Control of a wind park with doubly fed induction generators in support of power system stability in case of grid faults

Gabriele Michalke+, Anca D. Hansen*, Thomas Hartkopf+

+ University of Technology Darmstadt, Department of Renewable Energies, Institute for Electrical Power Systems, Landgraf-Georg-Strae 4, 64283 Darmstadt, Germany * Ris National Laboratory, Wind Energy Department, P.O. Box 49, DK-4000 Roskilde, Denmark Email: gmichalke@re.tu-darmstadt.de, anca.daniela.hansen@risoe.dk, thartkopf@re.tu-darmstadt.de

Abstract
The paper presents a control strategy for wind parks based on the doubly fed induction generator (DFIG) concept, which facilitates fault ride through of large DFIG wind farms and enables them to support the power system stability. The control system comprises a coordinated reactive power supply by means of the frequency converters as well as a damping controller in order to damp oscillations in the mechanical system. Simulations are performed by means of a detailed model of the wind park and the transmission grid in the power system simulation software DIgSILENT Power Factory. The results demonstrate, that the here presented control strategy of DFIG wind parks even facilitates a fault ride through of nearby connected wind power plants based on conventional technology.

Introduction

Due to the increasing penetration of wind power the power system operators are revising nowadays the grid codes in several countries, such as in Germany and Denmark [1], [2]. Basically these grid codes require newly installed wind turbines to support directly the power system in case of grid faults. The attention is thus on both, the wind farms fault ride through capability and its contribution to grid support. The fault ride through capability is required because a tripping of the large wind turbines in case of grid faults would cause a significant loss of electric power supply. Focus of this paper is to introduce a control strategy for wind parks based on the doubly fed induction generator (DFIG) concept, which facilitates fault ride through of large DFIG wind farms and enables them to support the power system stability. The variable speed wind turbine with doublyfed induction generator is recently the most commonly used wind turbine concept and is already applied in large wind parks as e.g. the Danish offshore wind park Horns Rev. The attractiveness of the DFIG for the use in wind turbines is due to the partial scale power converter in the rotor circuit causing an important financial advantage compared to generator concepts with full scale converter. The converter system offers also good potentialities for control, which can especially be exploited for fault ride through and grid support purposes. However, the financial advantage turns into a technical disadvantage in case of grid faults, as the partial scale power converter must be protected against high transient currents. The paper presents a control strategy of the DFIG wind turbine based on

[3], which allows fault ride through and facilitates reactive power supply and voltage re-establishment during grid faults. Furthermore an active damping controller is implemented in order to reduce the mechanical stresses of the turbine. An aggregated DFIG wind park model serves to analyse the interaction of wind park and power system. The investigations are carried out by means of a comprehensive dynamic simulation model of a DFIG wind park and a transmission system in the software DIgSILENT Power Factory.

The DFIG wind turbine concept

The system configuration of the DFIG-wind turbine concept is illustrated in Figure 1. Main characteristic of this concept is the partial scale power converter in the rotor circuit, while the stator is directly connected to the grid via the transformer.
Wind Model Aerodynamic Model Mechanical Model Generator Model with Rotor Side Converter Grid Side Converter Transformer Model External Grid

vwind

Taero

Gear

mech
DFIG

~
Crowbar

= ~
Controlsignaldq
Grid Side Converter Control

wtr
Speed Controller

Controlsignaldq
Rotor Side Converter Control

~ ~ ~

Control for normal operation Stage I

Generator
Overspeed Protection

PGrid
Damping

QGrid
Voltage Contol

UDC

QConverter
Fault control Stage II

Reactive Power Boosting

Generator

UGrid

Crowbarsignal

Figure 1 System configuration of the DFIG wind turbine with control structure.

