AGGREGATED MODEL OF LARGE-SCALE WIND FARMS FOR POWER SYSTEM SIMULATION SOFTWARE TOOLS

Abstract

A method of modeling an equivalent wind turbine generator (WTG) system for a wind farm having a plurality of WTG units includes determining an impact factor of each WTG unit of the plurality of WTG units, determining an equivalent single WTG unit model parameters of the wind farm based on the impact factor of each WTG unit, and determining an effective wind speed of the wind farm to use as the equivalent WTG input wind speed. The method produces a model of static and/or dynamic wind farm behavior. Additionally, a software configured to execute a method of modeling an equivalent wind turbine generator (WTG) system for a wind farm having a plurality of WTG units.

Description

FIELD OF THE INVENTION

The present invention generally relates to power system application and simulation software, and more specifically to power system analysis and simulation software for large-scale wind farms. By considering the impact factor and contribution of each of hundreds of wind turbine-generator units, a device or simulation software can be developed to have similar behavior of the large-scale wind farm as seen by a power grid from the point of common coupling.

BACKGROUND

Large-scale wind farms may consist of different types of wind turbine-generator (WTG) units including fixed-speed, doubly-fed induction generator, and permanent magnet synchronous generators. By increasing the penetration of large-scale wind farms into power systems, various aggregation methods are proposed to model a large-scale wind farm with an equivalent system for steady-state and dynamic analyses. An aggregation method for modeling of a large-scale wind farm reduces the computational burden of steady-state and dynamic analysis. A suitable model should be computationally efficient and adequately accurate for steady-state and dynamic behavior of the system under different operating conditions. Such a model should be easily used in steady-state and dynamic analyses.

A wind farm aggregated model is a reduced order system or an equivalent WTG unit that describes the electrical behavior of the wind farm seen from the point of common coupling to the grid. Such a model radically reduces the computational burdens and complexity in power system analysis that includes wind farms with hundreds of wind turbine generator (WTG) units. Rapid increase of large-scale wind farm installations necessitates to develop aggregated models for power system analysis. However, there is a trade-off between the accuracy of the aggregated model and its complexity that determines the performance and accuracy of the wind farm equivalent models.

Full, Zone, and Semi Aggregation (Agg.) methods are conventionally proposed to model WTGs in a wind farm by one or a few equivalent WTG units. Full Agg. method per-unitizes a WTG unit using its rated power as a base and then simply changes the base power to the rating of wind farm to develop the Full Agg. model for wind farm. This method assumes that the operating points and parameters of all WTGs are the same and a uniform wind speed distribution throughout the wind farm. This assumption is not valid for a real wind farm and can cause inaccuracy in dynamic and steady state analyses.

Zone Agg. uses the concept of Full Agg. method, however, it partitions the wind farm into several zones (clusters) with respect to various wind speed. Wang et al in develop an advanced time series wind turbines clustering method, based on a geometric template matching, to improve the accuracy of the Zone Agg. method. Also, an aggregated turbine and network impedance model has been presented in, in which a new sequence impedance model is developed for resonance analysis of wind farms. Zone aggregation increases the equivalent model accuracy at the cost of higher complexity that makes the model computationally inefficient for large-scale wind farms. Semi Agg. method replaces the wind farm generators by a single per unitized generator, however, it keeps the mechanical parts of WTGs intact. Semi Agg. method provides a relatively accurate method, however, it is also inefficient for large-scale wind farms since it models mechanical parts of hundreds of WTGs in details. Thus, challenges in developing an aggregated model of wind farms include the different wind speed zones in a large wind farm and different parameters of WTGs in a wind farm. These challenges have not been adequately addressed in existing aggregation methods.

Full aggregation, Zone aggregation, and Semi aggregation methods are conventional methods to aggregate wind farms. The Full aggregation method replaces WTGs in a wind farm by one WTG with averaging the WTGs parameters and assuming similar wind speed inputs. However, different wind speed zones and/or various machine parameters significantly increase the inaccuracy of the Full aggregation method. The Zone aggregation method uses zoning of the wind farm by their wind speed inputs and then aggregates every zone with the Full aggregation method. This significantly increases the accuracy of the aggregated model at the cost of higher complexity of the model which can be inefficient for large-scale wind farms. The Semi aggregation method replaces wind farm generators by one generator with averaging the parameters of generators, however, it keeps the mechanical part of WTGs intact. The Semi aggregation method provides an accurate method; however, it is also inefficient in large-scale wind farms since it needs to model all mechanical parts of WTG in details.

Increasing the penetration of large-scale windfarms into the power systems, necessitates an aggregation method to model a large-scale windfarm with an equivalent system to reduce computational burden for steady-state and dynamic analyses. Large-scale windfarms may consist of different types of WTG including fixed-speed, Doubly-Fed Induction Generator (DFIG), and Permanent Magnet Synchronous Generator (PMSG).

Applying an equivalent aggregated model method to calculate a large-scale DFIG windfarm is challenging and time consuming. To model a large-scale DFIG windfarm by one or a few WTGs using the state-space model of WTGs, the parameters of the windfarm model are determined to fit the best in the proposed set of optimization equations. Although the resulted model can be accurate, involving all the state-space equations of a large-scale windfarm in a repetitive optimizing solution leads to a high computational burden. This computational burden increases by increasing the number of WTGs. Moreover, the model accuracy varies in different windfarm operating points. The Full Aggregation (Full Agg) method models the windfarm by one equivalent WTG assuming similar wind speed inputs and WTG parameters in per unit for all WTGs. Parameters of this equivalent WTG is obtained by averaging the WTGs parameters in per unit while the equivalent apparent power is the summation of the windfarm. However, different wind speed zones and/or various machine parameters significantly decrease the accuracy of the Full Agg model. Other methods divide the windfarm to few zones considering their wind speed inputs and aggregate every zone by the Full Agg method. It significantly improves the accuracy of the aggregated model by the cost of higher model complexity which can be inefficient for the large-scale wind farms. The Semi Agg method keeps the mechanical part of WTGs intact, but the windfarm electrical side (generators and converters) is replaced by a single generator and converter by averaging the parameters similar to the Full Agg method. Semi Agg method provides an accurate model, however, it is also inefficient in the large scale windfarms since it needs to model all mechanical parts of the WTGs in detail.

Accordingly, there is a need for an aggregated model of wind farms that is highly accurate while being efficient and low-cost for large-scale wind farms. More specifically, there is a need for an aggregation model that is capable of quantifying the contribution of each WTG in a large-scale wind farm.

SUMMARY OF THE INVENTION

The present invention provides a method of modeling an equivalent wind turbine generator (WTG) system for a wind farm having a plurality of WTG units. The method includes determining an impact factor of each WTG unit of the plurality of WTG units, determining an equivalent single WTG unit model parameters of the wind farm based on the impact factor of each WTG unit, and determining an effective wind speed of the wind farm to use as the equivalent WTG input wind speed. The method produces a model of static and/or dynamic wind farm behavior. Additionally, a software system is provided that is configured to execute an inventive method of modeling an equivalent wind turbine generator (WTG) system for a wind farm having a plurality of WTG units.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is further detailed with respect to the following drawings that are intended to show certain aspects of the present of invention, but should not be construed as limit on the practice of the invention, wherein:

FIG. 1 shows a virtual wind farm based on an Asynchronous machine according to forms of the present disclosure;

FIG. 2 shows a virtual wind farm based on a Synchronous machine according to forms of the present disclosure;

FIG. 3A shows a schematic drawing of a two-zone wind farm according to embodiments of the present invention;

FIG. 3B shows a schematic diagram of a fixed-speed wind turbine generator having a collector system with capacitors to provide reactive power;

FIG. 4A shows a schematic of a two fixed-speed WTG wind farm having a collector system with capacitors to provide reactive power;

FIG. 4B shows a schematic of a semi aggregated two fixed-speed WTG wind farm having a collector system with capacitors to provide reactive power;

FIG. 5 is a schematic of a wind farm with 80 wind turbine generator units;

FIGS. 6A-6D are graphs showing active power dynamic behavior of system using Full aggregation, Zone aggregation, Semi aggregation and I.F aggregation models in four different scenarios covering the combinations of various wind speed inputs and different WTGs parameters in a wind farm system according to embodiments of the present invention;

FIGS. 7A-7D are graphs showing Full aggregation, Zone aggregation, Semi aggregation, and I.F. aggregation methods overall error and simulation time in four different scenarios covering the combinations of various wind speed inputs and different WTGs parameters in a wind farm system according to embodiments of the present invention;

FIG. 8 is an equivalent circuit diagram of the I.F. Agg. Model according to embodiments of the present invention;

FIG. 9 is a schematic drawing of a DFIG WTG according to embodiments of the present invention;

FIGS. 10A and 10B show a DFIG WTG and a Fixed-Speed WTG, respectively;

FIG. 11A shows windfarm d-q axis circuits;

FIG. 11B shows equivalent model d-q axis circuits of those shown in FIG. 11A;

FIG. 12 is a drawing showing rotating the d-q frame by δ₀;

FIG. 13 shows a schematic of a 4-WTGs windfarm;

FIG. 14 shows a schematic of a 20-WTGs windfarm;

FIG. 15 is a graph showing PCC voltage, phase A current and active and reactive power for FIG. 13 DFIG windfarm with similar WTG parameters and its Full Agg, Zone Agg and WD Agg models;

FIG. 16 is a graph showing PCC voltage, phase A current and active and reactive power for FIG. 13 DFIG windfarm with similar WTG parameters, its Full Agg and Zone Agg models with modified controllers and its WD Agg model;

FIG. 17 is a graph showing PCC voltage, phase A current and active and reactive power for FIG. 13 DFIG windfarm with unequal WTG parameters, its WD Agg model and its modified Full Agg and Zone Agg models;

FIG. 18 is a graph showing wind speed, phase A current and active and reactive power for FIG. 14 DFIG windfarm with unequal WTG parameters, its Full Agg and Zone Agg models with modified controllers and its WD Agg model; and

FIG. 19 is a graph showing PCC voltage, phase A current and active and reactive power for FIG. 13 fixed-speed windfarm with unequal WTG parameters, its modified Full Agg, Zone Agg and WD Agg models.

DETAILED DESCRIPTION OF THE INVENTION

The present invention has utility as a method for aggregately modeling a wind farm capable of quantifying the contribution of each WTG in a large-scale wind farm and as a power system simulation software for large-scale wind farms for considering the impact factor and contribution of each of hundreds of wind turbine-generator units making up the large-scale wind farm as seen by a power grid from the point of common coupling. The method of the present invention is highly accurate while being efficient and low-cost for large-scale wind farms.

The present invention provides an Impact Factor aggregation (I.F. Agg.) method that includes the contribution of each WTG unit, based on its parameters and operating point, within the equivalent model of wind farm. The method provides a computationally efficient model for a wind farm that significantly improves the accuracy compared with Full Agg. method. The reason is that I.F. Agg. method includes the effects of WTGs with different parameters and/or operating points. The inventive I.F. Agg. method analytically calculates the contribution of each WTG unit as a weight function in frequency domain. This technique allows one to determine the best set of equivalent model parameters to improve the model accuracy over the frequency range of interest. Most of existing methods develop and test the performance of their aggregated models mainly for fixed-speed wind farms to explain the main concept of the methods for the simplest wind farm configuration. Furthermore, the inventive method includes the effect of wind farm collector system in the equivalent model that is less considered and discussed in the other existing methods.

The present invention additionally provides a Weighted Dynamic (WD) model for a windfarm which has several advantages over the existing methods. In this method, the contribution of each WTG in the aggregated model is quantified which significantly increases the model accuracy compared to the Full Agg model. The derived set of aggregated dynamic equations results in a simpler modeling approach compared to equivalent admittance methods. Also, the windfarm equivalent WTG is obtained through a one-time calculation leading to a much lower computation burden compared to the optimization algorithms. Furthermore, the whole windfarm is modeled with a single WTG resulting in a simpler model compared to the Zone Agg and Semi Agg models. It is worth mentioning that the proposed equivalent turbine for the whole windfarm has not been clearly addressed in the existing literature which provides a simpler model and better insight regarding the system behavior.

The performance of the inventive method is evaluated based on the comparisons of a 4-WTGs DFIG windfarm and the obtained equivalent model. Moreover, a time-domain simulation of 20-WTGs DFIG windfarm with a variable wind speed curve is studied to verify the applicability of the proposed model in a more realistic scenario. Finally, a 4-WTGs fixed speed windfarm are studied to demonstrate the generality of the inventive method. Simulations results conducted under identical and unequal WTGs operating conditions demonstrate a superior or at least similar performance of the proposed method compared to the existing approaches. The inventive model results are compared with the Full Agg and Zone Agg models because the computational burden of these methods are the superior of the existing ones and comparable to the inventive method.

