Digitally controlled wind turbines in megawatt size
with doubly-fed induction generator without position sensor
The technology in wind turbines is in full progress and for every new turbine new technologies emerge. The purpose of this publication is to describe one of these technologies, the OptiSpeedTM, both in a popular and a more comprehensive form.
Most wind turbines use a so-called three-phase generator, also called an induction generator to generate alternating currents. A reason for choosing this type of generator is that it is very reliable, and tends to be comparatively inexpensive. The generator also has some mechanical properties, which are useful for wind turbines. The rotor of the induction generator with short circuit rotorwinding comprises a number of copper or aluminium bars, which are connected electrically by aluminium end rings.The speed of the induction generator will vary with the rotational force applied to it. In practice, the difference between the rotational speed at peak power and at idle is very small, about 1 per cent.
V80-2.0 MW Wind Turbine near Sörup
This difference in per cent of the synchronous speed is called the generator's slip. Thus a 4-pole generator will run idle at 1500 rpm if it is attached to a grid with a 50 Hz current. If the generator is producing at its maximum power, it will be running at 1515 rpm. This is why the induction generator is called an asynchronous generator. It operates asynchronous to the synchronous speed. It is a very useful mechanical property that the generator will increase or decrease its speed slightly if the torque varies. This means less wear and tear on the tower, gearbox and other components in the transmission line, i.e. lower peak torque, which is one of the most important reasons for using an asynchronous generator rather than a synchronous generator on a wind turbine which is directly connected to the electrical grid.
The slip in the induction generator, however, is a function of the (DC) resistance (measured in ohms) in the rotor windings of the generator. The higher resistance, the higher the slip. So one way of varying the slip is to vary the resistance in the rotor. In this way one may increase generator slip to e.g. 10%. On motors this is usually done by having a wound rotor, i.e. a rotor with copper wire windings which are connected in star, and connected with external variable resistors, plus an electronic control system to operate the resistors. The connection has usually been done with brushes and slip rings, which is a clear drawback over the elegantly simple technical design of a cage wound rotor machine. It also introduces parts, which wear down in the generator, and thus the generator requires extra maintenance. The Vestas OptiSlip function deals with that, avoiding the problem of introducing slip rings, brushes, external resistors, and maintenance altogether.
By mounting the external resistors on the rotor itself, and mounting the electronic control system on the rotor as well, you just have to communicate the amount of slip you need to the rotor. This communication is done very elegantly, using optical fibre communications.
The newest stage of the technology of the slip is called OptiSpeedTM , giving possibility of varying the speed up to 30%. In addition to the fact, that the previously mentioned advantages are further enhanced, it is also possible to design special operating strategies, where the possibility of operating at lower speed is utilised. This feature is used in connection with reduction of noise. Another advantage of the OptiSpeedTM feature is the possibility to exploit even more of the energy in the rotor and transfer it to the grid, not forgetting the high quality of the power delivered to the grid. The following chapters of this document gives a much more comprehensive description of the OptiSpeedTM advantages, together with a detailed look into the technology behind.
Released by the oil crisis and in view of the limited resources of the fossil sources of energy, there was set focus on the possible use of renewable energy sources for production of power. Reactor accidents and problems with the disposal as well as the irreversible damage of the biosphere by increasing CO2-load did not strengthen additionally the interest in new and exhaustive energy sources. In the meantime economical concepts within the area of the solar and wind energy arose. And the wind power technology had the largest focus, because of the relatively high efficiency converting the mechanical energy into electricity.
The nature of the wind energy production
With all advantages of the wind energy also disadvantages can be found. On the one hand it is the small power density of the wind, leading to those wide, material intensive wind rotors and on the other hand the heavily varying wind supply, which leads to fast fluctuations of the mechanical load, proportional in 3.rd power to the wind velocity. In order to be able to perform in both weak winds and storms, high demands is set, not only for the materials, but also to the dynamics of the controlling mechanisms. Exploiting wind energy turns up the basic problem that the power requirements usually do not correspond to the actual production. This problem can be minimised by feeding the power into a huge grid with alternative power sources, and in this way be able to match supply and demand. It becomes problematic however again for wind turbines in the megawatt class, because they are able to bring the wind gusts to the grid as large energy portions, causing voltage fluctuations.
The technical challenge
Due to the fast wind speed variations the demand rises for generators featuring variable speed. The result of such a feature, varying the rpm around the nominal point (e.g. 1500 rpm), is that no large additional portions of energy is put into the grid. The high varying wind energy (turbulence) can be stored for seconds as potential energy in the rotor blades, by changing their speed within an admissible area. This increased number of revolutions prevents fluctuations on the grid and relieves all the turbine mechanics. Furthermore, the load on the mechanism turning the blades angle of attack (pitch mechanism) is reduced. By changing the angle of the blade, this pitch control changes the impact of the wind on the rotor in such a way that it remains at the desired number of revolutions. The pitch mechanism must then go active only, if the rotor threatens to leave the admissible speed range. There are two technically relevant generator types meeting the request of varying number of revolutions. Both types are represented on the market.