The present investigations are based on the comprehensive dynamic simulation model of the

DFIG wind turbine concept, developed in the power system simulation software DIgSILENT Power Factory at Ris National Laboratory [4]. DIgSILENT Power Factory makes it possible to investigate the dynamic performance of grid connected wind turbines as a part of realistic electrical grid models. The electrical components of the system in Figure 1 are available in a DIgSILENT library as built-in blocks. All other parts necessary in the modelling of wind turbines such as wind model, aerodynamic model, mechanical model and a control has to be implemented by the user himself. Figure 1 provides also a quick overview on the control stages of DFIG based wind turbines: Stage I contains the control for normal operation, Stage II incloses the control, which is used in case of grid faults. The control of the DFIG is realized by means of two controllers, the rotor side converter controller and the grid side converter controller.

sented. Then details regarding the implemented fault control strategy are discussed.

3.1

Crowbar protection

2.1

DFIG normal operation control

In normal operation the aim of the rotor side converter controller is to control active and reactive power independently. The reference active power is provided by a maximum power tracking depending on the actual generator speed, which assures operation with maximum aerodynamic efficiency. During normal operation the reference reactive power is usually set to zero. For high wind speeds the speed of the turbine is limited to its rated speed, which implies indirectly that the power is limited to its rated value, too. This is realized by means of a speed controller interacting with the pitch mechanism. The aim of the grid side converter controller for normal operation is to maintain the DC-link capacitor voltage and to guarantee a converter operation with unity power factor. Detailed explanations about the control concept of the DFIG wind turbine concept for normal operation conditions can be found in [5].

In case of grid faults very high short circuit currents arise in the stator. These high transient currents are transferred to the rotor due to the inductive coupling between stator and rotor, implying a high damaging risk for the power electronics of the rotor side converter. Due to this reason a suitable protection of DFIGs is necessary. A simple protection method is to short-circuit the rotor via an external rotor resistant called crowbar (Figure 2) when high rotor currents are detected. The crowbar coupling protects the converter system and facilitates fault ride through, which enables the DFIG to contribute to power supply directly after fault clearing. When the crowbar protection is triggered, the rotor side converter is blocked and the DFIG behaves as a squirrel cage induction generator. This implies that the whole controllability of the DFIG is lost during crowbar coupling.

Gear
DFIG

Rotor Side Converter

Grid Side Converter

External Grid

~
Crowbar

= ~

Figure 2 3-phase short circuit at the point of common coupling of the DFIG.

~ ~ ~

3.2

Overspeed protection

DFIG fault control

Grid codes for wind turbines, which are specified by transmission system operators, demand, that newly installed wind turbines must not be disconnected in case of grid faults, as this would mean a disconnection of a significant amount of electrical power due to the increasing penetration of wind power. Thus, modern wind turbines are required to work as a conventional power plant does. Goal of the following section and main focus of the presented paper is to introduce a control strategy, which facilitates fault ride through of large DFIG offshore wind parks and enables the wind park to contribute to power system stability support. In the following protection issues of a DFIG based wind turbine under grid faults are briefly pre-

A short circuit, as illustrated in Figure 2, causes a voltage drop at the generator terminal, which in turn provokes an active power drop in the stator. As this reduces the electrical torque, too, the generator starts to accelerate. Because the DFIG behaves as a squirrel cage induction machine as long as the crowbar is triggered, the acceleration causes an increased reactive power consumption of the generator. Both, increased speed and increased reactive power demand have a negative impact on power system stability. Thus, it is necessary to prohibit a too strong acceleration by means of an overspeed protection. The pitch controller implemented in the presented model controls the generator speed and therefore it can directly be used as an overspeed protection. The pitch controller reacts to the acceleration of the drive train, which is due to the grid fault, and increases the pitch angle. The absorbed aerodynamic power of the turbine can so be reduced, which counteracts the acceleration process