It is to be understood that in instances where a range of values are provided that the range is intended to encompass not only the end point values of the range but also intermediate values of the range as explicitly being included within the range and varying by the last significant figure of the range. By way of example, a recited range of from 1 to 4 is intended to include 1-2, 1-3, 2-4, 3-4, and 1-4.

According to embodiments, the software element can be used for both steady-state and dynamic analyses of power systems including large-scale wind farms. The system significantly reduces the computational burden of a computer and its memory usage for power system simulation by replacing a large-scale wind farm including hundreds of wind turbine-generator units with an equivalent functional model. Compared to the existing aggregated models, the inventive functional model according to embodiments of the present invention supports a wind farm with different wind speed zones. It also supports modeling of a wind farm with different ratings of wind turbine generator units

Embodiments of present invention utilize Impact Factors (I.F.) of a WTG in a wind farm to quantify the contribution of each WTG in an aggregation model. According to embodiments, the I.F. aggregation method uses the frequency response technique to find the best match between the aggregated model parameters and wind farm based on d-q reference frame model of the wind farm generators. Using the I.F. concept results in the model having least amount of error and simulation time overall compared to Full aggregation, Zone aggregation and Semi aggregation methods for both steady-state and transient analyses. According to embodiments, the performance of the method is evaluated based on time-domain simulation of fixed-speed wind farm including 80 WTGs. The time-domain investigation compares the simulation results for the aggregation of the wind farm by Full aggregation, Zone aggregation, Semi aggregation and I.F. aggregation methods under four different scenarios. These test scenarios cover the combinations of various wind speed inputs and different WTGs parameters in the wind farm test system.

Furthermore, the inventive method includes the effect of wind farm collector system in the equivalent model that is less considered and discussed in the other existing methods. According to embodiments of the I.F. Agg. method, a wind farm including 80 WTGs is fully modeled using MATLAB/SIMULINK software tool. The time domain dynamic and steady state behavior of this wind farm is obtained and used as a reference to evaluate and compare different aggregation methods. Various test scenarios are defined including a combination WTGs with similar/different parameters and a wind farm with uniform/nonuniform wind speed distributions. The present invention also defines a normalized index to quantify computational burden, and accuracy to present superior features of the inventive I.F. Agg. method compared with the other methods.

FIG. 3A shows the diagram of a wind farm including tens of fixed-speed WTG units that are connected to a grid at the point of common coupling (PCC) through a collector system. In a large-scale wind farm, the wind speed distribution at WTG units is not necessarily uniform. To improve the accuracy, the wind farm can be partitioned in zones as shown in FIG. 3A. In partitioning of a large-scale wind farm, in addition to wind speed, several operating point parameters such as slip ratio, and active power must be used in a clustering algorithm. A fixed speed wind farm includes several WTG units with a given schematic as shown in FIG. 3B. The aerodynamic model of the turbine is shown in Equation 1.

P _m =C _p(λ,β)P_W=T_mω_r,  Equation 1:

where P_W=0.5ρπr²V_W³ the wind power. The parameters ρ, r, and V_W denote the air density, turbine radius, and the wind speed, respectively. C_p(λ,β) is the turbine coefficient that is a function of the pitch angle of the turbine blades, β, and the tip speed ratio, λ≙rω_l/V_W, where w_l is the turbine shaft speed. P_m, T_m, and w_r are the generator mechanical power, torque, and speed, respectively. For a given pitch angle, C_p can be estimated with a quadratic function as Equation 2.

$\begin{array}{cc}{C}_p\left(\lambda \right)=\frac{C_{\mathrm{pm}}}{{\lambda }_{\mathrm{opt}}^2}\left(2{\lambda }_{\mathrm{opt}}-\lambda \right)\lambda ,& \mathrm{Equation}2\end{array}$

where C_pm is the maximum of Cp that occurs at λ=λ_opt. The turbine coefficient in Equation 2 can be referred to generator shaft in Equation 3.

$\begin{array}{cc}{C}_p\left({\lambda }^{\prime }\right)=\frac{C_{\mathrm{pm}}}{{\lambda }_{\mathrm{opt}}^{{\prime }^2}}\left(2{\lambda }_{\mathrm{opt}}^{\prime }-{\lambda }^{\prime }\right){\lambda }^{\prime },& \mathrm{Equation}3\end{array}$

where the referred parameters are λ′=rw_r/V_W=Gλ and λ′_opt=Gλ_opt, and G_=w_r/w_l is gear box turns ratio in a WTG. Using this notation, the steady-state model of WTG at the generator side is shown in Equation 4.

T _m +T _e =Dω _r/ω_b,  Equation 4:

where T_e=(X_m²R_rs₀V_e)/ΔT_e is the generator electric torque, w_b is the based frequency, D is mechanical damping coefficient, and ΔT_e=[R_sR_r+s₀(X_m²−X_ssX_rr)]²+[R_rX_ss+s₀R_sX_rr]².

Vs is the effective voltage at PCC, R_s and R_r are the stator and rotor resistances, and X_m is the magnetizing reactance, respectively. The slip of induction generator at the operating point is s₀=(w_b−w_r)/w_b and the machine reactances are X_ss=X_m+X_ls and X_rr=X_m+X_lr, where X_ls and X_lr are leakage reactances of stator and rotor, respectively.

The schematics of equivalent system corresponding to Full Agg. method is depicted on FIG. 3B, in which the wind farm is replaced with a single unit WTG assuming that the WTG parameters and wind speed of all zones (zone 1 and 2 in FIG. 3A) are the same for all n WTG units within the farm. Based on these assumptions, the per unitized parameters of the generator in Full Agg. model is the same as that of calculated for a single WTG unit in section A, provided that S_beq=nS_b is used as the base power where S_b is the rated power of WTGs and S_beq is the rated power of the wind farm.

Two limitations of the Full Agg. model are ambiguities in the definition of mechanical parameters and modeling of real wind farms including machines with different ratings and various wind speeds within the zones. It has previously been proposed that the total mechanical power of the wind farm is calculated and applied to the equivalent generator without considering a model for wind turbine. The variable wind speed at different zones causes steady-state and dynamic errors when aggregated model for a large-scale wind farm is used. To mitigate the error of wind speed mismatch at different zone, the concept of equivalent effective wind speed of wind farm with unison WTGs is defined as Equation 5.

$\begin{array}{cc}V_W=\frac{1}{n}{\left(\sum _{k=1}^nV_{W_k}^3\right)}^{\text{?}},\text{?}\text{indicates text missing or illegible when filed}& \mathrm{Equation}5\end{array}$

where V_W is the wind speed that provides a power equals to the total power of wind farm.

Zone Agg. method partitions the wind farm into a few zones with respect to wind speed variations, and other operating point parameters of WTG units. Then, a Full Agg. model is associated to each zone to represent WTGs within the zone with an equivalent system as shown in FIG. 2(a). This method significantly improves the accuracy of equivalent models since each Full Agg. model is assigned to a zone with an almost fixed wind speed. However, improving the accuracy in this method is obtained at the cost of model complexity since the number of zones may be high in large-scale wind farms. Furthermore, the other limitations of Full Agg. method, including the inability to model WTGs with different parameters, will be directly induced to the Zone Agg. method.

To further improve the aggregated model accuracy, Semi Agg. method is proposed, in which the wind farm generators are represented with a single per unitized generator similar to the one in Full Agg. method. However, the wind turbines are individually modeled to calculate mechanical torque separately, as depicted on FIG. 4B. This method keeps the mechanical part of wind farm intact and only the generators are aggregated. Therefore, the method will become computationally inefficient for large-scale wind farms although it provides more accurate results compared with Full and Zone Agg. methods.

The inventive I.F. Agg. method quantifies the impact of each WTG within a wind farm to develop a more accurate equivalent system for the wind farm. This method introduces an equivalent WTG unit for the wind farm and determines its parameters based on weighted average of the WTGs within the farm. The weighting function is defined as the incremental ratio of WTG current to the wind farm current at PCC.

The weighted averaging technique can be analytically realized in frequency domain that needs the full model of WTG to be linearized about its operating point. Then, the technique will be used for WTG models in frequency domain to obtain the equivalent model parameters. An advantage of using I.F. Agg. method is that it can also define an equivalent RC model for the collector system of the wind farm. It will be shown that the weighted averaging technique significantly improves the accuracy of aggregation model while it remains computationally efficient and addresses the limitation of existing methods.

The first step of I.F. method needs to determine the operating point of WTG units. Based on Equation 1, the input mechanical power corresponding to the k-th WTG unit is

$P_{m_{k_0}}=C_{p_k}\left({\lambda }_{k_0}^{\prime },{\beta }_{k_0}\right)P_{w_{k_0}}$

where the subscript “0” signifies quantities at the operating point. For the sake of brevity, subscript k is removed within the rest this section till it is needed for merging the equations. The rated slip of high power induction generators is small (e.g. −0.005 for MW-scale generators), therefore, the mechanical speed of generator shaft can be approximated as r_r0=w_b/p where p is the number of pole pair of generator in a WTG. As the dynamic of pitch control system is slow compared with the power system dynamics, the pitch angle β₀ can be assumed constant corresponding to a fixed wind speed. Thus, P_m0 is given as P_m0=Cp(λ′₀)P_W0 where λ′₀≅rw_b/(p·V_W). Hence, to obtain the slop at WTG operating point, the per unitized P_m0 and w_r0=1−s₀ can be substituted in Equation 4 and solved it for so.

The next step in I.F. Agg. method is to obtain the impact factors of WTGs that are used for weighted averaging of machine parameters to obtain an equivalent WTG for the wind farm. This averaging can be appropriately performed in frequency domain to cover the frequency range of interest for power system studies. The linearized mechanical model of an squirrel cage induction generator is shown in Equation 6.

$\begin{array}{cc}\Delta T_m=X_mi_{\mathrm{dr}0}\Delta i_{\mathrm{qs}}-X_mi_{\mathrm{qr}0}\Delta i_{\mathrm{ds}}-X_mi_{\mathrm{ds}0}\Delta i_{\mathrm{qr}}++X_mi_{\mathrm{qs}0}\Delta i_{\mathrm{dr}}-2H\frac{d{\mathrm{\Delta \omega }}_r}{\mathrm{dt}}.& \mathrm{Equation}6\end{array}$

For a WTG and based on Equations 1-3, one obtains Equation 7.

$\begin{array}{cc}{{\omega }_{r0}\Delta T_m+T_{m0}{\mathrm{\Delta \omega }}_r=P_{W_0}\frac{\partial C_p}{\partial {\lambda }^{\prime }}}_{{\lambda }^{\prime }={\lambda }_0^{\prime }}{\mathrm{\Delta \lambda }}^{\prime }+C_{p0}\Delta P_W,& \mathrm{Equation}7\end{array}$

where Δλ′=rΔW_r/V_W0. Solving Equation 7 for ΔT_m yields Equation 8.

$\begin{array}{cc}\Delta T_m=e_m{\mathrm{\Delta \omega }}_r,e_m=\frac{1}{{\omega }_{r_0}}{(\frac{\partial C_p}{\partial {\lambda }^{\prime }}}_{{\lambda }^{\prime }={\lambda }_0^{\prime }}\frac{{\mathrm{rP}}_{W_0}}{V_{W_0}}-T_{m_0}).\mathrm{where}.& \mathrm{Equation}8\end{array}$

The linearized voltage equations of the machine in frequency domain are shown in Equations 9-12.