Synchronous generator is connected with its stator to DC-link converter system. The principle is a generator, which supplies a voltage and a frequency dependent on the number of revolutions. Since these do not fit directly a rigid grid, the output voltages from the generator must be transformed over a bridge rectifier into DC voltage (intermediate circuit). By means of a converter these DC voltages are transformed to AC and shaped regarding amplitude, frequency and phase into a voltage suitable for the grid. This principle enables a power production starting from a low number of revolutions up to the maximum speed. This wide speed range means a high energy yield. The entire amount of energy to the grid must be led however by one or more parallel operating converters. In order to control and limit the power output to the grid, the rotor field has to be controlled from its own converter through a set of slip rings. The complexity of this construction, and thereby the expenditure, affects the efficiency unfavourably.
Doubly-fed asynchronous generator
Doubly-fed three-phase generator is connected with its rotor to DC-link converter system and with its stator to the grid. The advantage is that the slip is proportional to the power flowing through the intermediate circuit. In order to prevent excessive load of the frequency converter, only speed fluctuations within the range of +/-30 % (slip s) around the rated speed (n0 = 1500 rpm) are allowed. The generator deliver over-synchronised (s < 0, n > n0) as well as under-synchronised (s > 0, n < n0) energy to the grid.
Active power at stator connectors:
PS = (1+s) Pmech
Active power at rotor connectors:
PR = -s PS
Active power sum:
PS + PR = Pmech
s = (wG PS wm)/wG
(PS = stator pole pairs)
Except from the losses, the main part of the energy is transferred to the grid via the stator terminals. Thus the static frequency
changer can be designed substantially smaller. In order to minimise the loss due to the rotor efficiency PR, the rotor energy is supplied
to the grid through the intermediate circuit and a controllable electric rectifier. By a relatively simple construction effort, a very high efficiency is obtained.
Reason for the choice of the double-fed three-phase generator
The OptiSlip concept was mentioned in the introduction. The OptiSlip concept is based on the induction generator with wound rotor circuit, and external rotor resistors mounted on the shaft. This concept has successfully been used on several thousands of turbines. At the OptiSpeedTM concept the same type of generator is used. The only difference is that the external resistors are replaced by highly reliable slip-rings. All the experience from the previous years of operation is retained, so a highly reliable generator system with wound rotor circuit is utilised in the OptiSpeedTM system. One new and very powerful feature in the OptiSpeedTM system is the active transmission oscillation damping system. This system constantly monitors oscillations in the speed of the generator, and if needed active damping of this is performed. This system helps to increase the lifetime of the drivetrain.
Fig. 3 depicts that Vestas structured the generator system according to the configuration presented in fig. 2. The shown directions of
the voltages and currents are in conformity with the following pictures.The stator of the generator can be switched to the grid both in
star and in delta mode. It keeps the current in the stator relatively low also at high yield. Up to approx. 800 kW the system operates
in star mode, and above and up to 2 MW in delta mode. This 2-way operation mode demands frequently connecting and disconnecting
to the grid, for the system to configure either star or delta mode. This switching process can occur quite frequently in the middle of
the capacity range, and it proves the advantage of being able to switch on an off rapidly. The present system is able to switch off in
3 grid periods and on again. In practice such quick switch on and off is prevented by invoking a hysteresis mechanism.
The complexity of the software solution made it necessary to utilise a 3- processor hardware construction (Fig 4). The host processor dedicated to the main functions and communication is supported by very fast DSPs in Master/Slave configuration for the control functions. Although the DSPs are floating point computers and programmed in C, the master DSP calculates and controls the rotor connected converter with a sample frequency of 5 kHz (switching frequency 2.5 kHz) and 10 kHz (switching frequency 5 kHz) for the converter on the grid side. In order to support necessary bandwidth for the reduction of current harmonics, the automatic controllers for the grid converter (Slave DSP) have an extra high clock frequency.
The dual port RAMs permits very effective communication between all processors. Over a RS232-interface the DSP system and thus the control for test and line-up can be operated independently of the host and the ArcNet connection. To be able to keep all parameters and variables in case of power failures, all data are stored in a static RAM. Time-critical monitoring and control functions are supported by means of logic devices (LSI). By a LSI the sample timing of the ADC is synchronised with the switch frequency and thereby the control frequency. This ensures a minimum self- and mutual disturbance of the measured values. The host communicates in the dual port RAM over an asynchronous handshake protocol with the master DSP.
