Document not found! Please try again

A Comprehensive LVRT Control Strategy for DFIG Wind ... - IEEE Xplore

2 downloads 1635 Views 2MB Size Report
Jul 18, 2013 - the ability of reactive power support of a doubly fed induction generator (DFIG) ... the DC-link voltage and rotor current are kept below the dan-.
3302

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 3, AUGUST 2013

A Comprehensive LVRT Control Strategy for DFIG Wind Turbines With Enhanced Reactive Power Support Dongliang Xie, Zhao Xu, Senior Member, IEEE, Lihui Yang, Jacob Østergaard, Senior Member, IEEE, Yusheng Xue, Member, IEEE, and Kit Po Wong, Fellow, IEEE

Abstract—The paper presents a new control strategy to enhance the ability of reactive power support of a doubly fed induction generator (DFIG) based wind turbine during serious voltage dips. The proposed strategy is an advanced low voltage ride through (LVRT) control scheme, with which a part of the captured wind energy during grid faults is stored temporarily in the rotor’s inertia energy and the remaining energy is available to the grid while the DC-link voltage and rotor current are kept below the dangerous levels. After grid fault clearance, the control strategy ensures smooth release of the rotor’s excessive inertia energy into the grid. Based on these designs, the DFIG’s reactive power capacity on the stator and the grid side converter is handled carefully to satisfy the new grid code requirements strictly. Simulation studies are presented and discussed. Index Terms—Doubly fed induction generator (DFIG), low voltage ride through (LVRT), power system fault, reactive power.

I. INTRODUCTION O minimize the negative effects of large-scale wind power integration on the reliability of power grids, different countries have defined different low voltage ride through (LVRT) requirements for wind turbines (WT) in their grid codes. Fig. 1(a) combines the LVRT requirements according to German, Scottish and Irish grid codes where WTs must stay connected when the terminal voltage remains inside or above the shadow area [1]. As illustrated in the figure, in Germany,

T

Manuscript received August 05, 2012; revised August 23, 2012, November 25, 2012, and January 05, 2013; accepted January 06, 2013. Date of publication February 04, 2013; date of current version July 18, 2013. This work was supported in part by Hong Kong RGC GRF grants no. 515110 and 528412, in part by Hong Kong Polytechnic University via grants 4-ZZ8Z and G-YK80, and in part by National Natural Science Foundation of China (No. 51207122). Paper no. TPWRS-00912-2012. D. Xie is with the Department of Electrical Engineering, Hong Kong Polytechnic University, Hong Kong, and also with State Grid Electric Power Research Institute, Nanjing, China (e-mail: [email protected]; [email protected]). Z. Xu is with the Department of Electrical Engineering, Hong Kong Polytechnic University, Hong Kong (e-mail: [email protected]). L. Yang is with State Key Laboratory of Electrical Insulation and Power Equipment, Xi’an Jiaotong University, Xi’an 710049, China (e-mail: [email protected]). J. Østergaard is with the Centre for Electric Power and Energy, Department of Electrical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark (e-mail: [email protected]). Y. Xue is with State Grid Electric Power Research Institute, Nanjing, China (e-mail: [email protected]). K. P. Wong is with the University of Western Australia, Crawley 6009, Western Australia (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TPWRS.2013.2240707

Fig. 1. Typical LVRT and reactive power requirements for wind turbines [1]–[4].

E.ON requires the LVRT to be much longer than 1 s. Other examples exist in USA (FERC) and Canada (AESO), where the required LVRT duration can be even longer, and the duration with the minimum voltage at 0.15 p.u. can be more than 500 ms [2]. Traditionally, in order to defend the grid against large disturbances on voltage and frequency, traditional thermal generators must fulfill the LVRT requirement as well as providing emergency power supports. With the increased wind penetration, grid connection requirements for WTs are getting reinforced to require reactive power support by WTs during grid faults as well [1]–[6]. For example, the Spanish grid code demands WTs to start voltage control within 20 ms after fault recognition by providing additional reactive current amounting to at least 2% for each percent of the voltage dip [3]. In Ireland [5] and the U.K. [6], WTs must provide the maximum reactive current during a fault. Fig. 1(b) plots the worst case of reactive current requirements according to Spanish [3] and German [4] codes where a WT’s reactive power support capability must remain inside or right to the shadow area. Doubly fed induction generator (DFIG) based WTs are widely used because of high efficiency, variable speed operation, as well as the use of converters with partial capacity that enables independent control on active and reactive power. To realize LVRT for DFIGs, several technical concerns arisen due to the grid faults must be properly addressed including the overcurrent in the stator and rotor circuits, the overvoltage of the DC-link connecting the rotor side converter (RSC) and the grid side converter (GSC), and the overloading of these converters. To meet the challenging requirements of reactive power support during voltage dips, the reactive power output from DFIGs needs to be increased as much as possible. The existing LVRT solutions for DFIGs can be generally categorized into two types including — The crowbar protection [7]–[11]: Additional crowbar resistors will be inserted into DFIG rotor circuits to limit

0885-8950/$31.00 © 2013 British Crown Copyright

XIE et al.: A COMPREHENSIVE LVRT CONTROL STRATEGY FOR DFIG WIND TURBINES WITH ENHANCED REACTIVE POWER SUPPORT

3303

Fig. 3. Different stages of voltage development during a typical fault.

