Power-Flow Analysis of Large Wind Power Plant Collector Systems With Remote Voltage Control Capability

Transcription

1 1 Power-Flow Analysis of Large Wind Power Plant Collector Systems With Remote Voltage Control Capability Mohamed Zakaria Kamh and Reza Iravani Abstract This paper presents and develops a new powerflow analysis approach of large wind farm collector systems. The wind turbine units, within the wind power plant, are controlled to remotely regulate the collector bus (or the point of interconnection (POI)) voltage. The developed power-flow algorithm determines the reactive power set points of the individual wind turbines to control the voltage at a remote bus. To incorporate the wind farm collector system in the bulk transmission power-flow analysis, an equivalent model of large wind power plants is developed. New bus types are defined and the necessary modifications to the conventional Newton-Raphson (N-R) power-flow formulation is presented. The proposed power-flow algorithm is implemented and tested on an existing wind farm. Index Terms AC Collector System, Doubly Fed Asynchronous Generator, Power-Flow Analysis, Remote Voltage Control, Wind Farm I. INTRODUCTION Over the past two decades, wind energy has gained the most attention as a clean, environmentally-friendly, and free source of electricity generation [1] [3]. Since 2002, the global annual installed wind capacity has been doubling every third year [4]. During 2009, the global installed wind capacity has increased by 37.5 GW, 950 MW of which is installed in Canada [4], [5]. Figure 1 depicts the growth in the Canadian cumulative wind energy installed capacity during the first decade of the third millennium [6]. Wind resources are usually located away from the load centers. As such, large wind farms are connected to the transmission networks [7], [8]. This results in security and stability problems that may propagate into the medium and low voltage networks. Transmission grid voltage instability comes on the top list of the wind energy interconnection challenges [9]. The severity of the problem depends on the type of wind turbine generating units (WTGU) deployed within the wind farm. Wind turbine units are classified into four types [10]: Type-1: Fixed speed unit with squirrel cage induction generator. Type-2: Semi-variable speed unit with wound rotor induction generator whose rotor circuit is closed through an electronically-controlled variable resistor. Type-3: Variable speed unit with doubly fed induction generator whose rotor is connected to the grid via a back-to-back voltage sourced converter (B2B VSC). The authors are with the Edward S. Rogers Sr. Department of Electrical and Computer Engineering, University of Toronto, Toronto, ON M5S 3G4 Canada ( mohamed.kamh@ieee.org; iravani@ecf.utoronto.ca). Fig. 1. Growth in the Canadian wind power installed capacity Type-4: Variable speed unit with permanent magnet synchronous generator (or squirrel cage induction generator) whose stator is connected to the grid via a full rated B2B VSC. Unless equipped with suitably rated reactive power compensators, wind farms comprising Type-1 and/or Type-2 wind turbines may lead to voltage instability since the associated induction generators require an external source of reactive power for excitation [10], [11]. On the other hand, Type-3 and Type-4 wind turbine units are less likely to cause voltage stability problems since they can provide ancillary services including voltage, reactive power, and power factor control. In addition to controlling the active and reactive power exchange with the grid, these services are used to remotely regulate the voltage at the collector bus via a supervisory control system imbedded within the wind farm energy management center [12]. The supervisory control system, also known as the WindVAR control system, (i) measures the collector bus voltage, (ii) evaluates the reactive power set points of individual wind turbine units, (iii) and controls the reactive power injection of the WTGU to maintain the collector bus voltage at a pre-specified value. Developing power flow models of large wind power plants with AC collector systems is addressed in the technical literature [2], [11], [13] [15]. However, these models assume the WTGU can only operate in either (i) constant power/power factor mode (PQ) or (ii) local voltage control mode (PV). This kind of modeling lacks the flexibility to incorporate the supervisory control action of remotely regulating the

