The Influence of Wind Power on the Small Signal Stability of a Power System T.R. Ayodele , A.A. Jimoh , J.L Munda, J.T Agee Department of Electrical Engineering. Tshwane University of Technology Pretoria, South Africa Phone: +27735605380, e-mail: AyodeleTR@tut.ac.za, JimohAA@tut.ac.za, MundaJL@tut.ac.za, AgeeJT@tut.ac.zaAbstract. This paper analyses the influence of wind power on cause instability and hence appropriate measures can bethe small signal stability of a power system. Factors like power taken.dispatch, generator technology, wind farm location and windpower integration level are considered. The oscillatory modes The different methods that have been used in thethat arise as a result of changes in system operating conditions literatures for analysing small signal stability of a largeare computed using modal analysis. Participation factors are power system include prony analysis, Fourier-methodalso calculated to determine the relative contribution of eachsystem state variables to a certain mode. Time domain , modal analysis method, time domain analysis andsimulations are carried out to validate the conclusion inferred Probabilistic method [4, 5]. The issue of constant speedfrom the modal analysis. wind turbine generator on power system oscillation was studied in , impact of large scale wind power wasKey words analysed in  using three area network and modal analysis of doubly-fed induction generator was presentedWind power, small signal stability, modal analysis, in  where the main focus is on the control mode.damping ratio, Power system. The objective of this paper is to identify the influence of1. Introduction wind power on the power system small signal stability considering factors like direction of power flow in theGrid integration of wind power today has achieved a tie-line, Transmission line length of the location of windconsiderable development. This fact is leading to a resources to the main grid, integration level and generatorgrowing concern about its influence on the operation of a technology.power system . Much of these concerns are centred onthe impact of wind power integration on the transient and The paper is organised as follows: section two presentsvoltage stability with less attention given to the small the model of wind generators, the methodology of thesignal stability. study is described in section three, the test system and simulation results in different scenario are given inSmall signal stability is considered when there are small section four while the conclusion is presented in sectionchanges in operating parameters of a power system . five.These changes can result in electromechanical oscillationbut mostly decay with time and thus the system comes 2. Characteristic and Modelling of Windback to stable operating point. However, if the system is Generatorsnot adequately damped, the oscillation can lead to loss ofsynchronism. The per unit d-q stator, rotor voltage equation and the flux linkage equation of an induction machine in aSmall signal stability studies are based on linearised synchronously rotating reference frame are given insystem around the operating point. The differential equations(1)-(4) . The generator convention is adoptedequations that describe the dynamic system are linearised i.e the stator and rotor currents are positive when flowingand the system eigenvalues are computed from the towards the network and both active and reactive powercharacteristic equation. Majorly, there are two kinds of are also positive when flowing towards the grid.oscillation usually considered in small signal stabilitystudies, the local area oscillations and the inter-area 1 Vqs = − rs i qs − ψ ds − ω pψ qs (1)oscillations. Participation of each generator’s state s 1variables to a particular mode of interest is computed Vds = − rs i ds + ψ qs − ω pψ ds (2) sfrom the participation matrix. The information obtained 1 Vqr = − rr i qr − sψ dr − ω pψ qrhere is used to identify the critical generators that can s (3) 1 Vdr = − rr idr + sψ qr − ωs pψ dr (4)
