7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms Constant Speed Turbines on a Grid with Variable Frequency – A Comparison in Terms of Energy Capture E. Troester Abstract--This paper describes a novel control concept for wind parks with just one central converter to control the frequency of the park grid. To assess this concept a comparison with conventional concepts in terms of energy capture is carried out. Basically this control concept is suitable for every type of constant speed turbine, but it has its biggest advantages in combination with a new type of generator, the permanent magnet induction generator. This type of generator allows to abandon the gearbox and converter on the turbine, thereby reducing the total number of parts and increasing the turbine’s reliability, which is a crucial parameter of offshore wind turbines. The results show that there is only a small reduction in energy yield (less than 4%) for the new control in comparison to the variable speed control. Together with the robust turbine design without gearbox and converter, the new control concept is a good option, particularly with regard to offshore applications. Fig. 1. Direct drive permanent magnet induction generator Index Terms--Constant Speed Turbine, Energy Capture Offshore, Permanent Magnet Induction Machine, Pitch Control, Variable Frequency I. NOMENCLATURE PMIG – Permanent Magnet Induction Generator II. INTRODUCTION Already several thousand megawatts of turbine power are planned on offshore locations near the European coasts. Due to the special conditions at sea there is a challenge to develop new concepts which meet the requirements of offshore wind parks. One of the major necessities is a reliable and low–maintenance system, leading to the objective of reducing the number of parts on the turbine as much as possible. The concept described in the paper uses a permanent magnet induction generator [1], which has no gearbox and no frequency converter, as the generator on the turbine (Fig 1). Thus, the number of parts is reduced and a maintenance- and oil–free system results. Eckehard Troester is with Energynautics GmbH, Mühlstrasse 51, 63225 Langen, Germany (e-mail: E.Troester@energynautics.com). Since there is neither a gearbox nor a converter included in the turbine, this type of turbine is a constant speed turbine when connected directly to the grid. This has the drawback that pitch control cannot be used and a lower energy yield has to be expected, as the rotation speed cannot be adapted to the wind speed, leading to a lower power coefficient. To overcome this drawback, the new control concept uses one central converter to control the park frequency and thereby the rotation speed of the turbine, in order to gain a better power coefficient and hence a higher energy yield. With this configuration, a grid with variable frequency is achieved. The frequency is controlled in such a way that the energy yield is maximized. The converter can either be placed onshore or, when using an HVDC-link, on a platform (Fig 2). Fig. 2. Constant speed turbines with central converter 7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms Both locations can easily be reached in the case of a fault in the converter. In addition, very good compliance to the requirements for grid connection (i.e., power factor, fault ride through etc.) can be achieved. At rated load, active stall and pitch control have been investigated. Active stall is well known for constant speed turbines. To enable pitch control, a special frequency control is introduced, but not without drawbacks: the mechanical power has to be reduced in order to limit the frequency and power dips happen, leading to a reduced energy yield. A comparison in terms of energy capture and location dependence has been carried out between this concept and other known control concepts (constant and variable speed control) using Matlab-Simulink as a simulation tool. The operating mode of the MPP-Controller is shown in Fig 3. Within 0.1 s the whole frequency spectrum is passed and the frequency which causes the highest power output is detected as optimum frequency (labeled with an arrow) and serves as reference input for the frequency controller. III. MODEL The here presented model is based on investigations of Hoffmann [2]. Data which is achieved with Hoffmann’s model is scaled up to a 2 MW wind turbine and adjusted to actual wind power plant performance. The model is advanced to a whole wind park model consisting of 10 turbines, which are controlled with one central frequency converter. Each turbine has a different wind time series, but with the same mean wind speed and turbulence. With a change in turbulence, deviations in the wind time series can be influenced: a high turbulence causes large deviations and smaller turbulence means small deviations. IV. CONTROL CONCEPT A. MPP-Tracking In order to adjust the speed of the turbines to the prevailing wind speed, the optimal frequency (which leads to optimal power output) has to be found. This is realized with MPP Tracking (Maximum Power Point Tracking). Two techniques are known for MPP tracking: The first technique is based on relative comparison. The frequency is changed in one direction in order to increase the power. If the power really increases, the frequency change can continue. In the case of a power reduction the frequency has to be changed in the other direction. The second technique works with absolute values. The whole frequency spectrum is scanned so that the point of maximum power can be detected. The frequency is then adjusted to this optimum. Both methods have benefits and drawbacks. The first technique with relative comparison is easy to implement but it holds the risk of detecting a local maximum only. The second “absolute” technique cannot directly be realized as described, since it is impossible to pass through all frequencies in a real wind park from time to time in order to find the maximum power. However, in order to take advantage