LVRT and Stability Enhancement of Grid-Tied Wind Farm Using DFIG-Based Wind Turbine

According to the grid code specifications, low voltage ride-through (LVRT) is one of the key factors for grid-tied wind farms (WFs). Since fixed-speed wind turbines with squirrel cage induction generators (FSWT-SCIGs) require an adequate quantity of reactive power throughout the transient period, conventional WF consisting of SCIG do not typically have LVRT capabilities that may cause instability in the power system. However, variable-speed wind turbines with doubly fed induction generators (VSWT-DFIGs) have an adequate amount of LVRT enhancement competency, and the active and reactive power transmitted to the grid can also be controlled. Moreover, DFIG is quite expensive because of its partial rating (AC/DC/AC) converter than SCIG. Accordingly, combined installation of both WFs could be an effective solution. Hence, this paper illustrated a new rotor-side converter (RSC) control scheme, which played a significant role in ensuring the LVRT aptitude for a wide range of hybrid WF consisting of both FSWT-SCIGs and VSWT-DFIGs. What is more, the proposed RSC controller of DFIG was configured to deliver an ample quantity of reactive power to the SCIG during the fault state to make the overall system stable. Simulation analyses were performed for both proposed and traditional controllers of RSC of the DFIG in the PSCAD/EMTDC environment to observe the proposed controller response. Overall, the presented control scheme could guarantee the LVRT aptitude of large-scale SCIG.


Introduction
Currently, it is even more evident that the significant burning of fossil fuels, price volatility, and rising awareness of environmental issues emphasized the importance of renewable and clean energy sources. Moreover, renewable energy sources are free, endless, clean, and readily available. With the installation of a 60.4 GW system in the wind industry, 2019 was a remarkable year as it was the second-biggest year in history. Moreover, the new installation introduced global cumulative wind power (WP) capacity up to 651 GW and represented year-over-year (YoY) growth of 19%. Furthermore, the global wind energy council (GWEC) estimated the addition of a new capacity for more than 355 GW over the next five years, with an annual installation of 71 GW until 2024 [1,2].
Such an enormous integration of WP in the power grid has severe consequences for transient stability. Concerning the stabilization of the system, a wind turbine (WT) needs to satisfy the grid code criteria based on the network condition. Low voltage ride-through (LVRT) is one of the vital grid code requirements for the transient analysis [3]. However, as per the LVRT criteria, WF must not only stay connected with the grid but also provide reactive power to maintain grid voltage throughout faults [4].
Fixed-speed wind turbines with squirrel cage induction generators (FSWT-SCIGs) are the most frequently operated WTs as they are easy to construct, robust, and costeffective [5]. Nevertheless, the SCIG is attached directly to the power grid and contains upon past experience to design fuzzy rules of the designer. To conclude from the above literature reviews, it is better to design a simple and robust controller of DFIG instead of any auxiliary devices for both WT generators and intelligent control to augment the LVRT competency and capacity of grid-tied SCIG based WF.
Thus, the prime contribution of this research is to design a new controller based on the PI controller to control the outer and inner loop of the RSC controller for the DFIG to augment LVRT functionality and enhance the capacity of SCIG based WFs. The outer loop for reactive power is controlled using a voltage-reactive power characteristics curve. The entire system was presented here, along with elaborate modeling and control techniques. Simulation studies were performed on an infinite bus power system network combined with large-scale FSWT-SCIG and small-scale VSWT-DFIG based WF to evaluate the efficacy of the proposed control strategy, and it was compared to that described in [33], which was comprised of a DFIG regulated by a conventional PI-based cascade control system. Therefore, the proposed control technique was found quite efficacious in enhancing LVRT capability and securing the stability of the power system. Furthermore, the installed SCIG's capacity could be increased.
The remaining part of this paper was categorized as follows. Section 2 covered the design of the power system, Section 3 covered the design of the WT, Section 4 covered the DFIG designing and proposed cascaded RSC control system, Section 5 covered the simulation results and a discussion of the effectiveness of the proposed system over conventional methods. Finally, Section 6 illustrated the conclusions. Figure 1 demonstrates the model of the power system considered in this research. The model was constituted of two WT (DFIG 15 MW and SCIG 35 MW) that were coupled to an infinite bus using 0.69 kV/6.6 kV and 6.6 kV/66 kV transformers and a double circuit transmission line. The required reactive power for the SCIG to operate in the steady-state condition was provided by the attached capacitor bank, and its value was chosen in such a way so that it operated at a unity power factor. Some intelligent control techniques are also employed in the RSC of the DFIG system to improve LVRT, for example, fuzzy logic controller [31] and model predictive controller [32]. Although the LVRT competency can be improved, the computational burden for both of these controllers is high. Moreover, the success of the fuzzy controller mainly depends upon past experience to design fuzzy rules of the designer. To conclude from the above literature reviews, it is better to design a simple and robust controller of DFIG instead of any auxiliary devices for both WT generators and intelligent control to augment the LVRT competency and capacity of grid-tied SCIG based WF.