2 2 Fig. 2. Erie Shores Wind Farm wind farm collector bus. Local bus voltage control (PV bus representation) is well defined and modeled in the Newton- Raphson power-flow analysis [16]. Power-flow models of remote voltage control are developed in rectangular coordinates and incorporated in the Newton-Raphson powerflow algorithm [17]. However, phasor coordinates are more prevalent for Newton-Raphson power-flow equations. The scope of this paper is to develop and implement a fast and efficient phasor coordinates-based Newton-Raphson power-flow algorithm to calculate the reactive power set points of individual variable-speed WTGU to remotely regulate the wind farm collector bus voltage. First, a simple wind farm equivalent model, adequate for bulk transmission networks power-flow analysis, is developed. Novel bus types are defined. Finally, the conventional N-R formulation is modified to incorporate the WindVAR control system into the power-flow equations. The proposed power-flow algorithm is implemented in the MATLAB R platform and tested on an existing wind farm, Erie Shores Wind Farm [18], [19]. The results of several case studies are reported and analyzed. II. TEST SYSTEM The wind farm test system used in this work is the Erie Shores Wind farm, located 25 km south of Tollsenburg, Ontario. It contributes to about 3% of the Canadian wind power installed capacity [19]. The wind farm name-plate capacity is 99 MW uniformly distributed over the the north shore of Lake Erie between Copenhagen and Clear Creek, Ontario. The farm comprises 66 GE 1.5 MW Type-3 variable speed wind turbines, equipped with the WindVAR control system to regulate the collector bus voltage [12]. The wind farm collector system is rated at 34.5 kv and comprises four radial string feeders (W1, W2, E1, and E2) connected to a 34.5 kv common collector bus. W1 and W2 are underground cables with 12 WTGU each. E1 and E2 are overhead lines with 22 and 20 WTGU, respectively [18]. The WTGU located on each feeder are lumped into 4 or 5 collecting points that are uniformly distributed every 3 km of the feeder path. The wind farm voltage is stepped up to 115 kv via a transformer located at the Port Burwell substation. The wind farm is then tapped into Hydro One s grid via a 29.8 km, 115 kv overhead line connected to Tollsenburg Junction, which in turn is 2 km away from Cranberry Junction [20]. A detailed schematic diagram of Erie Shores Wind Farm and the neighboring power system is depicted in Fig. 2. III. LARGE WIND FARMS EQUIVALENT MODEL To incorporate large wind farms in the transmission networks power-flow analysis, a simple wind farm model consisting of (i) an equivalent WTGU with its step up transformer, and (ii) an equivalent collector feeder can be deployed [12]. This section develops an equivalent wind farm model for large wind farms with radial string feeders, the most deployed topology for AC wind farm collector systems [21] [24]. However, this model can be extended to address other collector systems topologies such as single- /double-sided ring and star configurations [25]. The following assumptions are made: The collecting points are uniformly distributed along each feeder. The power injected into any collecting point along the same string feeder is almost the same. Consider the wind farm of Fig. 3(a), with n radial string feeders. To evaluate the wind farm equivalent model, the following steps are executed: 1) The WTGU of each string feeder are lumped into one equivalent unit, whose rating is equal to the sum of all the units along the feeder. Based on the aforementioned assumptions, the equivalent unit is located at the middle of the string feeder [26]. After executing this step, the resulting wind farm configuration is depicted in Fig. 3(b). 2) Assuming all the equivalent string feeders of Fig. 3(b) are connected in parallel. Thus, all the WTGU of the farm are lumped into one unit connected to an equivalent collector feeder whose admittance is twice the sum of the individual feeders admittances, Fig. 3(c). IV. PROPOSED POWER-FLOW MODEL OF REMOTE VOLTAGE REGULATION Local voltage control modeling (PV bus representation) is well established in the Newton-Raphson power-flow analysis

