Impact of DFIG-based Wind Generator on Dynamic Behavior of Power Systems during Over-Frequency Events

Transcription

1 Impact of DFIG-based Wind Generator on Dynamic Behavior of Systems during Over-Frequency Events Yonggang Zhang, Christian Klabunde, Martin Wolter Institute of Electric Systems Otto-von-Guericke-Universität Magdeburg Magdeburg, Germany Abstract For years, wind power penetration has increased in power systems and thus leads to less and less system inertia. In this paper, the impact of DFIG-based wind generator on dynamic behavior of power systems during over-frequency events is investigated. Mathematical models for traditional synchronous generators and DFIG-based wind turbine generators are presented to show how dynamic responses of power systems are influenced by increased wind penetration. A case study simulating % load decrease in different wind penetration s is carried out, where different wind penetration is realized by gradually replacing power generation from synchronous generators with same amount of DFIG-based wind power. Simulation results show that the power system tends to have higher rate of change of frequency (ROCOF) following a grid disturbance, as wind penetration increases. Keywords-wind penetration; DFIG-based wind generator; ROCOF; dynamic response I. NOMENCLATURE i, u complex current and voltage r, l resistance, inductance ψ complex flux-linkages f, ω, s frequency, angular speed, slip P, Q, U rms active power, reactive power and voltage M torque J inertia constant subscripts S, R stator, rotor a, b, c phase a, phase b, phase c f excitation d, q direct, quadrature axis component h, σ main field, leakage m, e mechanical, electrical II. INTRODUCTION During the last decades, both land-based and offshore wind power plants have been constantly installed and integrated into power grids. According to the Global Wind Energy Council (GWEC), by the end of,. GW of wind energy had been installed worldwide, and this number will be expected to reach 79. GW in []. In Germany, 7,7 wind turbines had been installed by the end of 6, in which.6 MW of wind power capacity was newly installed in 6. Among various types of wind turbines, the DFIG-based wind generator is the most popular one and feeds the majority of wind generation into power girds. For simplicity, in this paper, only the penetration of DFIG based wind generators will be discussed. As DFIG dominated wind penetration grows year by year, more conventional synchronous generators providing support of automatic frequency regulation may be dropped out of service. Since DFIG-based wind generators utilize power electronic converters to interact with power grids, its rotor speed could be decoupled from the grid frequency. In addition, DFIG-based wind generators generally operate according to the maximum power point tracking (MPPT) curve. Thus, no reserve power will be available. Therefore, the DFIG-based wind generators inherently provide little or no contribution to frequency support in power systems. In this situation, the increased wind penetration level may bring a system much closer to its operating limit and might lead to stability issues. Therefore, in the context of high DFIG dominated wind penetration, the analysis of power system dynamics has become a very important research issue, especially when serious gird disturbances occur. Various papers have been published on the impact of DFIGbased wind generators on the dynamic behavior of power systems, in which varying degrees of achievement have been presented [-8]. However, most of them have been addressing dynamic grid issues under under-frequency disturbances. In this paper, the impact of DFIG wind penetration level on dynamic responses of power systems under an over-frequency event has been studied. The penetration level is calculated as a function of the total wind power generation over the total load demand. The remainder of this paper is structured as follows: Section III describes the models of synchronous generator, DFIG-based