Design of the Power System
Thus, the prime contribution of this research is to design a new controller based on the PI controller to control the outer and inner loop of the RSC controller for the DFIG to augment LVRT functionality and enhance the capacity of SCIG based WFs. The outer loop for reactive power is controlled using a voltage-reactive power characteristics curve. The entire system was presented here, along with elaborate modeling and control techniques. Simulation studies were performed on an infinite bus power system network combined with large-scale FSWT-SCIG and small-scale VSWT-DFIG based WF to evaluate the efficacy of the proposed control strategy, and it was compared to that described in [33], which was comprised of a DFIG regulated by a conventional PI-based cascade control system. Therefore, the proposed control technique was found quite efficacious in enhancing LVRT capability and securing the stability of the power system. Furthermore, the installed SCIG's capacity could be increased.
The remaining part of this paper was categorized as follows. Section 2 covered the design of the power system, Section 3 covered the design of the WT, Section 4 covered the DFIG designing and proposed cascaded RSC control system, Section 5 covered the simulation results and a discussion of the effectiveness of the proposed system over conventional methods. Finally, Section 6 illustrated the conclusions. Figure 1 demonstrates the model of the power system considered in this research. The model was constituted of two WT (DFIG 15 MW and SCIG 35 MW) that were coupled to an infinite bus using 0.69 kV/6.6 kV and 6.6 kV/66 kV transformers and a double circuit transmission line. The required reactive power for the SCIG to operate in the steady-state condition was provided by the attached capacitor bank, and its value was chosen in such a way so that it operated at a unity power factor. The system had 100 MVA as base power and 50 Hz as rated frequency. All the necessary parameters of both generators are presented in Table 1. The system had 100 MVA as base power and 50 Hz as rated frequency. All the necessary parameters of both generators are presented in Table 1.

Design of the Wind Turbine and Drive Train Model
The aerodynamic power output in the WT model is as follows [8,33]: where P ω , ρ, R, V ω , and C p denoted as the extracted wind power, air density, rotor blade radius, wind speed, and power coefficient, respectively. C p can be calculated from the following equation [31]: where λ, T ω , and β represent the tip speed ratio, WT torque, and pitch angle, respectively. Furthermore, c 1 -c 6 present the characteristic coefficients of the WT [34], and ω r presents the rotational speed of the WT. Figure 2 illustrates the C p vs. λ characteristics that are achieved by solving Equation (2) with various β values. The optimum power coefficient (C popt ) and the optimum tip speed ratio (λ opt ) are equal to 0.48 and 8.1, respectively, when the β is equal to zero degrees. Figures 3 and 4 illustrate the designs of the FSWT and VSWT blade pitch control schemes, respectively [35]. The output power of SCIG was controlled by the pitch controller in FSWT so that it did not surpass the rated power and in VSWT, the pitch controller regulates the rotor speed of the DFIG so that it did not surpass the nominal condition. A first-order transfer function consisting of a pitch rate limiter characterized the pitch actuator control loop. Moreover, to control the tracking error, a PI controller was used.    Since the model of the two-mass shaft was adequate for the dynamic study of gridtied WTs, therefore, it was used in this research as indicated in Figure 5 [36]. The motion laws are defined as: Here, mechanical torque and equivalent turbine blade inertia transferred to the generator side is denoted as and , respectively. Electromagnetic torque is denoted as   Since the model of the two-mass shaft was adequate for the dynamic study of gridtied WTs, therefore, it was used in this research as indicated in Figure 5 [36]. The motion laws are defined as: Here, mechanical torque and equivalent turbine blade inertia transferred to the generator side is denoted as and , respectively. Electromagnetic torque is denoted as   Since the model of the two-mass shaft was adequate for the dynamic study of gridtied WTs, therefore, it was used in this research as indicated in Figure 5 [36]. The motion laws are defined as: Here, mechanical torque and equivalent turbine blade inertia transferred to the generator side is denoted as and , respectively. Electromagnetic torque is denoted as Since the model of the two-mass shaft was adequate for the dynamic study of grid-tied WTs, therefore, it was used in this research as indicated in Figure 5 [36]. The motion laws are defined as: , the rotational speed of turbine and generator are denoted as and , sequentially. represents the shaft angular displacement between the two ends. The stiffness of the shaft is denoted as , and stands for the generator inertia.   Figure 7 illustrates the arrangement of VSWT-DFIG together with its control strategy. Here, an aerodynamic WT scheme consisting of a drive train, pitch angle controller, wound rotor induction generator (WRIG), and an AC/DC/AC converter formed from two levels of insulated gate bipolar transistors (IGBTs) operated by the rotor side and grid side controllers comprised the overall system. Nevertheless, WRIG was connected to the WT, which transfigured the WP into electrical power. The rotor position ( ) and