$\begin{array}{cc}\Delta v_{\mathrm{qs}}=A_{\mathrm{qs}}\Delta i_{\mathrm{qs}}+\underset{-}{B_{\mathrm{qs}}}\Delta i_{\mathrm{ds}}+C_{\mathrm{qs}}\Delta i_{\mathrm{qr}}+D_{\mathrm{qs}}\Delta i_{\mathrm{dr}},& \mathrm{Equation}9\\ \Delta v_{\mathrm{ds}}=A_{\mathrm{ds}}\Delta i_{\mathrm{qs}}+B_{\mathrm{ds}}\Delta i_{\mathrm{ds}}+C_{\mathrm{ds}}\Delta i_{\mathrm{qr}}+D_{\mathrm{ds}}\Delta i_{\mathrm{dr}},& \mathrm{Equation}10\\ \Delta v_{\mathrm{qr}}=0=A_{\mathrm{qr}}\Delta i_{\mathrm{qs}}+B_{\mathrm{qr}}\Delta i_{\mathrm{ds}}+C_{\mathrm{qr}}\Delta i_{\mathrm{qr}}+D_{\mathrm{qr}}\Delta i_{\mathrm{dr}},& \mathrm{Equation}11\\ \Delta v_{\mathrm{dr}}=0=A_{\mathrm{dr}}\Delta i_{\mathrm{qs}}+B_{\mathrm{dr}}\Delta i_{\mathrm{ds}}+C_{\mathrm{dr}}\Delta i_{\mathrm{qr}}+D_{\mathrm{dr}}\Delta i_{\mathrm{dr}},& \mathrm{Equation}12\\ A_{\mathrm{qs}}=B_{\mathrm{ds}}=R_s+\frac{j\omega }{{\omega }_b}X_{\mathrm{ss}},B_{\mathrm{qs}}=-A_{\mathrm{ds}}=X_{\mathrm{ss}},C_{\mathrm{qs}}=D_{\mathrm{ds}}=\frac{j\omega }{{\omega }_b}X_m,D_{\mathrm{qs}}=-C_{\mathrm{ds}}=X_m,& \\ \mathrm{where}\mathrm{and}& \\ A_{\mathrm{qr}}=\frac{j\omega }{{\omega }_b}X_m-X_di_{\mathrm{dr}0},A_{\mathrm{dr}}=-s_0X_m+X_qi_{\mathrm{dr}0},B_{\mathrm{qr}}=s_0X_m+X_di_{\mathrm{qr}0},B_{\mathrm{dr}}=\frac{j\omega }{{\omega }_b}X_m-X_qi_{\mathrm{qr}0},C_{\mathrm{qr}}=R_r+\frac{j\omega }{{\omega }_b}X_{\mathrm{rr}}+X_di_{\mathrm{ds}0},C_{\mathrm{dr}}=-s_0X_{\mathrm{rr}}-X_qi_{\mathrm{ds}0},D_{\mathrm{qr}}=s_0X_{\mathrm{rr}}-X_di_{\mathrm{qs}0},D_{\mathrm{dr}}=R_r+\frac{j\omega }{{\omega }_b}X_{\mathrm{rr}}+X_qi_{\mathrm{qs}0}.& \end{array}$

Solving Equations 11 and 12 for Δi_qr and Δi_dr and substituting the solutions in Equations 9 and 10 yield Equations 13 and 14.

Δv_qs=α_q(jω)Δi_qs+β_q(jω)Δi_ds,  Equation 13:

Δv_ds=α_d(jω)Δi_qs+β_d(jω)Δi_ds,  Equation 14:

Finally, by solving Equation 14 for Δi_ds and substituting the solution in Equation 13, Δv_qsk for the k-th unit can be expressed in terms of two transfer functions K_k(jw) and G_k(jw) as given by Equation 15.

Δv_qsk=K_k(jω)Δi_qsk+G_k(jω)Δv_dsk,  Equation 15:

where

K _k(jω)=α_qk(jω)−β_qk(jω)α_dk(jω)/β_dk(jω),

G _k(jω)=β_qk(jω)/β_dk(jω).

The WTG units are connected in parallel through a collector system that is often design for negligible power losses (e.g. less than 2%) at rated power of wind farm. Thus, to develop equivalent system of a wind farm including n WTGs, it can be assumed that v_qsk≅v_qs and v_dsk≅v_ds for k=1, 2, . . . , n where v_qs and v_ds are dq-components of the wind farm at the point of common coupling. Thus, by applying summation over k=1, 2, . . . , n in Equation 15, the wind farm model in frequency domain can be expressed as Equation 16.

$\begin{array}{cc}n\Delta v_{\mathrm{qs}}=\Delta i_{\mathrm{qs}}\sum _{k=1}^n\frac{K_k\left(j\omega \right)\Delta i_{{\mathrm{qs}}_k}}{\Delta i_{\mathrm{qs}}}+\Delta v_{\mathrm{ds}}\sum _{k=1}^nG_k\left(j\omega \right),& \mathrm{Equation}16\end{array}$

The impact factor is defined as u_k≙Δi_qsk/Δi_qs at w=0, i.e. the dc gain of incremental current ratios since w=0 in dq frame corresponding to the fundamental frequency of the generator in time domain. Then, Equation 16 can be expressed as Equation 17.

$\begin{array}{cc}\Delta v_{\mathrm{qs}}=\frac{K^{\prime }\left(j\omega \right)}{n}\Delta i_{\mathrm{qs}}+\frac{G^{\prime }\left(j\omega \right)}{n}\Delta v_{\mathrm{ds}},\mathrm{where}K^{\prime }\left(j\omega \right)\stackrel{\Delta }{=}\sum _{k=1}^nK_ku_k,G^{\prime }\left(j\omega \right)\stackrel{\Delta }{=}\sum _{k=1}^nG_k.& \mathrm{Equation}17\end{array}$

Finally, by updating the base apparent power to the rating of wind farm, i.e. SWF=nSWTG, Equation 17 can be rearranged in wind farm per unit system as Equation 18.

Δv_qs=K′(jω)Δi_qs+G′(jω)Δv_ds,  Equation 18:

Based on Equation 18, the frequency domain model of wind farm is formulated similar to a single unit WTG as given in Equation 15. The schematic diagram of the equivalent circuit for this model is shown in FIG. 8. As the wind farm model is developed based on linearized model of WTGs, the parameters of equivalent model of wind farm in FIG. 8 can be readily determined using the impact factors u_k, 1, 2, . . . , n, e.g. X_rr^[pu]=Σ_k=1ⁿX_rrk^[pu]u_k, X_ss^[pu]=Σ_k=1ⁿX_ssk^[pu]u_k, etc. The main difference between the inventive I.F. Agg. and Full Agg. methods and in fact the key contribution of the I.F. Agg. is to calculate the parameters of equivalent model based on weighted averaging using impact factors u_k to provide more accurate results compared with existing methods.

An equivalent collector system and shunt capacitors can be defined with equivalent R_C and C_C as shown in FIG. 3B. R_C can be determined from the steady state power balance or it can be obtained based on weighted averaging method as R_C^[pu]=Σ_k=1ⁿR_Ck^[pu]u_k.

Using FIG. 3B and considering the steady state equivalent circuit of induction generator, the per phase reactive demand from grid for k-th unit is shown in Equation 19.

$\begin{array}{cc}{Q}_{d_k}=V_{s_k}^2\left(\frac{1}{X_{m_k}}+\frac{X_{\mathrm{ls}}+X_{\mathrm{lr}}}{{\left(\frac{R_r}{s}\right)}^2+{\left(X_{\mathrm{ls}}+X_{\mathrm{lr}}\right)}^2}-{\omega }_sC_{C_k}\right).& \mathrm{Equation}19\end{array}$

where V_sk is the effective voltage at the terminal of the k-th WTG unit. The equivalent capacitor C_C is obtained based on reactive power balance, as given by Equation 20.

$\begin{array}{cc}{Q}_d=V_s^2\left(\frac{1}{X_m}+\frac{X_{\mathrm{ls}}+X_{\mathrm{lr}}}{{\left(\frac{R_r}{s}\right)}^2+{\left(X_{\mathrm{ls}}+X_{\mathrm{lr}}\right)}^2}-{\omega }_sC_C\right)==\sum _{k=1}^nV_{s_k}^2\left(\frac{1}{X_{m_k}}+\frac{X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}}{{\left(\frac{R_{r_k}}{s_k}\right)}^2+{\left(X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}\right)}^2}-{\omega }_sC_{C_k}\right).& \mathrm{Equation}20\end{array}$

Considering a low loss collector system, V_sk≅V_s for k=1, 2, . . . , n. Thus, Equation 20 yields Equation 21.

$\begin{array}{cc}{C}_C=\frac{1}{{\omega }_sX_m}+\frac{\left(X_{\mathrm{ls}}+X_{\mathrm{lr}}\right)/{\omega }_s}{{\left(\frac{R_r}{s}\right)}^2+{\left(X_{\mathrm{ls}}+X_{\mathrm{lr}}\right)}^2}--\sum _{k=1}^n\left(\frac{1}{{\omega }_sX_{m_k}}+\frac{\left(X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}\right)/{\omega }_s}{{\left(\frac{R_{r_k}}{s_k}\right)}^2+{\left(X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}\right)}^2}-C_{C_k}\right).& \mathrm{Equation}21\end{array}$

To determine the parameters of equivalent wind turbine for a wind farm, assuming P_W=Σ_k=1ⁿP_Wk, Equation 22 is obtained.

$\begin{array}{cc}{P}_W=\frac{1}{2}\rho {\mathrm{AV}}_W^3=\sum _{k=1}^nP_{W_k}=\sum _{k=1}^n\frac{1}{2}\rho A_kV_{W_k}^3,& \mathrm{Equation}22\end{array}$

where A=Σ_k=1ⁿA_k is the equivalent surface of all WTGs. Thus, the equivalent radius, r, and wind speed, V_W, can be expressed as Equation 23.

$\begin{array}{cc}r={\left(\sum _{k=1}^nr_k^2\right)}^{\frac{1}{2}},V_W={\left(\sum _{k=1}^nA_kV_{W_k}^3/\sum _{k=1}^nA_k\right)}^{\text{?}}.\text{?}\text{indicates text missing or illegible when filed}& \mathrm{Equation}23\end{array}$

The equivalent mechanical power, P_m=Σ_k=1ⁿP_mk, and incremental power ΔP_m={dot over (Σ)}_k=1ⁿΔP_mk, equations yield Equation 24 and Equation 25.

$\begin{array}{cc}{C}_p\left({\lambda }^{\prime }\right)P_W=\sum _{k=1}^nC_{p_k}\left({\lambda }_k^{\prime }\right)P_{W_k}& \mathrm{Equation}24\\ P_W\frac{\partial C_p}{\partial {\lambda }^{\prime }}\frac{r}{V_W}\Delta {\omega }_r=\sum _{k=1}^nP_{W_k}\frac{\partial C_{p_k}}{\partial {\lambda }_k^{\prime }}\frac{r_k}{V_{W_k}}\Delta {\omega }_{r_k}.& \mathrm{Equation}25\end{array}$

Therefore, equivalent C_pm and λ′_opt in Equation 3 can be obtained from the simultaneous solutions of Equation 24 and Equation 25. Finally, λ_opt and gear-box ratio, G, for the equivalent wind turbine generator can be defined based on weighted average of λ_optk with respect to radii r_k for k=1, 2, . . . , n as Equation 26.

$\begin{array}{cc}\frac{{\lambda }_{\mathrm{opt}}}{r}\stackrel{\Delta }{=}\frac{1}{n}\sum _{k=1}^n\frac{{{\lambda }_{\mathrm{opt}}}_k}{r_k},G\stackrel{\Delta }{=}\frac{{\lambda }_{\mathrm{opt}}^{\prime }}{{\lambda }_{\mathrm{opt}}}.& \mathrm{Equation}26\end{array}$

Next, the performance and accuracy of the inventive I.F. Agg. method in comparison with Full, Zone, and Semi Agg. methods are compared using a fixed speed wind farm study system. According to embodiments, the system includes 80 WTG units as shown in FIG. 5 that are interconnected using a collector system with a given schematic in FIG. 3B. The wind farm is fully simulated using detailed parameters given in Table 1 and the simulation test results are used to generate the reference curves and points for calculation of the accuracy.

Furthermore, the simulation time of the 80-WTG wind farm is considered as the reference to compare the computational efficiency of different methods. Two types of generators with different parameters and ratings (Types I and II) are used in four different test scenarios A, B, C, and D, in which WTGs can have various wind speeds. The parameters of wind turbine generators Type I and II are listed in Table I and the details of test scenarios are as follows:

• • A. All WTG units are Type I with the same wind speeds, V_Wk=20 m/s for 1≤k≤80;
  • B. All WTG units are Type I with different wind speeds at

$V_{W_k}=\{\begin{array}{cc}20& 1\le k\le 20\\ 16& 21\le k\le 40\\ 12& 41\le k\le 60\\ 8& 61\le k\le 80\end{array};$

• • C. The first half of WTGs are Type I, the second half are Type II, and they all operate at same wind speeds, V_Wk=20 m/s for 1≤k≤80;
  • D. The first half of WTGs are Type I, the second half are Type II, and the wind speed of WTGs are:

$V_{W_k}=\{\begin{array}{cc}20& 1\le k\le 10\mathrm{or}41\le k\le 50\\ 16& 11\le k\le 20\mathrm{or}51\le k\le 60\\ 12& 21\le k\le 30\mathrm{or}61\le k\le 12\\ 8& 31\le k\le 40\mathrm{or}71\le k\le 80\end{array}.$

TABLE I
PARAMETERS OF TYPE I AND II WTG UNITS
Name of the
parameter	Type I	Type II	Units
GENERATOR
Sb	150	110	[kVA]
Vs	460	460	[V]
fs	60	60	[Hz]
H	0.3096	0.3175	[s]
D	0.0114	0.006839	[pu]
Rs	0.01282	0.01597	[pu]
Rr	0.00702	0.009103	[pu]
Xm	2.503	2.183	[pu]
Xss	2.55351	2.23942	[pu]
Xrr	2.55351	2.23942	[pu]
P (poles)	4	4	—
COLLECTOR & GRID
Rc	0.05	0.05	[pu]
Cc	2.63	2.63	[mF]
Vth	460	460	[V]
Rth	0.01	0.01	[pu]
TURBINE
VWb	12	11.33	[m/s]
r	10.6	10	[m]
G	23.55	23.55	—
Cpm	0.4	0.4	—
λopt	8	8	—

These four scenarios cover all events that can occur for a wind farm including different machine types and various wind speeds. The tests start at t₀=3 s by applying a small signal disturbance, that is a limited 3-phase connection to ground via resistances R_f=0.09 pu at PCC for 3 cycles. After removing this small signal disturbance, the wind farm operating point will be back to its prior operating point at t=3−.