Fig. 2. Schematic diagram of a DFIG WT system.

the overcurrent and the RSC will be disabled temporarily during fault periods. The detailed crowbar control strategy and its industrial applications have been described in [12]. Although the crowbar solution can guarantee successful fault ride through, it will however involve additional investments, energy dissipations by crowbar resistors, and particularly loss of control to active and reactive power outputs. When the crowbar is activated, the DFIG will absorb large amount of reactive power, which can be harmful to the grid. Therefore, some researches on the improvements of reactive power performance of the crowbar scheme have been conducted [8], [9]. — The demagnetizing method [13]–[17]: In order to decrease rotor current during a fault, demagnetizing methods try to eliminate transients of the induced electromagnetic force in the rotor circuit by controlling the RSC output to trace and counteract the oscillations of stator flux. However, most methods so far are too complicated to be practically implemented by industries. This is mainly due to the limitations in converter capacity, the concerns of the algorithm reliability, and the estimations involved for the control parameters like, e.g., rotor current, rotor flux and rotor voltages, etc. [13]–[15]. Recognizing the limitations of existing LVRT solutions, the authors have successfully developed an innovative control strategy featuring improved energy efficiency and algorithm simplicity [18]. By allowing the DFIG rotor to temporarily accelerate, wind energy continuously captured by a WT during a fault can be stored in the rotor as inertia energy, which will be released back to the grid after fault clearance. The simulation results indicate that the method in [18] can decrease the rotor current effectively. Its performance of limiting rotor current is even better than the crowbar protection, not to mention the fact that the latter dissipates the captured wind energy as heats during the faults. Therefore, the method in [18] is considered with high potentials of industrial applications in the future. Focusing on reinforcing the LVRT capability with reactive power supports, this paper will further develop the control strategy in [18] in order to fully utilize a DFIG’s potentials to generate power effectively and securely considering a wider range of fault scenarios. Different control schemes can be designed for different voltage stages. At stages of voltage falling and recovering, large rotor or stator current spikes due to electromagnetic transients should be suppressed. At

the stage of voltage stabilizing, the basic component of rotor current should be however controlled according to the depth of voltage dip. A staged reactive power controller is proposed to distribute reactive power support between the stator and grid (converter) sides automatically. The proposed control strategy has been implemented using MATLAB/Simulink for a practical 1.5-MW DFIG WT. It has been demonstrated that the new control strategy can fulfill the LVRT requirements and provide enhanced reactive supports to the grid in line with stringent grid code requirements. In addition, simulation results of various test scenarios have not only validated the effectiveness but also the superior performances of the proposed control strategy with respect to both active and reactive power control. II. MODELING OF A DFIG WIND TURBINE The schematic diagram of a grid-connected DFIG WT system is shown in Fig. 2. The system includes the wind turbine, the shaft system, the induction generator, the back-to-back PWM converters and the control system. Inside the DC-link, a dc-chopper is installed to protect the capacitor and converters from overvoltage [12]. The stator of the induction generator and the grid-side converter are directly connected to the grid synchronously. The control system consists of two subsystems for WT and DFIG respectively. The WT controller generates the rotor’s reference speed based on the optimum power-speed characteristic curve in order to capture maximum wind power, which is termed as maximum power point tracking (MPPT) control. During high winds, the pitch control will ensure the WT working at the rated speed until it is stopped at the cut-off wind speed. Based on vector control techniques, the DFIG controller operates rotor and grid side converters to regulate active and reactive power output of the DFIG independently according to the rotor’s reference speed and power factor requirement set by the grid code [19]. In the following analysis and simulation, the detailed fifth-order model of a DFIG WT is used [12]. The used MPPT control scheme is described in [18]. III. PROPOSED CONTROL STRATEGY FOR LVRT AND ENHANCED REACTIVE POWER SUPPORT A power system fault often consists of stages where voltage is falling, stabilizing and recovering, as illustrated in Fig. 3. At stage of voltage falling, the curve of rotor current often includes large spikes. At stage of voltage recovering, the transient rotor current is often of high amplitude. Based on this knowledge, special control schemes should be applied and catered for certain stages. The difficulties of LVRT control techniques for DFIG WTs exist in the rapid change of the stator flux caused by the step