3.3

Damping control

3.4

In stability analysis, as e.g. for stability investigations in case of grid faults, it is essential to approximate the drive train system by at least a two mass model [6]. The mechanical model implemented in the presented simulation model represents the large turbine mass connected via a flexible shaft to the much smaller generator mass. The transmission between the slowly rotating rotor and the fast rotating generator is modelled with an ideal gearbox. The low-speed shaft is characterised by a stiffness k and a damping c, while the high-speed shaft is assumed to be stiff. Thus, in case of grid faults the drive train behaves as a torsion spring, which is untwisted, due to the drop of the electrical torque in the generator. Because of this, heavy oscillations are excited in the mechanical system. In order to reduce the big mechanical stresses on the turbine, a damping controller (Figure 1) is implemented, which actively damps the speed fluctuations acting on the active power control. In order to verify the developed damping control a 3-phase short circuit at the point of common coupling of the DFIG wind turbine (Figure 2) is simulated. The short circuit happens at 0s and has a duration of 300ms. The DFIG wind turbine is connected to a grid, which is modelled as a Thevenin equivalent and the wind speed is kept constant during the simulation. Figure 3 depicts the performance of the damping controller. It shows the signals of rotor speed and mechanical torque. Since the mechanical quantities have bigger time constants than the electrical quantities, the fault response is much longer in the mechanical system. Thus, the mechanical signals are plotted for 10s although the fault clearing already happens at 300ms.
Rotor speed [rad/s]
1.82 1.80 1.78 1.76 1.74 1.72 1.70 -2.000
DIgSILENT

Voltage control of the rotor side converter

In order to utilize the good controllability of the DFIG for grid support in case of faults, the rotor side converter control is extended by a voltage control block. This controller regulates the voltage at point of common coupling (PCC) of the wind turbine or the wind park by adjusting the reactive power supply. The generator can so optionally provide inductive or capacitive reactive power. This can e.g. be utilized during reactive power imbalance in the system, when an inductive load or generation unit is coupled to the grid. During grid faults, as long as the crowbar is not triggered, the voltage controller can provide reactive power, which serves to re-establish the voltage level. Since the rotor side converter controller controls the reactive power, which is fed into the grid via the stator circuit, this converter can compensate a bigger reactive power demand than the grid side converter. Thus, the goal of the here presented control strategy is to control the reactive power demand mainly by means of the rotor side converter, while this task is only temporarily assigned to the grid side converter, when the rotor side converter is blocked by the crowbar. It is therefore necessary to develop a coordinated voltage control and reactive power supply between both converters [7]. Figure 4 illustrates as an example the voltage signal at the point of common coupling of the DFIG after an inductive load is connected to the grid. The voltage drops slightly below 1 p.u. due to the reactive power demand of the inductive load. The voltage remains at the reduced level when no voltage control by the rotor side converter is enabled (dashed line). However, by use of the voltage controller, the voltage can fast be re-established (solid line) and the reactive power demand of the inductive load can be met.
Voltage at PCC [p.u.]
1.01 1.00 0.99 0.98 0.97 0.96 -0.100

Inductive load coupling

0.40

2.800

5.200

7.600

[s]

10.00

Mech. torque [Nm]

3.0E+4

0.50

1.100

1.700

2.300

[s]

2.900

time [s] With voltage controller Without voltage controller

2.0E+4

1.0E+4

0.0E+0

-1.0E+4 -2.000

0.40

2.800

5.200

7.600

[s]

10.00

time [s] With damping controller Without damping controller

Figure 4 Performance of the rotor side converter voltage controller after inductive load coupling

3.5

Reactive power boosting of the grid side converter

Figure 3

Performance of the damping controller

Without the damping controller, the torsional oscillations are only slightly damped 10s after the grid fault. It is clearly visible that the oscillations decay very fast due to the included damping controller. Moreover the mechanical torque crosses only once through zero when the damping controller is used, which significantly reduces the mechanical stresses of the turbine.

