Electrical and Control Aspects of Offshore Wind Farms II (Erao II)

Transcription

1 ECN-C Electrical and Control Aspects of Offshore Wind Farms II (Erao II) Volume : Offshore wind farm case studies J.T.G. Pierik (ECN) J. Morren (TUD) E.J. Wiggelinkhuizen (ECN) S.W.H. de Haan (TUD) T.G. van Engelen (ECN) J. Bozelie (Neg-Micon) March 4 ECN-C- -4-5

2 Erao II, Volume : Offshore wind farm case studies Distribution Novem: J. t Hooft 5 TUD: J. Morren 6 S.W.H. de Haan 7 P. Bauer TenneT: W. Kling 3 J. Bozelie 4 Kema: R. de Groot 5 P. Vaessen 6 Essent: H. Slootweg 7 C. Houben 8 Continuon: M. Bongaerts 9 Nuon: M. van Riet Neg-Micon: A. Winnemuller ICE: J.C. Montero R. Jimenez 3 Sintef: J.O. Tande 4 Chalmers: O. Carlson 5 T. Thiringer 6 Risoe: P. Soerensen 7 VTT: B. Lemstrom 8 NREL: Y-H. Wan 9 E. Muljadi 3 Ineti: A. Lopez Estanquero 3 University College Dublin: A. Mullane 3 M. O Malley 33 UMIST: N. Jenkins 34 O. Anaya-Lara 35 Hydro Quebec: R. Gagnon 36 ECN: A.B.M. Hoff 37 C.A.M. van der Klein 38 W.C. Sinke 39 G.J.H. van Nes 4 G. Peppink 4 H.J.M. Beurskens 4 L.W.M.M. Rademakers 43 H. Snel 44 B.H. Hendriks 45 G.P. Corten 46 E.J. Wiggelinkhuizen 47 B.H. Bulder 48 P. Schaak 49 P. Heskes 5 P. Lako 5 T.J. de Lange 5 T.G. van Engelen 53 E.L. van der Hooft 54 J.T.G. Pierik 55 ECN Wind Energy Archive ECN Central Archive 7 ECN-C- -4-5

3 ABSTRT To investigate dynamic interaction of wind farms and the electrical grid, dynamic models of wind farms are needed. These models are not available however. The objective of the Erao- project has been () to develop these models, () to demonstrate their use by evaluating wind farms with different types of electrical systems and (3) to design and demonstrate controllers that can cope with grid code requirements. Four types of wind farm models have been developed based on different types of turbines: Constant Speed Stall turbine with directly coupled Induction Generator (CSS-IG); Constant Speed Stall turbine with Cluster Controlled induction generator operating in variable speed mode (CSS-CC); Variable Speed Pitch turbine with Doubly-Fed Induction Generator (VSP-DFIG); Variable Speed Pitch turbine with Permanent Magnet generator and full converter (VSP-PM). For each type of wind farm, three cases have been evaluated: normal operation including flicker production; response to a grid frequency dip; response to a grid voltage dip. For the wind farms which are able to support grid voltage or grid frequency, controllers for these purposes have been developed and demonstrated. Results and conclusions The response of a wind farm to a grid frequency dip strongly depends on the presence of a converter. A full converter, in the CSS-CC and VSP-PM wind farms, decouples the turbines from the disturbance. But also the system with a partial converter (VSP-DFIG) is hardly affected by the frequency dip. The constant speed wind farm (CSS-IG) on the other hand has serious problems with a frequency dip and the corresponding voltage dip. The constant speed wind farm can stay connected during the voltage dips that have been applied. The high amount of reactive power that is required by this wind farm during a voltage disturbances can be problematic. The Cluster Controlled wind farm (CSS-CC) can handle voltage dips if a resistor is placed in parallel to the dc-link capacitor and the surplus of energy during the voltage dip is dissipated. Wind farms using doubly-fed induction machines (VSP-DFIG) are the most problematic concept when voltage dips are considered. A solution is to provide a controlled by-pass for the high currents in the rotor. In the variable speed pitch wind farm with permanent magnet generators (VSP-PM) good voltage dip ride-through is achieved. The constant speed stall controlled wind farm (CSS-IG) can not assist in grid frequency control. The cluster controlled wind farm in the Erao- study is based on a stall controlled turbine (CSS-CC). Therefore it can not assist in grid frequency support either. Both variable speed pitch wind farms (VSP-DFIG and VSP-PM) can support grid frequency, which has been demonstrated by simulations. Only systems with converters are suitable for grid voltage control. The simulations demonstrate the feasibility of voltage control for wind farms with doubly-fed induction generators. There is no large difference between voltage control by wind farms with doubly-fed induction generators and voltage control by the other wind farms with IGBT converters: CSS-CC and VSP-PM. Recommendation With the completion of the wind farm models based on individual turbines, verification of models should now have a high priority. ECN-C

4 Erao II, Volume : Offshore wind farm case studies Keywords: wind farm models, wind farm dynamics, electrical systems, fault ride through, grid support Acknowledgement Erao- is a continuation of the Erao- project, in which a steady state (load flow) and economic model for offshore wind farms has been developed [9]. The Erao projects have been supported by the Dutch Agency for Energy and Environment (NOVEM) in the "Programma Duurzame Energie" of the Netherlands, executed by Novem by order of the Ministry of Economic Affairs. Novem project number: ECN project number: ECN-C- -4-5

