Enhancement of LVRT Ability of DFIG Wind Turbine by an Improved Protection Scheme with a Modified Advanced Nonlinear Control Loop

Abstract

One of the problems with the doubly-fed induction generator (DFIG) is its high vulnerability to network perturbations, notably voltage dips, because of its stator windings being coupled directly to the network. As the DFIG’s stator and rotor are electromagnetically mated, the stator current peak occurs during a voltage dip causing an inrush current to the critical converter back-to-back and an overload of the DC-link capacitor. For this purpose, a series of researchers have achieved a linear and non-linear controller with a crowbar-based protection scheme. With this type of protection, the Rotor Side Converter (RSC) is disconnected momentarily, and consequently, its control of both the active and reactive output power of the stator is totally lost, leading to incorrect power quality at the point of common coupling (PCC). In this document, a robust nonlinear controller by Advanced Backstepping with Integral Action Control (ABIAC) is initially employed to monitor the rotor and the network side converters under normal network operations. In the presence of a network fault, an improved protection scheme (IPS) is tacked on to the robust nonlinear control to help enforce the behavior of the DFIG system to be able to overcome the fault. The IPS, which is formed by a crowbar and an RL series circuit, is typically located in the space between the rotor coils and the RSC converter. Compared to a standard crowbar, the developed scheme is successful to limit the rotor transient current and DC-link voltage, also an RSC disengagement to rotor windings can be prevented during the fault. Furthermore, the controllers of both the RSC and the Network Side Converter (NSC) are modified to boost the supply voltage at the PCC. A comparative study is also performed between the IPS without and with modification of the reactive power control loops. The simulation results mean that with the modified controllers during the fault, the amount of reactive power sustainment with ABIAC at the PCC is optimized to 17.5 MVAr instead of 15 MVAr with proportional-integral control (PIC). Therefore, the voltage at the PCC is fort increased in order to comply with the voltage requirements of the farm and absolutely to maintain the connection to the network in case of voltage dip.

1. Introduction

The production of electricity is rapidly moving towards natural energy sources: photovoltaic, wind, biomass, hydroelectricity, and tides, because of the negative impact of the combustion of energy from fossil fuels on the biological environment. Wind-generated energy is one of the best sources of energy because it is usually available 24 h a day [1,2]. There are three main motivations for the DFIG to gain considerable importance for wind energy retrievers. Primarily, their rotor inverters can be designed to handle only 25 to 30 percent of the total generator power rating [3]. The second is that it is able to work over a wider range of wind speeds. Thirdly, it is capable of operating with a rotor speed between 30% above and 30% below synchronous speed [4]. This helps to capture more energy even when the wind speed is low.

In addition, the rotor converters may be commanded to take their turbine unit to a working location where the power extraction is optimal, by a technique named maximum power point tracking (MPPT). To achieve control of the turbine, a DFIG model is modified through the use of various transformations as the stator voltage orientation (SVO) [5] or the stator flux orientation (SFO) [6]. The SVO control in this document makes it possible to independently control the direct and quadrature current values of the rotor and the filter. The stator powers are controlled by the RSC to follow the active power reference and to adjust the unit power factor at the stator windings. On the other side of the DC-link capacitor, the NSC is used to keep the voltage of the DC-link at the rated value and to cancel the reactive power generated by the network filter. In the existing review of the published work, the orientation frameworks have been combined with the Proportional-Integral controller (PIC) [7] successfully, Sliding Mode Controller [8], Backstepping Controller [9,10], and Fuzzy Logic [11] to effectively manage the DFIG under stable network conditions. However, the DFIG is vulnerable to network perturbations, and the appearance of a network defect, such as a drop in the network voltage, causes fluctuations in the stator current, pulse in terms of both active and reactive stator power, and electromagnetic torque [12,13]. In addition, the existing powerful electromagnetic mated between the stator and the rotor makes the rotor inrush current readily spread to the fragile power inverters [14,15]. When this occurs, the DC-link condenser is surcharged, and the rotor inverters can be out of their safe area. As a result, the inverters are physically destroyed and power monitoring of the DFIG is unavailable even after the fault is removed [16]. The original approach to prevent converter destruction in the event of a network fault was to separate the DFIG from the network. Because of the massive integration of wind turbines today in the network, this practice places the restoration of the network voltage at risk. Therefore, network owners have initiated stringent network codes [17,18] which demand wind turbines to stay attached in the event of a network voltage dip and, in certain situations, to provide reactive power to the network to facilitate the restoration of the network voltage. To keep the wind turbine attached to the network if a voltage dip occurs, the wind system requires incorporating techniques to improve its Low-Voltage Ride Through (LVRT) ability.