The flux linkage equations are written as X2 X m X 2 rs m 1+ − S ω s To + 2 2 2 X r rs + X X r rs2 + X ψ qs = Xs i qs + X m i qr 1 (5) A = − To ψ ds = X s i ds + X m i dr (6) Sω T + X 2 rs m 1+ 2 XmX s o 2 2 X r rs + X 2 2 ψ qr = X r i qr + X m iqs (7) X r rs + X ψ dr = X r i dr + X m i ds (8) X 2 X m X2 rs m Xm 2 2 0 where, 2 Xr X r rs2 + X 1 Xr rs + X X s = X Ls + X m and X r = X Lr + X m B=− 2 To Xm rs X m X 2 Xm − 0 2 2 Xr Xr rs2 + X Xr rs2 + X The subscripts, s, and, r, denote the stator and the rotor quantities respectively, r, is the resistance, i, is thecurrent, ψ is the flux linkage per second, X, is the 3. Methodologyreactance, S is the slip, ωs is the synchronous speed, p isthe differential term (d(.)/dt). Xm, XLs and XLr are the Modal analysis method is adopted in this paper because itmagnetizing, stator leakage and rotor leakage reactances proves to be superior compared to other methods, as itrespectively. All the rotor quantities are referred to the provides additional information to identify the criticalstator. For squirrel cage induction generator Vqr = Vdr =0, generators that may be used to determine the location ofbecause the rotor is short-circuited. The electromagnetic eventually needed stabilizing devices in order totorque developed by the generator is written as influence the system damping efficiently. The dynamic behaviour of a large complex non linear system such as a power system can be represented by (12) and (13). Tem = 1 ωs ( X m i qs iidr − ids i qr ) (9) x = ƒ ( x, u, t ) & (12) y = g ( x, u, t ) (13)The mechanical coupling between the generator and theturbine with one mass system assumed is given by. where T 2H dω r = Tm − Tem (10) ( x = x1, x 2 ,....., x n ) dt T u = ( u1 , u 2 ,...., u n )Where H and Tm are the per unit turbine inertia andmechanical torque respectively. x is the state vector, u is the input vector and y is the output vector.A. Wind Generators Model for Stability Studies To study the small signal characteristic behaviour of aFor stability studies, voltage behind a transient dynamic system, its singular point is important;impedance model is usually adopted with some conclusion about stability can be drawn from a linearisedsimplification . Machine stator transients is neglected system. Therefore the non linear equation (12) and (13)i.e pψ qs = 0 , pψ ds = 0 . In this way, the stator are linearised using Taylor series. The linear result equation is written as (14)and (15) .differential voltage equation in (1) and (2) is simplified ∆ x = A∆ x + B ∆ u & (14)into an arithmetic equation. This is referred to as thirdorder reduced model in the literature. However, to obtain ∆y = C∆x + D∆u (15)this model, equations (1)-(8) in section two are Taking the Laplace transformation will result inrearranged and the following new terms are defined. equations (16) and (17) X2 X adj(sI − A)eqs = ( )ψ Xm Xr dr , eds = ( )ψ Xm Xr qr , X = Xs − m Xr , To = r ωs rr ∆x(s) = & det(sI − A) B∆u(s) (16)The state space model obtained is given by adj(sI − A) ∆y(s) = C B∆u(s) + Du(s) (17) x = Ax + Bu & (11) det(sI − A)where, The poles of ∆x(s) and ∆y(s) are the roots of the T T (x = eqs eds ) , ( u = Vqs Vds Vqr Vdr ) characteristic equation (18) det(sI − A) = 0 (18) Equation (18) can be re-written as (19) det(λI − A) = 0 (19)
The value of λ that satisfies equation (19) are theeigenvalues of matrix A. It contains information aboutthe response of the system to a small perturbation. Theeigenvalue can be real and/or complex. The complexvalues appear in conjugate pairs if A is real (20). λ i = σi ± jωi (20)Each eigenvalue represents a system mode and therelationship between this mode and the stability is givenby Lyapunov,s first method  σ(i) Stable system, then i <0 Figure 1: The system under study. σ(ii) Unstable system, then i >0 In the system under study, 439.96MW of power is(iii) When σi =0, then nothing can be said in general exported from area one to area two. The eigenvalues ofThe frequency of oscillation in, Hz, and the damping the BC are computed, 24 eigenvalues were obtained withratio are given by (21) and (22). negative real part. According to Lyapunov, this system is ω small signal stable. Six of the eigenvalues are complex, ƒ= i (21) 2π denoting an oscillatory mode. Only the results of the −σi oscillatory mode are given in table 1. ξ= (22) σi2 + ωi2 Table 1: Electromechanical mode of the study systemThe participation factor of mode i, can be computed Mode λ = σ ± jω ƒ ,Hz ξusing (23) and (24)  * -1.030 ± 9.002 1.433 0.114 ** -0.715 ± 8.065 1.284 0.088 p1i ∆ -0.069 ± 3.427 0.545 0.020 pi = p 2i (23) .... * Local mode in area one. p mi ** Local mode in area two. ∆ Inter area mode between area one and area two. Φ ki Ψ ik p ki = (24) n The modal bar diagram showing the speed participation ∑ Φ ki Ψ ik of the generators of the inter-area mode which has the k =1 least damping ratio is presented in figure 2m is the number of state variables, p ki is the participationfactor of the k th state variable into mode i, Φ ki is the i thelement of the k th right eigen vector of A, Ψ ik is the m thelement of the i th left-eigenvector of A.4. The System under Study and the Result of the Modal