2 wind generator, transmission network and static loads. Section IV describes simulations that compare three different s in varying DFIG wind power penetration levels and discusses the dynamic response characteristics of synchronous generators and DFIG-based wind generators. Section V summarizes the study. III. POWER SYSTEM MODELLING The power system model discussed here consists of synchronous generators, DFIG-based wind generators, transmission network and static load models. The nomenclature of all symbols used in the following equations is given at the beginning of the paper. A. Synchronous Generator Simple governor- and turbine model for thermal driven generators used for the PST6 6-Generator Test System is adopted in this study [9]. Electrical power of the generator can be expressed as: P G = Re{ U G I G } = Re{U I G } R a I G () Where U = U G (R a + j X d )I G U is the transient generator voltage, which is considered a constant value in the time scale of this study, R a is the resistance of stator coil, the equivalent circuit is shown in Fig.. U' R a +jx' d U G Fig. Diagram of generator equivalent circuit Equation of motion is given by: ] = [ ω R ] [ δ ] + [ ω R k m (P m + P e ) ] () [ δ Where p k m = = ω = const. J ω p T m S rg P e = Re{U I G } p is the number of pole pairs, P m is the input mechanical power and P e is the output electric power, S rg is the rated apparent power, ω Ṙ is the derivative of rotor speed to time dω R, dt δ is the transient power angle. B. DFIG Based Wind Generator The enhanced reduced order model of DFIG-based wind generator (ROM/E) is adopted in this paper. Using ROM/E for DFIG-based wind generators, the grid is modeled by algebraic equations, so that the simulation performance is much better than the alternative instantaneous value calculation based on full order models for both DFIG-based wind generator and grid as well [-]. I G Equations () - () represent the complete set of mathematical relationships that describe the dynamic behavior of DFIG-based wind generator. Voltage equations: [ u S u ] = [ r S ] [ i S ] + [ j ω S ψ R r R i R j (ω S ω R ) ] [ψ S S ] + [ ] () ψ R ψ R Flux equations: l h [ ψ S ] = [ l S ] [ i S ] () ψ R l h l R i R Where l S = l h + l σs and l R = l h + l σr Equation of motion: J dω R dt = M m M e () Where M e = ψ Sd i Sq + ψ Sq i Sd Voltage equations, flux equations and equation of motion constitute the th order model of DFIG-based wind generator that could be used for dynamic time domain simulations. The stator terminal voltage u S forms the link to the rest of the network. C. Transmission Network The transmission network model is described by the steadystate matrix equation: i K = Y KK u K (6) Where i K is the injection current vector to the network, u K is the nodal voltages vector and Y KK is the nodal admittance matrix. D. Static Load The static load model is represented by: Y L = U S L (7) IV. CASE STUDY A. Description of Test System The test system studied in this paper includes a center grid and two sub-grids. As presented in Fig. a), there are three conventional generators, three wind generators, three demand loads L, L and L, each with capacity of ( + j6) MVA, as well as load demand L with capacity of ( + j.6) MVA in the center grid. Two sub-grids have same grid structure, each of which contains 7 MW wind generation and ( + j) MVA load, as shown in Fig. b). The left sub-grid is connected to the center grid through node K and K, while the right subgrid is connected to the center grid through node K and K. System electricity demand is supplied by Gen, Gen, Gen, WT, WT, WT as well as wind generators in the left and right side sub-grids. All generators are equipped with voltage controllers. Three conventional generators deploy standard speed control and inherently participate to primary frequency control. Gen and Gen start to reduce active power when grid frequency is

3 . Hz or larger, whist Gen reduces active power immediately with over-frequency. Although it is technically possible for DFIG based wind generator to contribute to primary frequency control, it is for the best to prevent wind curtailment from the economic and environmental perspectives. Thus, the participation of wind generators in primary frequency control is not considered in this study. However, DFIG-based wind generators reduce active power when frequency is diverging positively from nominal frequency by. Hz. T kv/8kv T7 kv/kv L T kv/8kv Gen T kv/8kv WT T.8kV/kV Gen K T 8kV/kV T8 kv/kv K T 8kV/kV T9 kv/kv T 8kV/kV WT T6 kv/.8kv WT L K L T kv/.8kv Frequency dynamics of the test grid in %, % and 6% wind penetration s are presented in Fig.. It can be seen that the increased wind penetration level does not have much influence on post-disturbance peak frequency. But at the very beginning of the disturbance event, the rate of change of frequency (ROCOF) increases with wind penetration level. a) center grid K / K 8/ kv/kv kv/.69kv AL / AR.69kV/kV kv/kv (+j6)mva (+j6)mva kv/kv kv/.69kv,, (7+j.7)MVA, (+j6)mva 8/ BL / BR.69kV/kV kv/kv (+j6)mva (+j6)mva (+j6)mva.69kv/kv kv/kv (7+j.7)MVA (7+j.7)MVA (+j6)mva (7+j.7)MVA kv/kv kv/.69kv CL / CR (7+j.7)MVA.69kV/kV kv/kv K 9,9 B. Case Scenarios and Simulation Results Starting from an initial steady state of the base case without any wind generation connected, the test grid is subsequently put under stress with increased penetration levels of wind generation from % to %. In this