Design of the DFIG System
were computed using the WRIG rotor shaft. As shown in Figure 7, the stator terminal was firmly tied to the power grid, while the rotor was attached to the grid using RSC and GSC. The rating of the power converter was 30% of the WRIG. The pulse width modulation (PWM) approach was imposed in this research, and the carrier frequency was taken 3.0 kHz for AC/DC/AC converter. The DC-Link circuit contained a DC chopper. The comparator block controlled it, enabling the DC chopper switch whenever the DC-link voltage exceeded or equaled the predetermined limit (Vdc ≥ 1.15 pu). Moreover, the rated DC-Link voltage was 1.2 kV. Here, mechanical torque and equivalent turbine blade inertia transferred to the generator side is denoted as T t and H t , respectively. Electromagnetic torque is denoted as T e , the rotational speed of turbine and generator are denoted as ω t and ω g , sequentially. θ s represents the shaft angular displacement between the two ends. The stiffness of the shaft is denoted as K s , and H g stands for the generator inertia.
The curve of the maximum power point tracking (MPPT) for the VSWT-DFIG is illustrated in Figure 6 [33]. , the rotational speed of turbine and generator are denoted as and , sequentially. represents the shaft angular displacement between the two ends. The stiffness of the shaft is denoted as , and stands for the generator inertia.   Figure 7 illustrates the arrangement of VSWT-DFIG together with its control strategy. Here, an aerodynamic WT scheme consisting of a drive train, pitch angle controller, wound rotor induction generator (WRIG), and an AC/DC/AC converter formed from two levels of insulated gate bipolar transistors (IGBTs) operated by the rotor side and grid side controllers comprised the overall system. Nevertheless, WRIG was connected to the WT, which transfigured the WP into electrical power. The rotor position ( ) and

Design of the DFIG System
were computed using the WRIG rotor shaft. As shown in Figure 7, the stator terminal was firmly tied to the power grid, while the rotor was attached to the grid using RSC and GSC. The rating of the power converter was 30% of the WRIG. The pulse width modulation (PWM) approach was imposed in this research, and the carrier frequency was taken 3.0 kHz for AC/DC/AC converter. The DC-Link circuit contained a DC chopper. The comparator block controlled it, enabling the DC chopper switch whenever the DC-link voltage exceeded or equaled the predetermined limit (Vdc ≥ 1.15 pu). Moreover, the rated DC-Link voltage was 1.2 kV.  Figure 7 illustrates the arrangement of VSWT-DFIG together with its control strategy. Here, an aerodynamic WT scheme consisting of a drive train, pitch angle controller, wound rotor induction generator (WRIG), and an AC/DC/AC converter formed from two levels of insulated gate bipolar transistors (IGBTs) operated by the rotor side and grid side controllers comprised the overall system. Nevertheless, WRIG was connected to the WT, which transfigured the WP into electrical power. The rotor position (θ r ) and ω r . were computed using the WRIG rotor shaft. As shown in Figure 7, the stator terminal was firmly tied to the power grid, while the rotor was attached to the grid using RSC and GSC. The rating of the power converter was 30% of the WRIG. The pulse width modulation (PWM) approach was imposed in this research, and the carrier frequency was