AnalysisThe system under study consists of the popular two areanetwork in reference . The data used is the same with Figure 2: Speed participation bar plot of the inter-area modethe one given in . The system without wind powerintegration serves as the base case (BC). For the number A small disturbance of 1ms fault duration was created atof states to be limited in order to reduce the complexity time, t=0, at the middle of line 7-8A, the speed deviationof the problem and to improve usability of result, the of the generators is presented in figure (3). The plotsynchronous generators are not equipped with exciters. shows that the generators in area 1 oscillate against theThis is considered appropriate according to reference  generators in area 2 thereby confirming the result of theto gather a qualitative insight. It is however necessary to modal analysis.model the exciters when considering a detailedquantitative modelling of real power system .
1.007 SG1 Area One Local Mode Area Two Local Mode Inter-area Mode SG2 1.006 SG3 0.14 SG4 1.005 0.12 S e (p ) pe d u 1.004 0.1 Damping Ratio 1.003 0.08 1.002 0.06 1.001 0.04 1 0 1 2 3 4 5 Time (s) 0.02 Figure 3: Speed deviation of the 4 machines in the two area 100 150 200 250 300 350 400 450 500 network during 1ms fault duration at line (7-8A) Wind Power (MW) Figure 4: Influence of wind power penetration using SCIG onA. Wind Farm Integration the damping of electromechanical mode 0.14The wind farm (WF) is assumed to consist of 2MW, 0.120.69kV generators modelled as an aggregate wind farmin line with  to the required MW of power. Aggregate 0.1 Damping ratiomodel reduces simulation time required by detailed multi 0.08turbine system. It is also assumed that the wind farms are 0.06located far from the grid where the wind resources are 0.04located as the case for most real wind farms; therefore the 0.02WF is connected to the system at B 6 through a 25km 0 100 150 200 250 300 350 400 450 500line and two transformers (0.69/20kV, 20/230kV). The Wind power (MW) Area two local mode Area one local mode Inter-area modeWF consisting of squirrel cage induction generators(SCIG) is adequately compensated with shunt capacitor Figure 5: Influence of wind power penetration using DFIG onat 0.69kV and 20kV buses to ensure a satisfactory the damping of electromechanical modevoltage level at all the buses. The wind farm made up ofdoubly-fed induction generators (DFIG) is connected to From the result shown in figure 5, area one local mode isthe grid through a three phase transformer and power is positively influenced by the wind farm. The inter-areainjected into the grid at unity power factor. This is mode is slightly positively influenced while the influenceachieved through the pulse-with modulation frequency on area two local mode is not clear as it shows a constantconverter between the rotor of the generator and the grid. damping ratio for all wind power integration.The detailed model of the converter and its associatedcontrollers can be found in . 2) Influence of Tie-Line Power Flow on the Small Signal Stability of a Power System.The scenarios considered are discussed in the varioussubheadings The SG2 was now taken out completely and then substituted with the same amount of wind power so that 1) Influence of Wind Power Penetration on the about 50% of the power produced in area one is from the Oscillatory Modes of a Power System wind farm. The system became unstable when exporting power to area two. However, when the direction of powerA wind farm made up of SCIG was connected to bus 6. flow in the tie-line was changed (i.e interchanging theThe power from the wind farm was gradually increased load) so that power is imported from area two, theby 100MW while reducing the power output of SG2 by unstable modes at the right were pushed to the left side ofthe same amount so that the total power in area one the complex plane thus making the system stable. This isremains the same. The results show that the damping of depicted in figures 6 and 7 (only the critical modes arethe low frequency inter-area mode decreases while the shown on the complex plane for clarity.local modes are positively influenced. This is presentedin figure 4. The technology of the generators making up Similar results were obtained when SG4 in area two wasthe wind farm was changed to DFIG and the influence of totally replaced by wind farm. The results also show awind power penetration was again studied. The result is stable system for power import from area one to area twopresented in figure 5. and unstable operation for power export from area 2 to area 1.