Multiple LVRTs in the published works have been suggested. These can be classified according to the insertion of added protective equipment [19,20,21,22], the implementation of a reactive energy supply system [23,24,25], and the application of control strategies [26,27]. The commonly used protection equipment is the “crowbar” type resistor. It consists of a single set of a bank of resistors which are directly connected in parallel with the rotor windings through an IGBT (Insulated Gate Bipolar Transistor) commutator. During the fault, the rotor windings are not connected to the RSC and the collector rings are commutated to the crowbar resistors. This crowbar may be combined with other protection schemes like a DC-chopper connected to the DC-link condenser in parallel to reduce the overload of the condenser [28,29], and a dynamic series resistor (DSR) placed in series from the stator windings to increase the stator voltage [30,31]. Oftentimes, these types of protective equipment are used together with the PIC [32,33]. Under network faults, the PIC, due to the nonlinear behavior of the system at these conditions, is not able to manage the current surges effectively at the time of fault onset and removal. Furthermore, the crowbar is an excellent practice to protect the rotor side converter [34]. However, with a crowbar engaged, the vector control of RSC on the DFIG is unavailable, so the machine takes its current magnetization through the stator face. Consequently, a high value of reactive energy is consumed, which leads to a drop in the voltage of points common to the coupling (PCC) [35]. Additionally, the capacity of turbine generators to continue to be attached to the network under voltage dip faults is considered as the low voltage ride-through (LVRT) ability [36,37]. Usually, a specification of LVRT is given as the curve of voltage/time where the turbines should stay attached to the network. Figure 1 illustrates this paper’s considered LVRT curve, which requires wind generators to maintain attachment to the power system during a voltage dip fault where the PCC tension stays over the solid line in red [38].

Consequently, the problems of rotor high current and DC-link excess voltage must be overcome by the DFIG on the system to satisfy the fault-passing demands. Additionally, it should restrain oscillations of the electromagnetic couple in the time of transient reply for increasing the durability of the gear and the efficiency of the mechanism, as reported by [39]. Another key aspect addressed in this paper is the importance of maintaining the PCC voltage at a suitable level to the ensure uninterrupted operation of the DFIG-wind turbine plant during disturbances.

This paper presents a new method to protect and control DFIG wind turbines during voltage drops in the High Voltage (HV) power grid. This method, called IPS, uses an RL device added to the conventional crowbar circuit to improve the LVRT capability and limit the impact on the wind turbine behavior. At the same time, an advanced control strategy called ABIAC is proposed for the RSC and NSC controllers. This strategy uses stability in the Lyapunov sense [39] to better maintain the voltage at the PCC point during voltage drops. The results of ABIAC are compared with another control scheme called PIC [35]. Using IPS and ABIAC, there is no loss of control of the stator’s active and reactive power, which is advantageous compared to traditional methods. Another point of this contribution is that modifying the RSC and NSC reactive power control loops with ABIAC improves the LVRT capability of the DFIG and better supports the PCC voltage.

Later sections of the paper are arranged as listed below: The description and mathematical modeling of the wind power system (electrical part) are given in Section 2. The AABIAC-based robust control strategy is synthesized and applied to the wind energy system in Section 3. Section 4 presents our proposed DFIG excess rotor current protection scheme and its results in MATLAB/Simulink validation by employing a DFIG–2MW compared to the classic crowbar during a symmetric voltage dip in the network. In another development, the modification of the ABIAC loops of power reactive to support the voltage at the PCC is reported in Section 5, accompanied by a comparison of its results with PIC and a conclusion in Section 6.

2. System Description and Modeling

2.1. Description of the System

The system considered in this paper consists of a three-bladed turbine connected to the wound rotor induction generator or DFIG by means of a speed multiplier (G), the stator of the DFIG is directly connected to the three-phase 60 KV electrical grid with step-up transformers, while its rotor windings are coupled to the 60 KV grid through static variable frequency AC/DC/AC converters (AC: Alternatif Current and DC: Direct Current) consisting of the RSC, DC-link, NSC, and R_fL_f filter. Furthermore, a protection system is placed between the rotor and the RSC, together with a DFIG control unit by RSC and NSC is applied against disturbances in the form of a voltage dip on the electrical network (see Figure 2).

The power in the DC-link condenser and the total power output from the DFIG is the addition of the powers of the stator and the rotor, as indicated in the equations listed as follows.

$$\left\{\begin{array}{l}{P}_{dc}=P_f-P_r\\ P_{pcc}=P_s+P_f\\ Q_{pcc}=Q_f+Q_s\end{array}\right.$$

(1)

2.2. Power Model of DFIG and RSC in the Reference Framedirect-Quadrature (d-q)

Taking the synchronized rotation reference framework d-qat the angular velocity ω_s, the DFIG general dynamic model is found in (2)–(5) [40,41].The flux equations for the stator and rotor in terms of currents and inductances are written in (2) [42].

$$\left\{\begin{array}{l}{\psi }_{sd}=L_s{i}_{sd}+M{i}_{rd}\\ {\psi }_{sq}=L_s{i}_{sq}+M{i}_{rq}\\ {\psi }_{rd}=L_r{i}_{rq}+M{i}_{sd}\\ {\psi }_{rq}=L_r{i}_{rq}+M{i}_{sq}\end{array}\right.$$

(2)