3 3 Fig. 4. Five bus system with P and PQV busses P / Q are the active/reactive power mismatch vectors, and are given by (a) P = [ P 2, P 3, P 4, P 5 ], Q = [ Q 2, Q 3, Q 4 ], (2) where P i and Q i are the system state variables, and are evaluated similar to the conventional N-R power-flow formulation [16]. In (1), θ and V are the incremental variations in the system control variables, and are given by θ = [ θ 2, θ 3, θ 4, θ 5 ], V = [ V 3, V 4, V 5 ], (3) (b) (c) Fig. 3. Developing an equivalent wind farm model, (a) original wind farm configuration, (b) the equivalent configuration after executing the first step, (c) the final wind farm equivalent model. [16]. Remote voltage control is defined as regulating the voltage at Bus-k via adjusting the reactive power injected at Bus-m [17]. As such, the voltage (V k ) and active/reactive (P k /Q k ) power at Bus-k are specified quantities, while the reactive power injected at Bus-m (Q m ) is to be evaluated. Consequently, Bus-k is modeled as a PQV bus while Bus-m is represented as a P bus. A. Formulating the N-R Power-Flow Equations Consider the five-bus system of Fig. 4, in which the equivalent wind farm model of Section III is deployed. The individual WTGU are controlled to maintain Bus-2 voltage at a specified value. The following matrix equation represents the system power-flow equations: [ P Q ] = [ J11 J 12 J 21 J 22 ] [ θ V ]. (1) J 11, J 12, J 21, and J 22 are given by P 2 / θ 2 P 2 / θ 3 P 2 / θ 4 P 2 / θ 5 J 11 = P 3 / θ 2 P 3 / θ 3 P 3 / θ 4 P 3 / θ 5 P 4 / θ 2 P 4 / θ 3 P 4 / θ 4 P 4 / θ 5, P 5 / θ 2 P 5 / θ 3 P 5 / θ 4 P 5 / θ 5 P 2 / V 3 P 2 / V 4 P 2 / V 5 J 12 = P 3 / V 3 P 3 / V 4 P 3 / V 5 P 4 / V 3 P 4 / V 4 P 4 / V 5, P 5 / V 3 P 5 / V 4 P 5 / V 5 Q 2 / θ 2 Q 2 / θ 3 Q 2 / θ 4 Q 2 / θ 5 J 21 = Q 3 / θ 2 Q 3 / θ 3 Q 3 / θ 4 Q 3 / θ 5, Q 4 / θ 2 Q 4 / θ 3 Q 4 / θ 4 Q 4 / θ 5 Q 2 / V 3 Q 2 / V 4 Q 2 / V 5 J 22 = Q 3 / V 3 Q 3 / V 4 Q 3 / V 5, (4) Q 4 / V 3 Q 4 / V 4 Q 4 / V 5 where all the partial derivatives retain the same formulae of the conventional N-R power-flow [16]. Thus, (1)-(4) conclude the mandatory modifications to incorporate the remote voltage control into N-R power-flow equations, which are as follows: In the state variables vector, eliminate the reactive power mismatch associated with the P bus, i.e. Q 5 in the above example. In the control variables vector, eliminate the incremental variation in the voltage of the PQV bus, i.e., V 2 in the above example. In J 21 and J 22, eliminate the row corresponding to the P bus. Also, in J 12 and J 22, eliminate the column corresponding to the PQV bus.

4 4 B. Forcing the Reactive Power Limit of the P Bus Depending on the manufacturer s specifications, each variable speed wind turbine unit can supply/absorb a limited amount of reactive power. For example, the GE 1.5 MW wind turbine can operate in the range of 0.9 lagging/0.95 leading power factor. Thus, each GE 1.5 MW unit can inject up to 0.72 MVAr and absorb up to 0.49 MVAr [12]. These reactive power limits should not be violated at anytime. To incorporate the wind farm reactive power limits into the N-R power-flow algorithm, the reactive power injected by the P bus (Bus-m in this case) is evaluated, subsequent to each N-R power-flow iteration, using Q m = V m n i=1 V i Y ik sin (θ k θ i θ Yik ). (5) If the reactive power limits of the P bus are violated, then execute the following subroutine: 1) Set the reactive power of the violating P bus to the corresponding limit. 2) Modify the system Jacobean matrix and the state (control) variables vectors to incorporate the rows (columns) associated with the reactive power mismatch (incremental voltage variations) of the P (PQV) bus. 3) Change the P bus into PQ bus. 4) Change the PQV bus into PQ bus. 5) Proceed to the next iteration. C. The Proposed Power-Flow Algorithm Sections III, IV-A, and IV-B represent the main building blocks of the proposed power-flow algorithm. The algorithm is threefold. First, the wind farm equivalent model is developed and plugged in the system admittance matrix. Second, the Newton-Raphson power-flow equations are developed and solved to include the new control and state variables associated with the wind farm supervisory controller. Finally, the reactive power set points of the individual WTGU are evaluated. The complete algorithm is shown in Fig. 5. V. IMPLEMENTATION AND CASE STUDIES The algorithm of Fig. 5 is implemented in the MATLAB