subsection, simulation results of %, %, 6% wind penetration s will be presented and analyzed. Parameters for the feed-in power and rating capacity of each generator or generator group in the s are presented in Table I. FEED-IN POWER AND RATING CAPACITY OF EACH GENERATOR OR GENERATOR GROUP Scenario - % Wind Gen Gen Gen WT WT WT WEAa (MVA) Variables Scenario - % Wind (MVA) Scenario - 6% Wind (MVA) a. All generators in left and right side sub-grids Fig. Frequency curves of test grid in %, % and 6% wind penetration s In the of % wind penetration, the system frequency has a steep rise to the peak value of. Hz following the load shedding disturbance, after seconds of frequency adjustment, it tends to remain at a constant value of.9 Hz. The corresponding dynamic responses of all DFIGbased wind turbines and synchronous generators in this are shown in Fig.. It is shown that synchronous generators act to the load-shedding event immediately by reducing generation output, while wind generators only reduce active power output when frequency is greater than. Hz. According to the specifications on the frequency operating range in 6 E. ON grid code [], the post-disturbance peak frequency and steadystate frequency are still in safe operating range.,8 Gen,8 Gen,79 Since it is not practical to model every wind generator, aggregation techniques are applied for modelling a complete wind plant by an equivalent wind generator. 6% wind share, b) left / right side sub-grid Fig. Schematic diagram of the test system Scenarios % wind share,, (+j7.88)mva TABLE I. % wind share, f / Hz As shown in Fig. a), a load-shedding is simulated by disconnecting L branch from grid, which is applied in the simulation at time s. Since transient responses of power system elements to grid disturbances usually just last for a few seconds, here we set the simulation duration to seconds. As wind speed and load variations are relatively small with respect to the time interval of seconds, it is feasible to assume that wind speed and all loads stay unchanged. L Gen (7+j.7)MVA As power consumption and generation must be balanced at all times, the increase in wind penetration directly lead to less generation output of conventional generators. factors for conventional generators and wind generators are set as.8 and.9 respectively. Besides, to fit the changed power flow in increased wind penetration, transformer ratings are adjusted accordingly and parameters of transmission lines are varied. Gen,78,77,76,7

4 ,96 WT,9 WT,9 WT,9,9,9 Fig. Dynamic response curves of all generators in % wind penetration In the of % wind penetration, the system frequency has a steep rise to the point of. Hz following the % load shedding disturbance, after seconds of oscillation, it tends to remain at a constant value of. Hz. In contrast to the % wind penetration, the frequency ramp rate is larger, and the value of post-disturbance peak frequency and steady-state frequency in this are bigger. Corresponding dynamic responses of all generators in this % wind penetration are shown in Fig.. has almost no influence on transient responses of power grids during faults. However, through inertial response, synchronous generators can provide good grid support in transient status, and this so called inertial response comes originally from the generator s large overload capability. For DFIG-based wind generators, its dynamic response is mainly dependent on the coordinated control strategies of the converters used in its excitation system. In steady state, the generator primarily controls its rotor speed according to MPPT table to optimize power output from wind turbine, thus could not contribute inertial energy during under-frequency events [-6]. Although it is technical possible for DFIG based wind generator to provide primary reserve control during over-frequency events through adjusting pitch angle or generator side converter voltage [7], it is economically and environmentally not the best option to provide virtual system inertia by curtailing wind power.,8,97 Gen,8 Gen,78 Gen,76,7,8,8,8,7 Gen,79 Gen,7,78 Gen,77 WT,98,76 WT,7,7,96 WT,9,97,9 WT,96,9 WT,9 WT,9,9,9,9 Fig. 6 Dynamic response curves of all generators in 6% wind penetration Fig. Dynamic response curves of all generators in % wind penetration In the of 6% wind penetration, the postdisturbance peak frequency is about. Hz, which already poses threat to power system operation. In contrast to the % and % wind penetration s, the frequency ramp is faster, and the value of the post-disturbance peak frequency and steady-state frequency in this are bigger. Corresponding dynamic responses of all generators in this 6% wind penetration are shown in Fig. 6. C. Dynamic Reponse Analysis In most grid codes, DFIG-based