To investigate the performance and accuracy of the aggregation methods the Total Normalized Simulation Time and Error (TNSTE) criterion is defined and used to evaluate the proposed and existing methods. TNSTE consists of three main components as:

• • 1) Normalized simulation time, STE, that is defined as the ratio of detailed wind farm study system simulation time to the one for an aggregation method;
  • 2) Steady state error of the aggregated methods at their operating points, given by Equation 27.

$\begin{array}{cc}{e}_{\mathrm{ss}}=\frac{{\hat{f}}_{\mathrm{ss}}-f_{\mathrm{ss}}}{f_{\mathrm{ss}}}.& \mathrm{Equation}27\end{array}$

• •  fss in (27) denotes the steady state of f∈{V, I, P, Q} where V, I, P, and Q are the voltage, current, real, and reactive power at the point of common coupling of wind farm and grid. {circumflex over ( )}f represents the corresponding quantity to f_ss that is obtained from an aggregated model for the study state wind farm;
  • 3) Transient error of the aggregated methods at a specific operating point following to a small signal disturbance that is defined as Equation 28.

$\begin{array}{cc}{e}_{\mathrm{Trans}}=\frac{\frac{1}{T}{\int }_{t_0}^{T+t_0}\left(\hat{f}\left(t\right)-f\left(t\right)\right)\mathrm{dt}}{f_{\mathrm{ss}}\left(t_0^{-}\right)},& \mathrm{Equation}28\end{array}$

• •  where t0 is the fault time and f_ss (t₀⁻) is the steady state quantity prior to the small signal disturbance.

TNSTE is defined as the summation of these three normalized components as Equation 29.

TNSTE=STE+e_ss+e_Trans.  Equation 29:

The study system is simulated in MATLAB/SIMULINK software tool and the test results for active power following the small disturbance are depicted in FIGS. 6A-6d corresponding to the four test scenarios A, B, C, and D. For each scenario, the accuracy of the inventive I.F. Agg. method and one of the Full, Zone, and Semi Agg. methods are compared with the detailed 80-unit wind farm test results. The conclusions of active power simulations are:

• • 1) if the WTG parameters and wind speed of the units in a wind farm are similar (an unrealistic assumption), all equivalent aggregated systems accurately describe the wind farm behavior (FIG. 6A);
  • 2) The accuracy of Full Agg. method is highly sensitive to wind speed variations between the zones (FIGS. 6B and 6D);
  • 3) The Full Agg. method is also slightly sensitive to non-unison WTG units with different parameter. The Zone and Semi Agg. methods are more accurate compared with Full Agg. method in all test scenarios;
  • 4) The I.F. Agg. method provides more accurate results compared with Full and Semi Agg. methods in all test scenarios. However, Zone Agg. method in terms of active power matching is slightly more accurate in test scenarios (b) and (d).

The last conclusion is expected since in scenarios (b) and (d) the wind speed is different in four zones, thus, the Zone Agg. method that uses four equivalent WTGs corresponding to each zone provides better matching with reference in terms of active power. However, it will be shown in the next analysis (FIG. 6) that the advantages of I.F. Agg. method outweigh the Zone Agg. method when the computational efficiency and accuracy of matching other quantities (e.g. reactive power) are taken into account.

To further evaluating the performance and accuracy of the methods, simulation results are also studied for reactive power, current, and voltage of the wind farm and the results are compared based on TNSTE index as elaborated in Equations 27 to 29.

FIGS. 7A-7D show details of TNSTE components using color coded bars for each test scenario. The conclusion of test results are as follows:

• • 1) in all test scenarios the normalized simulation times (STE) for Full Agg. and I.F. methods (shown with dark gray color) are significantly smaller than Zone and Semi Agg. methods. Thus, I.F. and Full Agg. methods are the most computationally efficient ways for developing equivalent systems for wind farms;
  • 2) in case of uniform wind speed distribution (that is an unrealistic assumption) the normalized simulation time, TSE, is the dominant component in TNSTE index compared with the transient and steady state errors. Under this condition, I.F. and Full Agg. methods provide the best performance (FIGS. 7A and 7C);
  • 3) in case of nonuniform wind speed distribution the steady state and transient errors comprise the most part of TNSTE. Except for the Zone Agg. method, the steady state and transient errors are dominant components of TNSTE compared with normalized simulation time. Thus, when the wind speeds at the zones are different quantities, I.F. and Zone. Agg. method provide the best performances for aggregation of the wind farm (FIGS. 7B and 7D);
  • 4) The nonuniform wind speed distribution significantly increases the reactive power errors, as shown by light green and blue bars in FIGS. 7B and 7D. Thus, using only real power indicator is insufficient to evaluate the accuracy of aggregation methods;
  • 5) Comparing the scenarios A and C for steady state and transient errors show that change of generator parameters slightly increase the errors, however, these are more sensitive to wind speed variations as shown in FIGS. 7B and 7D.

The overall test results in FIGS. 7A-7D show that when the wind speed is changing between the zones (e.g. in large-scale wind farms) the I.F. and Zone Agg. methods are the top two effective aggregation methods. However, if the wind speed is almost the same for all WTG units (e.g. in small-scale wind farms (then I.S. and Full Agg. methods can be considered as the superior aggregation methods. Furthermore, clustering algorithms for large-scale wind farms need revising in selection their indicators, i.e. reactive power and computational efficiency must be taken into account.

Based on a newly defined impact factor of WTG units within a wind farm, this paper presents a systematic analytical method to develop an aggregated system for large-scale fixed-speed wind farms. The aggregated model is established based on linearized dq dynamic model of WTG in frequency domain. It also encompasses an equivalent circuit for the collector system of the wind farm that significantly improves the accuracy of the model specially in terms of reactive power balance. Conventional aggregation methods become highly inaccurate when the wind speed at different zones of a large-scale wind farm are unequal. The advantage of the proposed impact factor method is to improve the accuracy of the aggregated model by considering the contribution of each WTG in the equivalent system based on its operating point current. A study system including 80 WTG units is used for performance evaluation and verification of the method. The test results of the different test scenarios show the superior performance and accuracy of the proposed impact factor aggregation method specially for large-scale wind farms with different wind speed zones.

Furthermore, Xq and Xd in Equations 9-12 are:

$\begin{array}{cc}{X}_q=\left(X_mi_{\mathrm{qs}0}+X_{\mathrm{rr}}i_{\mathrm{qr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right),& \mathrm{Equation}30\\ X_d=\left(X_mi_{\mathrm{ds}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right),& \mathrm{Equation}31\\ {{\Delta }_{m0}=2\mathrm{Hj}\omega +\frac{\partial C_p}{\partial {\lambda }^{\prime }}}_{{\lambda }_0^{\prime }}\frac{{\mathrm{rP}}_{W_0}}{V_{W0}\left(1-s_0\right)}-\frac{P_{m0}}{{\left(1-s_0\right)}^2}.\mathrm{where}& \end{array}$

α_q,d and β_q,d in Equations 13 and 14 are:

$\begin{array}{cc}{\alpha }_{q,d}=A_{\mathrm{qs},\mathrm{ds}}+{\gamma }_qC_{\mathrm{qs},\mathrm{ds}}+{\gamma }_dD_{\mathrm{qs},\mathrm{ds}},& \mathrm{Equation}32\\ {\beta }_{q,d}=A_{\mathrm{qs},\mathrm{ds}}+{\gamma }_qC_{\mathrm{qs},\mathrm{ds}}+{\gamma }_dD_{\mathrm{qs},\mathrm{ds}},& \mathrm{Equation}33\\ \mathrm{where}& \\ {\gamma }_q=\left(\frac{A_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}A_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right),{\gamma }_d=\left(\frac{A_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}A_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right),& \\ {\lambda }_q=\left(\frac{B_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}B_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right),{\lambda }_d=\left(\frac{B_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}B_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right)& \end{array}$

The incremental speed is Δw_r=C(jw)Δi_qs, and C(jw) is:

$C\left(j\omega \right)=\frac{X_m\left(i_{\mathrm{dr}0}-i_{\mathrm{qr}0}C_1+i_{\mathrm{qs}0}C_3-i_{\mathrm{ds}0}C_2\right)}{e_m+2\mathrm{Hj}\omega },\mathrm{where}$ $C_1=-{\alpha }_d/{\beta }_d,C_2={\gamma }_q+{\lambda }_qC_1,\mathrm{and}C_3={\gamma }_d+{\lambda }_dC_1.$

Virtual Synchronous Machine (VSM) is an inverter connected to the grid which is controlled by a new control method. This new control method helps the inverter acts similar to a synchronous generator in aspect of delivering active and reactive power to the grid. Therefore, VSM contributes to the frequency stability of the grid by an virtual inertia provided in the control loop. Conventional VSM control approach use a fixed value for virtual inertia. But more advanced control techniques change the virtual inertia value accordingly to achieve desired behavior. But changing just one parameter of synchronous machine will result in moving the operation point from the nominal point. Moreover, it may move the system eigenvalues away from the realistic values. To prevent the following issues extra control loops and protection should be added to the system.

Impact Factor Aggregation method obtains the equivalent d-q model of the wind farm by considering the contribution of each wind turbine generator (WTG) in the model, and use these equation to control an inverter to act like a wind farm. By changing of the operation point, the virtual wind farm will be modeled by connecting or disconnecting of some WTGs while the remaining connected WTGs working near to their operation point. This method resolve the issues mentioned above automatically and let the system work without extra protection.

To calculate virtual wind farm parameters, every parameter obtained by electrical impact factors in per unit. For example, the calculated X_rr can be obtained by X_rr^[pu]=Σ_m=1ⁿu_rmX_rrm^[pu], and the base apparent power is obtained as S_b=Σ_j=1ⁿS_bj.

The electrical torque of induction machine can be found by Equation 35.

$T_e=\frac{X_m^2R_rs_0v_s}{{\Delta }_{T_e}}$ ${\Delta }_{T_s}={\left[R_sR_r+s_0\left(X_m^2-X_{\mathrm{ss}}X_{\mathrm{rr}}\right)\right]}^2+{\left[R_rX_{\mathrm{ss}}+s_0R_sX_{\mathrm{rr}}\right]}^2$ $\mathrm{where}.$

Steady-state stator d-q currents are shown by Equations 36 and 37.

$\begin{array}{cc}{i}_{\mathrm{qr}0}=-\frac{s_0^2X_mX_{\mathrm{rr}}}{s_0^2X_{\mathrm{rr}}^2+R_r^2}i_{\mathrm{qs}0}-\frac{s_0X_mR_r}{s_0^2X_{\mathrm{rr}}^2+R_r^2}i_{\mathrm{ds}0},i_{\mathrm{dr}0}=\frac{s_0X_mR_r}{s_0^2X_{\mathrm{rr}}^2+R_r^2}i_{\mathrm{qs}0}-\frac{s_0^2X_mX_{\mathrm{rr}}}{s_0^2X_{\mathrm{rr}}^2+R_r^2}i_{\mathrm{ds}0}& \mathrm{Equation}36\\ i_{\mathrm{qs}0}=\frac{v_{\mathrm{qs}0}-\frac{B_{q0}}{B_{d0}}v_{\mathrm{ds}0}}{A_{q0}-A_{d0}},i_{\mathrm{ds}0}=\frac{v_{\mathrm{ds}0}-\frac{A_{d0}}{A_{q0}}v_{\mathrm{qs}0}}{B_{d0}-B_{q0}}& \mathrm{Equation}37\end{array}$

where A_q0, B_q0, A_d0 and B_d0 are shown by Equation 38.

