This action might not be possible to undo. Are you sure you want to continue?
Detailed Modeling for Large Scale Wind Power Installations – A Real Project Case Study
Nayeem R. Ullah, Åke Larsson, Andreas Petersson, and Daniel Karlsson
Abstract--This paper reports on the modeling issues performed related to a feasibility study to investigate the possibilities to connect the 640 MW off-shore wind power farm, planned for Krieger’s Flak 30 km south of Trelleborg (Sweden), to the E.ON 130 kV subtransmission system. The aim of the entire study is to answer the question if such a connection is possible, and under what conditions; it is not meant as a design project. Following a general connection design discussion, the study comprises three major parts, fault current calculations, load flow calculations, and dynamic simulations. Concerning the modeling aspects, much effort has been put on details and scalability for the dynamic simulations. Index Terms--grid code, grid wind power connection, wind power modeling, grid wind power interaction, off-shore, wind power.
reduce the transmission and transformer losses. Therefore it seems to be a good idea to study the possibility to connect the 640 MW from Krieger’s Flak, directly to the E.ON 130 kV subtransmission system. Large off-shore power plants have a general problem to find suitable connection points on land, since the network is normally weak in the costal areas, except for the locations of large thermal plants or large industrial sites. Moreover, an appropriate work procedure of such a grid connection study, with respect to the local conditions, including the recent grid codes, of a large wind farm are hard to find in literature.
RIEGER’S FLAK, located in the Baltic Sea between Sweden, Denmark and Germany, some 30 km south of Trelleborg, see Fig. 1, offers a great opportunity for large scale off-shore wind power generation. Vattenfall has now the formal rights to harness the Swedish part of Krieger’s Flak, and a wind power installation of 640 MW is planned to be grid connected by 2010. Different grid connection alternatives are discussed and studied. The grid connection and the wind power farm properties have to comply with the recent National Grid Code concerning generation plants , formulated by the Swedish National Grid Company (Svenska Kraftnät). In other countries there are similar grid codes [2,3,4], and many transmission system operators publish their grid codes on their website. The fault ride-through capability for two types of wind turbine generators are investigated in . The Swedish National Grid Company has investigated a connection to the 400 kV level . The E.ON 130 kV subtransmission system is closer to the off-shore site, cables and equipment for 130 kV are cheaper than for 400 kV, and to some extent the power is needed on the 130 kV level. The south-west part of Sweden, where the wind power farm will be located is a typical load area, with very little generation. The 130 kV subtransmission system is operated as a meshed distribution system in the area. Introduction of new generation into the 130 kV system should
Krieger’s Flak location south of Sweden.
Nayeem R. Ullah is with Chalmers University, SWEDEN, Åke Larsson is with Vattenfall Power Consultants, SWEDEN, Andreas Petersson, and Daniel Karlsson are with Gothia Power AB, Aschebergsgatan 46, SE-411 33 Göteborg, SWEDEN, (e-mail: email@example.com).
Finding a strong connection point in the network is normally required, since the power input might be considerable. As shown in this study, the 130 kV system is on the limit to be able to take care of the 640 MW power. The initial approach was focused on complying with the grid code, but during the study it became clear that modern DFIG (Doubly-Fed Induction Generators) or full-power converter units are very robust, and will not loose the grid connection, even for backup clearance of nearby faults in the meshed 400 kV system. However the 130 kV system load currents and fault currents turned out to be too large for the present system, and reinforcements are necessary. As long as the voltage for all buses in the meshed 400 kV transmission system, are above the curve in Fig. 2, the wind power farm has to be grid connected. The background for the shape of the curve is that generation must not trip for a three-phase line fault close to a bus, cleared by the backup system (breaker-failure protection).