DFIG Model
The DFIG system was specified in the synchronous reference frame. The d and q-axis stator and rotor voltages are derived below: Here, the resistance and self-inductance of stator and rotor winding are illustrated as , , and , , sequentially. is denoted as mutual inductance. The currents of the stator and rotor are denoted as , , , and , respectively. Finally, is presented as the angular frequency of the grid.

DFIG Model
The DFIG system was specified in the synchronous reference frame. The d and q-axis stator and rotor voltages are derived below: Here, the resistance and self-inductance of stator and rotor winding are illustrated as R s , R r , and L st , L rt , sequentially. L m is denoted as mutual inductance. The currents of the stator and rotor are denoted as i sd , i sq , i rd , and i rq , respectively. Finally, ω s is presented as the angular frequency of the grid.

Proposed Rotor-Side Converter (RSC) Control Scheme
The RSC was controlled by the RSC controller, as presented in Figure 7. The control mechanism involved four PI controllers to balance numerous errors. The outer loops were responsible for extracting MPPT power from the DFIG and controlling reactive power. The inner loops regulated the dand q-axis currents. The turbine, rotor, and stator losses were taken into account as stated in the below equations to regulate active power.
Here, P s , P r , and P tl represent the stator, rotor, and turbine losses, respectively. By considering all these losses, the active power extracted could be more realistic.
In this study, the proposed controller focused on the maximization of reactive power injection. This was accomplished by regulating the reactive power based on the voltage dip at the common coupling point. The reference reactive power generated based on the voltage dip can be computed as follows [37,38]: where S represents the rated apparent power, V and V 0 represent the terminal voltage and pre-fault voltage (1.0 pu) of DFIG, respectively. Based on the equation, when terminal voltage was 1.0 pu at the steady-state period, the reactive power injection would be zero; thus, it ensured the unity power factor operation. Further, based on Equation (14), the voltage vs. reactive power characteristics curve was derived in Figure 8. Based on the diagram, the reactive power could be generated up to 0.5824 pu at the voltage level of 0.836 pu.

Proposed Rotor-Side Converter (RSC) Control Scheme
The RSC was controlled by the RSC controller, as presented in Figure 7. The control mechanism involved four PI controllers to balance numerous errors. The outer loops were responsible for extracting MPPT power from the DFIG and controlling reactive power. The inner loops regulated the d-and q-axis currents. The turbine, rotor, and stator losses were taken into account as stated in the below equations to regulate active power.
= 0.01 Here, Ps, Pr, and Ptl represent the stator, rotor, and turbine losses, respectively. By considering all these losses, the active power extracted could be more realistic.
In this study, the proposed controller focused on the maximization of reactive power injection. This was accomplished by regulating the reactive power based on the voltage dip at the common coupling point. The reference reactive power generated based on the voltage dip can be computed as follows [37,38]: where S represents the rated apparent power, and 0 represent the terminal voltage and pre-fault voltage (1.0 pu) of DFIG, respectively. Based on the equation, when terminal voltage was 1.0 pu at the steady-state period, the reactive power injection would be zero; thus, it ensured the unity power factor operation. Further, based on Equation (14), the voltage vs. reactive power characteristics curve was derived in Figure 8. Based on the diagram, the reactive power could be generated up to 0.5824 pu at the voltage level of 0.836 pu. During the fault period, the main objective was to maximize the reactive power injection rather than active power to limit the apparent power rating. Therefore, a comparator block was also embedded in the proposed control mechanism. This comparator sent a zero signal to make the active power transfer to gird zero when the terminal voltage dropped 90% of its rated value. The active power resumed its delivery to the grid when the terminal voltage was above 90%. By following this technique, the reactive power delivered could be enhanced, and effective LVRT competency could be ensured. During the fault period, the main objective was to maximize the reactive power injection rather than active power to limit the apparent power rating. Therefore, a comparator block was also embedded in the proposed control mechanism. This comparator sent a zero signal to make the active power transfer to gird zero when the terminal voltage dropped 90% of its rated value. The active power resumed its delivery to the grid when the terminal voltage was above 90%. By following this technique, the reactive power delivered could be enhanced, and effective LVRT competency could be ensured. Figure 7 also illustrates the GSC controller and compensates for various error signals, and it contains four PI controllers. Additionally, d-axis (I gd ) and q-axis (I gq ) current components regulated the GSC reactive power (Q) and DC-link voltage (V dc ), sequentially. Additionally, the reference reactive power was fixed to zero, and the reference DC-link voltage was fixed to 1.2 kV (1.0 pu).