of this method it is applied to a wind park model. For a model it is by all means possible to vary the frequency over a wide range in order to find the maximum power. The simulation result can finally be taken as reference input for the real wind park. Fig. 3. MPP-Tracking However, if this reference frequency would be applied to the wind park without a limit in the derivation of the frequency, very large power fluctuation would occur, which could even cause the turbine to work temporarily as a motor. This effect is also known for normal variable speed wind turbines, if the controller is made to adjust always the optimal tip speed ratio in the power optimization mode. Because of this the frequency change is limited, which causes a relatively slow tracing of the frequency to its reference value (Fig 4). The simulation presented here uses a frequency change of 0.1 Hz/s, which was empirically found by evaluation of simulation results. Fig. 4. Reference and actual frequency B. Power Control For power limitation in wind turbines two basic concepts are used: pitch control and stall control. Since the mechanical stresses are significantly reduced in pitch controlled systems, this control strategy prevails for wind turbines with high power output. The principle of stall control is much easier, however components in stall controlled wind turbines have to 7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms be designed much more robustly, which causes higher weights and probably higher prices. Thus the additional expenses for pitch actuators are sufficiently compensated by savings in material. An advancement of the stall control strategy is the active stall concept (active stall is known under artificial names depending on the manufacturer). Here the blades are pitched to larger angle of attack in order to enforce the stall effect. Hence, less powerful actuators can be used and it provides faster control. Like for pitch control the stresses can be significantly reduced for emergency stops and start up. A drawback of pitch control is that it acts comparatively slowly. This causes large power fluctuations due to wind gusts around rated wind speed. Thus a combination of a fixed speed system and power limitation with pitch control is not applied. However, this paper deals with a quasi fixed speed wind turbine concept combined with pitch control. A whole wind park consisting of “normal” fixed speed turbines provides variable speed in changing the frequency of the whole park. With this park-variable concept a change in the frequency allows to realize power limitation with pitch control. If one turbine demands power limitation it is detected and adjusted. This control strategy will be explained in the following paragraph. It is refrained from explaining the active stall concept in detail, because it is an approved solution for fixed speed wind turbines. C. Pitch Control Strategy Pitch control facilitates power limitation by changing the angle of attack. This causes a decrease in the power coefficient and limits the power which is absorbed by the rotor. Modern wind power plants have either mechanic or electrically driven pitch actuators, which are able to turn the blades with 6 up to 10 degrees per second. However, due to the dynamics of the wind this actuator speed is not sufficient to provide a reliable power limitation. Thus the speed of the turbine has to be increased in order to buffer the power temporarily as kinetic power. This indicates that pitch control can only be used for variable speed wind turbines (except for wound rotors with additional rotor resistance). However, the introduced park-variable concept allows operation at variable speed, as the frequency of the wind park can change. This in turn facilitates introducing a pitch controlled system. All wind turbines of the wind park have a common monitoring system. If one wind turbine reaches the power limitation mode, the frequency is increased so that the power is kept constant at rated power. Once the mechanical power can be limited with the pitch actuator the frequency can be reduced to its rated value (50 Hz). The features of this pitch controller can be explained with the help of the scenario shown in Fig. 5. As soon as the mechanical power exceeds the power limit of 2MW (at 67.2 s), the pitch actuator mechanism starts and the pitch angle increases. Due to the slip the electrical power follows with a delay. Once the electrical power reaches rated power, too, the controller starts to increase the frequency (at 67.5 s). At 68.2 s the mechanical power is reduced to its rated value and the pitch angle can be decreased. In the same way the frequency can be reduced to its initial value by keeping the electrical power larger than the mechanical power, so that the buffered kinetic energy can be relieved. At 69.2 s the initial conditions are finally obtained. Fig. 5. Power limitation Since each wind park consists of several wind power plants, two effects appear: 1. In a wind park with a larger number of turbines it might happen that power limitation mode is always actuated by one of the wind turbines. This would result in a steady increase of frequency (Fig 6). However, one can counteract this phenomenon if the mechanical power is limited with the pitch actuator system before rated power is reached. Again, the power can be buffered as kinetic power. 2. All wind turbines in the park have the same frequency. If the park frequency is increased the electrical power output decreases automatically. This is volitional for all turbines in power limitation mode but not for all the other turbines. Here a power dip appears (Fig 7). This is contrary to the aim of smooth power output. It causes unintentional power fluctuations and, in total, a lower energy capture. 