$\begin{array}{cc}{A}_{q0}=R_s+\frac{s_0X_m^2R_r}{R_r^2+s_0^2X_{\mathrm{rr}}^2},B_{q0}=X_{\mathrm{ss}}+\frac{s_0^2X_m^2X_{\mathrm{rr}}}{R_r^2+s_0^2X_{\mathrm{rr}}^2}& \mathrm{Equation}38\\ A_{d0}=-X_{\mathrm{ss}}+\frac{s_0^2X_m^2X_{\mathrm{rr}}}{R_r^2+s_0^2X_{\mathrm{rr}}^2},B_{d0}=R_s+\frac{s_0X_m^2R_r}{R_r^2+s_0^2X_{\mathrm{rr}}^2}& \end{array}$

Second-order equation of s₀ is

${\alpha }_2s\frac{2}{0}-{\alpha }_1s_0+{\alpha }_0=0,$

where α₂, α₁ and α₀ are shown by Equation 39.

α₂=DR_r+v_s², α₁=2DR_r+v_s², α₀=DR_r−P_mR_r  Equation 39:

Mechanical linearized equation of induction machine is shown by Equation 40.

ΔT_m=X_mi_dr0Δi_qs−X_mi_qr0Δi_ds−X_mi_ds0Δi_qr+X_mi_qs0Δi_dr−2HpΔω_r  Equation 40:

The resulted four electrical linearized equations by reducing Δw_r from d-q equations of Equations 41-44.

Δv_qs=A_qs(jω)Δi_qs+B_qs(jω)Δi_ds+C_qs(jω)Δi_qr+D_qs(jω)Δi_dr  Equation 41:

Δv_ds=A_ds(jω)Δi_qs+B_ds(jω)Δi_ds+C_ds(jω)Δi_qr+D_ds(jω)Δi_dr  Equation 42:

Δv_qr=A_qr(jω)Δi_qs+B_qr(jω)Δi_ds+C_qr(jω)Δi_qr+D_qr(jω)Δi_dr  Equation 43:

Δv_dr=A_dr(jω)Δi_qs+B_dr(jω)Δi_ds+C_dr(jω)Δi_qr+D_dr(jω)Δi_dr  Equation 44:

Where A_qs(jw), B_qs(jw), C_qs(jw), . . . are shown by Equation 45.

$\begin{array}{cc}{A}_{\mathrm{qs}}=R_s+\frac{j\omega }{{\omega }_b}X_{\mathrm{ss}},B_{\mathrm{qs}}=X_{\mathrm{ss}},C_{\mathrm{qs}}=\frac{j\omega }{{\omega }_b}X_m,D_{\mathrm{qs}}=X_m& \mathrm{Equation}45\\ A_{\mathrm{ds}}=-X_{\mathrm{ss}},B_{\mathrm{ds}}=R_s+\frac{j\omega }{{\omega }_b}X_{\mathrm{ss}},C_{\mathrm{ds}}=-X_m,D_{\mathrm{ds}}=\frac{j\omega }{{\omega }_b}X_m& \\ A_{\mathrm{qr}}=\frac{j\omega }{{\omega }_b}X_m-\left(X_mi_{\mathrm{ds}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{dr}0}& \\ B_{\mathrm{qr}}=s_0X_m+\left(X_mi_{\mathrm{ds}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{qr}0}& \\ C_{\mathrm{qr}}=R_r+\frac{j\omega }{{\omega }_b}X_{\mathrm{rr}}+\left(X_mi_{\mathrm{ds}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{ds}0}& \\ D_{\mathrm{qr}}=s_0X_{\mathrm{rr}}-\left(X_mi_{\mathrm{ds}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{qs}0}& \\ A_{\mathrm{dr}}=-s_0X_m+\left(X_mi_{\mathrm{qs}0}+X_{\mathrm{rr}}i_{\mathrm{qr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{dr}0}& \\ B_{\mathrm{dr}}=\frac{j\omega }{{\omega }_b}X_m-\left(X_mi_{\mathrm{qs}0}+X_{\mathrm{rr}}i_{\mathrm{qr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{qr}0}& \\ C_{\mathrm{dr}}=-s_0X_{\mathrm{rr}}-\left(X_mi_{\mathrm{qs}0}+X_{\mathrm{rr}}i_{\mathrm{qr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{ds}0}& \\ D_{\mathrm{dr}}=R_r+\frac{j\omega }{{\omega }_b}X_{\mathrm{rr}}+\left(X_mi_{\mathrm{qs}0}+X_{\mathrm{rr}}i_{\mathrm{dr}0}\right)\left(\frac{X_m}{{\Delta }_{m0}}\right)i_{\mathrm{qs}0}& \\ {\Delta }_{m0}=2\mathrm{Hj}\omega +\frac{\partial C_p}{\partial \lambda }\frac{{\mathrm{rP}}_W}{\left(G\right)V_W\left(1-s_0\right)}-\frac{P_{m0}}{{\left(1-s_0\right)}^2}& \\ \mathrm{where}& \end{array}$

The resulted two d-q linearized equation by reducing rotor and mechanical linearized equations is shown by Equation 46.

Δv_qs=α_q(jω)Δi_qs+β_q(jω)Δi_ds, Δv_ds=α_d(jω)Δi_qs+β_d(jω)Δi_ds   Equation 46:

where α_q(jw), β_q(jw), α_d(jw) and β_d(jw) are shown by Equation 47.

$\begin{array}{cc}{\alpha }_q=A_{\mathrm{qs}}+C_{\mathrm{qs}}\left(\frac{A_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}A_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right)+D_{\mathrm{qs}}\left(\frac{A_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}A_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right)& \mathrm{Equation}47\\ {\beta }_q=B_{\mathrm{qs}}+C_{\mathrm{qs}}\left(\frac{B_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}B_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right)+D_{\mathrm{qs}}\left(\frac{B_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}B_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right)& \\ {\alpha }_d=A_{\mathrm{ds}}+C_{\mathrm{ds}}\left(\frac{A_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}A_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right)+D_{\mathrm{ds}}\left(\frac{A_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}A_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right)& \\ {\beta }_d=B_{\mathrm{ds}}+C_{\mathrm{ds}}\left(\frac{B_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}B_{\mathrm{dr}}}{C_{\mathrm{qr}}-\frac{D_{\mathrm{qr}}}{D_{\mathrm{dr}}}C_{\mathrm{dr}}}\right)+D_{\mathrm{ds}}\left(\frac{B_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}B_{\mathrm{qr}}}{D_{\mathrm{dr}}-\frac{C_{\mathrm{dr}}}{C_{\mathrm{qr}}}D_{\mathrm{qr}}}\right)& \end{array}$

Therefore the electrical impact factors can be found using Equation 48.

$\begin{array}{cc}{u}_m=\frac{\Delta i_{\mathrm{qsm}}}{\Delta i_{\mathrm{qs}}}& \\ \mathrm{where}\text{:}& \\ \Delta v_{\mathrm{qs}}=K\left(j\omega \right)\Delta i_{\mathrm{qs}}+G\left(j\omega \right)\Delta v_{\mathrm{ds}},& \mathrm{Equation}49\\ K\left(j\omega \right)={\alpha }_q\left(j\omega \right)-\frac{{\beta }_q\left(j\omega \right){\alpha }_d\left(j\omega \right)}{{\beta }_d\left(j\omega \right)}& \\ G\left(j\omega \right)=\frac{{\beta }_q\left(j\omega \right)}{{\beta }_d\left(j\omega \right)}& \end{array}$

Relation between Δw_r and Δi_qs is Δw_r=C(jw) Δi_qs, where C(jw) is shown by Equation 50.

$\begin{array}{cc}C\left(j\omega \right)=\frac{X_mi_{\mathrm{dr}0}-X_mi_{\mathrm{qr}0}C_1\left(j\omega \right)}{C_0+2\mathrm{Hj}\omega }+\frac{-X_mi_{\mathrm{ds}0}C_2\left(j\omega \right)+X_mi_{\mathrm{qs}0}C_3\left(j\omega \right)}{C_0+2\mathrm{Hj}\omega }& \mathrm{Equation}50\\ \mathrm{where},& \\ C_0=\left(\frac{\partial C_p\left(\lambda ,\beta \right)}{\partial \lambda }\frac{{\mathrm{rP}}_W}{\left(G\right)V_W}-T_{m0}\right)\frac{1}{{\omega }_{r0}},C_1\left(j\omega \right)=-\frac{{\alpha }_d}{{\beta }_d}& \mathrm{Equation}51\\ C_2\left(j\omega \right)=\left(\frac{A_{\mathrm{qr}}\left(j\omega \right)-\frac{D_{\mathrm{qr}}\left(j\omega \right)}{D_{\mathrm{dr}}\left(j\omega \right)}A_{\mathrm{dr}}\left(j\omega \right)}{{\Delta }_q}\right)+\left(\frac{B_{\mathrm{qr}}\left(j\omega \right)-\frac{D_{\mathrm{qr}}\left(j\omega \right)}{D_{\mathrm{dr}}\left(j\omega \right)}{\beta }_{\mathrm{dr}}\left(j\omega \right)}{{\Delta }_q}\right)C_1\left(j\omega \right)& \\ C_3\left(j\omega \right)=\left(\frac{A_{\mathrm{dr}}\left(j\omega \right)-\frac{C_{\mathrm{dr}}\left(j\omega \right)}{C_{\mathrm{qr}}\left(j\omega \right)}A_{\mathrm{qr}}\left(j\omega \right)}{{\Delta }_d}\right)+\left(\frac{B_{\mathrm{dr}}\left(j\omega \right)-\frac{C_{\mathrm{dr}}\left(j\omega \right)}{C_{\mathrm{qr}}\left(j\omega \right)}{\beta }_{\mathrm{qr}}\left(j\omega \right)}{{\Delta }_d}\right)C_1\left(j\omega \right)& \end{array}$

Therefore, the mechanical impact factors can be found as Equation 52.

$\begin{array}{cc}{y}_k=\frac{\Delta {\omega }_{r_k}}{\Delta {\omega }_{r_{\mathrm{eq}}}}=\frac{C_k\left(j\omega \right)K\left(j\omega \right)}{K_k\left(j\omega \right)C\left(j\omega \right)}& \mathrm{Equation}52\end{array}$

Equivalent system mechanical relations are shown using Equations 53 and 54.

$\begin{array}{cc}\sum _{i=1}^nC_{\mathrm{pi}}P_{\mathrm{Wi}}=C_{\mathrm{peg}}P_{\mathrm{Weq}},\mathrm{where}P_{W_{\mathrm{eq}}}=\frac{1}{2}\rho A_{\mathrm{eq}}V_{\mathrm{Weq}}^3& \mathrm{Equation}53\\ \sum _{i=1}^nP_{\mathrm{Wi}}\frac{\partial C_{\mathrm{pi}}}{\partial {\lambda }_i^{\prime }}\frac{r_iy_i}{V_{\mathrm{Wi}}}=P_{W_{\mathrm{eq}}}\frac{\partial C_{\mathrm{peq}}}{\partial {\lambda }_{\mathrm{eq}}^{\prime }}\frac{r_{\mathrm{eq}}}{V_{\mathrm{Weq}}}& \mathrm{Equation}54\\ \mathrm{where}\text{:}& \\ {\lambda }^{\prime }=G\lambda ,A_{\mathrm{eq}}=\sum _{k=1}^nA_k,r_{\mathrm{eq}}=\sqrt{A_{\mathrm{eq}}/\pi }& \mathrm{Equation}55\\ C_{p_{\mathrm{eq}}}=C_{{\mathrm{pm}}_{\mathrm{eq}}}f\left({\lambda }^{\prime }\right),G_{\mathrm{eq}}=\frac{{\lambda }_{{\mathrm{opt}}_{\mathrm{eq}}}^{\prime }}{{\lambda }_{{\mathrm{opt}}_{\mathrm{eq}}}},& \\ {\lambda }_{{\mathrm{opt}}_{\mathrm{eq}}}=\frac{r}{n}\sum _{k=1}^n\frac{{\lambda }_{{\mathrm{opt}}_k}}{r_k}& \end{array}$

λ′_optea and C_pmeq can be found by help of Equations 53 and 54. Equivalent collector parameters can be found by Equations 56 and 57.