In contrast to the rotor side converter the grid side converter can stay active during grid faults, when the rotor side converter is blocked by the crowbar. The grid side converter can then be used as a STATCOM and contribute supplementary to reactive power supply. As shown in Figure 1 a reactive power boosting is implemented, which provides a reference reactive power signal to the grid side converter control. The reference reactive power for the grid side converter is set to its limits (+/-1 p.u.), so that the converter con-

DIgSILENT

tributes always with its maximum reactive power to voltage re-establishment. Figure 5 illustrates the signals of voltage at the point of common coupling (PCC) and the reactive power contribution of the grid side converter during a 3-phase short circuit at the PCC. The fault has a duration of 300ms and is followed by crowbar coupling in the DFIGs rotor circuit. After fault clearing the crowbar is still active for 100ms. In Figure 5 a) the voltage is plotted for two cases: with and without reactive power boosting of the grid side converter. During crowbar coupling the grid side converter, working like a STATCOM, can contribute to reactive power supply and improve the voltage level. Figure 5 b) compares the reactive power, which is provided by the grid side converter, with the reference reactive power. In this case the reference power of the reactive power boosting is set to 1p.u., which denotes that the grid side converter must feed its maximum possible reactive power into the grid. However, since the voltage level is significantly reduced during the fault, the grid side converters reactive power production capability is also reduced. It can be seen in Figure 5 b), that the reference reactive power cannot be provided during the fault. After fault clearing and thus after the voltage at the converter terminals has re-established, the grid side converter is able to provide the reference reactive power to the grid.
Voltage at PCC [p.u.]
1.20

4.1

Aggregated wind park model

The wind park model represents a large offshore wind farm consisting of 80 2MW DFIG wind turbines, in which all wind park turbines are combined to an aggregated wind park model. The aggregation method reduces the complexity and simulation time without compromising the accuracy of the simulation results [8] and is thus commonly used for system studies concerning the impact of large wind farms on the power system. This paper uses the aggregation approach presented in [8]. The 80 turbines are aggregated to one equivalent lumped wind turbine with rescaled power capacity according to the entire wind farm power. DIgSILENT Power Factory provides a direct built-in aggregation technique for the electrical system. The generator and the transformer can be directly modelled by a certain number of parallel machines, while the other components, as e.g. the power converter or the mechanical power of the turbine rotor have to be upscaled according to the wind farm power.

4.2

Transmission grid model

1.00

Fault incident Fault clearing Crowbar coupling

Crowbar deactivated

0.80

0.60

0.40 -0.100

0.06

0.22

0.38

0.54

[s]

0.70

a)
Grid side converter reactive power [p.u.]
0.30 0.00 -0.30 -0.60 -0.90 -1.20 -1.50 -0.100

0.06

0.22

0.38

0.54

[s]

0.70

b)
With reactive power boosting Reference reative power

time [s]

Without reactive power boosting

Figure 5 Performance of the grid side converter reactive power boosting a) Voltage at the point of common coupling (PCC) b) Reactive power boosting of the grid side converter

In order to investigate the impact of the large offshore wind farm on the power system, a generic simplified transmission system, is implemented in the software DIgSILENT Power Factory (Figure 6). The transmission model is based on a model, developed and provided by the Danish transmission system operator Energinet.dk [9]. The model embodies a generic representative transmission network and serves to simulate a realistic interaction between wind park and transmission grid during grid faults. The transmission system contains models for four central power plants and their control, several load centres and an aggregated model for local wind turbines with conventional technology based on asynchronous generators (Danish Concept). The transmission system model described in [9] is slightly modified in this paper. A 160 MW offshore wind park composed of exclusively DFIG wind turbines is used instead of the active stall wind farm of the original model.

Simulation model

The implemented control strategy for doubly fed induction generator based wind turbines in support of power system stability in case of grid faults is analysed in the following by means of an aggregated wind park model of a large offshore wind park, which is connected to a generic transmission system model.