At Krieger’s Flak. 4 shows the proposed locations of the two off-shore transformer substations and the internal grid of the wind farm. Therefore the work plan was carefully specified before the start up of the project. sea cables will connect to two off-shore transformer substations. each submarine cable will be joined to three single phase cables (1x1000 mm2 aluminum). 2. Even the on-shore cables will be able to carry 800 A / 200 MW at 145 kV. Fig. Two AC cables carrying 340 MW will be connected to the grid point Trelleborg N and two AC cables carrying 300 MW will be connected to the grid point Arrie. The 630 mm2 cable can carry 700 A / 45 MW and the 185 mm2 cable can carry 410 A / 27 MW. high wind power generation. 3. 3) Agreement on the initial conditions for the load flow calculations. see Fig. Off-shore. 5) Relevant faults and abnormal network switching conditions for the study were identified. how a study was performed and what the results were. It is very hard to give a good and general recipe on which studies to perform to insure a robust and efficient grid connection of large off-shore wind power farms. Each cable can carry a maximum current of 800 A which gives a transmission capacity of approximately 200 MW at 145 kV. . for the fault current calculations and for the dynamic simulations were established. and critical faults. high and low load conditions. Then it was continuously reviewed and updated. II. 4. Each off-shore transformer substation accommodates two transformers. three-core submarine cables (3x800 mm2 copper) will be used. But it is always possible to describe afterwards. stage 2010. as well as. WORK PROCEDURE A feasibility study for large off-shore wind power grid connection has never been performed in Sweden before. so no local experience on any relevant work procedure was available. because the conditions are so different from one situation to another. 3. was set up in PSS/E. the criteria for acceptance. Fiber optical cables for the SCADA system will be integrated into the submarine cable. Fig. The wind farm will be connected to the E. Layout of the internal grid at Krieger’s Flak. 7) Dynamic simulations were performed to see the voltage 47 Fig. as well as high and low exchange with neighbor areas. 6) An intermediate report was put together. Two different cross-sections will be used. as a basis for the rest of the study.DRPT2008 6-9 April 2008 Nanjing China 14 11 13 15 12 10 16 P2 17 6 8 18 9 1 7 P1 3 4 2 19 5 20 575 m 900 m Fig. 3x630 mm2 in each radial close to the transformer and 3x185 mm2 at the end of each radial. Grid connection of Krieger’s Flak using four AC sea cables. were established. On land. Swedish grid code requirement for large generation plants. The cables in the internal grid will operate at 36 kV. The grid connection is planned to use four AC sea cables. 2) Modeling of the power system: A power system model for southern Sweden.ON 130 kV grid at Arrie and Trelleborg N. 4) Load-flow calculations covering normal network operational conditions. Basically the work proceeded according to the following main items: 1) Specification and collection of data and other information necessary for the study. and a base load flow case and a base fault current network were established. abnormal switching conditions. The main purpose of this paper is to describe the work procedure and the results of the grid connection study of Krieger’s Flak wind farm. describing the initial load flow conditions.
new combined heat and power generation in Malmö (450 MW). 8) The detailed dynamic wind power model development work was performed by Chalmers University of Technology. and opening of all circuit-breakers of objects connected to the same busbar as the faulty object). A model of a full-power converter system was built up in PSCAD/EMTDC with a control similar to the one presented in . 11) Final dynamic simulations were performed based on 6) and 8) above. but both prototypes in the 5 MW range and mass-produced large wind turbines in the 3– 5 MW range are equipped with either a DFIG generator or a full-power converter. The results from these simulations were then compared to a simplified model of the wind park developed for PSS/E. 12) A draft report was submitted. During the voltage sag the active and reactive power is controlled to be zero. cleared from the breaker-failure protection (250 ms. A.DRPT2008 6-9 April 2008 Nanjing China profile at the point of wind power connection. Three types of variable-speed wind turbines were considered. 9) Fault current calculations were performed based on the intermediate report according to 6) above. There are not very many wind turbines in the 5 MW range available on the market today. DFIG system without a voltage sag ride-through system. and used as an input for the dynamic simulations . and additional land based wind power (200 MW was estimated). All clusters were then connected to the off-shore collection transformer and connected to the grid via a cable. even though there would be a possibility to support the grid with reactive power. transformers and cables) in the simulation package PSCAD/EMTDC (a simulation package utilizing instantaneous values). three-phase line fault. 10) Load flow calculations were performed according to 6) above. collection network. Already in an early stage it was concluded that the DFIG system without a voltage sag ride-through system would not fulfill the requirements in the Swedish grid code. and fault clearing sequences. 6) Minor network changes. This work procedure was specified on before hand. this system was not further considered. since this tool is used by the grid owner for this kind of studies. 3) Data for other new generation in the area. Controlling the rotor currents makes it possible to control the . and detailed comments were collected. DFIG system with a voltage sag ride-through system and a system utilizing a full-power converter system. The results obtained from the detailed model in PSCAD/EMTDC corresponds well with the results obtained with the simplified model in PSS/E. DFIG or fullpower converter units. INITIAL CONDITION DATA AND SETUP The present load flow and fault current network models were updated with known and planned changes up to year 2010. while fractions of a line are also used in fault current calculations. The turbines and one fourth of the wind park were modeled with details (power electronic. 48 5) Shut down of the Barsebäck nuclear power plant. This dimensioning fault is a nearby. 2) Data for unit transformers. The wind farm has to be represented in PSS/E. and connection cables. The wind turbines with a full-power converter system can be controlled fast and can handle voltage sags well . DFIG A Doubly Fed Induction Generator (DFIG) is an induction generator where the rotor currents can be controlled. It was realized that the fault current calculations were more time consuming than expected. and subjected to voltage sags. Hence. This voltage profile was then used by a research group at Chalmers University of Technology for the development of the detailed dynamic PSS/E model of 1/4 of the wind power farm (150/170 MW). The main reason is that more fictive buses are needed in the fault current network to take care of discontinuity points of the lines and mutual zero-sequence coupling between multi-circuit and parallel lines. a final report was assembled and submitted. the discussion and the comments. 