3304

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 3, AUGUST 2013

q-axis current in the rotor immediately. As a result, the increase of the rotor speed finally absorbs wind energy as the rotor inertia energy during the faults. Ignoring the friction losses, the stored mechanical energy in the generator rotor at time is relevant to the increase of rotor speed and can be expressed as follows: (1)

Fig. 4. Control scheme of the rotor side converter in normal operations.

change of grid voltage. This is the main reason for the observed rotor current oscillations. In view of this, the authors have tried to decrease the basic component of the rotor current through cutting off the energy injection into the rotor circuit in [18], a different way with practical merits to achieve the same goal of the crowbar protection. Once activated, however, the crowbar protection will convert the DFIG into a typical induction motor [12]. Many demagnetizing methods make full use of the control loops for both dand q-axis currents in the rotor circuit. Compared with these methods, the LVRT control strategy proposed by the authors in [18] reserves considerable potential reactive power capacity so that a well-designed reactive controller can largely enhance the reactive power delivery to the grid, which is the focus of this paper. The following sections will introduce the newly proposed control strategy in detail, which integrates the LVRT and reactive power control. Simulation validations as well as comparisons with other existing LVRT and reactive power control methods for DFIGs are given in Section IV. A. Rotor Side Control In normal operations, the decoupling of active and reactive power control is realized through the vector control technique with a reference frame oriented along the stator voltage vector [20], resulting in the relatively independent d- and q-axis control loops as shown in Fig. 4. The q-axis loop controls the electromagnetic torque whose input signal is given by a PI controller comparing the actual rotor speed with the reference one. The d-axis loop, corresponding to the control of magnetizing current, uses the desired reactive power as its input signal. The stator flux in Fig. 4 is estimated by the stator voltage [20]. At the stage of voltage recovering, starting when the voltage at the point of common coupling (PCC) increases above certain threshold, the excess inertia energy is sent out to the grid through a step-change signal given by the speed comparator in the q-axis loop. The stored inertia energy will increase due to the increase of the wind speed, the magnitude of voltage drop, or the fault duration time. The energy released in very short time duration will cause overcurrent if without any special control. Therefore, the terms and should be decreased in amplitudes at this stage to prevent overcurrent as soon as a voltage recovery is detected. During voltage dips, the power output of a DFIG is certainly decreased. Accordingly the power input should be reduced as well to restore the power balance, which can be done by reducing the electromagnetic torque. Known from Fig. 4, the decrease of electromagnetic torque will cause the decrease of

is the fault occurrence time, and is the rotor inwhere ertia. With high wind speeds, the pitch control will be immediately triggered once the turbine speed exceeds the rated value, so that the captured wind energy will be reduced effectively. Therefore the negative effect of increased mechanical stress to the WT’s construction can be little as the pitch control will prevent the rotor speed from over-speeding. Under variable speed mode, the MPPT control is realized by setting the generator rotor’s reference speed as [20] (2) where is the output power, is the electromagnetic torque, and is a construction dependent constant. To prevent overcurrent in the stator and rotor, the basic component of active current should be restricted. Accordingly, the maximum allowed active power injected into the generator during a fault is therefore restricted as below, based on (2) (3) is the RMS value of the current PCC voltage, and is the rated value of the voltage. The reference value of q-axis current is expressed as a function of the reference value of electromagnetic torque as in Fig. 4 [20] where

(4) is hard to be measured accurately, since it The stator flux changes rapidly and cannot be controlled directly. Equation (4) indicates that this will lead to errors in the reference value for the rotor current. The first-order partial derivative of on is as follows: (5) Because of (5), the impacts due to inaccurate stator flux could be reduced by decreasing the reference torque , which accordingly slows down the transformation from the mechanical and also magnetic energy to the electrical one in DFIGs. Based on this consideration and (3), the reference value of electromagnetic torque can therefore be derived as (6) is determined to effectively suppress where the factor the current spikes in the rotor and stator at the stage of voltage falling, and ensure safe LVRT. Considering a large may increase the rotor current, the typical value, 0–0.5, is chosen for this stage. At the stage of voltage stabilizing, a too small may

XIE et al.: A COMPREHENSIVE LVRT CONTROL STRATEGY FOR DFIG WIND TURBINES WITH ENHANCED REACTIVE POWER SUPPORT

3305

Fig. 6. Control scheme of the grid side converter in normal operations. Fig. 5. Control scheme of the rotor side converter during voltage dips.