$\begin{array}{cc}R_{C_{\mathrm{eq}}}^{\left[\mathrm{pu}\right]}=\sum _{k=1}^nR_{C_k}^{\left[\mathrm{pu}\right]}u_k& \mathrm{Equation}56\\ C_{C_{\mathrm{eq}}}=\frac{1}{{\omega }_sX_{m_{\mathrm{eq}}}}+\frac{\left(X_{{\mathrm{ls}}_{\mathrm{eq}}}+X_{{\mathrm{lr}}_{\mathrm{eq}}}\right)/{\omega }_s}{{\left(\frac{R_{r_{\mathrm{eq}}}}{s_{\mathrm{eq}}}\right)}^2+{\left(X_{{\mathrm{ls}}_{\mathrm{eq}}}+X_{{\mathrm{lr}}_{\mathrm{eq}}}\right)}^2}-\sum _{k=1}^n\left(\frac{1}{{\omega }_sX_{m_k}}+\frac{\left(X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}\right)/{\omega }_s}{{\left(\frac{R_{r_k}}{s_k}\right)}^2+{\left(X_{{\mathrm{ls}}_k}+X_{{\mathrm{lr}}_k}\right)}^2}-C_{C_k}\right)& \mathrm{Equation}57\end{array}$

Other required equations are:

$\begin{array}{cc}{V}_s=V_b\sqrt{\frac{R_s^2+{\omega }_e^2L_{\mathrm{ss}}^2}{R_s^2+{\omega }_b^2L_{\mathrm{ss}}^2}}& \mathrm{Equation}58\\ {\omega }_e=\frac{{\omega }_m+\sqrt{\max \left(0,{\omega }_m^2+X_{\mathrm{corr}}\right)}}{2}& \mathrm{Equation}59\\ X_{\mathrm{corr}}=3P\left(v_{\mathrm{qs}}i_{\mathrm{qs}}-2R_sI_s^2\right)/K_{\mathrm{tv}}& \mathrm{Equation}60\\ K_{\mathrm{tv}}=\frac{3\left(\frac{P}{2}\right)L_M^2R_rV_b^2}{R_r^2\left(R_s^2\right)+{\omega }_b^2L_{\mathrm{ss}}^2}& \mathrm{Equation}61\end{array}$

A large number of WTGs use induction generators with the stator directly connected to the grid. Due to the wide wind speed range, such induction machines operate at high slip away from their nominal speed. The high slip results in high rotor loss, low efficiency and heated rotor in WTGs with a squirrel cage rotor limiting the operating speed range and output power. Hence, the WTGs with a squirrel cage rotor that can efficiently operate close to the nominal speed are called fixed-speed WTGs. To expand the speed range of such induction machines, the rotor can be connect to the grid through an AC/DC/AC variable frequency converter forming a DFIG shown in FIGS. 9 and 10A. This converter delivers a portion of the rotor energy back to the grid resulting in a rotor loss reduction and a larger wind speed operating range.

FIG. 10B and FIG. 10A show a diagram of a fixed-speed and a DFIG WTG, respectively. Each WTG includes a wind turbine as the main mechanical part, generating mechanical power P_m and mechanical torque T_m according to Equation 62.

P _m =T _mω_m=½C_p(λ,β)ρAV_W³,  Equation 62:

where w_m is the mechanical speed of generator. The mechanical power is related to wind power by turbine coefficient C_p(λ, β). This factor depends on the structure of the wind turbine. β is the blade angle, λ=rw_l=V_W, C_p(λ, β) expressed by Equation 63, V_W is wind speed, ρ is air density, A is area covered by the blades, r is blades radius and w_l is blades rotational speed.

$\begin{array}{cc}{C}_p\left(\lambda ,\beta \right)=\frac{C_{\mathrm{pmax}}}{{\lambda }_{\mathrm{opt}}^2}\left(2{\lambda }_{\mathrm{opt}}-\lambda \right)\lambda .& \mathrm{Equation}63\end{array}$

While another control system can be applied, the present invention uses DFIG control approach, as shown in FIG. 9. In this control system, the Grid-Side Converter (GSC) controls the DC bus voltage to ensure the DC link voltage stability. A PI compensator uses the DC voltage error to generate the GSC d-axis reference current i*_dg. Also, GSC can provide a desired amount of reactive power Q*_g using the q-axis reference current i*_qg. An inner current control loop generates the gating signals for the GSC by means of i*_dg and i*_qg. The Rotor-Side Converter (RSC) controls the induction machine operating under a large range of wind speeds. RSC measures the rotor speed w_m and generates the torque reference T*_e through T*_e=K_optw²_m/G³, where K_opt is defined as Equation 64.

K _opt=½ρπr⁵C_pmax/λ_opt³,  Equation 64:

and G is the gear-box ratio. The T*_e signal is also used to form the rotor q-axis reference current i*_qr for the RSC current controllers by Equation 65.

$\begin{array}{cc}{i}_{\mathrm{qr}}^{*}=-\frac{T_e^{*}}{\frac{3}{2}p\frac{L_m}{L_s}{\psi }_s},& \mathrm{Equation}65\end{array}$

where p is the number of machine poles, L_m is the magnetizing inductance, L_s is the stator self inductance and |ψ_s| is stator linkage flux and estimated with

${\psi }_s=\sqrt{\frac{2}{3}}\frac{V_s}{{\omega }_s},$

where V_s is the rms value of the stator voltage and w_s is the synchronous speed. Also, the rotor d-axis reference current i*_dr is set to zero to use all of the RSC capacity for active power delivery from the rotor windings as the required reactive power for the induction machine is provided by the GSC. It is worth mentioning that, the generality of the inventive method is not limited by the i*_dr set value.

To find the dynamic model of the wind farm, a small-signal model of the WTG and windfarm should be derived. Small signal model of a WTG requires the steady-state calculations at the operating condition. The steady-state electro-mechanical relationship between the mechanical side and electrical side of a WTG are expressed as Equation 66.

T _m +T _e =Dω _m.  Equation 66:

T_e is the electrical torque and D is the mechanical damping of the induction machine. The steady state speed, w_m0, can be found with the assumption that the control system is stable so that T*_e=T_e under the steady-state and by substituting Equation 62 and T*_e=K_optw²_m into Equation 66 and solving it for w_m. Therefore, all other DFIG steady-state parameters can be expressed using Equation 67.

$\begin{array}{cc}{T}_{e_0}=K_{\mathrm{opt}}{\omega }_{m_0}^2,I_{\mathrm{qr}}=-\frac{T_{e_0}}{\frac{3}{2}p\frac{L_m}{L_s}{\psi }_s},I_{\mathrm{dr}}=0,& \mathrm{Equation}67\\ T_{m_0}=T_{e_0}+D{\omega }_{m_0},I_{\mathrm{qs}}=-\frac{L_m}{L_s}I_{\mathrm{qr}},I_{\mathrm{ds}}=\frac{{\psi }_s}{L_s}& \end{array}$

A squirrel cage induction machine T_e0≅X_m²R_rs₀|v_s|/ΔT_e in the steady state where ΔT_e=[R_sR_r+s₀(X_m²−X_ssX_rr)]²+[R_rX_ss+s₀R_sX_rr]² and, X_m, X_ss and X_r are the magnetizing inductance, stator and rotor self-inductances respectively. R_s and R_r are the stator and rotor resistances and s₀ is the slip. By substituting Equation 62 into Equation 66 and solving that for w_m, the w_m0 can also be found for a fixed-speed WTG. By applying V_qr=V_dr=0 for squirrel cage rotors and w_m0 in the induction machine steady-state equations, the other steady-state parameters such as Ids, I_qs, I_dr and I_qs can be found.

Next, the windfarm equivalent electrical side (generator and control system) and mechanical part (the equivalent turbine) separately.

Each WTG has a set of small-signal equations which form d and q axis circuits shown in FIG. 3. The objective is to find an equivalent WTG with the same structure of the individual WTGs whose dynamic behavior is close to the overall windfarm from the grid point of view. Therefore, the windfarm equivalent WTG should also has a set of small signal equations and the d and q axis circuits similar to the individual WTG as in Equation 68.

$\begin{array}{cc}{\tilde{v}}_{{\mathrm{ds}}_{\mathrm{eq}}}={\tilde{v}}_{\mathrm{ds}},{\tilde{v}}_{{\mathrm{ds}}_{\mathrm{eq}}}={\tilde{v}}_{\mathrm{qs}}& \mathrm{Equation}68\\ {\tilde{i}}_{{\mathrm{ds}}_{\mathrm{eq}}}=\sum _{k=1}^n{\tilde{i}}_{{\mathrm{ds}}_k},{\tilde{i}}_{{\mathrm{qs}}_{\mathrm{eq}}}=\sum _{k=1}^n{\tilde{i}}_{{\mathrm{qs}}_k}.& \end{array}$

Therefore, to find the interaction of the windfarm with the grid and its equivalent model, the differential equation relating {tilde over (ι)}_dqs and v_dqs should be derived. Therefore, first the WTG mechanical small-signal equation for {tilde over (w)}_m is derived as a function of {tilde over (ι)}_dqr and {tilde over (ι)}_dqs. Then, the WTG small-signal equations is used to derive {tilde over (ι)}_dqr as a function of {tilde over (ι)}_dqs and {tilde over (v)}_dqs. Finally, applying the resulted equations to the stator small signal equations, {tilde over (ι)}_qs can be found as a function of {tilde over (v)}_dqs.

As P_W=½C_p(λ,β)ρZV_W³ is constant, Equation 62 is linearized as Equation 69.

{tilde over (P)} _m ={tilde over (T)} _mω_m0+T_m0{tilde over (ω)}_m={tilde over (C)}_p(λ,β)P_W,  Equation 69:

where β is constant by Equation 70.

$\begin{array}{cc}{\tilde{C}}_p\left(\lambda ,\beta \right)=\frac{\partial C_p}{\partial \lambda }{}_{\lambda ={\lambda }_0}\tilde{\lambda }=\frac{\partial C_p}{\partial \lambda }{}_{\lambda ={\lambda }_0}\frac{r{\tilde{\omega }}_m}{{\mathrm{GV}}_W}& \mathrm{Equation}70\end{array}$

Considering Equation 70 and solving Equation 69 for {tilde over (T)}_m yields Equation 71.

$\begin{array}{cc}{\tilde{T}}_m=e_m{\tilde{\omega }}_me_m=\left(\frac{\partial C_p}{\partial \lambda }{|}_{\lambda ={\lambda }_0}\frac{{\mathrm{rP}}_W}{{\mathrm{GV}}_W}-T_{m0}\right)/{\omega }_{m_0}.\mathrm{where}& \mathrm{Equation}71\end{array}$

The mechanical linearized equation of the induction machine can be expressed as Equation 72.

$\begin{array}{cc}{\tilde{T}}_m=X_mI_{\mathrm{dr}}{\tilde{i}}_{\mathrm{qs}}-X_mI_{\mathrm{qr}}{\tilde{i}}_{\mathrm{ds}}-X_mI_{\mathrm{ds}}{\tilde{i}}_{\mathrm{qr}}-X_mI_{\mathrm{qs}}{\tilde{i}}_{\mathrm{dr}}-2H\frac{d}{\mathrm{dt}}{\tilde{\omega }}_m.& \mathrm{Equation}72\end{array}$

Now by substituting Equation 71 into Equation 72 {tilde over (w)}_m can be found in terms of {tilde over (ι)}_dqs and {tilde over (ι)}_dqr. Assuming a very fast controller results in T*_e=T_e and {tilde over (ι)}*_dqr={tilde over (ι)}_dqr, and substituting Equation 71 in the linearized controller equation of Equation 65 yields Equation 73.

$\begin{array}{cc}{\tilde{i}}_{\mathrm{qr}}=\frac{e_m}{\frac{3}{2}p\frac{L_m}{L_s}{\psi }_s}{\tilde{\omega }}_m.& \mathrm{Equation}73\end{array}$

{tilde over (ι)}_qr can be found as a function of only {tilde over (ι)}_dqs because {tilde over (w)}_m is as function of {tilde over (ι)}_dqs and {tilde over (ι)}_dqr, while {tilde over (ι)}_dr=0 as i*_dr=0. Therefore, the stator linearized equations can be derived as Equation 74.

$\begin{array}{cc}{\tilde{v}}_{\mathrm{ds}}=\left(R_s+L_s\frac{d}{\mathrm{dt}}\right){\tilde{i}}_{\mathrm{ds}}-{\omega }_sL_s{\tilde{i}}_{\mathrm{qs}}+L_m\frac{d}{\mathrm{dt}}{\tilde{i}}_{\mathrm{dr}}-{\omega }_sL_m{\tilde{i}}_{\mathrm{qr}}{\tilde{v}}_{\mathrm{qs}}={\omega }_sL_s{\tilde{i}}_{\mathrm{ds}}+\left(R_s+L_s\frac{d}{\mathrm{dt}}\right){\tilde{i}}_{\mathrm{qs}}+{\omega }_sL_m{\tilde{i}}_{\mathrm{dr}}+L_m\frac{d}{\mathrm{dt}}{\tilde{i}}_{\mathrm{qr}},& \mathrm{Equation}74\end{array}$

which yields Equation 75.