DIgSILENT

750MW

SG

2500MW

SG
Fault incident
Voltage at the PCC [p.u.]
1.25 1.00 0.75 0.50 0.25 0.00 -0.100

Voltage at the PCC

Fault clearing

400kV

400kV

400kV

L
135kV 135kV

0.50

1.100

1.700

2.300

[s]

2.900

Wind turbine of the DFIG offshore wind park

L
135kV SG 1500MW

L Load Centre

Crowbar coupling
Voltage at the WFT [p.u.]
1.25 1.00 0.75 0.50 0.25 0.00 -0.100

Crowbar deactivated

L
1400MW SG

135kV

PCC
135kV

0.50

1.100

1.700

2.300

[s]

2.900

WFT
Power [MW]

2.50 2.00 1.50 1.00 0.50 0.00

Central Power Plant 500MW Local Wind Turbines based on conventional technology 160MW

-0.50 -0.100

0.50

1.100

1.700

2.300

[s]

2.900

Reactive power [MVar]

2.00 1.50 1.00 0.50 0.00 -0.50 -1.00 -0.100 0.50 1.100 1.700 2.300 [s] 2.900

DFIG Offshore Wind Park

Figure 6 Simplified illustration of the transmission system model.

Reactive power boosting of the grid side converter Case II

Reactive power supply of the rotor side converter Case II

Local wind turbine

Generator speed [p.u.]

Grid support of large wind farms

1.100 1.075 1.050 1.025 1.000 0.975 -0.100

In this section the developed control strategy of the DFIG wind turbine is assessed and analysed by means of simulations. A worst-case scenario is simulated in order to demonstrate the ability of the presented control strategy of the DFIG wind park in terms of grid support during faults. A 3-phase short circuit at the PCC (see Figure 6) with a duration of 300ms is performed. In the moment of the fault, the wind farm is assumed to work at its rated conditions. Due to its location close to the wind farm and its long duration, the fault scenario denotes a severe fault and critical situation for the grid and the coupled wind turbines. The simulation results are shown in Figure 7: The voltage at the point of common coupling (PCC) The voltage at the wind farm terminal (WFT) The active and reactive power of one representative turbine of the DFIG wind park The generator speed of a local wind turbine Two simulations can be compared in Figure 7: Case I (dashed line) shows the results without fault control strategy of the DFIG wind farm Case II (solid line) shows the results with fault control strategy of the DFIG wind farm

0.50

1.100

1.700

2.300

[s]

2.900

time [s]

Case I : DFIG wind park without fault control Case II : DFIG Windpark with fault control

Figure 7 Simulation results after a 3-phase short circuit at the PCC

The short circuit causes a significant voltage drop at the PCC and at the wind farm terminal. This implies a drop in the generators active power production. Due to the arising high transient currents the crowbar is coupled directly after the fault incident. The voltage drop causes also a reduction of the flux in the DFIG and the generator is demagnetized. In Figure 7 the demagnetisation is visible as a reactive power peak in the moment of the fault. Because of the active power drop, which causes a reduction of the electromagnetic torque, the induction generators of the local wind turbines start to accelerate. It can be found in case I that heavy generator speed fluctuations are excited in the conventional (local) wind turbines. These fluctuations are transferred to the voltage at the PCC. After the fault is cleared, the reactive power demand of the asynchronous generators of the local wind turbines has increased due to the acceleration. Moreover, also the DFIG wind park consumes reactive power during crowbar coupling, which influences the voltage re-establishment negatively. A tripping of the local wind turbines during grid faults, as it is common prac-