not ensure the rotor speed within the normal range, especially under the high wind scenarios with the required LVRT duration of more than 1 s as shown in Fig. 1(a). With the proposed control, DFIG will continue the active power production during faults which is encouraged by the British grid code [6]. The rotor-side control scheme during voltage dips is illustrated in Fig. 5. The new scheme in Fig. 5 is still a first-order system. Only the measurements of electric variables e.g. the grid voltage are involved. Though the inertia energy is changing during this control process, there is no inertia response [21], [22] considered, measured or directly controlled. Therefore only a small time delay is possibly involved in the closed control loop. PI controllers with large bandwidths would be better for dealing with the fast oscillations of rotor current during voltage dips. However a larger bandwidth controller often has lower stability. The bandwidth of a first-order system can be estimated from the 10%–90% rise time as follows [23]: (7) In the following simulations, PI controllers with a bandwidth of about 35 Hz, i.e., the rise time of 10 ms, are enough to achieve good performances. Similar to the demagnetizing methods, the control performance could also be affected if the voltage magnitude of RSC reaches its limit. Since the proposed method will not fully stop the DFIG from converting the mechanical energy to the electrical one during faults, the RSC voltage limit may not be exceeded at all while keeping the rotor current below 2 p.u. during the short durations of LVRT.

Fig. 7. Control scheme of the grid side converter during voltage dips.

(9) (10) where , and are the currents of the three branches connecting the DC-link circuit and the RSC, the GSC and the capacitor , respectively. is the d-axis component of the grid-side circuit current. is the voltage cross the capacitor . is the instantaneous active power injecting into the DC-link from the rotor side. , nominally 0.75, is the factor of PWM modulation depth [20]. During voltage dips, intense electromagnetic transients will cause large . The PI controller in Fig. 6 cannot follow the rapid changes of the current in time. A compensation term is required to be added to represent the instantaneous variations in . The term is used as the compensation signal in [24], but the stator voltage may reduce to zero, consequently resulting in overcompensation. Based on (8) and (9), the term is therefore added to the d-axis control loop as shown in Fig. 7, in order to smooth the fluctuations of DC-link voltage.

B. Grid Side Control The reference frame of the GSC controller is also oriented along the stator voltage vector based on the vector control technique. The d-axis loop controls the active power in order to maintain a stable DC-link voltage. The q-axis loop controls reactive power generated from the grid side as illustrated in Fig. 6. Under normal operations, the grid-side circuit current is much smaller than the rotor circuit one. However, the grid-side circuit current will rise significantly in low voltage events, where the GSC is an important delivery channel for the remaining active power generated in the rotor circuit, and also a reactive power source. According to Kirchhoff’s circuit laws, the DC-link voltage and related branch currents as denoted in Fig. 2, satisfy (8)

C. Reactive Power Control Due to limited capacity of DFIG converters, the reactive power reference of the GSC is kept zero in normal operations, in order to decrease the current and losses in both the converters. Meanwhile, the response time of GSC is usually faster than RSC [12]. Therefore, the coordination for reactive power generation between stator and GSC can be realized by using the following staged control scheme, where the output signal of the reactive power controller on the RSC is the input to the one of the GSC. Therefore, the needed reactive power control due to PCC voltage variations can be distributed automatically between both sides. It firstly uses the RSC to generate reactive power until the fault becomes so serious that the GSC is also required to supply additional reactive power.

3306

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 3, AUGUST 2013

Fig. 8. Reactive power control scheme of a DFIG WT.

Fig. 10. Schematic diagram of the crowbar protection circuit.

Fig. 9. Single line chart of the test system.

The LVRT is a sort of power system transient security problems under large disturbances, where voltage drops and needed reactive supports can be uncertain and rather fluctuating. Therefore it is hard to obtain U-Q characteristic curves in advance for power grids during those transients. The curves are often used to determine a schedule to distribute reactive power demands among different control resources in normal operations. Consequently, a PCC-voltage feedback control is proposed for WTs to enhance reactive power support in this paper, as illustrated in Fig. 8. In Fig. 8, denotes the reference RMS value of the PCC voltage. and must be lower than the maximum allowed reactive power of the stator and grid side accordingly. IV. SIMULATION RESULTS As shown in Fig. 9, a simple test system has been established consisting of a DFIG and a small transmission system. The DFIG system is based on the models presented in Section II. As shown in Fig. 9, a three-phase symmetric fault happens at the location denoted by the arrow, causing voltage dips at the PCC bus. The following simulations were done in the Matlab/ Simulink platform with all models built using standard blocks. Major model parameters can be found in the Appendix. For comparisons, the DFIG WT has been simulated with 3 different control strategies considering possible scenarios with different voltage dips. The focus is on their ability of LVRT and reactive power support. These strategies are: Strategy A: A crowbar protection equipped with a DC chopper is illustrated in Fig. 10 [12]. The crowbar resistors are all . Strategy B: A fast demagnetizing method proposed in [15]. The control schemes of RSC are given in Fig. 11 for voltage dips and Fig. 4 for normal operations, respectively. is connected to the signal in Fig. 4, while which is a modification from [15] in order to enhance reactive support by this strategy. Strategy C: The proposed control strategy described in Section III. The settings of and in different stages of voltage development are listed in Table I.