5 EXECUTIVE SUMMARY In The Netherlands offshore wind power is on the brink of implementation. Plans exist for two offshore wind farms of about MW, located and 5 km from the coast of the province of North Holland. In 3 an investigation has been started to quantify the effect of 6 MW offshore wind power on the high voltage grid. Only the steady state behaviour has been considered, resulting in suggestions for grid reinforcement. This investigation needs to be complemented by a study on the dynamic interaction of wind power and the electrical grid. Objective of Erao- To investigate dynamic interaction of wind farms and the electrical grid, dynamic models of wind farms are needed. These models will be of great help in the evaluation of the behaviour of wind power during normal grid operation as well as during grid faults and in the design of controllers that enable wind farms to support the grid. Dynamic models of wind farms, including the relevant electrical components and sections of the grid, are not available however. The objective of the Erao- project is () to develop these models, () to demonstrate their use by evaluating wind farms with different types of electrical systems and (3) to design and demonstrate controllers that can cope with grid code requirements. Part : Model development The wind farm models are based on models of electrical components and controllers developed in this project and already existing models of wind, rotor, tower, mechanical drive train and pitch controller. The modelled electrical components and controllers are: induction generator doubly-fed induction generator permanent magnet generator IGBT converter and converter controller transformer cable synchronous generator consumer load wind farm controller for grid frequency support converter controller for grid voltage support A simple grid model and a model of the flicker meter has also been developed. An important aspect of dynamic models of electrical systems is computational speed. Electrical transients have very small time constants, resulting in small time steps and long computation time. In Erao- special attention has been paid to computational speed. An important increase in speed can be realised by the use of the dq-transformation, which has been applied to all models of electrical components in the Erao- component library. Volume of this report gives a mathematical derivation of the electrical component models, followed by the implementation of the models in Simulink, a computer program suitable for dynamic simulation. Turbines are modelled by connecting the electrical component models to ECN-C

6 Erao II, Volume : Offshore wind farm case studies the models of the rotor, tower, mechanical drive train and pitch controller. In the second step, individual turbine models are connected by cable models to produce the wind farm model. Results and conclusions from model development Dynamic models of wind farms based on individual turbine models are large and complicated. The number of state variables is high and some of the time constants are small, leading to a relatively long simulation time. The level of detail is high however, which makes these models suitable for the evaluation of wind farm dynamics and wind farm-grid interaction as well as for the design of controllers. The application of the dq-transformation significantly reduces the simulation time during normal operation of the wind farm, when transients from electrical switching operation have died out. Part : Model demonstration The second part of the Erao- project demonstrates the use of the developed wind farm models. In a number of case studies, four types of wind farms have been compared. The wind farm types use different turbines and different control methods, viz.: Constant Speed Stall turbine with directly coupled Induction Generator (CSS-IG, reference case); Constant Speed Stall turbine with Cluster Controlled induction generator operating in variable speed mode (CSS-CC); Variable Speed Pitch turbine with Doubly Fed Induction Generator (VSP-DFIG); Variable Speed Pitch turbine with Permanent Magnet generator and full converter (VSP- PM). The layout of a proposed offshore wind farm, the Near Shore Wind farm (NSW), has been taken as reference. The Near Shore Wind Farm is planned in the North Sea near the town of Egmond in The Netherlands. One string of turbines has been modelled with each of the four types of turbines. A simplified grid model has been included to enable simulation of wind farm-grid interaction. For each type of wind farm, three cases have been evaluated: normal operation including flicker production; response to a grid frequency dip; response to a grid voltage dip. For the wind farms which are able to support grid voltage or grid frequency, a converter controller or a wind farm controller suitable for this purpose has been developed and demonstrated. Volume of this report describes the case study results. Results and conclusions from case studies Normal operation of the wind farms has been simulated by the response to a wind gust. The simulations demonstrated proper operation of the generator and converter models, the converter controllers and proper overall behaviour of the wind farm. A limited flicker evaluation has been executed. Instantaneous flicker values have been determined over the complete range of operating conditions for the four types of wind farms. Flicker 6 ECN-C- -4-5