Additionally, stator and rotor winding voltages and currents are described by the following equation [43,44]:

$$\left\{\begin{array}{l}{v}_{sd}=R_s{i}_{sd}+\frac{d{\psi }_{sd}}{dt}-{\omega }_s{\psi }_{sq}\\ v_{sq}=R_s{i}_{sq}+\frac{d{\psi }_{sq}}{dt}+{\omega }_s{\psi }_{sd}\\ v_{rd}=R_r{i}_{rd}+\frac{d{\psi }_{rd}}{dt}-{\omega }_{slip}{\psi }_{rq}\\ v_{rq}=R_r{i}_{rq}+\frac{d{\psi }_{rq}}{dt}+{\omega }_{slip}{\psi }_{rd}\end{array}\right.$$

(3)

where the slip speed (ω_slip) is assumed to be the result of the difference of angular speed of the stator (ω_s) and the number of pole pairs (p) multiplied by the speed of the rotor (ω_m), as in (4).

$${\omega }_{slip}={\omega }_s-p{\omega }_m$$

(4)

Furthermore, the instantaneous stator active and reactive powers in terms of current and voltage are given in (5).

$$\left\{\begin{array}{l}{P}_s=1.5(v_{sd}{i}_{sd}+v_{sq}{i}_{sq})\\ Q_s=1.5(v_{sq}{i}_{sd}-v_{sd}{i}_{sq})\\ P_r=1.5(v_{rd}{i}_{rd}+v_{rq}{i}_{rq})\\ Q_r=1.5(v_{rq}{i}_{rd}-v_{rd}{i}_{rq})\end{array}\right.$$

(5)

In order to control the active and reactive powers separately, we use a phase-locked loop (PLL) to determine the direction of the stator voltage by orienting the reference frame’s d-axis with the stator voltage’s direction. It signifies the result v_sq = 0. Under the premise of a stable network connection to the DFIG, the derivatives of ψ_sd and ψ_sq can be considered zero.

If we neglect the stator resistor, this assumption is reasonable for heavy-duty wind turbines, so v_sd = −ω_sψ_sq. The stator power dynamics can be deduced from the previous equations and expressed in (6):

$$\left\{\begin{array}{l}{\dot{P}}_s=1.5\frac{{\omega }_r{\psi }_{s}M}{L_s}(\frac{v_{rd}}{\sigma L_r}-\frac{R_r{i}_{rd}}{\sigma L_r}+{\omega }_r{i}_{rq}+\frac{M{\omega }_r{\psi }_s}{\sigma L_r{L}_s})\\ {\dot{Q}}_s=-1.5\frac{{\omega }_r{\psi }_{s}M}{L_s}(\frac{v_{rq}}{\sigma L_r}-\frac{R_r{i}_{rq}}{\sigma L_r}-{\omega }_r{i}_{rd})\end{array}\right.$$

(6)

where σ = 1 − M²/(L_sL_r) is the leakage factor.

2.3. Power Model of NSC in the Reference Frame d-q

However, we concentrate in this subsection on the power modeling of the RSC to the electrical network through the R_fL_f filter. The voltages and powers generated by the NSC after the d-q transformation are defined by the following:

$$\left\{\begin{array}{l}{v}_{fd}=R_f{i}_{fd}+L_f\frac{d{i}_{fd}}{dt}-{\omega }_s{L}_f{i}_{fq}+v_{pcc\_d}\\ v_{fq}=R_f{i}_{fq}+L_f\frac{d{i}_{fq}}{dt}+{\omega }_s{L}_f{i}_{fd}+v_{pcc\_q}\end{array}\right.$$

(7)

$$\left\{\begin{array}{l}{P}_{dc}=P_f-P_r\\ P_f=1.5(v_{pcc\_d}{i}_{fd}+v_{pcc\_q}{i}_{fq})\\ Q_f=1.5(v_{pcc\_q}{i}_{fd}-v_{pcc\_d}{i}_{fq})\end{array}\right.$$

(8)

where P_dc is the active power delivered to the terminals of the DC-link condenser.

The orientation of the reference frame d-axis is the same as the angle position for the mains voltage. Since the voltage amplitude of the network is kept constant, v_{pcc_d} is constant (=V_pcc) and v_{pcc_q} is zero (=0). Thus, the power dynamics generated by the NSC can be obtained from the previous equations and expressed in (9).

$$\left\{\begin{array}{l}\frac{d\left(\frac{V_{dc}^2}{2}\right)}{dt}=\frac{1}{C_{dc}}(P_f-P_r)\\ {\dot{P}}_f=1.5V_{pcc}(-\frac{R_f}{L_f}{i}_{fd}-{\omega }_s{i}_{fq}+\frac{V_{pcc}}{L_f}-\frac{v_{fd}}{L_f})\\ {\dot{Q}}_f=-1.5V_{pcc}(-\frac{R_f}{L_f}{i}_{fq}-{\omega }_s{i}_{fd}-\frac{v_{fq}}{L_f})\end{array}\right.$$

(9)