Fig. 11. Control scheme of the rotor side converter in Strategy B during voltage dips.

TABLE I PARAMETER SETTINGS IN THE PROPOSED CONTROL STRATEGY

“-” stands for not applicable

For a fair comparison, the same reactive power control strategy included in Strategy C is applied to Strategies A and B. For all the three strategies, the protection thresholds of rotor and stator currents are both set to 1.5 p.u., and the DC-chopper resistor is selected as 0.5 p.u. Based on the worst profile, i.e., the lower boundary in Fig. 1(a), three scenarios have been simulated where low voltage faults last for 1733 ms, 625 ms, or 150 ms if voltage drops 50%, 85%, or 100%, respectively. According to Fig. 1(b), the capacitive reactive current of WTs should be no less than 90% of the rated value in the above three scenarios. A. Scenarios With Longer Time Voltage Dips Fig. 12 plots transient time-response curves of terminal voltage and active power etc. of the tested DFIG WT system with Strategies A, B, and C. The first scenario has a 50% voltage drop, fault duration time of 1733 ms and low wind speed (at 8 m/s). The fault occurs at 4 s and ends at 5.733 s. The ratings for both the generation stator and rotor are listed in the Appendix. Fig. 12 shows that all the control strategies can fulfill the LVRT requirements, e.g., the stator, rotor and grid-side circuit current should be lower than 2 times the rated value, the variance of DC-link voltage must be less than 15% of its rated value, the rotor speed should be within a range from 70% to 130% of the synchronous speed (i.e., 1 p.u.), and the apparent power of both the converters need to be below 30% of the DFIG’s rated power

XIE et al.: A COMPREHENSIVE LVRT CONTROL STRATEGY FOR DFIG WIND TURBINES WITH ENHANCED REACTIVE POWER SUPPORT

3307

Fig. 13. Simulated transient responses of the studied DFIG WT at 50% voltage drop and low wind speed (8 m/s) for control strategies A (crowbar), B (demagnetizing method) and C (the proposed strategy). (a) Variation of total reactive power. (b) Variation of reactive power from the stator side. (c) Variation of reactive power from the GSC. (d) Variation of capacitive reactive current.

stator side during the fault. Strategies B and C provide 50% rated reactive power from the stator side in Fig. 13(b) and give almost 1 p.u. (100% of the rated value) reactive current shown in Fig. 13(d). However, the current Strategy A can provide is much less than the required minimum value, i.e., 90% of the rated value, even if the reactive power of 0.3 p.u. is generated from its GSC. B. Scenarios With Shorter Time Voltage Dips

Fig. 12. Simulated transient responses of the studied DFIG WT at 50% voltage drop and low wind speed (8 m/s) for control strategies A (crowbar), B (demagnetizing method), and C (the proposed strategy). (a) Variation of terminal voltage. (b) Variation of active power. (c) Variation of rotor current. (d)Variation of DC-link voltage. (e) Variation of grid-side circuit current. (f) Variation of rotor speed. (g) Variation of the voltage on RSC.

[12]. The peak value of rotor current in Strategy C is much less than the others, seen in Fig. 12(c), thanks to the control scheme applied at the stage of voltage recovering. These three control strategies lead to more obvious differences in power curves. With Strategy C, over 0.1 p.u. of active power is delivered to the grid during the fault. Strategy A absorbs active power when voltage drops, because the DFIG at this moment has the same behavior as a typical induction motor. This also leads to the differences in the rotor speed in Fig. 12(f), where Strategy C largely suppresses the increase of the rotor speed, compared with the other two strategies. Fig. 12(g) shows the simulated voltage on the RSC, where the voltage magnitude of both Strategies B and C is lower than the rated value, 0.3 p.u., under the scenario. Strategy A introduces relatively high voltage on the rotor because the RSC is disconnected from the rotor circuit. Fig. 13 shows the results of reactive power. Behaving like an induction motor, Strategy A absorbs reactive power from the