7 values of a single turbine have been compared to the values of a string of twelve turbines under the same operating conditions and fictitious grid parameters. The constant speed stall wind farm generates the highest flicker, the flicker production of the wind farms with partial and full converter is lower. Wind farm response to grid frequency and grid voltage dips The response of a wind farm to a grid frequency dip (5 Hz, sec) strongly depends on the presence of a converter. A full converter, in the case of the CSS-CC and VSP-PM wind farms decouples the turbines from the disturbance. But also the system with a partial converter (VSP- DFIG) is hardly affected by the frequency dip due to the effective adjustment of the rotor currents by the rotor converter. The constant speed system on the other hand has serious problems with a frequency dip and the corresponding voltage dip: depending on the depth and the conditions at the start of the dip, current, power and reactive power peaks may exceed rated values and may lead to a wind farm shut down. The farm with constant speed stall turbines and directly connected induction generators (CSS- IG) can stay connected during the voltage dips that have been applied (3%- sec, 5%-.5 sec and 85%-. sec). High currents are flowing during the voltage drop. Due to the high thermal capacity of the induction machine these currents will be no problem. The currents may trigger protective devices in the grid. The high amount of reactive power that is required by the wind farm during a voltage disturbances can be more problematic. When the dip lasts too long this may lead to voltage collapse. The Cluster Controlled wind farm (CSS-CC) can handle voltage dips if a resistor is placed in parallel to the dc-link capacitor and the surplus of energy during the voltage dip is dissipated. Wind farms using doubly-fed induction machines (VSP-DFIG) are the most problematic concept when voltage dips are considered. Large currents will flow in the rotor circuits and in the converters. Due to the limited thermal capacity of the power electronic devices in the converters, these currents may destroy the converters. A possible solution is to limit the high currents in the rotor by providing a by-pass over a set of resistors connected to the rotor windings. With these resistors it is possible to survive grid faults without disconnecting the turbine from the grid. One of the case studies demonstrates this solution. Manufacturers of DFIG systems are working on this solution and are making progress in meeting the voltage ride-through requirement. In the variable speed pitch wind farm with permanent magnet generators (VSP-PM), all the essential parameters can be controlled. Therefore good voltage dip ride-through can be achieved. The power supplied by the generator is reduced by the controllers during the dip. This is required because otherwise the current in the converters or the dc-link voltage becomes too high. To avoid overspeeding the pitch controller is activated. Wind farms assisting grid frequency or grid voltage The constant speed stall controlled wind farm (CSS-IG) can not assist in grid frequency control. The cluster controlled wind farm in the Erao- study is based on a stall controlled turbine (CSS-CC). It can not control aerodynamic power directly and therefore it can not assist in grid frequency support either. Both variable speed pitch wind farms (VSP-DFIG and VSP-PM) can be controlled to support grid frequency, which has been demonstrated by simulations. The controller consists of two parts: delta-control to realise a power margin and frequency feed-back to act on a frequency deviation. Since frequency control capability for wind farms implies maintaining a power margin, this feature may not be cost-efficient. Only systems with converters are suitable for grid voltage control. Different voltage and reac- ECN-C

8 Erao II, Volume : Offshore wind farm case studies tive power control strategies have been investigated for the VSP-DFIG wind farm. It has been shown that it is possible to control the power factor and that the wind farm can follow reactive power setpoints. Two voltage control options have been investigated. In the first option each turbine controls the voltage at its own terminal, in the second option the voltage at the grid connection point is controlled. Droop control has been implemented on each turbine. With this type of control, the wind farm behaviour during voltage deviations is similar to conventional power plant behaviour. The results depend on the X/R ratio of the grid: low X/R ratios require large amounts of reactive power to control the voltage and the wind farm converters are limited in current and thus in reactive power. Nonetheless, the simulations demonstrate the feasibility of voltage control for wind farms with doubly-fed induction generators. There is no large difference between voltage control by wind farms with doubly-fed induction generators and voltage control by the other wind farms with IGBT converters: CSS-CC and VSP-PM. This has been demonstrated by simulations with a cluster of CSS-CC turbines and a string of VSP-PM turbines. The results are similar to those of the VSP-DFIG wind farm. Economic evaluation The load flow program and the database with electrical and economic parameters developed in the Erao- project has been used in an economic evaluation of the four wind farm electrical systems. For a wind regime representative of the North Sea, the power production including the electrical losses, has been determined for the layout of the Near Shore Wind farm. This results in the contribution of the electrical system to the Levelised Production Costs (LPC). The VSP-DFIG farm performs best:.4 Eurocent/kWh. The CSS-IG farm is of the same magnitude:.6 Eurocent/kWh, while the other two farms have relatively expensive electrical systems:.6 Eurocent/kWh (VSP-PM) and 4.57 Eurocent/kWh (CSS-CC). The high price for the Cluster Controlled system is caused by the expensive converters. Recommendations With the completion of the wind farm models based on individual turbines, verification of models should now have a high priority. The Erao-3 project has been started with model validation as one of the objectives. For the incorporation of dynamic models of wind farms in models of national grids, the complexity of the wind farm models has to be reduced. Aggregated wind farm models, in which all turbines are represented by a single equivalent model are more suitable for this purpose. However, aggregated models loose the wide range of applicability of the wind farm models based on individual turbine models. It is recommended to develop aggregated wind farm models, tailored to application in power system models. The wind farm models developed in Erao- can serve as reference in the development of these aggregated models. Systems with cables to shore have not been included in the Erao- case studies. The Erao- component library includes all models necessary to investigate connections, with the exception of the thyristor converter. This converter however, is a less likely option for the connection of offshore wind farms than the IGBT converter, due to its limited controllability and large footprint. connections are currently more expensive than, but may offer a number of advantages. It is recommended to include these systems in a future study and for comparison purpose also develop a thyristor converter model. 8 ECN-C- -4-5

9 Nomenclature volume Symbols Subscripts C capacitance a aerodynamic f frequency conv converter i current dc dc-link L inductance e electrical P (active) power g grid Q reactive power m mechanical R resistance r rotor T torque s stator u voltage v voltage V Velocity w angular velocity Z impedance Table : Base values for per unit calculation CSS-IG CSS-CC VSP-DFIG VSP-PM Parameter Value Parameter Value Parameter Value Parameter Value Ps.75 MW Pconv, Pg 3*Ps Pconv, Pr 75 kw Ps, Pconv.5 MW Vs 96 V Udc 39 V Vconv, Vr 67 V Vs, Vconv 4 V wm *pi Udc V Udc 45 V Pg *Ps Vg 34 kv Parameters for CSS-CC, VSP-DFIG and VSP-PM are equal to that of CSS-IG, unless otherwise stated. ECN-C