13) Based on the draft report. In load flow it is always the characteristics of the entire line that is of interest. Dynamic Wind Power Model The main purpose of the dynamic models was to estimate the response of the wind turbines to symmetrical voltage sags and the model should be used for dynamic simulations in PSS/E. such as Lillgrund wind power farm (110 MW). PSS/E is a phasor based simulation package with a typical time step of 10 ms. step-up transformers. Also the dynamic data file had to be updated. ELECTRICAL SYSTEM IN WIND TURBINES Krieger’s Flak is planned for wind turbines with a rated power of 5 MW. This is considered to be conservative and would not put any “demands” on reactive power production during the voltage sag. For DFIG units the initial short-circuit current is estimated to a maximum of 5 times the rated current  and for the full-power converter alternative the initial short-circuit current is estimated to a maximum of 1. an oral presentation and discussion session was also held. to verify that the grid code was fulfilled for all relevant 400 kV and 130 kV faults. IV. A. The model of the full-power converter system consists of a current source that supplies the dc-link of the converter with power from the turbine and a power electronic inverter connected to the grid. III. and only minor adjustments were made during the work. and reinforcement of other nuclear plants. when a dimensioning fault was applied in the meshed 400 kV system. Each cluster of wind turbines (up to seven turbines in one cluster) in the park were aggregated to one unit. The following power system changes were estimated and agreed on: 1) Data for the generators at Krieger’s Flak.41 times the rated current. It is interesting to note that utilities often have different network models for load flow calculations and for fault current calculations. and that the dynamic simulations were less time consuming than expected. 4) A slight load increase in the area was estimated.
in principle. Fig. the turbine injects reactive power in order to support the grid. an aggregated model of seven wind turbines was developed and compared with simulations of seven single wind turbines with a detailed representation. for instance. the “optimal” control of active and reactive power during a grid fault is beyond the scope of this paper. B. if it is too detailed it will 49 . GE and Gamesa. Due to the high voltages induced in the rotor windings during grid faults. Siemens uses an induction generator and Vensys uses a gearless permanent magnet generator. 7 shows a PSCAD/EMTDC simulation of the response of a single wind turbine with a full-power converter to a voltage sag.e. This results in a variable generator speed and the possibility to control active and reactive power. the turbine handles the sag well. For the lay-out of the system see Fig. This is done using a crow-bar. 6. Full-Power Converter Using a full-power converter connected between the generator and the grid makes it possible for the generator to operate at a frequency separate from the grid frequency. if the power from the turbine is larger than the maximum deliverable power to the grid the excess power is dissipated in the “braking resistor. The grid-side converter is vector controlled and synchronized to the grid through a phase-locked loop (PLL). This means. the result will not be accurate. The currents are controlled by a frequency converter connected to the rotor circuit of the generator. and thereby also the power delivered to the grid. More details of the modeling and control can be found in  and . the converter is infinitively fast in terms of power system dynamics. 6 shows a model of a wind turbine equipped with a fullpower converter. short-circuits leading to voltage sags in the connected grid. This system is used by many wind turbine manufacturers. The generator is modeled as a current source and the converter is modeled with six transistors. The control of the crowbar varies from one manufacturer to another. (i) first. At the dc-link there is also a transistor connected to a “braking resistor” that dissipates power in case of a large voltage sag. i. The DFIG system introduces some additional technical risks compared to the full-power converter system due to the slip rings and the three-winding transformer that are used. The remaining voltage is approximately 5 %. Both slip rings and three-winding transformers have a bad track record in off-shore wind turbines. it is necessary to short-circuit the rotor. It is also important to have the objective of the model in mind when modeling. before the sag. It should be pointed out that since the converter can control active and reactive currents very quickly it is possible to control the converter in numerous ways during a voltage sag.e. Fig. Looking at the technical specifications from different wind turbine manufacturers it seems as DFIGs are more sensitive to variations in voltage and frequency than generators equipped with a full-power converter. The converter has a fast inner current control loop that controls active and reactive currents. DFIG with a converter and an active crowbar on the rotor circuit. In Fig. Fig. Model of the electrical system for a full size converter wind turbine. The wind turbine consists of a current source that supplies the dc-link with the power given by the turbine.95 inductive. 7. 5. the converter is programmed to inject as much reactive power as possible in order to keep up the voltage. V. Moreover. Grid filter Grid Fig. Wind turbine model Fig. take too much computer time to run the simulations and if it is too simple. Enercon uses a gearless synchronous generator. (ii) second. an aggregated model of the wind farm in PSS/E was developed and compared with a more detailed one in PSCAD/EMTDC. 7. the fast transients are not included in the PSS/E model. 6 shows the layout of the system model. Of course. A. The figure also shows that it is possible to control the current from the turbine very quickly. i. 5. The dynamic model derived in this paper will be used for studies of faults. As can be seen in Fig. The aggregation results are similar to the results of the detailed modeling. a detailed model of a wind turbine with fullpower converter was developed in PSCAD/EMTDC. There may be other ways of controlling the converter that can be more or less beneficial for the turbine or the grid. 7. During the simulation in Fig. It is also possible to use different types of generators. the turbine is operated close to rated power with a power factor of 0. DYNAMIC MODEL OF THE WIND FARM WITH FULL-POWER CONVERTER WIND TURBINES When simulating a large wind farm. it is important to estimate the need of complexity of the model. Krieger’s Flak wind farm has been simulated in three steps.DRPT2008 6-9 April 2008 Nanjing China rotational speed. that slow wind variations are of minor importance since short-circuits are cleared relatively fast. The rated power of the converter is proportional to the speed variation.” The main task of the gridside converters is to control the dc-link voltage. which means that. Pt . such as Vestas. However. and (iii) finally. vdc .