The testing scenario in Fig. 14 has an 85% voltage drop, fault duration of 625 ms and high wind speed (at 13 m/s). High wind speed and ultra-low fault voltage contribute to obviously high oscillations in rotor current as shown in Fig. 14(c), leading to the increase of current in the rotor and grid sides as well as the DC-link voltage. Strategy C has lower and smoother rotor current than Strategies A and B at the stage of voltage recovering. The increases of rotor speed in the three strategies are all acceptable and far below 130% of the synchronous speed as shown in Fig. 14(f). Fig. 14(g) still shows that the voltage on the RSC is below its rated value in both Strategies B and C. In Fig. 15, where the results about reactive power and current are shown, only Strategy C meets the requirement with reasonably high capacitive current. The peak reactive current Strategy A can provide is around 80% of the rated value. For Strategy B, it is around 75%. Compared with the results in Fig. 13, the performance of Strategy B in providing reactive power is reduced, due to the conflict of control purposes between operating the rotor current to trace the EMF’s oscillations and generating smooth reactive current. Fig. 16 displays the rotor current and reactive current supplied by the DFIG with the three control strategies when the voltage at the PCC drops to zero. The LVRT requirement can be fulfilled by all the strategies and only Strategy C can satisfy the requirement of capacitive reactive current. Strategy B gives a result around 80% of the rated value with high fluctuations,

3308

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 3, AUGUST 2013

Fig. 15. Simulated transient responses of the studied DFIG WT at 85% voltage drop and high wind speed (13 m/s) for control strategies A (crowbar), B (demagnetizing method) and C (the proposed strategy). (a) Variation of total reactive power. (b) Variation of capacitive reactive current.

Fig. 14. Simulated transient responses of the studied DFIG WT at 85% voltage drop and high wind speed (13 m/s) for control strategies A (crowbar), B (demagnetizing method) and C (the proposed strategy). (a) Variation of terminal voltage. (b) Variation of active power. (c) Variation of rotor current. (d) Variation of DC-link voltage. (e) Variation of grid-side circuit current. (f) Variation of rotor speed. (g) Variation of the voltage on RSC.

and Strategy A supplies almost zero reactive power during the fault. C. Robustness of the Proposed Strategy This section shows the effect of different scenarios and control delays on the performance of the proposed control strategy. Fig. 17 gives the simulated rotor currents when shifting the starting time of the fault to the moment one quarter or half of a cycle after the original one. The resultant rotor current profiles of high similarity indicate that the proposed method is not sensitive to the timing of a fault within a voltage cycle. The proposed strategy is a first-order closed-loop system. It can use the same signals that trigger the crowbar protection to realize the switching between the normal and the new LVRT control schemes shown in Figs. 5 and 7. Fig. 18 compares the rotor current of the proposed strategy (Strategy C) with the crowbar protection (Strategy A) if the fault detection delays for 10 ms. The current spikes at the stage of voltage falling

Fig. 16. Simulated transient responses of the studied DFIG WT at 100% voltage drop and high wind speed (13 m/s) for control strategies A (crowbar), B (demagnetizing method) and C (the proposed strategy). (a) Variation of rotor current. (b) Variation of capacitive reactive current.

increase in both strategies inevitably, but the magnitudes are almost at the same level. V. CONCLUSIONS The proposed control strategy has the following features that distinguish it from other existing LVRT methods: Multi-stage switching: The control topology or parameters will change at different stages of fault voltage development, instead of the simple on-off switching in the crowbar protection. According to above simulations, the design provides better performance for all the stages, e.g., slower rotor acceleration at the stage of voltage stabilizing,

XIE et al.: A COMPREHENSIVE LVRT CONTROL STRATEGY FOR DFIG WIND TURBINES WITH ENHANCED REACTIVE POWER SUPPORT

3309

continuously operating resource of active and reactive power and fulfill stringent grid code requirements during various low voltage events. Being a simple control strategy without additional hardware required, the proposed control is more superior to the existing methods including the classical crowbar protection both in performance and in the improvement of grid reliability. APPENDIX

Fig. 17. Simulated rotor current of the studied DFIG WT at 85% voltage drop and high wind speed (13 m/s) for different starting time of a fault.

Fig. 18. Simulated rotor current of the studied DFIG WT at 85% voltage drop and high wind speed (13 m/s) for control strategies A (crowbar) and C (the proposed strategy) when the fault detection delays for 10 ms.

and smaller rotor current at the stage of voltage falling and recovering; Exclusive reactive power control for RSC: The control loop for the magnetizing current is excluded from the LVRT part of the proposed method, which is often used to generate a component of rotor current in many demagnetizing methods during LVRT. Larger reactive current is therefore obtained in the new design; Scenario-independent GSC control: Different from [24], the compensation term added to the GSC controller, in order to trace the rapid change of terminal voltage, can theoretically work under all LVRT scenarios including the ones with serious voltage dips; Feed-back and staged reactive power control: Simulation results prove that the reactive power of a DFIG-based WT can be well and securely controlled by the proposed novel controller during grid faults, which have not been thoroughly studied in other LVRT methods anyhow; No filtering component involved: There is no low-pass, band-pass or high-pass filters used in the proposed method. Those filters are often seen in many demagnetizing methods, in order to identify specific components from electro-magnetic variables, which sometimes affect the control performance significantly due to the introduced delay. The simulation results showed that the proposed control strategy can enable a safe LVRT of a DFIG based WT as a