3. Advanced Nonlinear Control Strategy

3.1. Power Control on the Rotor Side

Concerning the power inverter on the rotor input side, the control of the active and reactive stator power will be conceived. The ABIAC ensures that the stator’s active and reactive output power (=0 KVAr) follows its references. The ABIAC is an advanced variant of the classic backstepping controller and is rendered more reliable by considering the integral value of the variable of error. It is used for speed and reactive power control effectively by [41]. It was found to be robust in the face of parameter variations. Using the following variables, Equation (6) required for the design of the controller are made compact for presentation purposes: α = 1.5(Mω_rψ_s)/L_s, γ = 1/(σL_r), τ = R_r/(σL_r) and δ = (Mω_rψ_s)/(σL_sL_r). Then, the Equation (8) are transformed into (10).

$$\left\{\begin{array}{l}{\dot{P}}_s=\alpha (\gamma v_{rd}-\tau i_{rd}+{\omega }_r{i}_{rq}+\delta )\\ {\dot{Q}}_s=-\alpha (\gamma v_{rq}-\tau i_{rq}-{\omega }_r{i}_{rd})\end{array}\right.$$

(10)

Stator Active and Reactive Power Control

On the stator side of the generator, the reactive power reference (Q_s^*) is made to equal zero to guarantee a unity power factor, however during the voltage drop of the network, when the DFIG must deliver reactive power to the network to help it recover its voltage, Q_s^* is non-zero. Additionally, the active power reference (P_s^*) is always given through the MPPT strategy and is realized by [40,41,42,43,44].

Step 1: The tracking error of the active power and its derivative are expressed in (11) and (12).

$${\epsilon }_1=P_s-P_s^{\ast }+k_p{\displaystyle \int \left(P_s-P_s^{\ast }\right)dt}$$

(11)

$${\dot{\epsilon }}_1=\alpha (\gamma v_{rd}-\tau i_{rd}+{\omega }_r{i}_{rq}+\delta )+k_p(P_s-P_s^{\ast })$$

(12)

A Lyapunov function is defined as follows.

$$V_1=\frac{1}{2}{\epsilon }_1^2$$

(13)

As a positive Lyapunov function for a stable error that converges to zero, the derivative of V₁ should be strongly negative. Therefore, condition (14) must be verified to make the stator active power stably track its reference variable.

$${\dot{V}}_1={\epsilon }_1{\dot{\epsilon }}_1=-K_1{\epsilon }_1^2$$

(14)

By replacing (12) in (14), we deduce the controller output and express it in (15).

$$v_{rd}=\frac{1}{\gamma }\left[\frac{1}{\alpha }(-K_1{\epsilon }_1-k_p(P_s-P_s^{*}))+\tau i_{rd}-{\omega }_r{i}_{rq}-\delta )\right]$$

(15)

Step 2: The tracking error of the reactive power and its derivative are expressed in (16) and (17).

$${\epsilon }_2=Q_s-Q_s^{\ast }+k_q{\displaystyle \int \left(Q_s-Q_s^{\ast }\right)dt}$$

(16)

$${\dot{\epsilon }}_2=-\alpha (\gamma v_{rq}-\tau i_{rq}-{\omega }_r{i}_{rd})+k_q(Q_s-Q_s^{\ast })$$

(17)

To ensure both the stability of the error in a limited time and to guarantee the convergence of Qs to Q_s^*, we use the Lyapunov function and the resulting stability condition in (18) and (19), respectively.

$$V_2=\frac{1}{2}{\epsilon }_2^2$$

(18)

$${\dot{V}}_2={\epsilon }_2{\dot{\epsilon }}_2=-K_2{\epsilon }_2^2$$

(19)

By replacing (17) in (19), we deduce the controller output and express it in (20).

$$v_{rq}=\frac{1}{\gamma }\left[-\frac{1}{\alpha }(-K_2{\epsilon }_2-k_q(Q_s-Q_s^{*}))+\tau i_q+{\omega }_r{i}_{rd})\right]$$

(20)

The block design of the RSC controller is shown in Figure 3 below and comprises the following:

▪

Analog comparators to determine the difference between the set point (reference) and the measurement (output).

▪

Sensors to be able to measure voltage and current.

▪

Equation blocks (control law) to generate the commands.

▪

Blocks of direct and inverse Park transformation (123 ---$>$dq and dq ---$>$123)

▪

A Space Vector Pulse Width Modulation (SVPWM) block to elaborate the six pulses to control the IGBT switches.

▪

A block Phase-Locked Loop (PLL) to make the stator electrical angle.

▪

Calculation blocks of some important values for the RSC controller.

3.2. Power Control on the Network Side

Network-side power control aims to stabilize the DC-link tension and cancel the reactive power in the R_fL_f filter. The resulting compact power equations are reported in (21):

$$\left\{\begin{array}{l}\frac{dz}{dt}=d(P_f-P_r)\\ {\dot{P}}_f=a(-\tau i_{fd}-{\omega }_s{i}_{fq}+f{V}_{pcc}-f{v}_{fd})\\ {\dot{Q}}_f=-a(-\tau i_{fq}-{\omega }_s{i}_{fd}-f{v}_{fq})\end{array}\right.$$

(21)

where z = (V_dc²)/2, a = 1.5 Vpcc, τ = R_f/L_f, f = 1/Lf and d = 1/C_dc. The used controller is the ABIAC.