wind generators are required to have the same or similar behavior as synchronous generators during and immediately after grid faults. But they have complete different technology in operation and control. For synchronous generators, the excitation control is quite slow and A machine contributes inertial energy if it accelerates in response to a frequency rise. The average acceleration of a system or the rate of change of frequency (ROCOF) at the moment of power imbalance depends on the total inertia of the entire interconnected system. Since DFIG based wind generators don t contribute inertial energy at the moment of disturbance event, power system in the 6% wind penetration has less system inertia than in the % and % wind penetration s. This means that the test grid in 6% has higher ROCOF and becomes more sensitive to grid disturbances. V. CONCLUSIONS In this paper, the impact of penetration level of DFIG-based wind generators on dynamic behavior of power systems during over-frequency events is investigated. Mathematical models for traditional synchronous generators and DFIG-based wind turbine generators are presented to theoretically show their dynamic responses to system disturbance. In the case study, test grid is subsequently put under stress with increased wind penetration from % to %, simulation results of %, %, 6% wind penetration s show that under an around %

5 load shedding event, the increased wind penetration by replacing synchronous generator with DFIG-based wind generator tends to decrease system inertia and thus increase the ROCOF for the specific test grid. However, this conclusion should be treated carefully, as the results are still very limited. The impact of wind power on the dynamic behavior of power systems needs to be elaborately studied further. REFERENCES [] Global Wind Energy Council (GWEC), Global Wind Report, Apr. 6, [] O. Anaya-Lara, F.M. Hughes, N. Jenkins, G. Strbac,"Influence of Wind Farms on Systems Dynamics and Stability,"Wind Engineering, vol., no., pp. 7-7, 6. [] J. Rasmussen, P. Jorgensen, M.T. Palsson, K. Uhlen,"Wind Impact on Transient and Voltage Stability of the System in Eastern Denmark,"8 IAESTD Int. Conf and Energy Systems, Marina del Ray, USA, pp.-, Oct.. [] L. Meegahapola, D. Flynn, "Impact on transient and frequency stability for a power system at very high wind penetration", and Energy Society General Meeting IEEE, pp. -8,, ISSN [] W. Qing, X. Ancheng, B. Tianshu, Z. Yuanjie, "Impact of DFIG-based wind farm on transient stability of single machine infinite bus system", and Energy Engineering Conference (APPEEC) IEEE PES Asia-Pacific Kowloon, pp. -,. [6] Tiezheng Huang, Jie Ma, Wei Wang, "Study on transient voltage stability of wind farm incorporated system with reactive power compensation device", System Technology (POWERCON) International Conference on, pp. 6-66,. [7] S. Libao, D. Shiqiang, N. Yixin, Y. Liangzhong, M. Bazargan, "Transient stability of power systems with high penetration of DFIG based wind farms", & Energy Society General Meeting 9. PES 9. IEEE Calgary AB, pp. -6, 9. [8] S.W. Liu, G.Y. Li, M. Zhou, " system transient stability analysis with integration of DFIGs based on center of inertia", CSEE Journal of and Energy Systems, vol., no., pp. -9, Jun. 6, ISSN 96-. [9] P. Kundur, system stability and control, New York: McGraw-Hill, 99. [] J. Kretschmann, H. Wrede, S. Engelhardt, and I Erlich, "Enhanced reduced order model of wind turbines with DFiG for power system stability studies", International and Energy Conference, Nov. 8-9, 6, Malaysia. [] L.B. Shi, Z. Xu, J. Hao and Y.X. Ni, "Modelling analysis of transient stability simulation with high penetration of grid-connected wind farms of DFIG type", Wind Energy, vol., pp. -, Mar. 7. [] I. Erlich, F. Shewarega, "Modeling of Wind Turbines Equipped with Doubly-Fed Induction Machines for System Stability Studies", PSCE, October 9 November, 6. [] I. Erlich, J. Kretschmann, J. Fortmann, S. Engelhardt, H. Wrede, "Modeling of Wind Turbines Based on Doubly-Fed Induction Generators for System Stability Studies", Systems IEEE Transactions on, vol., pp , 7, ISSN [] E.ON Netz GmbH, Grid Code for high and extra hight voltage, pp.-, st April 6. [] A. Mohammed, B. Tarek, " System Transient Stability Analysis with High Wind ", International Electrical Engineering Journal (IEEJ), vol., no., pp. 97-9,, ISSN [6] C. Feltes, S. Engelhardt, J. Kretschmann, J. Fortmann and I. Erlich, "Comparison of the Grid Support Capability of DFIG-based Wind Farms and Conventional Plants with Synchronous Generators," in IEEE PES '9 GM, 9, pp. -7. [7] Jörn Runge, Modellierung von Windenergieanlagen für die Netzberechnung, Berichte aus der Elektrotechnik: Shaker Verlag, pp. -8, 8.