In order to verify the aggregated wind turbine model. however. if the turbines are operated at different power levels the results from the aggregated turbine and the cluster would produce basically the same results. sCphi sUdc PLATTFORM 1 2 3 1. 1 2 3 1.0 s WT6 ac #2 #1 COUPLED PI SECTION s Cphi sUdc 1 2 3 1. Responses to a voltage dip for a full size converter wind turbine. a weighted mean value of the cable length was used.0 s WT5 ac #2 #1 COUPLED PI SECTION s Cphi sUdc 1 2 3 1. all turbines are operated at rated power. This is a large system that demands a lot of computer capacity. An EMTDC Model of the electrical system of a cluster with seven full-size converter wind turbines.0 s WT4 ac #2 #1 COUPLED PI SECTION s Cphi Fig.0 s WT7 ac #2 #1 COUPLED PI SECTION s Cphi sUdc 1 2 3 1.0 s WT1 ac #2 #1 COUPLED PI SECTION PLATTFORM #1 #2 Fig. the per-unit values of the turbine and transformer were kept constant. Moreover. 7. a simulation of the response of the aggregated model and the cluster with seven turbines to a voltage sag is shown in Fig. 9. 8. 8 shows an example of a cluster containing seven wind turbine models and the equivalent aggregated wind turbine model is shown in Fig.DRPT2008 6-9 April 2008 Nanjing China s Cphi sUdc 1 2 3 1. while the cable impedance per meter was not changed. Aggregation of a cluster One fourth of the wind park consists of 32 wind turbines. As seen in the figure. Fig.0 s WT1 ac #2 #1 COUPLED #1 PI SECTION #2 Fig. An aggregated EMTDC model of seven wind turbines. 10. When aggregating a cluster. In this simulation. 50 . In order to decrease the need for computer capacity each wind turbine cluster must be aggregated to one equivalent wind turbine with a transformer and a cable.0 s WT3 ac #2 #1 COUPLED PI SECTION s Cphi sUdc WT2 s ac 1 2 3 #2 #1 1.0 s Cphi sUdc COUPLED PI SECTION 1 2 3 1. the two models produced close to identical results. 9. including transformers and cables. sUdc B.
all clusters will be replaced with the equivalent aggregated wind turbine.0 1. However.DRPT2008 6-9 April 2008 Nanjing China fault 1. 12 was approximately 0%. Note that during the sag. The voltage sag was accomplished with a three-phase short-circuit (with impedance) at the connection point. Before the fault. The response of the 32 wind turbines to a voltage sag is shown in Fig.0 fault fault 1. 11. In the following text. 12. 12. red line aggregated model. The response to a voltage dip at rated power operation of seven wind turbines. where reactive current was injected in order to support the grid. i. Due to this approach. 11 shows the equivalent PSCAD/EMTDC model of one fourth of the wind park (160 MW). These oscillations were due to the excitation of the (long) cable when the voltage returned after the sag.95 inductive. 7.e. Blue line is the response at the point of connection to the on-shore grid. The response to a voltage sag at rated power operation of 32 wind turbines. in contrast to the simulations presented in Fig. This was done in order to model the wind power farm response to the sag more conservatively. the wind turbines of the wind farm operated at nominal power and at a power factor of 0. red line is the low voltage side of the platform transformer. Fig. This means that the total numbers of turbines is 32 with a total rated power of 160 MW. Blue line seven wind turbine model. The model of the wind farm with five aggregated clusters and the on-shore connection.0 Fig. the active and reactive power was controlled down to zero. Fig. Two clusters correspond to seven turbines and three clusters correspond to six wind turbines. Modeling of 32 wind turbines The model of one fourth of the Krieger’s flak wind park consists of five aggregated wind turbine clusters. (aggregated model). C. The figure also shows that there were large oscillations in the voltage after the sag. the results from the analysis will be on the safe side when applying for grid connection. The remaining voltage in Fig. 12 shows that the wind turbines responded as expected. the wind turbine was programmed to not inject any active or reactive current. Fig. 51 . the objective of this work is not to study these fast transients and therefore this phenomenon is considered for future work. 10. Fig.