10 Erao II, Volume : Offshore wind farm case studies. ECN-C- -4-5

11 CONTENTS Introduction 5 Reference wind farm NSW 6 3 Normal operation 9 3. Constant speed stall wind farm Cluster controlled wind farm Variable speed pitch controlled wind farm with DFIG Variable speed pitch controlled wind farm with PM Conclusion Flicker evaluation Constant speed stall wind farm Cluster controlled wind farm Variable speed pitch controlled wind farm with DFIG Variable speed pitch controlled wind farm with PM Conclusion Frequency dip response of wind farms Constant speed stall wind farm Cluster controlled wind farm Variable speed pitch controlled wind farm with DFIG Variable speed pitch controlled wind farm with PM Conclusion Voltage dip behaviour of wind turbines Introduction Voltage dips Simulation setup Constant speed stall wind turbine Introduction Dip of 3% - seconds Dip of 5% -.5 seconds Dip of 85% -. seconds Final remarks Cluster controlled wind turbine Introduction Dip of 3% - seconds ECN-C- -4-5

12 Erao II, Volume : Offshore wind farm case studies Dip of 5% -.5 seconds Dip of 85% -. seconds Discussion Variable speed pitch controlled wind turbine with DFIG Introduction Dip of 3% - seconds Dip of 5% -.5 seconds Dip of 85% -. seconds Discussion Variable speed pitch controlled wind turbine with PM Introduction Dip of 3% - seconds Dip of 5% -.5 seconds Dip of 85% -. seconds Discussion Conclusions and recommendations Grid frequency support by wind farms 9 7. Introduction Constant speed stall wind farm Cluster controlled wind farm Variable speed pitch controlled turbine with PM Delta control Grid frequency control support Variable speed pitch controlled turbine with DFIG Conclusion Grid voltage control by wind farms Introduction Voltage control strategies Simulation setup Introduction Doubly-Fed Induction generator Permanent magnet generator Cluster coupled induction machines Constant power factor control Reactive power setpoint Voltage control per turbine Voltage control at connection point ECN-C- -4-5

13 CONTENTS 8.8 Droop control Permanent magnet generator and cluster coupled induction machines Discussion Weak and strong grids X/R ratio Maximum current rating of wind turbines Distributed versus centralised control Conclusion and recommendations Economic analysis 9 Conclusions and remarks. Conclusions Remarks and recommendations A EeFarm load flow results 7 B Data for dynamic modelling of wind turbines 3 ECN-C

14 Erao II, Volume : Offshore wind farm case studies. 4 ECN-C- -4-5

15 INTRODUCTION For the problem description and the objectives of the Erao- project is referred to the introduction of Volume of this report. Volume of the Erao- report focuses on the dynamic behaviour of the four types of wind farm models in a number of a case studies. The four types of wind farm that have been considered are: Constant Speed Stall turbine with directly coupled induction generator (CSS-IG, reference case); Constant Speed Stall turbine with Cluster Controlled induction generator operating in variable speed mode (CSS-CC); Variable Speed Pitch turbine with Doubly Fed Induction Generator (VSP-DFIG); Variable Speed Pitch turbine with Permanent Magnet Generator with full converter (VSP- PM). The dynamic models of the wind farms have been used to: investigate wind farm behaviour during normal operation (for instance to evaluate flicker production or to examine dynamic interaction between interconnected turbines and between the wind farm and the grid); investigate wind farm response to deviations of the grid voltage and frequency; develop wind farm controllers. These applications are demonstrated in case studies in which the layout of a proposed Dutch offshore wind farm, the Near Shore Wind farm (NSW), has been taken as reference. The Near Shore Wind Farm is planned in the North Sea near the town of Egmond in The Netherlands. One string of the farm, consisting of turbines has been modelled with the four types of turbines. For each type of wind farm, three cases are evaluated: () normal operation including flicker production, () the behaviour during a grid frequency and (3) grid voltage dip. For those systems that are able to support grid voltage or grid frequency, a wind farm controller suitable for these purposes has been developed and demonstrated. Table gives an overview of all evaluations. Table : Erao- case study evaluations CSS-IG VSP-DFIG VSP-PM CSS-CC Normal operation X X X X Flicker X X X X Frequency dip X X X X Voltage dip X X X X Frequency support - X X X Voltage support - X X X ECN-C

16 REFERENCE WIND FARM NSW The Near Shore Wind Farm (NSW), planned near Egmond (see figure ), has been chosen as reference system for the Erao- evaluations. Figure gives the NSW layout. The farm consists of three sets of twelve NM9.75MW turbines connected by three cables to the 5 kv substation in IJmuiden. The cables in the farm and to shore are rated at 34 kv. In the farm two types of cables are used, depending on the loading at a given location. For the connection from the wind farm to the transformer in the substation also two types of cables are used, one for the submarine section of the route and a second type for the on-land route. Between the two sections, a relay station is located (see bottom part of figure ). Figure 3 gives the steady state characteristics of the NM9 turbine, a variable speed pitch controlled turbine. Based on the data of this turbine (including a design of the pitch control by ECN), the cable and transformer data, the ECN-TUD program EeFarm has been used to calculate the load flow over the full range of wind speeds: see figure 4. The load flow results (voltage, current, active and reactive power at all nodes) for full load for the four types of wind farms are listed in appendix A. NSWP ECN Figure : Near Shore Wind Park (NSWP) location 6 ECN-C- -4-5

