Chapter 10. Wind Turbine Control [DRAFT TEXT CBPRICE 4-12-11] [REVISED 09-12-11] [REVISED 11-12-11] Introduction Energy generation by wind turbines seems to be a popular alternative source of energy for the future though there continues to be a debate about this. At the moment Denmark produces around 20% of its energy from wind turbines, which is actually the agreed target for the EU as a whole by 2020. The most common form of turbines are horizontal axis such as the CART31 experimental turbine shown in Figure 1. This shows the Figure 1. Figure 2. main components: the tower, the rotor and the nacelle. The latter houses the low-speed shaft connected to the rotor, an up-speeding gearbox connected via a high-speed shaft to the electricity generator. This is illustrated in Figure2. The available wind energy is partially captured by the rotor where it appears as kinetic energy of rotation which then is converted to electrical energy. There is also a yaw motor and gearbox which points the rotor into the wind, and many turbine designs have pitch rotators which control the angle of the rotor blades. Wind turbines may be designed to be of fixed or variable pitch. The latter while are more expensive are becoming the preferred design for larger machines as will be explained below Wind turbines may be designed to be fixed speed or variable speed (of rotation). Fixed speed turbines are easier to interface with the electrical grid, since the grid distributes electricity at a particular frequency of 50 Hertz which translates to a fixed speed of turbine rotation. However, variable speed turbines are able to extract around 2.3% more energy from the wind and are the design preferred by the wind industry, despite 1 Details obtained from the National Renewable Energy Laboratory (Colorado, US) and from personal conversations with Ervin Bossanyi, Principal Engineer, Turbine Engineering, GL Garrad Hassan, Bristol UK. requiring additional interface electronics to match the variable generator speed with the fixed grid frequency of 50 Hertz. Figure 3 (left) shows a graph of the power generated by the CART3 turbine as a function of wind speed experienced. There are three clear “regions” of operation. In Region 1, the wind speed is below 2.5 m/s and at these speeds the power generated is less than the loss of power in the machinery, so the turbine is not allowed to rotate. In Region 2 where the wind speed is between 2.5 and 11.7 m/s the variable speed turbine can capture more wind power than the fixed speed turbine. Finally region 3 is associated with wind speeds from 11.7 to 20 m/s and here the power absorbed and generated is limited to the operating point to ensure safe mechanical and electrical load limits. If the wind speed should exceed 20 m/s then the control systems may not cope, so the turbine is shut down. Figure 3 (right) shows two curves, the maximum power available for extraction from the wind and the actual power which can be extracted for various wind speeds for the CART3 turbine. The latter is less than the theoretical maximum, the reasons for this difference are explored below. The maximum amount of power which can be absorbed from the wind is not 100% of the available wind power, rather it is only 59% as shown by Betz [cite]. Imagine if the turbine absorbed 100% of the wind energy, that would mean that the wind leaving the rotor would have no kinetic energy, is it would stop! This would prevent any more wind from passing through the rotor, so this can’t happen. Let’s start the maths by considering the power available in moving wind. Look at the volume of wind passing through a rotor in a certain time βt (Figure 4). This volume is Aβl where βl is the length of the wind passing through in this time. The mass of the air is mass = density x volume, i.e., ππ΄βπ and since kinetic energy is ½ x mass x speed squared, the kinetic energy in this volume is 1 2 ππ΄βππ£ 2 but since power is energy transferred in time, for time βt this transfer becomes 1 2 βπ ππ΄ βπ‘ π£ 2 which is simply 1 2 ππ΄π£ 3 The power captured by the turbine is less than this, and depends on the individual rotor design. The efficiency of power capture is known as the turbine’s power coefficient πΆπ which is established by measurement, so we finally have the following expression for the power captured 1 π = 2 πππ 2 πΆπ π£ 3 (1) where R is the radius of the rotor. We shall come back to this expression later. We will need an expression for the torque exerted on the rotor by the wind; this is obtained by dividing the power by the rotational speed giving us 1 1 ππ = 2 πππ 2 πΆπ π£ 3 π (2) But please remember that the power coefficient πΆπ is not a fixed number, rather it is a function of wind speed and turbine rotation speed. The Turbine Controller The controller’s job is to keep the