aggregated models. Fault currents are calculated for three-phase. VII. during normal operation conditions. For high voltage components the following short-circuit current criteria have to be checked: 1) The interrupting capacity for circuit-breakers: The circuitbreaker must be able to break the short-circuit current.ON fault current network was used. This means that the earth fault current might exceed the three-phase fault current. 3) The initial short-circuit current: The mechanical stress on the component. with appropriate updates and modifications. The transformer zero sequence impedance was set equal to the positive sequence impedance.000 EURO. 14. 17 circuit-breakers have to be able to interrupt short-circuit currents. A very rough cost estimate for a 130 kV. at the most unfavorable voltage phase angle position.DRPT2008 6-9 April 2008 Nanjing China D. the fault currents exceed the limits in some cases. with no additional impedance in the neutral. and all 130 kV windings on new transformers. 50 kA circuit-breaker is 100. as well as coming small sized land based wind power (this is regarded as the worst case). FAULT CURRENT CALCULATIONS Fault current calculations were performed with PSS/E to check if the connection of the wind power farm would result in fault currents exceeding the design criterion for any component in the nearby region. and phase-phase-earth metallic faults. With a number of new power plants to be connected in the nearby region. The current references were low-pass filtered with a first-order filter in order to attain a “smoother” current source. A more detailed dimensioning check is recommended. The most critical parts of the 130 kV regional network is shown in Fig. due to the connection of the wind power farm. The earth fault current can be reduced with a low impedance reactor in some transformer neutrals.20. Most equipment in the area around the connection points. seen from the 130 kV side. The 130 kV subtransmission system is effectively earthed. phase-phase. . The short-circuit currents in the area around the planned connection points will exceed the dimensioning values for a number of components. which appear immediately after a short-circuit. In the most exposed substation (Sege in Fig. The current source injected the active and reactive currents. resulting in the short-circuit current flow through the breaker. New thermal generator xd’’ was estimated to 0. operated at close to nominal power and at a power factor of 0. Arrie and Trelleborg N. active and reactive power both at the platform and at the point of connection. in the study. The reference values in the PSS/E model were low-pass filtered with a bandwidth corresponding to a 10–90% rise time of 50 ms. The current references were given by the desired active and the reactive power and were lowpass filtered before they were injected into the grid. The RMS values are calculated with a smoothing time constant of 10 ms. again. Fig. the earth fault currents will be high. Fig. POWER FLOW CALCULATIONS Power flow calculations are performed to check if any components might be overloaded. have been modeled as directly earthed. A table showing the results is presented in Appendix I. 4) The size and duration of the short-circuit current: The thermal stress on the high voltage component exposed to a short-circuit current is determined from the size and duration of the fault current. Modeling of 32 wind turbines with PSS/E In PSS/E. For phase-phase-earth faults both the maximum phase current and the maximum earth current are noted. is determined from 52 the maximum currents. due to the current forces. The turbines were.95 inductive. 13. will be the recommended strategy for a more detailed analysis and system upgrade. Fig. the aggregated 32 wind turbines were modeled as a controlled current source. It is found that the maximum current flows in the earth connection for phase-phase-earth faults. Since all additional 130 kV transformers have been modeled as directly earthed. A combination of efforts to reduce the earth fault currents and reinforcements to increase the equipment ability to withstand the short-circuit currents. 2) The making capacity for circuit-breakers: The circuitbreaker must be able to switch-on to a fault. as well as for reduced transmission capacity network situations. phase-earth. The figure shows the voltage. Comparison of simulations with an EMTDC model (red line) and a PSS/E model (blue line) of a 32 wind turbine wind farm. are designed for 40 kA fault current. The E. 13 shows a comparison between the PSS/E model and the model developed in PSCAD/EMTDC. VI. 13 shows that the responses of the PSS/E model and the PSCAD/EMTDC model are basically the same for a fault in the grid leading to a voltage sag. 14). The circuit-breaker is supposed to be the most expensive component in the bay.