Simulation Results
PSCAD/EMTDC software was used to perform the simulation analysis on the system depicted in Figure 1. Moreover, to analyze the simulation performance, the most severe triple-line-to-ground (3LG) fault was appraised to be a network disruption near the infinite bus with a simulation run time of 10 s that is illustrated in Figure 1. The fault arose at 0.1 s and the total fault time span was 0.1 s. At 0.2 s, the circuit breakers (CBs) on the faulted line were tripped to detach it from the power grid. The fault was assumed to have been cleared, and the CBs were reclosed at 1.0 s. Since the transient stability analysis period was so short, it was assumed that the wind speed did not vary significantly during this interval. As a consequence, the wind speed utilized to individual WT remained constant.
To justify the feasibility of the proposed PI-controlled RSC controller of DFIG, simulations were conducted for two cases: Case 1: Conventional cascaded scheme of RSC of DFIG presented in [33]. Case 2: Proposed RSC controller of DFIG illustrated in Section 4. The responses to reactive power from DFIG and SCIG are indicated in Figures 9 and 10. It could be observed that during the transition phase, the DFIG provided SCIG with an adequate percentage of reactive power in Case 2. Therefore, the voltage at the connection point recovered more rapidly to the rated condition in Case 2, and the system became stable, which was presented in Figure 11. The reactive power in Case 2 rose significantly as soon as the terminal voltage decayed at 0.1 s (approximately), and it supplied around 12 MVAR until the terminal voltage reached the normal condition, as indicated in Figures 9  and 11. When the terminal voltage resumed to 1.0 pu, the reactive power also reached its pre-fault condition.  Figure 7 also illustrates the GSC controller and compensates for various error signals, and it contains four PI controllers. Additionally, d-axis (Igd) and q-axis (Igq) current components regulated the GSC reactive power (Q) and DC-link voltage (Vdc), sequentially. Additionally, the reference reactive power was fixed to zero, and the reference DC-link voltage was fixed to 1.2 kV (1.0 pu).