Control of DC-Link Tension and Reactive Energy

The DC-link tension (V_dc) control to keep it steady will be realized in two steps. The first step is calculating the active power of the filter’s (P_f) reference value, which makes the DC-link tension reach its reference value. The active power reference value is then sent to the second step where an advanced backstepping with integral action controls the active power of the filter to monitor it. Step three consists in synthesizing the advanced backstepping with integral action based control law to control the reactive energy (Q_f) of the filter toward its zero reference value.

Step 1: The tracking error of DC-link tension and its derivative are expressed in (22) and (23).

$${\epsilon }_1=z-z_{ref}+k_{dc}{\displaystyle \int \left(z-z_{ref}\right)dt}$$

(22)

$${\dot{\epsilon }}_1=d{P}_f-d{P}_r-{\dot{z}}_{ref}+k_{dc}\left(z-z_{ref}\right)$$

(23)

A similar Lyapunov stability approach for convergence and stability of the primary error will be used. For the positive quadratic Lyapunov function in (24) to be derived in (25), and to make sure that the derived function is negative, the requirement in (25) must be correct.

$$V_1({\epsilon }_1)=\frac{1}{2}{\epsilon }_1^2$$

(24)

$${\dot{V}}_1({\epsilon }_1)={\epsilon }_1(d{P}_f-d{P}_r-{\dot{z}}_{ref}+k_{dc}\left(z-z_{ref}\right))=-K_1{\epsilon }_1^2$$

(25)

The true value of reference for the following step is then obtained through (25) and written in (26).

$$P_{f\_ref}=\frac{1}{d}(-K_1{\epsilon }_1+d{P}_r+{\dot{z}}_{ref}-k_{dc}\left(z-z_{ref}\right))={\alpha }_p$$

(26)

Step 2: The second error variable between the active power measured and its reference value calculated in the first step may be defined as follows.

$${\epsilon }_2=P_f-{\alpha }_p+k_{\alpha }{\displaystyle \int (P_f-{\alpha }_p)}dt$$

(27)

$${\dot{\epsilon }}_2=a(-\tau i_{fd}-{\omega }_s{i}_{fq}+f{V}_{pcc}-f{v}_{fd})-{\dot{\alpha }}_p+k_{\alpha }(P_f-{\alpha }_p)$$

(28)

In addition, the function of the Lyapunov and its derivative are given in (29) and (30) respectively.

$$V_2({\epsilon }_1,{\epsilon }_2)=V_1+\frac{1}{2}{\epsilon }_2^2$$

(29)

$${\dot{V}}_2({\epsilon }_1,{\epsilon }_2)=-K_1{\epsilon }_1^2+{\epsilon }_2{\dot{\epsilon }}_2$$

(30)

So that the positive Lyapunov function derivative is kept strictly negative, the final control variable is chosen as presented in (31).

$$v_{fd}=\frac{1}{f}\left(\tau i_{fd}+{\omega }_s{i}_{fq}-f{V}_{pcc}-\frac{1}{a}\left(-K_2{\epsilon }_2+{\dot{\alpha }}_p-k_{\alpha }(P_f-{\alpha }_p)\right)\right)$$

(31)

Step 3: The tracking error of power reactive generated by NSC and its derivative are expressed in (32) and (33).

$${\epsilon }_3=Q_f-Q_f^{*}+k_{qf}{\displaystyle \int \left(Q_f-Q_f^{*}\right)}dt$$

(32)

$${\dot{\epsilon }}_3=-a(-\tau i_{fq}-{\omega }_s{i}_{fd}-f{v}_{fq})+k_{qf}\left(Q_f-Q_f^{*}\right)$$

(33)

The function quadratic of the Lyapunov and its derivative concerning the error variable are given in (34) and (35).

$$V_3({\epsilon }_3)=\frac{1}{2}{\epsilon }_3^2$$

(34)

$${\dot{V}}_3({\epsilon }_3)={\epsilon }_3{\dot{\epsilon }}_3=-K_3{\epsilon }_3^2$$

(35)

To make sure our positive Lyapunov function derivative is negative, we express this as follows.

$${\dot{V}}_3={\epsilon }_3\left(-a(-\tau i_{fq}-{\omega }_s{i}_{fd}-f{v}_{fq})+k_{qf}\left(Q_f-Q_f^{*}\right)\right)=-K_3{\epsilon }_3^2$$

(36)

So, with this consideration, the final control output is given in (37).

$$v_{fq}=-\frac{1}{f}\left(\tau i_{fq}+{\omega }_s{i}_{fd}-\frac{1}{a}\left(-K_3{\epsilon }_3-k_{qf}\left(Q_f-Q_f^{*}\right)\right)\right)$$

(37)

The block design of the NSC controller is shown in Figure 4 below and comprises the following:

▪

Analog comparators to determine the difference between the set point (reference) and the measurement (output).

▪

Sensors to be able to measure voltage and current.

▪

Equation blocks (control law) to generate the commands.

▪

Blocks of direct and inverse Park transformation (123 ---$>$dq and dq ---$>$123)

▪

A Space Vector Pulse Width Modulation (SVPWM) block to elaborate the six pulses to control the IGBT switches.