The reactive power balance and voltage control principles have also to be reviewed. have to be reviewed.0 max exp.6 34. To take care of the reactive power control in the load flow studies voltage controlled mechanically switched shunt reactors were added in Arrie and Trelleborg N (2x50 Mvar and 2x35 Mvar. The critical parameter for the maximal loading is the distance to ground (h1<h2). The most critical fault is a nearby 400 kV three-phase line fault in the meshed system.0 37. The network losses are affected by the wind power generation.3 max exp.ON regional 130 kV network are shown in Table I. without changing the wire or the tower. and they are not allowed to consume any reactive power during high load conditions. VIII. load and wind variations over.8 57. different exchange levels.0 49. the overload on the remaining system will be considerable. during low load and high wind power generation. A more thorough investigation of the wind power generation impact on the losses in different parts of the system is recommended. for different load levels. The wind power connections are not allowed to supply any reactive power into the grid during low load conditions. and high wind power generation. such as lines and transformers. however. 79. due to the wind power generation. the losses in the regional grids might increase. It is reasonable to believe that a combination of reinforcements of line capacity and wind power generation restrictions during outages will be the best way to avoid overload.DRPT2008 6-9 April 2008 Nanjing China Fig. 15. The voltage levels in the 130 kV system does not seem to be problematic.4 83. if no actions are taken. The 130 kV network around Arrie and Trelleborg N. DYNAMIC SIMULATIONS Dynamic simulations were performed to make sure that the wind power farm remains connected to the grid for faults specified in the grid code issued by the national grid operator. as traditional synchronous generators. In all other cases the network losses are increased. Such an upgrade will remove all normal switching stage power transfer limitations. In such a study. tapchanger controlled shunt reactor will roughly cost around 1 million EURO. Low load. Although south of Sweden is a typical load center and generation in the load center relieves the 400 kV transmission grid. This has not happened before. Exchange of the insulator chain arrangement to increase the distance to ground. By the change of arrangement this distance can be increased. The capacity of the lines Sege-Arrie-StjärneholmTrelleborg N 1) and 2) can be increased by exchange of the insulator chain arrangement. The results from the load flow calculations show that some overhead lines will be overloaded during normal operational conditions. a year should be taken into account. This kind of wind power units does not go “out-of-step”. The total losses in the E.9 max imp. 93. Table I shows that the regional network losses are reduced for medium wind power generation at high load conditions and a high export level. respectively). and thereby the loadability. During low load conditions and high wind power generation. and the tap-changer controls together with the routines for reactive power control. The critical parameter is the minimum line conductor distance to ground. The total loss increase can be as much as 30 MW. The cost is roughly estimated to 800. 15.1 87. and different levels of wind power generation. Svenska Kraftnät. for such a power flow direction. A more smooth .3 83. From a power flow point of view it is possible to connect the Krieger’s Flak wind power generation to the 130 kV substations in Arrie and Trelleborg N. During forced and planned outages of network components. voltage control will be achieved with a thyristor or tapchanger controlled reactor. cleared by the breaker-failure 53 Fig. 55. 100 Mvar. Since modern doubly-fed induction generators or full-power converters are presumed for the wind power farm. Table I: Network losses and wind power generation [MW] 100% WP 40% WP 0% WP High load. the power flows from the 130 kV subtransmission system up to the 400 kV level. there will not be any problems with the fault ridethrough capability.6 max imp. Such a 145 kV. overloaded elements have been identified and a combination of reinforcements and wind power generation restrictions is proposed.0 80. High load. A more detailed study on the effect of such generation restrictions is recommended to find the suitable combination of restrictions and reinforcements. according to Fig.000 EURO. The most critical lines are Sege-Arrie-Stjärneholm-Trelleborg N 1) and 2). 99. at least. Low load. 14.