{tilde over (v)} _ds=α_ddĩ_ds+α_dqĩ_qs

{tilde over (v)} _qs=α_qdĩ_ds+α_qqĩ_qs,  Equation 75:

where α_dd, α_dq, α_qd and α_qq are a function of both WTG and calculated steady-state parameters. Replacing

$\frac{d}{\mathrm{dt}}$

with jw, the frequency response of α_dd, α_dq, α_qd and α_qq can be found for an arbitrary w. The final frequency response occurs at w=0. Thus, using α_dd, α_dq, α_qd and α_qq at w=0, the desired {tilde over (v)}_dqs, {tilde over (ι)}_dqs can be found. To find the same set of equations for a fixed-speed WTG the same steps can be followed by considering {tilde over (v)}_dqr=0 due to its squirrel cage rotor structure. Finally, rewriting Equation 75 for {tilde over (ι)}_dqs yields Equation 76.

ĩ _ds =y _dd {tilde over (v)} _ds +y _dq {tilde over (v)} _qs

ĩ _qs =y _qd {tilde over (v)} _ds +y _qq {tilde over (v)} _qs,  Equation 76:

where details of y_ddqq is given in Equation 89 below.

Considering Equation 76 for individual WTGs, the contribution of each WTG to the windfarm current injection to the grid can be found. The inventive method determines the equivalent WTG parameters by the weighted average of WTGs parameters, where these weights are determined by the contribution of each WTG injected current to the grid. As FIGS. 11A and 11B show, the current of each WTG and windfarm can be projected to d and q axis circuits. As shown in FIG. 12, to make derivations more straightforward, the dq axis is rotated to d′q′ so that the overall current injected to the grid does not have any q—component. Rearranging Equation 76 in a new d-q frame where Σ_k=1ⁿĩ′_qsk=0 and for a desired {tilde over (v)}′_ds=|ΔV_s| and {tilde over (v)}′_qs=0 yields Equation 77.

ĩ′ _ds=(y_dd cos²(δ₀)+y_dd sin²(δ₀)−(y_dq+y_qd)sin(δ₀)cos(δ₀)){tilde over (v)}′_ds

ĩ′ _qs=(y_qd cos²(δ₀)−y_dq sin²(δ₀)+(y_dd−y_qq)sin(δ₀)cos(δ₀)){tilde over (v)}′_ds,  Equation 77:

where δ₀ is the phase difference between the two d-q reference frames as shown in the FIG. 12. Therefore, the weight for the dynamic equations of m-th WTG is defined as Equation 78.

$\begin{array}{cc}{\mu }_m=\frac{{\tilde{i}}_{{\mathrm{ds}}_m}^{\prime }}{{\sum }_{k=1}^n{\tilde{i}}_{{\mathrm{ds}}_k}^{\prime }}.& \mathrm{Equation}78\end{array}$

The equivalent machine and control parameters can be found by the weighted average of the windfarm parameters in per unit while the equivalent apparent power of the equivalent WTG is the summation of the windfarm WTGs apparent powers. For example, the equivalent L_meq can be achieved as Equation 79.

$\begin{array}{cc}{L}_{m_{\mathrm{eq}}}^{\mathrm{pu}}=\sum _{k=1}^n{\mu }_kL_{m_k}^{\mathrm{pu}},\mathrm{and}S_{\mathrm{eq}}=\sum _{k=1}^nS_k.& \mathrm{Equation}79\end{array}$

Note that the control parameters can also be per unitized by their equations. For example, K_p unit is [V/A] and is similar to an impedance.

To find an equivalent turbine and the equivalent wind speed, a few facts should be considered. First, the area which is covered by the equivalent turbine should be equal to the summation of the area that is covered by the al the WTGs in the windfarm combined, as shown by Equation 80.

$\begin{array}{cc}{A}_{\mathrm{eq}}=\sum _{k=1}^nA_k.& \mathrm{Equation}80\end{array}$

Second, the amount of wind power in the area is independent of windfarm structure, yielding Equation 81.

$\begin{array}{cc}{P}_{W_{\mathrm{eq}}}=\sum _{k=1}^nP_{W_k}\Rightarrow V_{W_{\mathrm{eq}}}=\sqrt[3]{\left(\sum _{k=1}^nA_kV_{W_k}^3\right)/A_{\mathrm{eq}}}.& \mathrm{Equation}81\end{array}$

Third, the equivalent mechanical power P_m generated by the equivalent turbine should also be equal to the summation of windfarm generated mechanical power per Equation 82.

$\begin{array}{cc}{P}_{m_{\mathrm{eq}}}=\sum _{k=1}^nP_{m_k}.& \mathrm{Equation}82\end{array}$

To have more realistic situation one can assume T_eeq=Σ_k=1ⁿT_ek in steady-state which yields Equation 83.

$\begin{array}{cc}{G}_{\mathrm{eq}}=\sqrt[3]{{\omega }_{m_{0_{\mathrm{eq}}}}K_{{\mathrm{opt}}_{\mathrm{eq}}}/\left(\sum _{k=1}^n{\omega }_{m_{0_k}}^2K_{{\mathrm{opt}}_k}/G_k^3\right)}.& \mathrm{Equation}83\end{array}$

It should be noted that any speed ratio G_eqp_eq between the mechanical and electrical side can be used as long as Equation 82 is satisfied and it will not limit the generality of the method. To find

${\omega }_{m_{{\underset{.}{o}}_{\mathrm{eq}}}},$

Equation 66 can be used similar to the steady state derivation by considering Equation 84.

$\begin{array}{cc}{I}_{{\mathrm{qs}}_{\mathrm{eq}}}=\sum _{k=1}^nI_{{\mathrm{qs}}_k},I_{{\mathrm{qr}}_{\mathrm{eq}}}=-\frac{L_{s_{\mathrm{eq}}}}{l_{m_{\mathrm{eq}}}}I_{{\mathrm{qs}}_{\mathrm{eq}}},T_{e_{0_{\mathrm{eq}}}}=-\frac{3}{2}p_{\mathrm{eq}}\frac{L_{m_{\mathrm{eq}}}}{L_{s_{\mathrm{eq}}}}{\psi }_{s_{\mathrm{eq}}}I_{{\mathrm{qr}}_{\mathrm{eq}}},T_{m_{0_{\mathrm{eq}}}}=\frac{P_{m_{0_{\mathrm{eq}}}}}{{\omega }_{m_{0_{\mathrm{eq}}}}}.& \mathrm{Equation}84\end{array}$

Finally, considering and K_opteq=½ρπr_eq⁵C_pmaxeq/λ_opteq³ and Equation 82, the equivalent turbine curve C_pmaxeq and λ_opteq can be found by Equation 85.

$\begin{array}{cc}{C}_{p_{\mathrm{eq}}}P_{W_{\mathrm{eq}}}=\sum _{k=1}^nC_{\mathrm{pk}}P_{W_k}K_{{\mathrm{opt}}_{\mathrm{eq}}}^{\mathrm{pu}}=\sum _{k=1}^n{\mu }_kK_{{\mathrm{opt}}_k}^{\mathrm{pu}}.& \mathrm{Equation}85\end{array}$

It is worth nothing that Equation 88 below can also be found for a fixed-speed windfarm by following the same steps that led to Equation 85 equivalent turbine equations for a DFIG windfarm. Solving Equation 85 and Equation 88 yields equivalent turbine λ_opteq and C_pmaxeq for a DFIG windfarm and a fixed speed windfarm respectively. Now using equivalent generator, controllers and turbine parameters, an equivalent WTG can be achieved for the whole windfarm.

A 4-WTGs DFIG windfarm shown in FIG. 13 and a large scale 20-WTGs DFIG windfarm shown in FIG. 14 are studied under various operating scenarios. The simulation parameters are shown in the Tables IV, V, and VI. The rest of DFIG windfarm parameters are calculated using Equation 87 below. Also, a 4-WTGs fixed-speed windfarm with a similar configuration in FIG. 13 is studied to verify the generality of the inventive method. The parameters for this simulation are provided in Table VII.

In 4-WTGs DFIG simulations, a 0.2 pu voltage sag is applied at t=3 s and is cleared at t=5 s. To study the transients responses, two scenarios (A,B) are considered where the WTGs have similar and unequal parameters. Similarly, for fixed-speed windfarm simulations, the voltage sag is considered to start at t=ls and end at t=2 s. To compare and verify the proposed method, all windfarms are aggregated with the Full Agg, Zone Agg and the proposed WD Agg methods. A simple small windfarm is chosen for the first two scenarios to simplify the comparison between the inventive equivalent WTG and other existing methods. Moreover, 20-WTGs large scale windfarm is chosen as the third scenario to verify the generality and applicability of the inventive method in real life cases. All scenarios are simulated by MATLABnSimulink 2019b.

Scenario A: 4-WTGs DFIG Windfarm with Equal Parameters. When all WTGs parameters are equal, all voltage nodes have similar dynamic equations. Hence, the nodes can be connected to each other which put similar components of the WTGs in parallel. For example, the L_meq for an accurate equivalent model will be L_ml//L_ml// : : : //L_mn. The same rule applies to L_s, R_s and C_DC. Table II shows the calculated equivalent parameters for the mentioned components in all methods. Table II results verify the accuracy of WD method in existing methods assumption because the equivalent impedances are following the rule as mentioned. Also, controller parameters like K_p and K_i which have impedance unit type have ¼ value of the detailed system. It seems they are behaving similar to the real impedance components in parallel. The same justification is true for the control parameter K_opt and turbine shaft parameter D which have admittance unit type in WD Agg. Therefore, Table II results also verify the accuracy of aggregating controller and turbine parameters in per unit with WD method.

TABLE II
EQUIVALENT ROTOR-SIDE CONVERTER
AND TURBINE PARAMETERS
	Name of the
	parameter	Full Agg	Zone Agg	WD Agg
	Cpmax	0.48	0.48, 0.48	0.2729
	λopt	8.1	8.1, 8.1	13.42
	Kopt × 10−5	72.684	12.849, 12.849	9.0855
	Ki	122.9	245.8, 245.8	122.9
	Kp	0.1443	0.2886, 0.2886	0.1443
	Lm[mH]	0.625	1.25, 1.25	0.625
	Ls[mH]	0.64675	1.3, 1.3	0.64675
	Rs[mΩ]	0.65	1.3, 1.3	0.65
	CDC[F]	0.32	0.16, 0.16	0.32
	D	0.004	0.002, 0.002	0.004

FIG. 15 shows the PCC voltage, phase A current, active and reactive power for all aggregation methods. As it shows, the Full Agg model has oscillatory current due to a very large K_opt. Therefore, in Full Agg approach, if the number of WTGs is increased and the same equation are used to design the control parameters, the system model may exhibit oscillatory or even unstable behavior. To solve this issue there are three options: First, dividing the WTGs into few smaller zones and use the Zone Agg which results in higher model complexity. Second K_opteq can be redesigned by K_opteq=Σ_k=1ⁿK_optk which leads to inaccuracy of the model because it violates the designing Equation 64 between K_opt and turbine parameters C_pmaxeq and λ_opteq. Third, the current controller parameters K_i and K_p can be redesigned by trial and error method to stabilize the model. FIG. 16 shows the same results for all methods with modified Full Agg and Zone Agg with the third option. As it shows, the Full Agg model is stable now but with a slower controller. The second option is studied in Scenario C. Both FIGS. 15 and 16 show better performance for the WD Agg model compared to the Full and Zone Agg models.

TABLE III
CALCULATED ERROR INDEXES FOR SCENARIO A AND B
	Current	Active Power	Reactive power
Error Index	Sc. A	Sc. B	Sc. A	Sc. B	Sc. A	Sc. B
errFull/errWD	3.34	4.11	7.12	7.47	11.74	7.67
errZone/errWD	3.11	3.74	3.21	4.01	5.40	3.62
errWD/errWD	1	1	1	1	1	1

Scenario B: 4-WTGs DFIG Windfarm with Unequal Parameters. Turbine shadow effect and different rated powers for WTGs are considered in this scenario. It should be noted that by different apparent power for WTGs, other parameters are also different in per unit as shown in Table V. FIG. 17 shows the PCC voltage, phase A current, active and reactive power for all aggregation methods. WD Agg model shows better performance compared to the Full and Zone Agg models. To measure the model accuracy, an error index can be defined as the integral of the absolute value of the difference between the model and the detailed system responses using Equation 86.

err=∫ _{t=t 0} ^{t=t 1} (|F_detailed−F_model|)dt,  Equation 86:

where F can be any voltage, current or other response curves. This error index is calculated for phase A current, active and reactive power curves in all methods between t₀=2 to t₁=10 and normalized by the minimum error which was WD in all cases. The results are illustrated in the Table III. It can be observed that WD model is at least 3 times more accurate in all scenarios while it has half of the complexity compared to the Zone Agg model.