Figure 6: SG2 totally replaced by wind farm with power exported from area 1 to area 2 Figure 9: IEEE 9 bus benchmark 16 eigenvalues were computed first for SCIG technology. All the eigenvalues have negative real part for the three scenarios considered indicating that the system is small signal stable in all the cases. Electromechanical modes are picked out and presented in table 2 according to the rule relating to frequency of oscillation i.e 0.2< ƒ <2.5Hz . Table 2: electromechanical eigenvalues, frequency and damping ratio of IEEE 9 bus using SCIG technology Figure 7: SG2 totally replaced by wind farm with power SCIG λ = σ ± jω ƒ ,Hz ξ imported from area 2 to area 1 3SG -0.785 ± 10.990 1.749 0.071 WF at B2 -1.239 ± 15.073 2.399 0.0623) Impact of Transmission Line Length on the Small WF at B3 -0.867 ± 8.292 1.320 0.104Signal Stability of a Power System From table 2 it can be inferred that wind farm (SCIG) canMost wind resources are found very far from the city have both positive and negative influence on the dampingwhere access to main grid is not available. The influence of the oscillatory mode depending on the location. Whenof the length of the transmission line to the main grid on WF is connected to bus 2, the damping ratio of thethe small signal stability is investigated. From the result oscillatory mode decreases. The oscillatory modes arein figure 8, it seems the length of the transmission line better damped when the wind farm is connected to bus 3,does not have influence on both the local and the inter- ( ξ >10%).area mode. 0.2 The wind farm technology was again changed to 1.6 Frequency of Oscillation (Hz) 0.15 1.4 DFIG. The elecromechanical modes are presented in table 3.Damping Ratio 0.1 1.2 1 Table3: Electromechanical modes (DFIG) 0.05 0.8 DFIG λ = σ ± jω ƒ ,Hz ξ 0 0.6 3SG -0.785 ± 10.990 1.749 0.071 -0.05 0.4 Inter area mode WF at B2 -1.654 ± 17.578 2.398 0.094 -0.1 40 60 80 100 120 140 0.2 40 60 80 100 120 140 WF at B3 -1.002 ± 13.675 2.177 0.074 Distance of wind farm from the main grid (Km) Distance of wind farm from the main grid (Km) Local mode area one Local mode area two Inter-area mode The wind farm (DFIG) has a better damping in all the locations considered. The participation factors of theFigure 8: Influence of transmission length on the damping ratio electromechanical modes are presented in table 4. and the frequency of electromechanical mode Table 4: Participation factor 4) Change in Wind Generator Location DFI variables Magni Angle λ = σ ± jω G tudeThe IEEE 9 bus test system was used to study the WF SG1 speed 0.953 -170.63influence of change in wind generator locations on the at -1.654 ± 17.578 SG3 speed 0.051 -165.72 B2small signal stability of a power system. Synchronous WF SG2 speed 0.921 -173.6generators at bus 2 and bus 3 were taken out in turn and at -1.002 ± 13.675 SG2 phi 1 0substituted with the same amount of wind farm. B3
The modal phasor diagram for DFIG wind farm at bus 3 (IRDP fund) and ESKOM (TESP) for the support of thisis presented in figure 10 research. References  A. Mendonca and J. A. P. Lopes, "Impact of large scale wind power integration on small signal stability," in Future Power Systems, 2005 International Conference on, 2005, pp. 5 pp.-5.  P. Kundur, "Power System Stability and Control,The EPRI Power System Engineering Series." vol. 1: McGraw-Hill,Inc, New York, 1994.  