R platform. The developed program is applied to the Erie Shores Wind Farm. The equivalent power system is depicted in Fig. 6. The developed program is used to evaluate the reactive power set point of the individual wind turbine units to maintain the voltage at the HV collector hub (Bus-5) within a range of pu, or until the reactive power limit of the individual WTGU is reached. The results are depicted in Fig. 7. For each case, the corresponding total system losses are evaluated, Fig. 8. As concluded from Fig. 7, the reactive power injected by the wind farm can provide voltage support at Bus-5 up to pu. However, if the voltage at this bus is to be regulated at 1.05 pu, then additional capacitor banks should be connected to the MV collector hub. These banks should be rated not less than 15.5 MVAr. This value is calculated by (i) relaxing the reactive power constraint in the algorithm of Fig. 5, (ii) evaluating the reactive power injected by the wind farm to maintain the voltage of the HV collector hub at Fig. 5. Proposed Power-Flow Algorithm

5 5 Fig. 6. Single line diagram of the equivalent Erie Shores Wind Farm and the neighboring power system Fig. 7. Reactive power set point of the individual WTGU Fig. 8. Total system losses 1.05 pu, and finally (iii) subtracting the calculated reactive power from the wind farm reactive power limit. Another application of the developed software is to find the optimal reactive power set point for the individual WTGU. As indicated in Fig. 8, the total system losses are reduced as the HV collector hub voltage is improved. However, if the voltage at Bus-5 exceeds 1.01 pu, the total losses start to increase. Thus, combining Fig. Fig. 7 and 8 concludes the optimal reactive power set point for the individual WTGU, in terms of minimizing the system losses, is MVAR. VI. CONCLUSION This paper develops a power-flow model of large scale wind farms comprising variable speed wind turbine units with supervisory control capability. The proposed model is incorporated in a power-flow algorithm, especially designed to evaluate the reactive power set points of the individual wind turbines within the wind power plant. The power-flow algorithm is implemented in MATLAB R and applied to an existing Canadian wind power plant, Erie Shores Wind Farm. The developed program successfully (i) evaluates the WTGU reactive power set points, (ii) estimates the rating of an additional reactive power compensation device to extend the range of the wind farm supervisory control, and (iii) calculates the wind farm optimal reactive power set point to minimize the total system losses. The developed software is an essential building block in the wind farm energy management center. REFERENCES [1] U. Eminoglu, B. Dursun, and M. H. Hocaoglu, Incorporation of a New Wind Turbine Generating System Model into Distribution Systems Load Flow Analysis, Wind Energy Journal, vol. 12, no. 4, pp , [2] U. Eminoglu, Modeling and Application of Wind Turbine Generating System (WTGS) to Distribution Systems, Renewable Energy Journal, vol. 34, no. 11, pp , Nov [3] V. Akhmatov, Induction Generators for Wind Power. Multi-Science Publishing Company, Ltd., [4] World Wind Energy Report, worldwindenergyreport2009 s.pdf, World Wind Energy Association, March [5] Global Installed Wind Power Capacity Report, /Annex %20stats%20PR% pdf, Global Wind Energy Council, [6] Canada Wind Power Installed Capacity, ty e.pdf, Canadian Wind Energy Association, [7] Y. Chi, Y. Liu, W. Wang., and H. Dai, Voltage Stability Analysis of Wind Farm Integration into Transmission Network, in Proc. International Conference on Power System Technology, PowerCon 2006, 2006, pp [8] A. Beekmann, J. Marques, E. Quitmann, and S. Wachtel, Wind Energy Converters with FACTS Capabilities For Optimized Integration of Wind Power into Transmission and Distribution Systems, in CIGRE/IEEE PES Joint Symposium on Integration of Wide-Scale Renewable Resources Into the Power Delivery System,, , pp [9] M. Palsson, T. Toftevaag, K. Uhlen, and J. Tande, Large-Scale Wind Power Integration and Voltage Stability Limits in Regional Networks, in IEEE Power Engineering Society Summer Meeting, vol. 2, 2002, pp [10] IEEE PES Wind Plant Collector System Design Working Group, Characteristics of Wind Turbine Generators for Wind Power Plants, in IEEE Power Energy Society General Meeting, PES 09, July 2009, pp. 1 5.