turbine on its power curve (Figure 3). In Region 2, the controller must ensure that the turbine extracts the maximum power from the wind given that the turbine’s rotational speed may vary. In Region 3, the controller must ensure that the turbine extracts a fixed amount of power, even though the available wind power may exceed this design limit. Two typical control “loops” are illustrated in Figure 5. The “Torque” controller is used in Region 2 and it adjusts the turbine rotational speed π to track the wind speed v to maximize the extracted power. The “Pitch” controller is used in Region 3 and changes the pitch of the rotor blades, so that a constant amount of power is extracted from the wind as the available power increases with increasing wind speed. The Torque Controller When the wind is interacting with the rotor blades there are two speeds which are important. The first is the speed of the wind and the second is the speed of the tip of the rotor blade. These are shown on Figure 5A. Let’s say that the wind is blowing with a certain speed. If the rotor tip speed is larger than this then the rotor is seen by the wind as a solid disk of obstruction since in each interval of time, the rotor is making many revolutions while the wind is moving a small distance through it. The rotor is clearly not extracting an optimal amount of power from the wind. On the other hand if the tip speed is small, then the rotor allows too much wind to pass without having an impact on the turbine and again there is a sub-optimal extraction of power. Clearly there has to be an optimal relationship between wind speed and rotor tip speed. Each rotor design has its own optimal relationship, and this is described as a power coefficient curve which is obtained by experiment. Figure 5A about here. This curve is a function of the ratio of the rotor tip speed and the wind speed. This ratio is called the “tip speed ratio” π and is defined as π= ππ (3) π£ This curve shows how the power coefficient πΆπ but varies with π. This is plotted in Figure 6 for the CART3 turbine. The power coefficient tells you the fraction of the power which can be extracted from the wind for various values of tip speed ratio. πΆπ Figure 6. π You can see here that the wind turbine operates most efficiently when the tip speed ratio is around 6. In this case, πΆπ is around 0.45, which means that 45% of the wind energy is being absorbed by this machine. This compares favourably with the Betz limit of 59% mentioned above. So the turbine control systems should adjust the tip speed ratio to be the optimal value of 6, for all operating conditions. We can use the definition of π to simplify the expression for the torque on the rotor produced by the wind (see above) to 1 πΆπ ππ = 2 πππ 5 π3 π2 (4) Of course the rotor experiences an opposite torque provided by the generator, since the generator is extracting energy from the rotor. Let’s call this torque ππΊ . It is the difference between these two torques that causes the rotor to change its rotational velocity so that it can track the changing wind velocity when it is in Region 2. We use this difference to control the rotor speed to maximize the actual turbine πΆπ . Also glance back to Figure 2 where you will notice the torque sensors on the high speed and low speed shafts. So we can write down the equation for the rate of change of rotational velocity of the rotor as ππ ππ‘ 1 = π½ (ππ − ππΊ ) (5) where J is the moment of inertia of the rotating system. This expression is the rotational equivalent of Newton’s second law π = πΉ ⁄π for translational motion. You can see that when ππΊ = ππΊπ then the right hand side of this expression is zero, so the rotational velocity does not increase and the rotor and wind speeds are matched. Now the question is how to construct an expression for ππΊ which we can then use to control the generator torque which will achieve an optimal extraction of wind power. This occurs at the maximum value of πΆπ . We need to construct an expression which when subtracted from expression (4) will give us zero. Clearly we need ππΊ to be proportional to π2 since it must work for all values of the rotational velocity. So we have ππΊ = πΎπ2 and now all we have to do is choose the value of K. Well, remember this expression must look like expression (4), so it is easy to see that there is only one possible choice for K and this is 1 πΎ = 2 πππ 5 πΆππππ₯ π3πππ‘ where πΆππππ₯ is the maximum value of the turbine coefficient and ππππ‘ is the associated “optimal” value of the wind tip ratio. In this case expression (5) becomes ππ ππ‘ 1 πΆπ = 2π½ πππ 5 π2 (π3 − πΆππππ₯ π3πππ‘ ) (6) You can clearly see that when the turbine is working most efficiently where πΆπ will have the maximum value πΆππππ₯ and π will have its optimum value ππππ‘ then the