D. Thakur and N. Mithulananthan, "Influence of Constant Speed Wind Turbine Generator on Power System Oscillation," Electric Power Components and Systems, vol. 37, pp. 478 - 494, 2009. Figure 10: Phasor modal plot of the electromechanical mode  R. C. Burchett and G. T. Heydt, "Probabilistic when wind farm is connected to bus 3 Methods For Power System Dynamic Stability Studies," Power Apparatus and Systems, IEEEFrom the participation factor computed, it can be seen Transactions on, vol. PAS-97, pp. 695-702, 1978.that wind farm itself does not contribute to the  Z. Xu, Z. Y. Dong, and P. Zhang, "Probabilistic small signal analysis using Monte Carlo simulation," inelectromechanical modes. Power Engineering Society General Meeting, 2005. IEEE, 2005, pp. 1658-1664 Vol. 2.  L. Li, S. Liangliang, L. Wenxia, and J. Sheng, "Modal5. Conclusion analysis concerning the control mode of doubly-fed induction generator," in Sustainable PowerThe influence of wind power on the small signal stability Generation and Supply, 2009. SUPERGEN 09.of a power system has been studied on four machine two International Conference on, 2009, pp. 1-6.  P. C. Krause, O.Wasynczuk, and S. D. Sudhoff,area network and the IEEE 9 bus system as a benchmark. "Analysis of Electric Machinery and Drive SystemsFrom the studies, wind power has positive influence on Second Edition," A John Wiley andthe damping of local area modes. DFIG has a better Sons,Inc.Publicaton, 2002.  F. Mei and B. C. Pal, "Modelling and small-signaldamping influence on the inter-area mode than the SCIG. analysis of a grid connected doubly-fed inductionThe direction of wind power dispatch on the stressed tie- generator," in Power Engineering Society Generalline has influence on the small signal stability of a power Meeting, 2005. IEEE, 2005, pp. 2101-2108 Vol. 3.system. In this study, the system is small signal unstable  J. G. Slootweg and W. L. Kling, "The Impact of Wind Power Generation on Power System Oscillations,"when about 50% of power produced in an area is Electric Power System Research, vol. 67, pp. 9-20,exported over the weak tie-line. However, the system is 2003.stable for power import.  J. Conroy and R. Watson, "Aggregate Modelling of Wind Farms Containing Full-Converter Wind Turbine Generators with Permanent MagnetThe length of the transmission grid from the wind Synchronous Machines: Transient Stability Studies,"location to the main grid seems not to have an influence IET Renewable Power Generation, vol. 3, pp. 39-52,on both the damping and frequency of oscillation of both 2009.  M. A. Poller, "Doubly-fed induction machine modelsthe local area and the inter-area modes. The location of for stability assessment of wind farms," in Powerwind farm made of SCIG technology can have both Tech Conference Proceedings, 2003 IEEE Bologna,positive and negative influence on the damping of the 2003, p. 6 pp. Vol.3.oscillatory mode depending on where the wind farm is  W. Chen, S. Libao, W. Liming, and N. Yixin, "Small signal stability analysis considering grid-connectedlocated. However, with doubly-fed induction generator wind farms of DFIG type," in Power and Energytechnology, the wind farm is seen to have better damping Society General Meeting - Conversion and Deliveryin all the locations. of Electrical Energy in the 21st Century, 2008 IEEE, 2008, pp. 1-6.In this paper, a qualitative study mainly is carried out onsmall test systems. Further investigation is necessary forlarge power systems.AcknowledgementThe authors want to acknowledge the TshwaneUniversity of Technology, National Research Foundation