The following gives the parameters used in the studied DFIG WT model. WT: Rated wind speed is 12 m/s; inertia constant ; damping coefficient ; shaft stiffness coefficient ; time constant of the pitch servo ; DFIG: Base power, voltage and current are 1.5 MW, 575 V and 1505 A accordingly; stator rated voltage and current are 575 V and 1505 A separately; rotor rated voltage and current are 172.5 V and 1505 A accordingly; rotor rated speed is 1.1 p.u. (the synchronous speed is 1 p.u.); inertia constant ; friction coefficient ; stator resistance ; rotor resistance ; stator leakage inductance ; rotor leakage inductance ; mutual inductance ; the rated voltage on DC-link is 1200 V; the filtering inductance on the grid side . REFERENCES [1] W. Christiansen and D. R. Johnsen, Analysis of Requirements in Selected Grid Codes, Section of Electric Power Engineering, Technical University of Denmark (DTU), 2006, Tech. Rep. [2] M. Tsili and S. Papathanassiou, “A review of grid code technical requirements for wind farms,” IET Renew. Power Gener., vol. 3, no. 3, pp. 308–332, Mar. 2009. [3] Requisitos de respuesta frente a huecos de tension de las instalaciones de produccion de regimen especial, PO 12.3. REE, Spain, Nov. 2005. [4] Grid Code—High and Extra High Voltage, E.ON Netz GmbH. Bayreuth, Germany, Apr. 2006. [5] EirGrid Grid Code, ver. 3.5, EirGrid plc, Ireland, Mar. 2011. [6] The Grid Code, no. 4, Rev. 13, National Grid Electricity Transmission plc, U.K., Jun. 2012. [7] L. Peng, B. Francois, and Y. Li, “Improved crowbar control strategy of DFIG based wind turbines for grid fault ride-through,” in Proc. IEEE 24th Annual Applied Power Electronics Conf. Expo., Washington, DC, USA, Feb. 2009, pp. 1932–1938. [8] A. H. Kasem, E. F. EI-Saadany, H. H. EI-Tamaly, and M. A. A. Wahab, “An improved fault ride-through strategy for doubly fed induction generator-based wind turbines,” IET Renew. Power Gener., vol. 2, no. 4, pp. 201–214, Dec. 2008. [9] J. López, E. Gubía, E. Olea, J. Ruiz, and L. Marroyo, “Ride through of wind turbines with doubly fed induction generator under symmetrical voltage dips,” IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4246–4254, Oct. 2009. [10] M. Kayikçi and J. V. Milanović, “Assessing transient response of DFIG-based wind plants-the influence of model simplifications and parameters,” IEEE Trans. Power Syst., vol. 23, no. 2, pp. 545–554, May 2008. [11] J. Yang, J. E. Fletcher, and J. O’Reilly, “A series-dynamic-resistorbased converter protection scheme for doubly-fed induction generator during various fault conditions,” IEEE Trans. Energy Convers., vol. 25, no. 2, pp. 422–432, Jun. 2010. [12] V. Akhmatov, Induction Generators for Wind Power. Brentwood, CA, USA: Multi-Science, 2005. [13] D. Xiang, L. Ran, P. J. Tavner, and S. Yang, “Control of a doubly fed induction generator in a wind turbine during grid fault ride-through,” IEEE Trans. Energy Convers., vol. 21, no. 3, pp. 652–662, Sep. 2006.