bracket in (6) will be zero and the rotational velocity will not change. The Pitch Controller [future] Coding the Torque Controller omega theta tsr power Cp Wind velocity, set by console command or else Rotational (angular) velocity of the rotor Angle of the rotor at any given time Tip speed ratio Power generated Power coefficient. Fraction of available G yep G G G G G yep yep yep yep yep Default Value Declared? π π π P πΆπ Global or Local windV Meaning v Parameter Variable Math Symbol The class provided contains a number of variables which have been declared and others which have not. These are listed in the table below. You should use the declared variables in your code, and invent names for those which have not been declared. Default values are also shown. ππ ππΊ J R π πΆππππ₯ ππππ‘ torqueW torqueG error J R rho beta CpMax tsrOptim vCutIn vRated vCutOut omegaInit windVInit wind power absorbed Torque on the rotor from the wind Torque on the rotor from the generator Difference between the wind and the generator torques Rotational inertia of the turbine Blade radius Density of air at the rotor height Pitch angle of the rotor blades (not used) Maximum value of the power coefficient for this turbine Value of the tip speed ratio corresponding to the maximum value of the power coefficient L L G no no yep G G G G G yep yep yep yep yep G yep 5.9 Wind speed at which the turbine becomes operational Wind speed at the end of Region 2. Wind speed when the turbine is halted. G yep 3.1 G G yep 11.7 yep 20 Initial rotational velocity in the simulation Initial wind speed in the simulation. G yep 4.0 G yep 8.0 The code template provided contains the following lines: (1) To limit the simulation to Region 2 if(bLimitToRegionII) { if(windV < vCutIn) windV = 0.0; if(windV > vRated) windV = vRated; } (2) To calculate Cp via a look-up table Cp = lookupPowerCoefficient(tsr,beta); (3) To calculate the generated power power = torqueG*omega; You must insert additional code to compute the following: torqueW = torqueG = error = omega += 644877 20 0.98 0.064577 0.4528 Modelling the Wind Speed To conduct a simulation of a wind turbine response to varying wind speeds we could take two approaches. First we could record actual wind speed changes over time and use this actual data as an input into our simulation. This record will provide us with a distribution of wind speeds over time. While this may sound acceptable, it is not generalizable, ie it uses just one data set and does not allow us to explore various wind speed distributions. To conduct an investigation for various wind speed distributions, we need a model of a generic wind speed distribution. This is the second approach. Let’s reflect on our experience of wind speed. We can surely agree on two factors. First wind does not have a steady speed, but it changes with time, we have gusts. Second, there are strong winds and weaker winds (both with gusts), so we experience average wind speeds, which relate to the Beaufort scale. Therefore any model of wind speed needs to take these two factors into account; the model should be able to represent an average wind speed, but also to represent gusts. Since wind is complex and non-deterministic any model must be stochastic and therefore give us information about the probability (or chance) of experiencing any particular speed. One good model of a wind speed distribution is the Rayleigh distribution. This tells us, given a certain average wind speed, the chance of experiencing a variation around this average. Have a look at Figure 7. Here the series1 blue dots show actual recorded data from a research paper2. Analysis of the authors’ data reveals that the average wind speed is 4.81 m/s. The measured distribution of wind speed values is interesting. It seems that there are more recorded values above the average than below. In other words, when wind is blowing at some average speed, the gusts seem to be larger than the average. I guess this is what we experience? The series2 red line shows the results of the Rayleigh model which I have calculated for you. There is reasonable agreement, not perfect, but certainly good enough to use Rayleigh’s model in our simulations, rather than relying on actual data recorded at one time at one place. The Rayleigh model takes as its input just one parameter, the mean (average) wind speed π£ππ£ . Given this, it predicts the chance of the wind having any other gust speed v. The expression to calculate this is as follows π(π£) = 2 π π£ −[π4(π£π£ )2 ] ππ£ π 2 2 π£ππ£ Shimada, T., and H. Kawamura (2005), Statistical compartmentalization of surface wind field over coastal seas using high-resolution SAR-derived winds, Geophys. Res. Lett., 32, L05607, doi:10.1029/2004GL022231 0.18 0.16 0.14 0.12 0.1 Series1 0.08 Series2 0.06 0.04 0.02 0 0 5 10 15