to find the self inductances of the stator and rotor windings. For the determination of the induc- tances between the stator and rotor windings the motion of the rotor has to be taken into account.

4.2 Extensions to the inductance calculation

The following paragraphs give an overview of the effects considered in the proposed model. a) Linear rise of the magnetomotive force: Assuming that the magnetomotive force rises linearly between the slot openings of width b s the winding function from (13) becomes (Joksimovic´ , Durovic´ , Obradovic´ , 1999):

n i ¼

8

>

>

>

<

>

>

>

:

i 1 b

1;

i 2

s

b

s

0;

þ

þ

2 ; for

for 2 ; for else

1

1

i 1 b s G

2

i 1 þ b s G’ ’ i 2 b s G

2

2

i 2 þ b s

2

i 2 b s

2

i 2 þ b s

2

ð16Þ

b) Slotting: The effect of slotting is taken into account by a

fictitious air gap widening in the area of the slots. The proposed model implements an approach described in (Elkner, 1986), which uses conformal mapping to solve the field equations. The following parameter presentation is the starting point:

2

x

b

s

and

¼

2

p

8

>

<

>

:

arcsin

q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9 p ffiffiffi s
q
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
9
p
ffiffiffi
s
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
>
þ 2d 0
1
þ 2d 0 2 u
u
u
b
b
s
s
þ d 0 ln q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
p
ffiffiffi
b
=
>
1 þ 2 d 0
s
2d 0
;
1
þ 2d 0 2 u
u
b
s
b
b
s
s
B
¼ 2d 0
1
l ¼
r
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B max
b s
2d 0
1 þ
2 u
b s

ð17aÞ

ð17bÞ

while x is the position along the average radius within a slot pitch, b s is the slot width, d 0 is the geometric air gap and u is an auxiliary variable. With the ratio l it is possible to define a fictitious air gap function:

d N ¼ d 0

l d 0

ð18Þ

c) Skewing: To consider skewing effects an inductance per unit

length to each axial section is assigned.

L

0

i

; j ¼ L i ; j

l

ð19Þ

With the rotor position angle r , the skewing angle g and the position j along the stator core one can write ( Joksimovic´ , Durovic´ , Obradovic´ , 1999):

L

0

i ; j ;skew ð r ; j Þ ¼ L i ; j

0

r þ j g

l

ð20Þ

The total inductance results from the integration over the core length:

þ

l

2

L i ;j ; skew ð r Þ ¼ Z

l

2

L

0

i

;j ;skew ðr ; j Þd j

ð21Þ

d) Saturation: Following to ( Moreira, Lipo , 1992) the main field

saturation can be modeled as modulation of the air gap permeance with twice the flux density wave frequency. Therefore the air gap function becomes:

d ¼ k sat; 1

Mai 2011 | 128. Jahrgang

d 0 þ d N s þ d N r

1 þ k sat ; 2

cosð2 pð h ÞÞ

ð22Þ

ORIGINALARBEITEN

while d N s and d N r are the air gap function due to the stator and rotor slotting, p is the number of pole pairs and h is the angle of the main flux. The factor k sat; 1 considers the MMF-drop in the iron and k sat; 2 the main field saturation. These factors are derived from finite element calculations. e) Modified winding function: A variable air gap function requires the introduction of the modified winding function approach to satisfy the condition for the zero divergence magnetic field (Faiz, Tabatabaei, 2002). Hence, (14) has to be modified:

N j ¼ n j hn j d 1 i

hd 1 i

ð23Þ

Due to the dependencies of the air gap function, the inductances are a function of the rotor position angle and the main flux angle:

2p

L i ;j ðr ; ’ h Þ ¼ w i w j m 0 rl Z

0

n i ð’; ’ r Þ N j ð’; ’ r Þ d ð’; ’ r ; ’ h Þ

d

ð24Þ

4.3 Voltage equations

To reduce the computational effort it is necessary to summarize the single inductances: the inductances of one pole per phase carry the same current and can be summarized to partial windings. Those partial windings can be summarized to phase windings. With the aid of the winding resistance and the leakage inductance

the phase-voltage equations, written in hyper matrix notation,

become:

u¼ R i þ L s

di

d t þ

v r dL dr

þ v s

di

dL i þ L dt

dh

ð25Þ

The voltage equation (25) requires the inductances as well as the

derivative of the inductances with respect to the rotor position angle

and the main flux angle. The voltage equations are solved numeri- cally in a simulation. These inductance matrices are calculated before the simulation and stored in a look up table.