Scenario C: Large-Scale 20-WTGs DFIG Windfarm. A large-scale windfarm including 20 DFIG WTGs with variable wind speed curve is also studied for all aggregation methods. The system specification is shown in the Table VI. FIG. 18 shows the wind speed, phase A current, active and reactive power curves for the detailed system and all aggregation models. This figure shows a superior performance for WD Agg model compared to the Full and Zone Agg models especially in the transient behavior. Also, calculating the error index defined in Equation 86 for all active power curves shows that the Full Agg and Zone Agg models have higher error about 2.09 and 2.02 times the WD Agg model, respectively. The results achieved from all scenarios demonstrate a superior performance of the proposed WD model compared to the existing methods.

Scenario D: 4-WTGs Fixed-Speed Windfarm. Fixed-speed windfarms are getting obsolete due to their limited operating range, high power loss and the need for a static reactive power compensation. However, they are still functional in the existing energy system and play a considerable role in the power system dynamic behavior. Therefore, the inventive model for the fixed-speed induction machine based windfarms is studied. The parameters of the simulation for fixed-speed windfarm with a similar configuration in FIG. 13 is shown in Table VII. FIG. 19 shows the PCC voltage, phase A current, active and reactive power for all aggregation methods. As Table VII shows, the WTGs parameters are unequal. FIG. 19 shows that Zone and WD Agg provide much accurate models compared to the Full Ag considering shadow effect and unequal parameters for WTGs. It is worth noting that similar to Full Agg, WD Agg models the windfarm with one WTG but with better performance. Using the error index in Equation 86 for the active power curves shows that Full Agg and Zone Agg models have 5.64 and 2.02 times higher error than WD Agg model, respectively, in this scenario. Zone Agg has better performance compared to the Full Agg model at the cost of higher complexity compared to the WD Agg. To have an accurate Zone Agg model for the large-scale windfarms with realistic conditions, a large number of zones should be selected which results in a very high complex model. On the other hand, using a low number of zones will result in higher error than WD model similar to the Full model. Moreover, selected zones should be rearranged by a small change such as wind direction in the windfarm. In conclusion, WD Agg is superior compared to the Full and Zone Agg considering both accuracy and simplicity.

Using machine d-q equations, the present invention provides a systematic and simple method to model large-scale induction machine based windfarms by one WTG. This equivalent WTG d-q model is obtained by quantifying the contribution of each WTG to the windfarm using Weighted Dynamics (WD). Performance of the inventive model is evaluated through simulating and studying a 4-WTGs and a large-scale 20WTGs windfarms in four different scenarios of various wind speeds and WTGs parameters. It is shown that the error of the inventive WD Agg method is at least 2 times less than Full Agg and Zone Agg models. Also, presenting a single equivalent WTG through a one-time simple calculations, results in significantly less computational burden and model complexity compared to equivalent admittance, optimization methods and Semi. Agg models. It is shown that the inventive WD Agg method is adequately accurate in both transients and steady-state responses and it can be readily used for modeling large-scale windfarms to reduce the overall computational burden of the system.

The DFIG controller parameters are designed by Equation 87.

$\begin{array}{cc}\sigma =1-L_m^2/L_sL_r,{\alpha }_1=-L_m/L_s,{\alpha }_2=L_r-L_m^2/L_s{\tau }_i=\sigma L_r/R_r,{\omega }_{\mathrm{ni}}=100/{\tau }_i,{\omega }_{\mathrm{nn}}=1/{\tau }_nK_{p_{\mathrm{idq}}}=2{\omega }_{\mathrm{ni}}\sigma L_r-R_r,K_{i_{\mathrm{idq}}}={\omega }_{\mathrm{ni}}^2L_r\sigma K_{\mathrm{pg}}=-K_{\mathrm{qg}}=1/\left(\frac{3}{2}V_s\sqrt{\frac{2}{3}}\right),{\tau }_{\mathrm{ig}}=L_g/R_g,{\omega }_{\mathrm{nig}}=2\pi fK_{p_{\mathrm{idqg}}}=2{\omega }_{\mathrm{nig}}L_g-R_g,K_{i_{\mathrm{idqg}}}={\omega }_{\mathrm{nig}}^2L_g& \mathrm{Equation}87\end{array}$

The set of equations to find the equivalent turbine parameters for a fixed-speed windfarm are shown in Equation 88.

$\begin{array}{cc}{C}_{p_{\mathrm{eq}}}P_{W_{\mathrm{eq}}}=\sum _{k=1}^nC_{\mathrm{pk}}P_{W_k}{\lambda }_{{\mathrm{opt}}_{\mathrm{eq}}}=\sum _{k=1}^n{\mu }_k{\lambda }_{{\mathrm{opt}}_k}& \mathrm{Equation}88\end{array}$

Equation 76 coefficients y_dd, y_dq, y_qd and y_qq can be found by Equation 89.

y _dd=α_qq/Δ_y, y_dq=−α_dq/Δ_yy_qd=−α_qd/Δ_y

y _qq=α_dd/Δ_y, where: Δ_y=α_ddα_qq−α_dqα_qd.  Equation 89:

The grid voltage v_s=690 [v], grid frequency f=50 [Hz], DC-Link voltage V_DC=1500 [v] and switching frequency f_sw=4 [kHz] for the A, B and C scenarios. The rest of Scenario A, B and C parameters can be found in Tables IV, V and VI respectively. The grid voltage v_s=575 [v] and grid frequency f=60 [Hz] for the Scenario D. The rest of Scenario D parameters can be found in Table VII.

TABLE IV
4-WTGs DFIG WINDFARM, SCENARIO A SPECIFICATIONS
Parameter	WTG 1	WTG 2	WTG 3	WTG 4	Unit
S	2	2	2	2	[MV A]
Lsi	87	87	87	87	[μH]
Lm	2.5	2.5	2.5	2.5	[mH]
Rs	2.6	2.6	2.6	2.6	[mΩ]
Rr	2.9	2.9	2.9	2.9	[mΩ]
p	2	2	2	2	—
J	127	127	127	127	[kgm2]
D	0.001	0.001	0.001	0.001	—
τn	12.5	12.5	12.5	12.5	[ms]
G	100	100	100	100	—
r	42	42	42	42	[m]
Cpmax	048	0.48	0.48	0.48	—
λopt	8.1	8.1	8.1	8.1	—
VW	10	10	10	10	[m/s]
CDC	80	80	80	80	[mF]
Lg	0.4	0.4	0.4	0.4	[mH]
Rg	20	20	20	20	[μΩ]
Kpu	103	103	103	103	—
Kiv	3 × 105	3 × 105	3 × 105	3 × 105	—

TABLE V
4-WTGs DFIG WINDFARM, SCENARIO B SPECIFICATIONS
Parameter	WTG 1	WTG 2	WTG 3	WTG 4	Unit
S	2	2	1	1	[MV A]
Lsi	87	87	87	87	[μH]
Lm	2.5	2.5	2.5	2.5	[mH]
Rs	2.6	2.6	2.6	2.6	[mΩ]
Rr	2.9	2.9	2.9	2.9	[mΩ]
p	2	2	2	2	—
J	127	127	63.5	63.5	[kgm2]
D	0.001	0.001	0.001	0.001	—
τn	12.5	12.5	12.5	12.5	[ms]
G	100	100	100	100	—
r	42	42	42	42	[m]
Cpmax	0.48	0.48	0.48	0.48	—
λopt	8.1	8.1	8.1	8.1	—
VW	11	10	9	8	[m/s]
CDC	80	80	80	80	[mF]
Lg	0.4	0.4	0.4	0.4	[mH]
Rg	20	20	20	20	[mΩ]
Kpu	103	103	103	103	—
Kiv	3 × 105	3 × 105	3 × 105	3 × 105	—

TABLE VI
4-WTGs DFIG WINDFARM, SCENARIO C SPECIFICATIONS
Parameter	WTG 1-5 & 11-15	WTG 6-10 & 16-20	Unit
S	2	1	[MV A]
Lsi	87	87	[μH]
Lm	2.5	2.5	[mH]
Rs	2.6	2.6	[mΩ]
Rr	2.9	2.9	[mΩ]
p	2	2	—
J	127	63.5	[kgm2]
D	0.001	0.001	—
τn	12.5	12.5	[ms]
G	100	100	—
r	42	29.7	[m]
Cpmax	0.48	0.48	—
λopt	8.1	5.73	—
CDC	80	80	[mF]
Lg	0.4	0.4	[mH]
Rg	20	20	[mΩ]
Kpu	103	103	—
Kiv	3 × 105	3 × 105	—

TABLE VII
4-WTGs DFIG WINDFARM, SCENARIO D SPECIFICATIONS
Parameter	WTG 1	WTG 2	WTG 3	WTG 4	Unit
S	149.2	149.2	74.6	74.6	[kV A]
Lsi	284	284	284	284	[μH]
Lm	14.25	14.25	14.25	14.25	[mH]
Rs	24.75	24.75	24.75	24.75	[mΩ]
Rr	13.29	13.29	13.29	13.29	[mΩ]
p	2	2	2	2	—
J	2.6	2.6	1.3	1.3	[kgm2]
D	0.06346	0.06346	0.06346	0.06346	—
G	100	100	100	100	—
r	42	42	42	42	[m]
Cpmax	0.48	0.48	0.48	0.48	—
λopt	8.1	8.1	8.1	8.1	—
VW	11.5	10.5	11	10	[m/s]
Zone	1	2	1	2	—

While at least one exemplary embodiment has been presented in the foregoing description and attached appendix, it should be appreciated that a vast number of variations exist. It should also be appreciated that the exemplary embodiment or exemplary embodiments are only examples, and are not intended to limit the scope, applicability, or configuration of the described embodiments in any way. Rather, the foregoing description and incorporated references will provide those skilled in the art with a convenient roadmap for implementing the exemplary embodiment or exemplary embodiments. It should be understood that various changes may be made in the function and arrangement of elements without departing from the scope as set forth in the appended claims and the legal equivalents thereof.

Claims

1. A method of modeling an equivalent wind turbine generator (WTG) system for a wind farm having a plurality of WTG units, the method comprising: determining an impact factor of each WTG unit of the plurality of WTG units;determining an equivalent single WTG unit model parameters of the wind farm based on the impact factor of each WTG unit; anddetermining an effective wind speed of the wind farm to use as the equivalent WTG input wind speed.

2. The method of claim 1 wherein the method produces a model of static wind farm behavior.

3. The method of claim 1 wherein the method produces a model of dynamic wind farm behavior.

4. The method of claim 1 wherein the plurality of WTG units are fixed speed units.

5. The method of claim 1 wherein a frequency response technique is employed to determine the impact factor of each WTG unit.

6. The method of claim 1 wherein the wind farm features a plurality of wind speed inputs.

7. The method of claim 6 wherein the plurality of wind speed inputs are in a plurality of locations throughout the wind farm.

8. The method of claim 1 wherein the plurality of WTG units feature a plurality of machine parameters.

9. The method of claim 8 further comprising determining an effect of plurality of machine parameters on each WTG unit of the plurality of WTG units.

10. The method of claim 1 wherein the impact factor of a WTG unit is the proportion of the WTG unit output current increment to a total wind farm output current increment.

11. The method of claim 1 wherein the mechanical input of the equivalent WTG unit is the sum of a total mechanical input of all WTG units in the wind farm.

12. The method of claim 1 wherein the mechanical input increment of the equivalent WTG unit is a sum of a total mechanical input increment of all WTG units in the wind farm.

13. The method of claim 1 further comprising determining an equivalent collector system model parameter of the wind farm based on the impact factor and an equilibrium point of each WTG unit.

14. The method of claim 1 wherein determining an equivalent single WTG unit model parameters of the wind farm based on the impact factor of each WTG unit includes first determining an equivalent electrical side of the windfarm including at least one generator or converter based on associated impact factors, and then determining an equivalent mechanical side of the windfarm including at least one wind turbine based on the equivalent electrical side and the impact factors.

15. A computer comprising machine readable medium having stored thereon one or more sequences of instructions configured to execute the method of claim 1.

16. A method of modeling an equivalent system for parallel systems having a plurality of units, the method comprising: determining an impact factor of each unit proportional to its operating point of the plurality of units;determining an equivalent single unit model parameters of the overall parallel system based on the impact factor of each unit; anddetermining an effective input of the overall parallel system to use as the equivalent input.

17. The method of claim 16 wherein the parallel systems include any of parallel converters, PV farms, battery banks and renewables.