Stockholm. Dahlgren. preferably in combination with measures to reduce the currents. Gothia Power. ACKNOWLEDGMENT The authors of this paper want to thank the colleagues at Vattenfall. FUTURE WORK Throughout this paper the need for more detailed work has been proposed concerning: 1) Quantification and specification of reinforcements to handle a) the increased fault currents and b) the load currents. Voltage step-changes due to load rejection. Sweden. The fault analysis was more time consuming than expected and the dynamic simulation part of the study was less time consuming than expected.  W. “Recorded Fault Ride-Through Capability for Two Types of Wind Turbine Generators”. Trip of 1/4 of Krieger’s Flak. Åke Larsson and Richard Larsson. Billund. There is a study going on. stated by the National Grid Operator. are required. Jan-Åke Pettersson. Although the total losses in the system decrease.” Svenska Kraftnät. specified the whole project and contributed throughout the work with valuable material. Svenska Kraftnät Regulation. Ola Carlson and Nayeem Ullah. for the worst case N-1 fault. the losses in the local subtransmission system might increase considerably. “Analysis of requirements in selected Grid Codes”.  S. May 2004. Johnsen. Beside Krieger’s Flak. is critical for the power quality with respect to other customers connected to the same busbar.ON. which is considered as most acceptable. Fig. to the 130 kV network in the southwest of Sweden. comments and corrections. Rep. N-328. In the design phase of the wind power farm. was the coordinator for the different parts of the project. It was concluded that DFIG and full-power converter wind power units easily could comply with the grid code. D. and Ulf Lager. due to the wind power farm installation. for valuable help and a fruitful co-operation in a nice and friendly atmosphere. corresponding to 170 MW connected to Trelleborg N. 16.  UCTE Position Paper on “Integrating wind power in the European power systems – prerequisites for successful anf organic growth. REFERENCES  Affärsverket svenska kraftnäts föreskrifter och allmänna råd om driftsäkerhetsteknisk utformning av produktionsanläggningar. it has to be decided if the wind power units should produce reactive power. are shown to be very moderate. Denmark. Chalmers. especially concerning situations with power injection to the 400 kV grid. All contributions are gratefully acknowledged. cleared by the ordinary fault clearing system have been investigated. results in a 2% voltage change on the 130 kV bus. Å. Åke Juntti.  E. CONCLUSION The study reported in this paper has shown that it is possible to connect such a large off-shore wind power farm as Krieger’s Flak. of International Workshop on Largescale Integration of Wind Power and Transmission Networks for Offshore Wind Farms. due to the wind power generation. due to outages. Juntti. XI. Project Report. Sep. see Fig. Peter Ellerth. planned for 1000 MW. Göran Loman. active power (P) and reactive power (Q) at the wind power farm connection point for the most severe fault. investigating around 10 different land connection alternatives.T. to support the grid during faults. Voltage (V). Christiansen. X. 2006.” in Proc. there are many large potential offshore wind power generation sites under consideration.  A. Samuelsson. Also three-phase faults in the meshed 130 kV system. Martin Randrup. one of the largest. even though the voltage on the connected 130 kV bus will be as low as 7% of the voltage immediately before the fault occurrence. O. “Grid requirements challenges for wind turbines. and Sture Lindahl. SwedPower. The initial fault current from a DFIG unit is approximately four times the initial fault current from a full-power converter unit. 795/2003/AN40.DRPT2008 6-9 April 2008 Nanjing China protection. 2) Review of the reactive power balance and the voltage control principles. C. New Wind Power Conference. Dec. Bolik. Johnny Malmgren. 2005 (in Swedish). 2003. although minor reinforcements and some generation restrictions. ”Nätanalys för anslutning av Kriegers Flak. During the fault the wind power units reduce their active and reactive power output to zero. 2005 (in Swedish) . SvKFS 2005:2. Klippel. Helsinki. around the coasts of southern Sweden. 54 . Tomas Johannesson. The voltage step-change after the most severe trip of wind power generation. Norgren. and E. 3) A quantification and evaluation of the change in losses in different parts of the network. IX. XII. Among these is “Södra Midsjöbanken”. according to the N-1 criterion. 2006. 640 MW. ABEL consult. 16. Ørsted DTU. May 22-23. Tech.
D.25 Öresundsverket II 23.9 11.20 (gives for the DFIG turbine the maximum estimated fault current of 5 times the rated current) for Krieger’s Flak Fault type 3-phase 1-phase 2-phase 2-phase-earth [kA] X0/X+ [kA] [kA] [kA] Iph-max Ij Fault location\Stage 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 Sege 36. Sweden. no. Power Eng.8 10.5 44.0 28.72 1. [Online]. Dec.9 17.2 7.4 9.2 1.0 0.1 10.  A.7 14.3 20. 2005 (in Swedish).5 14.0 7.91 0.0 16.8 7.4 15.30 Bunkeflo 15.04 1. Thiringer.9 1.9 17.93 1.2 34.0 0.97 Svedala 17.1 8.76 1.84 1.publ/Abstracts/20 05/AndreasPhD.2 9.5 8.8 11.  A. L. APPENDIX I – FAULT CURRENTS IN DIFFERENT SUBSTATIONS Fault Current at maximum short-circuit capacity.4 21.02 Arrie 30.9 7.86 Käglinge 20.2 19.6 12.9 8.6 35.58 Vellinge 14.0 1.7 10. ”Modellering av vindkraftparken Kriegers Flak – för felfallsstudier i överliggande nät.8 24.9 10.7 13.4 26.0 2.pdf  R.1 11.4 11.7 22.0 8.07 Svedala 17.4 36.7 20.8 12.1 39.9 16.99 1.09 1.90 Trelleborg N 21.8 12.0 17.36 Stjärneholm 13.0 7.8 9.1 15.7 20.6 11.6 44.6 11.6 18.1 13.0 16.2 21.5 20. 