4.4 Model of the entire wind power plant

Figure 6 shows the basic configuration of the investigated power

plant. The stator filters 1 and 2 and the inverter filter 2 are passive

RLC notch filters. Stator filter 1 and 2 is tuned to the pulse frequency of the rotor side inverter, the inverter filter 2 on those of the grid sided inverter. To decouple the line side inverter form the grid, a LC-low pass (inverter filter 1) is used. Possible parallel

resonances have to be regarded. Hence, a damped filter design is preferred.

Turbine DFIG Grid Transformer Stator Stator filter 1 filter 2 Inverter 3∼ = Inverter Inverter
Turbine
DFIG
Grid
Transformer
Stator
Stator
filter 1
filter 2
Inverter
3∼
=
Inverter
Inverter
filter 1
filter 2
= 3∼
f r,T =2250 Hz
f G,T =4500 Hz

Fig. 6. Structure of the entire wind power plant

The additional filter-, inverter- and transformer-components have to be considered in the entire plant model. Two methods for the

calculation of the harmonic current emitted on the grid connection point are developed and compared.

© Springer-Verlag

heft 5.2011

177

ORIGINALARBEITEN

1) Method in time domain : In this method the filter and the transformer are represented by frequency-independent, concen- trated components, based on differential equations. The inverter at the rotor and grid side is modeled as a symmetrical, sinusoidal pulse-width modulated voltage source, the grid as a fundamental voltage with internal impedance. For the grid side inverter the fundamental voltage is calculated with the model presented in Sect. 2. For the rotor side inverter a rotor current controller is implemented. Finally, all the differential equations of the modeled components are coupled and solved in the time domain using numerical techniques. 2) Method in frequency domain : In this case the grid current is determined by means of complex transfer functions. Firstly, the current spectrum emitted by the machine is calculated. Therefore the stator windings are supplied with fundamental voltage; the voltage at the rotor side is given as a PWM-voltage. For the calculation of the grid current the transmission behavi- our from the stator current to the grid current is determined with the classical network theory. Furthermore, also the influ- ence of the grid side inverte r (PWM-voltage) and the grid voltage need to be taken into account. Lastly, the linear super- position of the single sources yields the grid current. Since only the machine equations are solved in the time domain, this method consumes less computing time compared to the method in IV-D1.

4.5 Simulation results of the machine model To verify the machine model the simulation results of two different machines are compared with finite element calculations and measurements. 1) Flux Densities: In Fig. 7 the radial air gap flux density obtained with the winding function approach (WFA) is compared with finite element calculations. Only the stator winding is current fed. The iron in the finite element model is modeled with m r ¼ 50; 000. The finite element analysis is performed for different permeabilities of the stator slot wedges. For m r ¼ 1 there is a good agreement between the two methods. For m r ¼ 5 the field weakening due to the slot openings reduces significantly. In the proposed model the influence of the slot wedges is considered by weighting factors for the stator and rotor slots:

ð26aÞ d N s ;corr ¼ k corr; N s d Ns ð26bÞ d N
ð26aÞ
d N s ;corr
¼ k corr; N s d Ns
ð26bÞ
d N r ;corr ¼ k corr; N r d Nr
0.4
FEM µ r =
1
0.2
FEM µ r = 5
WFA
WFA corrected
0
−0.2
−0.4
−0.6
−0.8
−1.0
−1.2
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
B r (T)

j(rad)

Fig. 7. Radial air gap flux density for one pole pitch

178 heft 5.2011

© Springer-Verlag

G. Ofner et al. DFIG for wind turbines – modeling the steady state

2) Determination of the occurring frequencies: The n th rotor field in the rotor fixed coordinate system g is:

with

b

r

b n ; r ðg ; t Þ ¼ B n sinðn g sv s t Þ

¼ v s t ð1 sÞ þ g

p

ð27Þ

ð28Þ

the rotor field in the stator fixed coordinate system becomes:

b

b n s ;r ð’; t Þ ¼ B n sin

n

s þ n ð1 sÞ

p

v s t

This yields the appearing frequencies in the stator current:

ð1 sÞ f s

f n ;s ¼

s þ n

p

ð29Þ

ð30Þ

Figure 8 shows the stator frequencies as a function of the slip for different ordinal numbers.