▪

A block Phase-Locked Loop (PLL) to make the stator electrical angle.

▪

Calculation blocks of some important values for the RSC controller.

3.3. DFIG during Voltage Dip

When the voltage dip is detected, a sudden change in the stator AC tension appears as demonstrated by [35] in (38).

$$\Delta V_s=M\left(1-\frac{L_s{L}_r}{M^2}\right)\frac{d{I}_r}{dt}+\frac{L_s}{L_r}\Delta V_r$$

(38)

To maintain dIr/dt = 0 through the voltage dip, a significant step change in the rotor tension ∆V_r must occur for tracking changes in stator tension. However, the RSC is not able to deliver the ∆V_r required because the maximum voltage capability of the RSC is only approximately 30%. As a result, a significant excess current will occur across the rotor, so protection of the rotor current is required.

4. Protection against Rotor Excess Current

A crowbar is the most popular DFIG safeguard that is temporarily deployed. Enabling the crowbar in the event of a network voltage dip inhibits the operation of the RSC, resulting in a loss of control of the generator. As a result, the intense reactive energy request on the stator removes the PCC tension while the generator vibrations degrade all mechanical components of the wind turbine.

4.1. Improved Protection Scheme (IPS)

A schematic of the improved protection and switch control flowchart against rotor current to enhance the ability of the DFIG-LVRT in the event of a network voltage dip is illustrated in Figure 5 and Figure 6 below. Incorporating an RL series circuit enhances the classical protection scheme based on a crowbar device (Figure 5a), as illustrated in Figure 6a. First, the IPS is switched at point C to connect with the rotor coils, while the RSC is switched at point S if the current of the rotor exceeds its limit level. Consequently, the imperfection current of the rotor ($I_r^f$) is separated into the current to the crowbar (I_crw) which is defined by the value of the IPS crowbar resistance ($R_{crw}^{IPS}$) and the current to the RSC (I_rsc) which is related to the series impedance (Z_series) value. By ignoring switching maneuvers during the fault clearing and tripping delay, the RSC stays attached to the coils of the rotor during the full operation of the DFIG. Additionally, when the IPS is enabled, the generator’s rotor windings are partially shorted by the crowbar and partially reconnect to the RSC. The generator’s strategic stator power control by the RSC controller is also retained, offering an advantage over the classic crowbar. In addition, such a protection scheme consists of the rotor coils being partially shorted by the crowbar and partially linked to the RSC converter through the Z_series circuit, which places the generator into a momentary mode. This represents a medium state, where part of the generator behaves as a rotor-wound induction generator and part as a squirrel-cage induction generator(SCIG) during the fault with the IPS in action.

The IPS is able to perform as a regular crowbar by selecting a very high series impedance with $R_{crw}^{IPS}$=$R_{crw}^c$(classic crowbar resistance) used, but this depends on the values of the parameters on the series impedance (R,L) and the IPS crowbar resistance ($R_{crw}^{IPS}$). On the other hand, if a very modest impedance device is used with a choice of heavy-important resistance of the crowbar, the demand current from the rotor flows to the RSC via the impedance circuit, possibly damaging the RSC. Therefore, it is required to preserve perfect crowbar resistance and RL circuit values to achieve the IPS objectives.

4.2. Results of the Simulation

Our proposed IPS is examined in this subsection using model studies in the MATLAB/Simulink interface. A DFIG-based wind system with two converters (RSC and NSC) spaced by a continuous bus is modeled as a two-mass system where the DFIG component values are listed in

Table 1. Simulation parameters for a DFIG [2].

Parameter	Value
Rated power	Pn = 2 MW
Stator nominal voltage	Vsn = 690 V
Nominal frequency	f = 50 Hz
Number of pole pairs	p = 2
Stator resistance	Rs = 0.023 p.u
Rotor resistance	Rr = 0.016 p.u
Stator inductance	Ls = 0.18 p.u
Rotor inductance	Lr = 0.18 p.u
Mutual inductance	M = 2.9 p.u
DC-link voltage	Vdc = 1.2 kV

. A short-circuit three-phase fault of more than 30 km distance away on the PCC was tripped at t = 7.5 s and removed at t = 7.62 s through isolation of the failure location. To illustrate the results of the executed responses for one DFIG system, a comparison is made between the two cases, one with the classic crowbar protection scheme(case I) and the other with the improved protection scheme (Case II). Thus, two different responses are compared for both cases, one with no protection and the other with protection. The limit value (I_r-thr) concerning the rotor current is set at 1.5 p.u (per-unit) to assure that any chosen protection system will be enabled during critical failure situations. Through various validation of simulation exercises and with the test and error approach, the best parameters for the classical crowbar resistance ($R_{crw}^c$) and the IPS ($R_{crw}^{IPS}$, R, and L) are found. These optimal values are reported in

Table 2. Optimal values of crowbar and IPS parameters.