6 6 [11], Reactive Power Compensation for Wind Power Plants, in IEEE Power and Energy Society General Meeting, PES 09, , pp [12] N.W. Miller, W.W. Price, and J.J. Sanchez-Gasca, Dynamic Modeling of GE 1.5 and 3.6 Wind Turbine-Generators, available at: [13] A. E. Feijoo and J. Cidras, Modeling of Wind Farms in the Load Flow Analysis, IEEE Trans. Power Systems, vol. 15, no. 1, pp , Feb [14] G. Coath and M. Al-Dabbagh, Effect of Steady-State Wind Turbine Generator Models on Power Flow Convergence and Voltage Stability Limit, in Proc. Australasian Universities Power Engineering Conference- AUPEC05, [15] K. Divya and P. Rao, Models for Wind Turbine Generating Systems and Their Application in Load Flow Studies, Electric Power Systems Research, vol. 76, no. 6, pp , June [16] J. D. Glover and M. S. Sarma, Power System Analysis and Design. the Wadsworth Group, [17] P. Garcia, J. Pereira, and S. Carneiro, Voltage Control Devices Models for Distribution Power Flow Analysis, IEEE Trans. Power Systems,, vol. 16, no. 4, pp , Nov [18] System Impact Assessment Report for the Erie Shores Wind Development Project, SIA pdf, Independent Electricity System Operator (IESO), April [19] Erie Shores Wind Farm Fact Sheet, Macquarie Power & Infrastructure Income Fund. [20] S. Auddy, R. Varma, and M. Dang, Field Validation of a Doubly Fed Induction Generator (DFIG) Model, 2007, pp [21] P. Christiansen and J. Kristoffersen, The wind Farm Main Controller and the Remote Control System of the Horns Rev Offshore Wind Farm, in Proc. Fourth Int. Workshop on Large-Scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms, Denmark, Oct [22] T. Ackermann, Transmission Systems for Offshore Wind Farms, IEEE Power Engineering Review, vol. 22, no. 12, pp , [23] J. Smith, B. Zavadil, and C. Bryan, Engineering Design and Integration Experience from Cape Wind 420 MW Offshore Wind Farm, in Proc. Fourth Int. Workshop on Large-Scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms, Denmark, Oct [24] A. Tabesh and R. Iravani, Transient Behavior of a Fixed-Speed Grid-Connected Wind Farm, in Proc. Int. Conf. on Power Systems Transients, IPST 05, Montreal, Canada, June [25] G. Quinonez-Varela, G. Ault, O. Anaya-Lara, and J. McDonald, Electrical Collector System Options for Large Offshore Wind Farms, IET Renewable Power Generation, vol. 1, no. 2, pp , June [26] W. Kersting, Distribution System Modeling and Analysis, 2nd ed. New York: Taylor & Francis Group, Mohamed Zakaria Kamh (S 08) received the B.Sc. (Honors) and M.Sc. degrees from Ain Shams University, Cairo, Egypt in 2003 and 2007 respectively, both in electrical engineering. Since 2008, he has been with the Electrical and Computer Engineering Department, University of Toronto, Toronto, ON, Canada, where he is currently pursuing his Ph.D. degree. Since 2009, he has been a member of the IEEE Power and Energy Society. He served as a power system specialist and senior electrical engineer with well reputed electromechanical consulting firms from 2003 to His research interests include power systems, distributed and renewable energy resources, smart grids, and virtual power plants. His personal website is Reza Iravani (F 03) received the B.Sc. degree in electrical engineering from Tehran Polytechnic University, Tehran, Iran, in 1976, and the M.Sc. and Ph.D. degrees in electrical engineering from the University of Manitoba,Winnipeg, MB, Canada, in 1981 and 1985, respectively. He is currently a Professor at the University of Toronto, Toronto, ON, Canada. His research interests include power electronics and applications of power electronics in power systems.