2006.0 11.7 1. Energy Conversion.7 8. Ullah.” Div.6 11...8 31.0 32.06 1.9 29. A. “Analysis.7 14.1 11.3 18.9 13.5 16.0 5.02 1.2 19.1 18.00 Käglinge 21.70 1.8 14.9 28.4 10.1 13.7 8.0 11..1 16. Ottersten. Göteborg.8 0.3 12.81 Arrie 28.1 11.2 11.7 16.0 21.5 20. Tech.9 9.5 33.8 18. Available: http://www.24 1.5 0.4 1.95 1.3 6.2 18.8 13. pp.2 18.1 12.3 37. Petersson.9 7.20 Stjärneholm 12.9 22. Pietiläinen. Petersson and K.2 16.6 11.2 6.7 13.0 9.6 29.5 30.5 11.5 0.39 1.3 39.2 24.5 10.97 1.2 10.6 1.1 31.5 22.2 1.0 10.4 17.2 14.8 26.2 10.1 6.7 37.5 15.8 36.87 1.2 32. “Voltage sag response of PWM inverters for variable-speed wind turbines.0 12.2 26.8 1.2 12. “Modeling and experimental verification of Grid Interaction of a DFIG wind turbine.53 Öresundsverket II 26. of Technol.7 0.32 55 .. T.79 Trelleborg N 20.2 11. Modeling and Control of Doubly-Fed Induction Generators for Wind Turbines.5 7. of Technol.. pp.7 36.1 27.9 18.3 10. of Elect.54 Öresunsdverket I 36.1 16.75 0.4 0.3 10. Dec.5 9.5 18. 4.5 12.6 44.elteknik. Göteborg.0 29.3 15.1 9.5 12.” Ph.4 18.1 25.20 Bunkeflo 14.8 20.8 13.61 Fault Current at normal short-circuit capacity.6 0.8 17. Rep.6 11.3 28. vol.3 2.2 8.7 14. 1.5 18. Petru.5 23.1 14.1 29. 878–886. Chalmers Univ.04 2.9 29. and xd’’=0.5 21. Dept.7 7. Feb.8 11.6 9.2 25.3 8.47 Öresunsdverket I 32. Petersson.2 13.0 17.8 22. 20.6 22.7 0. 16.20 (gives for the DFIG turbine the maximum fault current of 5 times the rated current) for Krieger’s Flak Fault type 3-phase 1-phase 2-phase 2-phase-earth [kA] X0/X+ [kA] [kA] [kA] Iph-max Ij Fault location\Stage 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 2010 2005 Sege 42.DRPT2008 6-9 April 2008 Nanjing China  A.2 1.11 1.0 12.” IEEE Trans. vol.9 6.chalmers.se/Publikationer/EMKE.8 9. 2006.2 5.7 26.05 1.8 22. 6–14.7 7.0 25. and xd’’=0.73 Östra Klagstorp 11.1 20.2 2.8 14.2 22.2 19.8 18.5 8. Chalmers Univ.6 10.5 9.8 6.1 17.5 16.4 13.0 47.1 7.4 9.2 21. Harnefors and T.” EPE Journal. 2005.5 29.8 13.83 Östra Klagstorp 12.8 6. no. R.41 1.1 11. dissertation. of Energy and Environ.38 1. Petersson and N.74 Vellinge 15.1 2.3 7.3 15.2 6.8 25.5 14.2 7.1 9.5 7.6 14.1 11.
He received the M.. M’06) was born in 1974 in Halmstad. Sweden. From 1997 to 2000 he worked as a system engineer at Kockums AB. Sweden. towards a Ph. Daniel Karlsson (M’1991. Sweden. degree. Currently. His work has comprised theoretical investigations at academic level. Licentiate. regulatory requirements. Göteborg. Malmö. Karlsson is a member of Cigré and a senior member of IEEE. Göteborg. Karlsson hold a position as Application Senior Specialist at ABB Automation Technology Products. His area of research interest includes analysis. on-load tap-changers and generator reactive power limitations. where he is currently working towards the Ph. respectively. measurements and evaluation. Sweden. He is now working as a consultant with Gothia Power AB. Between 2000 and 2005 he worked as a research assistant at the Department of Electric Power Engineering.D. in 2002. degree and Licentiate in electric power engineering from Chalmers University of Technology.Sc. He serves as a reader at Chalmers University of Technology and he has supervised a number of diploma-workers and Ph. Through the years he has been active in several Cigré and IEEE working groups. in 1997.Sc. degree in electrical and electronic engineering from Bangladesh University of Engineering and Technology (BUET).D. and Ph. degrees in electrical engineering from Chalmers University of Technology. 56 . SM’1997) is a principal engineer at Gothia Power. He received his Ph. power quality. Sweden. in 2004 and 2006. Chalmers University of Technology. and the M. He is also Vattenfall’s project manager for the electrical system of Lillgrund wind power plant (a 110 MW offshore wind farm located in Öresund which is under construction). His research focused on the power quality of wind turbines. with the auxiliary supply and degaussing system on submarines. Between 1985 and April 1999 he worked as an analysis engineer at the Power System Analysis Group within the Operation Department of the Sydkraft utility. His work has been in the protection and power system analysis area and the research has been on voltage stability and collapse phenomena with emphasis on the influence of loads. Sweden. Andreas Petersson (S’03. he works for Vattenfall Power Consultants with design of electrical systems for wind farms.D. grid design. Dr.Sc. Åke Larsson recieved in 2000 his Ph. From 1994 until he left Sydkraft in 1999 he was appointed Power System Expert and promoted Chief Engineer. Dhaka. His current research interests include grid integration of wind energy converters. modeling and control of wind turbines. Göteborg. D students at Swedish universities. respectively.D. D in Electrical Engineering from Chalmers University in Sweden 1992. from Chalmers University of Technology. as well as extensive field measurements in power systems. He has broad experience in wind power.DRPT2008 6-9 April 2008 Nanjing China XIII. 2003 and 2005. Most recently Dr. degree with the Division of Electric Power Engineering. Bangladesh. BIOGRAPHIES Nayeem Rahmat Ullah (S'04) received the B.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.