Parameter	Value
Classic crowbar resistance	${\mathrm{R}}_{\mathrm{c}\mathrm{r}\mathrm{w}}^{\mathrm{c}}$ = 0.25 Ω
IPS crowbar resistance	${\mathrm{R}}_{\mathrm{c}\mathrm{r}\mathrm{w}}^{\mathrm{I}\mathrm{P}\mathrm{S}}$ = 0.18 Ω
Resistance of the series impedance	R = 12 Ω
Inductance of the series impedance	L = 30 H

and were considered consistently across all simulations. In Figure 7, the validation results for Cases I and II are presented.

From Figure 7a, the root mean square (rms) inrush rotor current for the two cases without any protection is about 5.2 kA, which is a very high value of the safety limit, damages the converter on the rotor side, and increases the tension on the DC link. However, after activation of the classic crowbar (case I) and IPS (case II), the value of the current rotor is lowered to a security limit of 3.0 kA and 3.49 kV, for each. Furthermore, both crowbar and IPS schemes result in a reduced current of RSC to less than 0.1 kA and 2.9 kA, exactly, as shown in Figure 7b, which makes sure that the RSC converter is not destroyed. Of course, the crowbar consumes up to 2.52 kA of current at the time of the network voltage dip trip if just the crowbar is operated, and with IPS operated, the current consumed is 1.52 kA by the IPS crowbar (Figure 7c).

It should be noted that the principal reason for using the classic crowbar is to be able to offer protection to DFIGs and secure the RSC terminal to the rotor coils by rapidly dissipating the rotor overcurrent across the grounded crowbar resistor. Using the IPS design, the goal is to maintain the RSC connected to the rotor coils. The IPS incorporates an integrated device similar to a crowbar that detects and redirects excess current from the rotor to the ground through its resistance. This action partially mitigates the excess current, ensuring that the RSC current remains within a normal operating range.

Such tolerance safeties are available in the power electronics conceptions to manage transient states. As an illustration, if the RSC limit current is two times (2 × 1.775 kA), then the DFIG can operate without any network interruption. However, the major disadvantage of the classic crowbar is that the DFIG loses its RSC controller as the crowbar operates. With IPS, the deactivation of the RSC is prevented and therefore the passage of the DFIG is improved.

The tension of the DC-link is kept in its safe range as well (V_dc < 1.35 p.u). Figure 7d shows the tension response of the DC-link circuit just previous and next to the independent application of the two schemes. The absence of a protective device resulted in a DC-link voltage of 2 kV. However, when either the classic crowbar (in case I) or the IPS (in case II) was implemented, the DC-link tension was effectively limited to a maximum of 1.34 kV and 1.4 kV, respectively.

In addition, the DFIG operation with the classic crowbar results in a temporarily higher demand for reactive energy at the PCC. In effect, enabling the crowbar causes the DFIG to become a SCIG, and would need reactive energy from the network. The demand for reactive energy occurs precisely at the opportune moment when it is needed by the PCC to maintain its tension level. Consequently, a depression of the PCC voltage at the time of tripping and fault elimination is also detected. The reactive energy provided to the PCC is depicted in Figure 8a. While voltage dip is occurring under both I and II situations, an estimated mean of 0.25 MVAr is provided to the PCC, a high overshoot, and a long response time at the moment of the network tension recovery with no type of protection. Its value is 0.5 MVAr when the crowbar is first engaged (at voltage dip initiation) and 0.26 MVAr after intermediate disengagement (before voltage dip removal). When deactivating the crowbar in the middle of the operation, RSC is enabled and the generator’s reactive energy need is minimized as the DFIG actually gains its control of the RSC converter and returns to its initial mode temporarily (SCIG to DFIG) in advance of the crowbar re-enables upon voltage dip elimination due to switching transients. With IPS, there is no blockage of the RSC and therefore the control of the power is maintained and the surplus reactive current is successfully discharged to the PCC through the stator and the NSC, but no transient regime improvement is given.

In case II of Figure 8a, the reactive energy of the PCC at the beginning and during the voltage dip is different from zero with IPS. In this situation, the RSC and NSC both operate in a way that retains their pre-default manner of control, in which unity power factor functioning is favored. As a result of the high reactive power request at the PCC and the low ratings of the converters, a certain amount of reactive power circulates to the PCC to maintain its tension. However, the generator command is not inoperative, and consequently, the PCC tension stays approximately equal to the one unprotected, as indicated in Figure 8b, case II.

5. Modification of the Nonlinear Control Strategy

The reactive power and voltage behavior of the PCC during a network perturbation is depicted in Figure 4. We can observe that if an IPS is enabled during a voltage dip, the RSC is not disabled and the voltage behavior of the PCC is almost the same as without any protection. Consequently, the NSC and RSC controllers should be altered to provide better reactive power support to the PCC.

5.1. Control of Reactive Power

Disregarding stator resistance and given an orientation of the vector stator flux, the expression of the reactive stator power (Q_s) was shown in (10). The reactive power of the stator can therefore be regulated directly by the RSC reactive power controller.