1400 ν = −5p ν +7p = 1200 ν −11p = ν +13p = 1000
1400
ν
=
−5p
ν +7p
=
1200
ν −11p
=
ν +13p
=
1000
800
600
400
200
0
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Slip s
fn ,s (Hz)

Fig. 8. Appearing stator frequencies. The fundamental frequency is 50 Hz

Because of the slot numbers of the investigated machines no additional frequency components show up due to the slotting. The frequencies caused by the main field saturation can be determined with:

f s ;sat ¼

s 2 þ n

p

ð1 sÞ f s

ð31Þ

The dominant spectral components caused by the rotor side

inverter are:

whereby

 

l

¼

or

 

l

¼

f r ; PWM ¼ l f r ;T

1; 3; 5 ;

2; 4; 6 ;

m

m

m f r

¼

2; 4; 6 ;

¼ 1; 3; 5 ;

ð32Þ

ð33aÞ

ð33bÞ

For the calculation of the resulting stator frequency the mechanical motion of the rotor must be taken into account:

ð34Þ

f s ; PWM ¼ f r ; PWM ð1 sÞf s

The sign of f r ;PWM depends on whether the rotor three phase system builds a positive or negative sequence. Figure 9 illustrates the trans- mission of these frequencies in the stator.

e &i elektrotechnik und informationstechnik

G. Ofner et al. DFIG for wind turbines – modeling the steady state

10 I r 9 I 2170 Hz 2330 Hz s 8 7 6 (−2230 +
10
I
r
9
I
2170 Hz
2330 Hz
s
8
7
6
(−2230 + 60) Hz
(2270 + 60) Hz
5
4
2230 Hz 2270 Hz
3
2
1
0
2100
2150
2200
2250
2300
2350
2400
Line current |I s (j w )| (A)

f (Hz)

Fig. 9. Transmission of the inverter frequencies in the stator ( s = 20%, f r,T = 2250 Hz)

3) Comparison with measurements: The measurement shown in Fig. 10 was carried out with short-circuited rotor windings at a constant speed and rated stator voltage. Note that the stator wind- ing of the investigated machine is D-connected. Therefore the fre- quency components due to saturation do not show up. Those components cause circulating currents within the winding and in- duce additional losses.

4.6 Simulation results of the model of the entire plant The grid current spectrum emitted by the configuration in Fig. 6 is demonstrated in Fig. 11. The filters were included in the calculations. The results of the two suggested methods show a good accordance. Spectral components below 2200 Hz are generated by the discrete winding distribution and the slotting effects. In addition, the inverter causes higher frequency components at multiples of the pulse frequency.

20 Measurement 18 Simulation 16 14 12 10 8 6 4 2 0 0 200
20
Measurement
18
Simulation
16
14
12
10
8
6
4
2
0
0
200
400
600
800
1000
1200
1400
1600
1800
2000
Line current | I s (j w ,)| (A)

f (Hz)

Fig. 10. Frequency spectrum of the stator current. Fundamental current: Measurement = 2212 A; Simulation = 2293 A

5. Conclusion The presented model provides a solid and versatile foundation for the steady state simulation of DFIGs. Furthermore, a polyharmonic machine model is developed and the influence of the rotor side

Mai 2011 | 128. Jahrgang

ORIGINALARBEITEN

35 Simulation, time domain Simulation, frequency domain 30 25 20 15 10 5 0 0
35
Simulation, time domain
Simulation, frequency domain
30
25
20
15
10
5
0
0
1000
2000
3000
4000
5000
6000
7000
8000
9000 10000
Line current | IG (j w ) | (A)

f (Hz)

Fig. 11. Frequency spectrum of the grid current. Operating point:

U G = 690 V, n = 1200 min 1 , cosðG Þ = 0,95. Fundamental current:

Simulation, time domain = 2743 A; Simulation, frequency domain = 2758 A

converter in doubly fed operation is demonstrated. Same measure- ments are carried out and are compared with the simulation results. The simple machine model gives a good overview of the basic machine behavior under doubly fed operation. To predict accurately machine currents and voltages, this simple model has to be extended with some non linear effects. With the improved model a closed form solution of the machine currents and voltages is not possible. In order to avoid this difficulty, an iterative solver approach was implemented. The calculated rotor currents show good agreement with the measured values. It can be clearly seen, that the main field saturation has an essential impact. Results show that strong saturation of the main field increases the necessary rotor current when the wind power plant has to deliver reactive power to the grid while operating at overvoltage conditions. In order to assess the power quality behaviour of a DFIG a poly- harmonic machine model based on a modified winding function

approach is used. The characteristic emitted frequencies are shown

and the influence of the rotor side converter is demonstrated. Simulation results of the current spectrum show an acceptable accordance with measurements.