17 REFERENCE WIND FARM NSW R.S. Figure : Near Shore Wind Fark layout ECN-C

18 Erao II, Volume : Offshore wind farm case studies Pel (kw) Vw (m/s) Pel (kw) N (rpm) Torque (knm) 5 5 Torque (knm) Vw (m/s) N (rpm) Figure 3: NM9 power and torque curves EeFarm ver.. 7 NSW Egmond WF Pel (MW) Vw (m/s) Ploss (MW) Vw (m/s) 95 Eff (%) Vw (m/s) Figure 4: NSW total power, electrical losses and efficiency calculated by EeFarm ( ECN- TUD) 8 ECN-C- -4-5

19 3 NORMAL OPERATION The first step in the case study will be to illustrate the response of the four types of wind farms during normal operation. To characterize the wind farm dynamic behaviour, a wind gust has been chosen from cut-in to rated wind speed and back again. To all turbines in the farm the same gust is applied, but with a small time delay. The total time simulated will be sec. The wind farm is connected to a grid model which consists of a MW synchronous generator with frequency and voltage control, two consumer loads (5 and 65 MW) and a 5 km 4 kv cable connecting to the 5 kv grid. A detailed description of the models can be found in Volume of this report. 3. Constant speed stall wind farm 96 V 34 kv 5 kv Figure 5: CSS-IG wind farm: electrical layout of a string of CSS turbines Figure 6 shows the response of turbine, 6 and of the CSS-IG farm to a wind speed increase from 4 to 5 m/s and back again. The reactive power demand of the turbines increases during the gust and is not supplied by capacitors but by the cables connecting the farm to the HV grid and the HV grid. It can be profitable to absorb (part of) the reactive power production of the cables by the wind farm. The total active and reactive power of the string, as well as current and voltage at the 5 kv side of the wind farm transformer are plotted in figure 7. ECN-C

20 Erao II, Volume : Offshore wind farm case studies 5 Turbine, 6, 3 plot CSSout Vw (m/s) 5 Paero (MW) Q tur (MVA) slip ( ) x Figure 6: Response of first turbine in CSS-IG wind farm to wind gust 3 Park plot CSSout Pel park (MW) Q park (MVA) I park (A) 5 V park (kv) Figure 7: Power, reactive power, current and voltage at the 5 kv side of the CSS-IG wind farm transformer ECN-C- -4-5

21 3 NORMAL OPERATION Grid plot CSSout Pel grid (MW) P cons, (MW) freq (Hz) 5 V exc (V) Figure 8: Response of grid to wind gust in CSS-IG wind farm Figure 8 shows how the changes in wind power ( to 5 MW) are compensated by the frequency controller of the synchronous machine representing the HV grid. The frequency stays within a band of.5 Hz around the 5 Hz setpoint. The voltage regulator on the synchronous machine compensates for the reactive power changes, the voltage at the HV side of the park transformer is kept within a band near 5 kv (Figure 7). ECN-C- -4-5

22 Erao II, Volume : Offshore wind farm case studies 3. Cluster controlled wind farm ω 34 kv ω3 34 kv ω 34 kv ω4 34 kv 34 kv 5 kv Figure 9: CC wind farm: electrical layout of four strings of 3 cluster controlled turbines Figure 9 shows the layout of a string of the near shore wind in cluster controlled mode. The string is divided into four clusters of three wind turbines each, connected to a single back-toback converter. The converters are connected through 34 kv submarine cables to the 34/5 kv transformer in the transformer station on shore. In the Simulink model, the cluster controlled wind farm consists of one cluster, the 34 kv cable to the transformer, the 5 kv transformer and a simplified grid model. For the details of the component models is referred to Volume of this report. For the cluster controlled wind turbine, power limitation by either stall or pitch control can be chosen. In principle, both options are technically feasible. The effect of cluster controlled operation on the aerodynamic power of both options is illustrated in figure. On the left the power-wind speed curves for -6 Hz operation of a constant speed stall turbine are plotted. 6 Hz operation postpones stall and can lead to tripping of the turbine due to excess power. Therefore, 5 Hz will mark the upper speed limit for the cluster controlled stall turbines. On the right hand side of figure the power-wind speed curves for -6 Hz operation of a variable speed pitch turbine are plotted for pitch angle zero. Exceeding the rated aerodynamic power will be prevented by pitching the blades and the rotational speed of the cluster generators does not need to be limited to 5 Hz. ECN-C- -4-5