For improvement of the PCC tension, the reactive energy controller RSC is adjusted. When operating in continuous operation, the supervisor is operated to keep the output reactive power at zero, taking priority over the transfer of maximum active power to the network (unit power factor operation). Furthermore, in order to get the reactive power reference, a comparison is made between the nominal voltage of the PCC and its current value, and the error is sent to the controller PI (operation in reactive power). Both operating modes are toggled according to the PCC voltage level. As an example, if the voltage of the PCC drops under 85% of the standard value, the link for the reference of reactive power is changed starting at point N to point F, as illustrated in Figure 9a. The NSC works in a reference frame oriented toward the AC voltage of the network where the DC-link voltage and the reactive power are directly controlled by BIAC, like in Section 3. The level of reactive power generated can be fine-tuned to offer support to the Point of Common Coupling (PCC). Consequently, the device that regulates the reactive power of the NSC is equally adapted and has two operating modes, as displayed in Figure 9b. Under this situation, the NSC functions both as a source of reactive power through the transitional state and as a unity control mode of power factor in continuous operation.

The management of reactive power is generally limited by the nominal values of the converters. The limits are set by the rotor and the maximum NSC currents. If the voltage at the rotor surpasses the limits, the RSC is reinitialized and the amplitude of the voltage is restricted by means of the limiter block. Likewise, the NSC reactive power controller works to protect the inverter’s components by incorporating a flow restriction block to ensure that the inverter operates only at its rated power. Thus, the converters handle the reactive power of the PCC and do not exceed their nominal values.

5.2. Results of the Simulation

In order to prove the IPS performance with and without the use of the revised reactive power controllers. A 10-turbine DFIG wind farm with a nominal power rating of 2 MW each is depicted in Figure 10. A short-circuit three-phase fault of more than 30 km distant from the PCC has been initiated at t = 7.5 s and removed at t = 7.62 s through isolation of the failure section. After this, all phases of voltage dip were detected across the generator, leading to strong rotor and stator currents causing the improvement protection scheme to be activated and the reactive power controllers modified to take on reactive power from the PCC to improve the voltage. The simulation results for the reference values of reactive power, if the DFIG system is checked both before and after the modification of the controller, are illustrated in Figure 11.

When operating without the redesigned controller, the fundamental values of both converters are all set to zero, however, with the revised controller, the RSC and NSC base values increase to approximately 0.75 and 0.98 p.u, respectively, with the fault. With IPS enabled, the reactive power transfer and PCC tension response comparisons with and without the modified control are displayed. Figure 12a shows that the peak reactive power supplied for the PCC without the tweaked controllers is low but with the tweaked ABIAC, the PCC reactive power support is boosted up to 17.5 MVAr with limitation overshoot instead of 15 MVAr with PIC. This shows that with the proposed control, the wind farm can support the network with a great quantity of reactive power. The tension responses of the PCC throughout the error with the edited controller are displayed in Figure 12b. If the control is not modified, the tension of the PCC drops into the trip region (red curve) of the considered network curve (dotted line) with a transient poor regime (response time = 50 ms) at the moment of the network tension recovery because the RSC and NSC function in the unit power factor mode. Using the altered reactive energy controllers, with the ABIAC, the voltage restitution of the PCC during the voltage dip is sufficiently improved (green curve) with good transient regiment the moment of the network tension recovery (response time = 15 ms) that with the PIC (blue curve) which represents modest restitution of the network voltage during fault and a response time at the moment of network voltage recovery of 48 ms, which better meets the tension requirements of the wind farm to stay online in the event of a network voltage dip.

Figure 13, Figure 14, Figure 15 and Figure 16 show all the command signals in the simulations of the proposed strategy (ABIAC + IPS) of control of the wind power system with and without voltage dips on the electrical network.

6. Conclusions

In this paper, an improved LVRT technique for DFIG wind turbines is suggested, to enhance the stability of the wind farm’s behavior against a grid disturbance such as a voltage dip, and then remains connected and contributes to the electrical network with no disruption to its equipment. The suggested IPS is installed between RSC and the rotor coils. It incorporates a classic crowbar with a series RL circuit. The IPS works in the same manner as the crowbar in order to simultaneously secure the DC-link capacitor and converter and keeps the RSC attached to the windings of the rotor, thus preventing loss of generator control. With ideal IPS parameters for the series RL device and its embedded crowbar, the DFIG becomes a partial SCIG and remains a partial DFIG. Consequently, caution should be exercised in selecting parameters to comply with the request of any protection device while keeping the RSC working.

This paper demonstrated that when working with the IPS, we can improve the capacity of the DFIG to overcome the voltage dip. In addition, the altered reactive power regulators based in ABIAC to enhance the PCC tension were described in detail as well. This modification of the control strategy and IPS could lead to a better enhancement of the PCC tension while the network is perturbed and conform to the chosen network code. Hence, the DFIG wind system efficiency during network voltage dip can be enhanced with IPS and revised reactive power regulators. To extend this research work, future projects are envisaged:

-

Implement this MADA-based wind turbine protection technique experimentally and illustrate the limitations of the different techniques against each other.

-

Develop the study to verify the proper protection of the MADA against all other types of disturbances that may occur on the high voltage network and propose solutions in this sense.