tice for conventional wind turbines today, is justifiable in this case. However, the tripping causes a significant loss of active power production, as the conventional wind turbines cannot immediately be reconnected after the fault is cleared. In contrast to case I, case II shows the results, which could be achieved, when the wind park is equipped with the new developed control strategy in the simulated fault case. As soon as the crowbar is connected, the reactive power boosting of the grid side converter is activated, which improves the voltage level during the fault. The active power signal has a similar characteristic in both cases. However, an improved voltage level causes an improvement of the active power, too. After crowbar decoupling, the voltage control of the rotor side converter starts to work. Figure 7 depicts the reactive power supply of the total wind park so that the voltage can much faster be reestablished. This has a positive impact on the behaviour of the local wind turbines with conventional technique, too. The speed fluctuations are significantly reduced due to the control system of the DFIG wind park. The simulation results of case II show, that a tripping of the local wind turbines is no longer necessary if the DFIG is equipped with the developed control strategy. The control system of the nearby connected DFIG wind farm even facilitates a fault ride through of the conventional wind turbines. A significant loss of active power production can so be avoided. Since the simulated grid fault can be understood as a worst-case scenario, it can be concluded, that the developed control strategy will also have a positive impact in case of other grid faults. It has been shown, that a large DFIG wind farm equipped with the here presented control strategy can support the power system stability.

control can even improve the behaviour of local wind turbines with conventional technology under grid faults, which are connected in vicinity to the DFIG offshore wind farm. A disconnection of the conventional wind turbines in case of grid faults is not necessary anymore. This means, that significant power losses due to disconnection under grid faults could be avoided. The presented control strategy facilitates fault ride through of DFIG wind farms and even enables them to improve the power system stability in case of grid faults.

Literature

Conclusion

This paper presents a control strategy of large wind farms equipped with doubly-fed induction generators (DFIG), which serves to improve the behaviour of a DFIG wind farm during grid faults and enables it to contribute to power system stability support. The control strategy is based on the design of a proper coordination between three controllers: the damping controller, the rotor side converter voltage controller and the reactive power boosting of the grid side converter. A damping controller damps the mechanical oscillations excited by grid faults. The grid voltage is controlled by the rotor side converter as long as it is not blocked by the protection system, otherwise the grid side converter is taking over the voltage control. A damage of the converter system due to too high fault currents is prevented by the crowbar protection. Simulation results in DIgSILENT Power Factory exemplify how DFIG wind farms with such control strategy participate to re-establish the voltage during grid faults. The

[1] E.ON. Netz GmbH, Netzanschlussregeln, Hoch- und Hchstspannung, www.eonnetz.com, 2006. [2] Energinet.dk, Wind Turbines Connected to Grids with Voltage above 100kV, Technical regulation TF 3.2.5, www.energinet.dk [3] Akhmatov V., Variable-speed wind turbines with doubly-fed induction generators. Part IV: Uninterrupted operation features at grid faults with converter control coordination, Wind Engineering, Vol. 27, Nr. 6, 2003. [4] Hansen A.D., Jauch C., Srensen P., Iov F., Blaabjerg F., Dynamic wind turbine models in power system simulation tool DIgSILENT, Risoe Report R-1400(EN), Ris National Laboratory, Denmark, 2003. [5] Gail G., Hansen A.D., Hartkopf T., Controller design and analysis of a variable speed wind turbine with doubly-fed induction generator, Conference Paper, European Wind Energy Conference Athen, Greece, 2006. [6] Akhmatov V., Analysis of Dynamic Behaviour of Electric Power Systems with Large Amount of Wind Power, PhD thesis, restad-DTU, Denmark, 2003. [7] Hansen A.D., Michalke G., Srensen P., Lund T., Iov F., Co-ordinated voltage control of DFIG wind turbines in uninterrupted operation during grid faults, submitted to Wind Energy, 2006. [8] Pller M., Achilles S., Aggregated Wind Park Models for Analyzing Power System Dynamics, Forth International Workshop on Large Scale Integration of Wind Power and Transmission Networks, 20th and 21st October, Denmark, 2003. [9] Akhmatov V., Nielsen A., A small test model for the transmission grid with a large offshore wind farm for education and research at Technical University of Denmark., Wind Engineering, Vol 3, No. 3, 2006.