23 3 NORMAL OPERATION 6 Power speed curves CSS turbine 6 Power speed curves VSP turbine (Bladhoek gr.) 6 Hz Hz 4 4 Paero (MW) 3 5 Hz Paero (MW) 3 5 Hz 4 Hz 4 Hz 3 Hz 3 Hz Hz Vw (m/s) Hz Vw (m/s) Figure : CCS (left) and VSP (right) turbine steady state power curves At low wind speed, decreasing rotational speed below rated will increase aerodynamic efficiency compared to constant speed operation: the, 3 and 4 Hz curves are above the 5 Hz power curve. The combination of high wind speed and low rotational speed reduces the aerodynamic power compared to constant speed operation. For a turbine with pitch control, this effect can be reduced by pitching at below rated wind speeds. Since individual wind speeds at the individual turbines in a cluster will differ, there will always be some mismatch, leading to a lower overall aerodynamic efficiency compared to individual variable speed. This reduction in energy yield has been estimated at.4% [7]. Hopefully, this is compensated by a cost reduction due to a smaller number of converters. A second aspect to consider for cluster control is the effect on turbine power and torque variations. The question is whether these variations will decrease or increase compared to constant speed operation. Individual turbine values as well as overall cluster behaviour can be compared. The cluster controlled wind farm in this evaluation will be based on a constant speed stall turbine (CSS-CC). If a variable speed turbine is chosen, a modification of the pitch control is required, especially if an attempt is made to increase efficiency by blade pitching below rated speed wind speed. This modification is outside the scope of this study. Speed control of a cluster will be based on measured wind speed(s). In the simulations, the wind speed at turbine has been chosen, but a different choice may prove to be more efficient. The speed of the turbines in a cluster is controlled for constant lambda operation compared to this wind speed, between an upper and a lower limit. The turbine speed is dictated by the frequency of the rectifier (turbine side converter). Since this results in a reduced frequency in the stator of the induction machine, the amplitude of the stator voltage is reduced proportionally to this frequency. This is necessary since the stator impedance is proportional to the stator frequency and a decrease in frequency would otherwise lead to high currents and possibly the activation of the thermal protection. The stator voltage is determined by a feed forward controller on the turbine side converter, which directly sets the amplitude of the voltage (see for details the Simulink blocks for the CSS-CC system in Volume ). Therefore, the control strategy of the turbine side converter in the cluster controlled system is completely different from the converters in the DFIG and PM systems, which control the current. ECN-C

24 Erao II, Volume : Offshore wind farm case studies Figure illustrates the combined effect of reduced frequency and voltage: the power-slip and torque-slip curves are similar in shape, only the pull-out power is reduced. The reactive power consumption is reduced. = 5 Hz, 96 V, = 4 Hz, 768 V x 4 amstat6 5.5 Pel (MW) ids (A) Slip ( ) 5 5 iqs (A) 3 x 4 Tel (Nm) Slip ( ) Reactive Power (MVAr) Slip ( ) Figure : Steady state curves for an induction machine at 5Hz, 96V and 4Hz, 768V 4 ECN-C- -4-5

25 3 NORMAL OPERATION vw, vw, vw3 (m/s) 5 5 CC turb,, 3 n, n, n3 (rpm) CC normal oper Pa, Pa, Pa3 (MW) plot CC a.m 9 Jan vs, vs, vs3 (V) 4 8 is, is, is3 (A) s, s, s3 (%) Figure : Cluster controlled wind farm: normal operation, turbines, and 3 Figure demonstrates normal operating conditions of a wind farm with one cluster of 3 turbines. A gust from 4 to 5 m/s passes the turbines with a small delay. The wind speed at turbine determines the rotor speed setpoint. Rotor speed and stator voltage vary proportionally. Power and slip vary accordingly. The turbine speed controller attempts to maintain a tip speed ratio of 5 below rated wind speed, compared to the wind speed at turbine. Figure 3 shows the total power of the cluster, the reactive power, the currents, voltages, current and the cluster frequency. The response of the grid to the changing operating conditions of the cluster can be seen in figure 4. ECN-C

26 Erao II, Volume : Offshore wind farm case studies CC converter CC normal oper plot CC a.m 9 Jan 4 P cluster (WM) Q cluster (kva) 5 5 idg, iqg (A) vdg, vqg (kv) udc (kv) f cluster (Hz) Figure 3: Cluster controlled wind farm: normal operation, converter values Grid plot CSSout 8 8 P SM (MW) 6 4 P cons (MW) freq (Hz) 5 V exc (V) Figure 4: Cluster controlled wind farm: normal operation, grid values 6 ECN-C- -4-5

27 3.3 Variable speed pitch controlled wind farm with DFIG 3 NORMAL OPERATION 96 V 34 kv 69 V 5 kv Figure 5: DFIG wind farm: electrical layout of a string of DFIG turbines Figure 5 gives the layout of a string of turbines equiped with doubly fed induction generators. Normal operation of this wind farm is demonstrated by the response to a wind gust passing through the farm. Figure 6 gives the rotor effective wind speed at turbine, 6 and in the string. The aerodynamic powers P a, P a6 and P a and slip of the turbines s, s6 and s follow the wind speed changes. The - converter on the generator rotor controls the generator torque and the reactive power of the stator (reactive power setpoint zero) by adjusting the d- and q-current in the rotor (for the details see Volume ). The grid side - converter controls the -voltage and the reactive power to the grid (i d setpoint zero, corresponding to a practically zero reactive power) by adjusting the d- and q-current to the grid. The total reactive powers Q, Q6 and Q (stator plus grid side converter) are plotted in the upper right part of figure 6. ECN-C