3310

[14] G. Abad, M. A. Rodríguez, J. Poza, and J. M. Canales, “Direct torque control for doubly fed induction machine-based wind turbines under voltage dips and without crowbar protection,” IEEE Trans. Energy Convers., vol. 25, no. 2, pp. 586–588, Jun. 2010. [15] L. Yu, G. Chen, D. Cao, and G. Wu, “Low voltage ride-through control of doubly fed induction generator during symmetric voltage sag,” Elect. Mach. Control, vol. 14, no. 7, pp. 1–6, Jul. 2010. [16] S. Hu, X. Lin, Y. Kang, and X. Zou, “An improved low-voltage ridethrough control strategy of doubly fed induction generator during grid faults,” IEEE Trans. Power Electron., vol. 26, no. 12, pp. 3653–3665, Dec. 2011. [17] J. Liang, W. Qiao, and R. G. Harley, “Feed-forward transient current control for low-voltage ride-through enhancement of DFIG wind turbines,” IEEE Trans. Energy Convers., vol. 25, no. 3, pp. 836–843, Sep. 2010. [18] L. Yang, Z. Xu, J. Østergaard, Z. Y. Dong, and K. P. Wong, “Advanced control strategy of DFIG wind turbines for power system fault ride through,” IEEE Trans. Power Syst., vol. 27, no. 2, pp. 713–722, May 2012. [19] M. Fujimitsu, T. Komatsu, K. Koyanagi, K. Hu, and R. Yokoyama, “Modeling of doubly-fed adjustable-speed machine for analytical studies on long-term dynamics of power system,” in Proc. PowerCon 2000, Int. Conf. Power System Technology, Dec. 2005, vol. 4–7, pp. 25–30. [20] R. Pena, J. C. Clare, and G. M. Asher, “Doubly fed induction generator using back-to-back PWM converters and its application to variable speed wind-energy generation,” IEE Proc. Elect. Power Appl., vol. 143, no. 3, pp. 231–241, May 1996. [21] L. Holdsworth, J. Ekanayake, and N. Jenkins, “Power system frequency response from fixed speed and doubly fed induction generator based wind turbines,” Wind Energy, vol. 7, pp. 21–35, 2004. [22] A. Mullane and M. O’Malley, “The inertial response of induction-machine-based wind turbines,” IEEE Trans. Power Syst., vol. 20, no. 3, pp. 1496–1503, Aug. 2005. [23] L. Harnefors and H. Nee, “Model-based current control of AC machines using the internal model control method,” IEEE Trans. Ind. Appl., vol. 34, no. 1, pp. 133–141, Jan./Feb. 1998. [24] J. Yao, H. Li, Y. Liao, and Z. Chen, “An improved control strategy of limiting the DC-link voltage fluctuation for a doubly fed induction wind generator,” IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1205–1213, May 2008. Dongliang Xie received the M.Sc. degree from Royal Institute of Technology (KTH), Sweden, and the Ph.D. degree in electrical engineering from Southeast University, China, in 2012. During 2011–2012, he was a research associate in Hong Kong Polytechnic University. He is currently working for State Grid Electric Power Research Institute (SGEPRI), State Grid Corporation of China. His research interests include analysis, simulation and control for smart grid architectures and essentials consisting of renewable power generation, power market and power system interactions, and demand elasticity.

Zhao Xu (M’06–SM’12) received the Ph.D. degree in electrical engineering from The University of Queensland, Brisbane, Australia, in 2006. From 2006–2009, he was an Assistant and later Associate Professor with the Centre for Electric Technology, Technical University of Denmark. Since 2010, he joined Hong Kong Polytechnic University. His research interests include demand side, grid integration of wind power, electricity market planning and management. Dr. Xu is now serving as the Secretary of PES/IAS/ PELS/IES Joint Chapter, IEEE Hong Kong Section.

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 3, AUGUST 2013

Lihui Yang received the Ph.D. degree in electrical engineering from Xi’an Jiaotong University, China, in 2010. During 2008–2009, she was a visiting Ph.D. student with the Centre for Electric Technology, Technical University of Denmark. She is currently an Assistant Professor with Xi’an Jiaotong University. Her research interests include stability and control of wind power generation.

Jacob Østergaard (M’95–SM’09) is Professor and Head of the Centre for Electric Power and Energy, Department of Electrical Engineering, Technical University of Denmark. His research interests include integration of renewable energy, control architecture for future power system, and demand side. Prof. Østergaard is serving in several professional organizations including the EU SmartGrids advisory council.

Yusheng Xue (M’87) received the Ph.D. degree in electrical engineering from the University of Liege, Belgium, in 1987. He is now the Honorary President of State Grid Electric Power Research Institute (SGEPRI), State Grid Corporation of China. His research interests include nonlinear stability, control and power system automation. Dr. Xue became a Member of the Chinese Academy of Engineering in 1995.

Kit Po Wong (M’87–SM’90–F’02) received the M.Sc, Ph.D., and higher doctorate D.Eng. degrees from the University of Manchester, Institute of Science and Technology, Manchester, U.K., in 1972, 1974, and 2001, respectively. He is a Winthrop Professor at with the Department of Electrical, Electronic and Computer Engineering, The University of Western Australia. His current research interests include smart grid, power system analysis, planning and operations. Prof. Wong received three Sir John Madsen Medals (1981, 1982, and 1988) from the Institution of Engineers Australia, the 1999 Outstanding Engineer Award from IEEE Power Chapter Western Australia, and the 2000 IEEE Third Millennium Award. He was General Chairman of IEEE/CSEE PowerCon2000, IEE APSCOM 2003 and APSCOM 2009 international conferences. He was an Editor-in-Chief of IEE Proceedings in Generation, Transmission and Distribution and Editor (Electrical) of the Transactions of Hong Kong Institution of Engineers. He is serving as an Editor-in-Chief of IEEE PES Letters.