References

Elkner, R. (1986): Berechnung des gesa¨ ttigten Leerlauffeldes bei DS-Asynchronmaschinen mit Nutungseinfluss. Master Thesis, Vienna University of Technology. Faiz, J., Tabatabaei, I. (2002): Extension of winding function theory for nonuniform air gap in electric machinery. IEEE Transactions on Magnetics, 38 (6): 3654–3657. Joksimovic´ , M., Durovic´ , D. D., Obradovic´ , B. A. (1999): Skew and linear rise of MMF across slot modelling – winding function approach. IEEE Transactions on Energy Conversion, 14 (3): 315–320. Ko¨ nig, O. (2009): Modellierung einer doppelt gespeisten Asynchronmaschine fur Windkraftanlagen der MW-Leistungsklasse. Master Thesis, Graz University of Technology. Kral, C. (1999): Modellbildung und Betriebsverhalten einer Asynchronmaschine mit defektem Rotorstab im La¨ uferka¨ fig einschließlich Detektion durch Vienna Monitoring Method. PhD Thesis, Technische Unversitat Wien. Luo, X., Liao, Y., Toliyat, H. A., El-Antably, A., Lipo, T. A. (1995): Multiple coupled circuit modeling of induction machines. IEEE Transactions on Industry Applications, 31 (2):

311–318.

Mu¨ ller, S., Deicke, M., De Doncker, R. W. (2002): Doubly fed induction generator systems for wind turbines. Industry Applications Magazine, IEEE, 8: 26–33. Moreira, J. C., Lipo, T. A. (1992): Modeling of saturated ac machines including air gap flux harmonic components. IEEE Transactions on Industry Applications, 28 (2):

343–349.

© Springer-Verlag

heft 5.2011

179

ORIGINALARBEITEN

Oberretl, K. (2007): Losses, torques and magnetic noise in induction motors with static converter supply, taking multiple armature reaction and slot openings into account. IEEE Transactions on Electric Power Applications, 1 (4): 517–531. Ofner, G., Ko¨ nig, O., Dannerer, G., Seebacher, R. (2010): Steady state modeling of doubly fed induction generators for mega watt class wind turbines. 19th International Conference on Electrical Machines, Rome, Italy, Sep. 2010.

Authors

on Electrical Machines, Rome, Italy, Sep. 2010. Authors Georg Ofner was born in Silandro/Schlanders, Italy, in

Georg Ofner was born in Silandro/Schlanders, Italy, in 1975. He received the MSc and PhD de- gree in Electrical Engineering from the Graz University of Technology, Austria, in 2001 and 2005, respectively. Since 2005 he has been working at the ELIN Motoren GmbH in Preding/Weiz, Austria. He works as R&D engineer and his current research interests include finite element modeling and the design of electrical machines.

element modeling and the design of electrical machines. Paul Handgruber received the MSc degree in electrical

Paul Handgruber received the MSc degree in electrical engi- neering from the Graz University of Tech- nology, Austria, in 2010. Currently, he is a PhD student and a member of the Christian Doppler Laboratory for Multiphysical Simu- lation, Analysis and Design of Electrical Ma- chines located at the Institute for Fundamentals and Theory in Electrical En- gineering, Graz University of Technology. His special fields of interest include modeling and simulation of electrical machines.

180 heft 5.2011

© Springer-Verlag

G. Ofner et al. DFIG for wind turbines – modeling the steady state

Ostovic´ , V. (1989): Dynamics of saturated electric machines. Springer-Verlag. Seinsch, H. O. (1992): Oberfelderscheinungen in Drehfeldmaschinen. B.G. Teubner Stuttgart. Wilkes, J. (2010): Wind in power, 2009 European statistics. The European Wind Energy Association.

European statistics. The European Wind Energy Association. Roland Seebacher was born in Lienz/Osttirol, Austria in

Roland Seebacher was born in Lienz/Osttirol, Austria in 1961. He received his MSc and PhD degrees in electrical engineering from the Graz Univer- sity of Technology, Austria, in 1991 and 1996, respectively. Currently he is an assis- tant professor at the Institute of Electrical Drives and Machines. His research interests are in the area of modeling and control of electric machines and drives.

e &i elektrotechnik und informationstechnik