28 Erao II, Volume : Offshore wind farm case studies DFIG turb, 6,. plot DFIG norm op.m vw, vw6, vw (m/s) Q, Q6, Q (MW) Pa, Pa6, Pa (MW) s, s6, s (%) Figure 6: DFIG wind farm, normal operation, turbines, and 3 5 DFIG normal oper plot DFIG norm op.m.5 P park (MW) 5 Q park (MVA) idp, iqp (A) vdp, vqp (kv) Figure 7: DFIG wind farm, normal operation, string of turbines The total electric power produced by the string of turbines is plotted in figure 7. A peak 8 ECN-C- -4-5

29 3 NORMAL OPERATION power of about MW is reached. The park reactive power at the low voltage side of the 34kV-5kV transformer fluctuates between -.5 and.5 MVA, inversely proportional to the produced electric power. The fluctuations are caused by the inductivity of the turbine transformers and the cables inside the wind farm. The d-component of the 34kV voltage is relatively small, the q-component of the string current is therefore almost proportional to the string power, and the d-component to the reactive power. In chapter 8 is demonstrated how the converter can be controlled to realise zero reactive power at the point of common coupling. 75 Grid 7 plot DFIG norm op.m P SM (MW) P cons (MW) freq (Hz) dv exc (V) Figure 8: DFIG wind farm, normal operation, grid values In figure 8 the response of the grid model to the changing wind power is illustrated. The synchronous machine adjusts the power to maintain the grid frequency. The frequency deviations are small, in spite of the relatively small rotating mass in the grid. Consumer load is constant and the synchronous machine adjust the exciter voltage to keep the grid voltage constant. ECN-C

30 Erao II, Volume : Offshore wind farm case studies 3.4 Variable speed pitch controlled wind farm with PM 34 kv 5 kv Figure 9: VSP-PM wind farm: electrical layout of a string of PM turbines Figure 9 gives the layout of a string of turbines equipped with permanent magnet generators. Normal operation of this wind farm is demonstrated by the response to a wind gust passing through the farm. Figure gives the rotor effective wind speed at turbine, 6 and in the string. The aerodynamic powers P a, P a6 and P a and rotational speeds of the turbines s, s6 and s follow the wind speed changes. The - converter on the generator controls the generator torque and the reactive power of the stator (reactive power setpoint zero) by adjusting the d- and q-current in the stator. The grid side - converter controls the -voltage and the reactive power to the grid (reactive power setpoint zero) by adjusting the d- and q-current to the grid. The reactive power of turbine, 6 and (Q, Q6 and Q) is plotted in the upper right part of figure. 3 ECN-C- -4-5

31 3 NORMAL OPERATION 6 PM turb, 6,. PM norm oper vw, vw6, vw (m/s) Q, Q6, Q (MW) Pa, Pa6, Pa (MW) plot PM norm op.m s, s (%) Figure : PM wind farm: normal operation, turbines, 6 and 3 PM norm oper plot PM norm op.m 5 P park (WM) 5 Q park (kva) idp, iqp (A) 5 vdp, vqp (kv) Figure : PM wind farm: normal operation, string of turbines The total electric power produced by the string of turbines is plotted in figure. A peak ECN-C

32 Erao II, Volume : Offshore wind farm case studies power of 3 MW is reached. The park reactive power at the low voltage side of the 34kV- 5kV transformer fluctuates between - and.5 MVA, inversely proportional to the produced electric power. The range is wider that of the farm equipped with DFIG. The fluctuation is caused by the inductivity of the turbine transformers and the cables inside the wind farm. 7 Grid 7 plot PM norm op.m 65 6 P SM (MW) P cons (MW) freq (Hz) 5 dv exc (V) Figure : PM wind farm: normal operation, grid values In figure the response of the grid model to the changing wind power is illustrated. The synchronous machine adjusts the power to maintain the grid frequency. The frequency deviations are small. Consumer load is constant and the synchronous machine adjusts the exciter voltage to keep the grid voltage constant. 3.5 Conclusion Normal operation of four types of wind farms has been demonstrated by their response to a wind gust. The cluster controlled system has been equiped with a different machine side converter control strategy than the VSP-DFIG and VSP-PM system due to different requirements. The simulation showed proper operation of the generator and converter models and the converter controllers. 3 ECN-C- -4-5

33 4 FLICKER EVALUATION The currents and voltages calculated in the previous chapter during the first s of normal operation have been used to calculate instantaneous flicker values for a single turbine, cluster or the wind farm. For the details of the flicker calculation is referred to Volume of the Erao- report. The short circuit power for the flicker calculation is adapted to the case, i.e. 5 times the rated power of the turbine, the cluster or string (farm). A fictitious grid angle of 3 o has been chosen. The sample frequency in the wind farm calculations was 4 Hz. Flicker values are binned during intervals of 6 s, resulting in 4 values per binning period, except for the final interval. 4. Constant speed stall wind farm 5 CSS, normal operation, Turbine, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 3: Binned instantaneous flicker values for a CSS-IG turbine Binned instantaneous flicker values for the CSS turbine and for a string of CSS turbines are plotted in figure 3 and 4. In this example, turbine rotors are synchronized, due to identical initial values and the relatively short simulated time span. The flicker level of the farm is lower than the level of the individual turbine, caused by averaging effects. Although, the complete operating range of the turbine is included in the s normal operation, a full flicker evaluation requires a longer simulation. For a similar constant speed stall turbine, the NW46, the flicker levels were measured: a Pst between.5 and.5 was found for below rated operation and.5 to.45 for rated operation. The calculated values for the CSS turbine correspond well to the below rated operation, calculated rated operation values are lower and may be caused by differences in turbine properties. ECN-C