7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms Fig. 6. Frequency of 5 and 10 turbines Fig 6 shows the described effect (effect 1): In a wind park with 5 turbines the frequency is temporarily increased but can be led back to 50 Hz. For comparison a wind park with 10 turbines is shown. Here one turbine always causes power limitation mode so that the frequency shows a steady increase without going back to its rated value. In order to counteract this phenomenon the mechanical power is reduced depending on the frequency, which is shown in Fig 7. At around 6 s the mechanical power exceeds its rated value, so that at first the pitch angle increases and later on also the frequency. In order to be able to lead the frequency back to its rated value, the mechanical power is limited slightly below rated power. At 7.3 s this power amounts to 1900 kW. However, the electrical power output is kept at 2 MW, so that the difference between mechanical and electrical power is drawn from rotational power and the frequency can be reduced to its rated value (50 Hz). The smaller the difference between the frequency and its rated value is, the smaller is the difference of mechanical power and rated power. If it were possible to keep the electrical power output constant to rated power, there would be no loss in energy. Fig. 7. Mechanical power reduction The power dip caused by the increased frequency (effect 2) is shown in Fig 8. The power of two wind turbines is compared during a time of increasing frequency. Furthermore the power of the total park, consisting of 10 turbines, is plotted here. At 70.5 s wind turbine 2 reaches power limitation, the frequency is increased in order to keep the electrical power P2 at rated power. This causes in turn a loss of 600 kW of turbine 1. At the same time the total power of the wind park decreases because not only turbine 1 is affected by this incident but also all the 8 other turbines in the wind park. The total wind park suffers from a power loss of 5 MW and decreases to 15 MW. At 72.0 s wind turbine 1 now reaches the power limitation mode. The frequency has to be increased again now, in order to limit P1. And again the total wind park power is reduced to 15 MW. 7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms The following figure illustrates the computed power curves, simulated with a turbulence of 10 %. Fig. 8. Power curve It is noticeable that the constant speed turbines (CST) have less power at partial load, which is due to the lack of adjustment to the prevailing wind speed. Near the transition from partial load to rated power both stall controlled concepts (CST and PCS) show a slightly lower power output than the variable speed concept (VST). However, the PCP concept yields the smallest values under rated conditions. Due to the pitch control (see above) the average power is even below rated power. The next figure shows the energy loss of all concepts compared to the variable speed concept (VST) with a turbulence of 10 % and a Weibull parameter of 2. Fig. 8. Power dip V. ENERGY YIELD In order to compute the energy yield of the different concepts, 5 minute simulations of the wind are performed with mean wind speeds between 3 m/s up to 25 m/s. The power outputs of these simulations serve to determine a power curve. This power curve is finally assessed with a Weibull distribution and the energy capture of one year results. Powercurve 0 4 5 5 0 5 5 6 0 6 5 Energy 0 6 6 4 5 5 0 8 5 5 8 Windspeed 6 0 6 5 4 0 4 5 5 0 5 5 6 0 6 5 9 Power Location Weibull Distribution Fig. 8. Procedure to calculate energy yield Fig. 8. Energy yield in comparison to the variable speed turbine Four different concepts are compared x Constant Speed Turbine (f=50Hz) + Active Stall (CST) x Variable Speed Turbine + Pitch (VST) x Proposed Concept (10 Turbines) + Active Stall (PCS) x Proposed Concept (10 Turbines) + Pitch (PCP) The loss in energy of the constant speed concept (CST) compared to the variable speed concept (VST) decreases with higher average wind speeds. Here the turbines are more often in power limitation mode, where the power output is the same. The PCP concept shows the opposite, because it has a lower power output at higher wind speeds. The PCS concept has an energy loss of less than 4 % and even smaller values for high wind speeds which are found for offshore sites. 7th International Workshop on Large Scale Integration of Wind Power and on Transmission Networks for Offshore Wind Farms VI. CONCLUSIONS A comparison in terms of energy capture and location dependence has been carried out between constant speed turbines on a grid with variable frequency with pitch and active stall and other known control concepts (constant and variable speed control) using Matlab-Simulink as a simulation tool. MPP-tracking control and pitch control have been investigated. The simulations show that pitch control works but drawbacks in terms of energy yield have to be taken into account, so that pitch control cannot be recommended. However, the results for active stall control show that there is only a small reduction in energy yield (less than 4%) for the new concept in comparison to the variable speed control. Together with the robust turbine design without gearbox and converter, the new control concept with active stall control seems to be a good option, particularly with regard to offshore applications. VII. REFERENCES [1] [2] Hagenkort B., Hartkopf T., Binder A., Jöckel S., “Modelling a Direct Drive Permanent Magnet Induction Machine” Proc. ICEM 2000, Helsinki University of Technology, Vol. 3, pp 1495-1499. Hoffmann, R.: "A comparison of control concepts for wind turbines in terms of energy capture", D17 Darmstädter Dissertation, 2002 VIII. BIOGRAPHY Eckehard Troester was born in Marburg in Germany, on December 7, 1975. He holds a Master of Electrical Engineering from Darmstadt University of Technology, Germany. He is currently working on his PhD, which will be finished in 2008. His research focuses on electrical power systems, renewable energies and electrical machines, especially wind power generators. He has worked as a scientific assistant at the Institute of Renewable Energies, Darmstadt. Since 2007 he works for Energynautics.