Simulation Results
PSCAD/EMTDC software was used to perform the simulation analysis on the system depicted in Figure 1. Moreover, to analyze the simulation performance, the most severe triple-line-to-ground (3LG) fault was appraised to be a network disruption near the infinite bus with a simulation run time of 10 s that is illustrated in Figure 1. The fault arose at 0.1 s and the total fault time span was 0.1 s. At 0.2 s, the circuit breakers (CBs) on the faulted line were tripped to detach it from the power grid. The fault was assumed to have been cleared, and the CBs were reclosed at 1.0 s. Since the transient stability analysis period was so short, it was assumed that the wind speed did not vary significantly during this interval. As a consequence, the wind speed utilized to individual WT remained constant.
To justify the feasibility of the proposed PI-controlled RSC controller of DFIG, simulations were conducted for two cases: Case 1: Conventional cascaded scheme of RSC of DFIG presented in [33]. Case 2: Proposed RSC controller of DFIG illustrated in Section 4. The responses to reactive power from DFIG and SCIG are indicated in Figures 9 and  10. It could be observed that during the transition phase, the DFIG provided SCIG with an adequate percentage of reactive power in Case 2. Therefore, the voltage at the connection point recovered more rapidly to the rated condition in Case 2, and the system became stable, which was presented in Figure 11. The reactive power in Case 2 rose significantly as soon as the terminal voltage decayed at 0.1 s (approximately), and it supplied around 12 MVAR until the terminal voltage reached the normal condition, as indicated in Figures  9 and 11. When the terminal voltage resumed to 1.0 pu, the reactive power also reached its pre-fault condition.    However, in Case 1, as described in Figures 9 and 10, the DFIG could not supply SCIG with a sufficient amount of reactive power. As a consequence, the terminal voltage, in this case, could not regain the rated value. Hence, the grid code requirement was not fulfilled in Case 1, and the system was unstable. Figure 12 depicts the rotor speed response of the SCIG, which revealed that in Case 2, the response was stable. However, in Case 1, it was unstable. This was because the capacitor bank, which was attached to the terminal of SCIG, was not enough to inject sufficient reactive power at the transient state. Thus, the developed electromagnetic torque decreased, and the rotor speed increased, as indicated in Case 1. However, in Case 2, the proposed method of DFIG helped to inject reactive power from the DFIG and made the rotor speed stable. Figures 13 and 14 represent the reactive and active power response of the connection point. It was seen from both figures; the reactive and active powers were stable in Case 2.  However, in Case 1, as described in Figures 9 and 10, the DFIG could not supply SCIG with a sufficient amount of reactive power. As a consequence, the terminal voltage, in this case, could not regain the rated value. Hence, the grid code requirement was not fulfilled in Case 1, and the system was unstable. Figure 12 depicts the rotor speed response of the SCIG, which revealed that in Case 2, the response was stable. However, in Case 1, it was unstable. This was because the capacitor bank, which was attached to the terminal of SCIG, was not enough to inject sufficient reactive power at the transient state. Thus, the developed electromagnetic torque decreased, and the rotor speed increased, as indicated in Case 1. However, in Case 2, the proposed method of DFIG helped to inject reactive power from the DFIG and made the rotor speed stable. Figures 13 and 14 represent the reactive and active power response of the connection point. It was seen from both figures; the reactive and active powers were stable in Case 2. However, in Case 1, as described in Figures 9 and 10, the DFIG could not supply SCIG with a sufficient amount of reactive power. As a consequence, the terminal voltage, in this case, could not regain the rated value. Hence, the grid code requirement was not fulfilled in Case 1, and the system was unstable. Figure 12 depicts the rotor speed response of the SCIG, which revealed that in Case 2, the response was stable. However, in Case 1, it was unstable. This was because the capacitor bank, which was attached to the terminal of SCIG, was not enough to inject sufficient reactive power at the transient state. Thus, the developed electromagnetic torque decreased, and the rotor speed increased, as indicated in Case 1. However, in Case 2, the proposed method of DFIG helped to inject reactive power from the DFIG and made the rotor speed stable.        Figure 17 illustrated the rotor speed response of DFIG.     Figure 17 illustrated the rotor speed response of DFIG.     Depending on the above observations, it was possible to conclude that the proposed DFIG controller could effectively stabilize the grid-tied WF during transient periods. Furthermore, it ensured LVRT capability.

Conclusions
The current study proposed partial integration of the DFIG with the proposed rotorside controller based on the PI controller to improve the LVRT effectiveness of the SCIG-    Depending on the above observations, it was possible to conclude that the proposed DFIG controller could effectively stabilize the grid-tied WF during transient periods. Furthermore, it ensured LVRT capability.

Conclusions
The current study proposed partial integration of the DFIG with the proposed rotorside controller based on the PI controller to improve the LVRT effectiveness of the SCIG- Depending on the above observations, it was possible to conclude that the proposed DFIG controller could effectively stabilize the grid-tied WF during transient periods. Furthermore, it ensured LVRT capability.

Conclusions
The current study proposed partial integration of the DFIG with the proposed rotorside controller based on the PI controller to improve the LVRT effectiveness of the SCIGbased WF. Furthermore, a comparison of the suggested and conventional rotor side controllers was performed. According to the simulation outcomes and performance evaluations, the following aspects of the proposed approach were remarkable: (1) The proposed controller of DFIG with a minimum power rating could stabilize the SCIG with a higher power rating during a fault state. Therefore, the combined installation of the proposed DFIG consisting of relatively low power rating with SCIG based WF could enhance the LVRT capability as well as the system stability.
In future work, the frequency control method of hybrid WF could be a strong candidate.