34 Erao II, Volume : Offshore wind farm case studies 5 CSS, normal operation, turb, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 4: Binned instantaneous flicker values for a wind farm of CSS-IG turbines 4. Cluster controlled wind farm 5 CC, normal operation, one cluster Nr. of samples per bin Pst ( ) Figure 5: Binned instantaneous flicker values for a wind farm of three cluster controlled turbines The flicker level of a variable speed turbine is expected to be lower than the level of a constant speed turbine, if the turbine control is well tuned. Cluster control is a special case of variable speed operation; all turbines in a cluster have the same speed. A flicker calculation for individual turbines is not possible because the voltage is not intended to be constant but changes in a wide band in relation to the speed. Therefore, only the flicker produced by the cluster is calculated, over the same period and in the same way as for the CSS-IG wind farm. Comparable values require taking the rated power of both cases into account: the cluster only consists of 3 34 ECN-C- -4-5

35 4 FLICKER EVALUATION turbines while the CSS-IG farm contains turbines. Binned instantaneous flicker values for a CSS-CC cluster are plotted in figure 5. The flicker level of the CSS-CC farm is below the CSS-IG farm level. The variation in flicker level of the CSS-CC farm is small. 4.3 Variable speed pitch controlled wind farm with DFIG 5 DFIG, normal operation, Turbine, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 6: Binned instantaneous flicker values for a DFIG turbine 5 DFIG, normal operation, turb, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 7: Binned instantaneous flicker values for a farm of DFIG turbine In the turbine equipped with doubly fed induction generator, the stator is connected to the grid and the rotor to a converter. The control of the rotor side converter will to a large extent determine the turbine short term dynamics as well as the flicker level. In the simulated DFIG ECN-C

36 Erao II, Volume : Offshore wind farm case studies turbine flux oriented control is applied, which should be able to accurately control stator power and reactive power. The flicker level of this system is expected to be low. This is confirmed by figures 6 and 7. Farm flicker level shows a small reduction with respect to the individual turbine level. 4.4 Variable speed pitch controlled wind farm with PM 5 PM, normal operation, Turbine, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 8: Binned instantaneous flicker values for a PM turbine 5 PM, normal operation, turb, grid angle 3 o Nr. of samples per bin Instantaneous flicker ( ) Figure 9: Binned instantaneous flicker values for a farm of PM turbine The wind turbine with permanent magnet generator incorporates a full converter connected to the stator of the generator. Turbine and grid are largely decoupled and, if well controlled, low 36 ECN-C- -4-5

37 flicker levels result. This is confirmed by the results in figure 8 and 9. 4 FLICKER EVALUATION 4.5 Conclusion Only a limited flicker evaluation could be executed within the framework of the Erao- project. In a relatively short simulation instantaneous flicker values were determined over the complete range of operating conditions for the four types of wind farms. Flicker values of a single turbine were compared to the values of a string of twelve turbines under the same operating conditions and fictitious grid parameters. As could be expected, the directly connected constant speed stall turbine (CSS-IG) generates the highest flicker, while the systems with partial and full converter perform better and similar. As expected, increasing the number of turbines reduces the flicker level. ECN-C

38 5 FREQUENCY DIP RESPONSE OF WIND FARMS 5. Constant speed stall wind farm A frequency dip has been simulated by a change of the frequency setpoint of the synchronous machine which simulates the grid (figure 3). At t= s the setpoint is decreased to 45 Hz and at s it is changed to the normal value of 5 Hz. This change is frequency is far beond any expected change, in magnitude as well as rate, but has been chosen to demonstrate the wind farm behaviour more clearly. In reality, grid frequency deviations take time due to inertia in the electric system and immediately action by grid primary control. 8 Grid 4 plot CSSout 6 Pel (MW) Pcons, (MW) freq (Hz) V exc (V) Figure 3: CSS-IG wind farm: Frequency dip response of the grid The simulated frequency dip results in a voltage dip (V park in figure 3), which is corrected by the synchronous machine voltage controller V exc. The initial voltage dip is of the same magnitude as the freqency dip: about %. The total consumer load P cons and P cons is about MW, which is supplied partly by the synchronous machine and partly by the wind farm. During the frequency dip all turbines operate at an average wind speed V w of m/s and the aerodynamic power P aero is about.5 MW (see figure 3). The aerodynamic power does not change significantly due to the change in frequency. The reactive power consumption oscillates, corresponding to a current peak (figure 3). The slip oscillates as well and the sign of the slip reverses. After a few seconds a new steady state is reached. The turbines have decreased in speed and the slip is back to its original value. The park power (P el in figure 3) strongly oscillated and includes a power reversal, corresponding to the oscillating slip. The frequency and voltage changes also change the power consumption of the grid. The turbine and park currents show large peaks and these could result in wind farm shut down, depending on the settings of the protection equipment. Protection measures are not included in the wind 38 ECN-C- -4-5