Theoretical basis of the control structures
The two following pictures show the electrical circuit diagrams of the generator, the grid connected frequency converter with the
choke coil, the transformer and the grid. Additionally the current and voltage vectors with the appropriate angles are drawn in. Thus the basis for the development of the control algorithms is determined.
Over a PLL from the mains voltage UL the phase angle g is the basis for the rotating stator field co-ordinate system. Within that, red drawn rotor current components IRd and IRq are defined (Fig. 5a). This stator field chart is identical to the mains voltage chart in Fig. 5b. The shift of these co-ordinate systems around p/2 opposite gG leads to the fact that in the rotating co-ordinate system all active sizes in the q-axis and all reactive values in the d-axis are represented. Thus it is forming the basis for the control of the converter. The measured rotor current IR runs with slip frequency wR and the angle e off the mechanically rotating rotor and therefore always an AC with variable frequency is also stationary. With a rotor position giver the slip angle s is found; from r and gG or g is calculated (formula in Fig. 5a) and thus could the measured rotor current of the components IRd and IRq be determined.
Control of the static rotor side inverter
Without rotor position sensor the slip angle needed for the transformation can be calculated from the difference of e - l. The electrical angle e is simply calculated from the measured rotor current values IRa, IRb (the sliprings carry this mechanical transformation out),
while the angle from the rotor current values measured in the rotor vector chart only can be calculated indirectly over the generator model (Fig. 6). The generator model supplies the rotor current values ISd, ISq with detailed stator current values IRd, IRq in the rotary
stator field chart. From these it is easy to calculate l. Fig. 5b depicts the chart for the grid-side rotary frequency converter and for
the stationary earth connection. Both co-ordinate systems have their congruent correspondences in the rotary stator-field and stationary stator co-ordinate system in Fig. 5a. The equality of the co-ordinate systems leads to using the same angle and makes it
very easy to calculate the levels of the controls distributed on different processors together.
The feeding power into the grid is regulated via the rotor current (Fig. 6 on the right above). The control of the rotor current takes place in the co-ordinate system rotating with slip frequency wR, according to its transformation of the slip angle s. Since the measured rotor currents have the slip frequency as electrical frequency likewise, it leads to the fact that the automatic controllers in the stationary position see only direct currents and the automatic controller interpretation thereby becomes very simple. The allocation of the rotor current in field building d-component and a moment building q-component enables the cascade overlay to the rotor current control with a grid effect, e.g. an automatic grid reactive power controller (PMG, QMG).
In the lower half of fig. 6 is seen the calculation of the slip angle SIGMA from the grid, indispensable of the co-ordinate conversion to the rotor current (IL, IR). Due to the relatively high level of harmonics in the rotor current, the angle calculation is filtered. This leads to a substantial phase shift, which is dependent on the slip frequency. By means of a differentiating the feedback of the calculated slip angle SIGMA the slip frequency OMEGAR is calculated and after priority of these with the filter time constants TI_IR, an arc tangent operation leads to the current angle of the phase shift. Via addition to the calculated angle value (EPS) a compensation of the tracking caused by the filter takes place. In order to calculate the stationary angle LAMDA, the stator current must be calculated. From this the rotor current IRQ_SNL, IRD_SNL in stator field co-ordinates can be calculated by means of a simple generator model (in fig. 5a it is the red drawn components IRq und IRd). These currents may not be equated with IRQ and IRD in the fig. 6, since IRQ and IRD are components of the rotor slip angle co-ordinate system.
Control of the static grid side inverter
The control of the grid side static frequency inverter (Fig. 7) takes place in the Slave DSP. Apart from the entry of the grid and the stator voltage angle (gG,gS) the intermediate circuit voltage UDC on constantly 800 V is controlled here. According to Fig. 5b the difference in phase angle j of current and mains voltage is regulated to zero. The active power is thus transported in both directions power by the grid side static frequency inverter. The control operates in mains voltage co-ordinates, so that the automatic controllers have to do it in the stationary status only with equal sizes. Over a superior control also directed reactive power over IRD_REF can be placed and be implemented through reactive power compensation (phase shifters).
Synchronising to the grid without rotor position sensor
Before the generator can be switched to the grid, it shall be ensured that the voltages correspond regarding amplitude, frequency and phase. The larger the deviation at the connecting is, the higher balance current flows. It can lead to monitoring the current and to power-off of the system. In order to avoid any disturbance of the grid, the differential voltage between generator stator and the grid is kept on an insignificant level. This is not easy to implement due to the harmonics in the generator voltage. Fig. 8 depicts steps in synchronisation to the grid. In order to enable the synchronisation, the voltages between the lines 1 and 2 of the grid and the stator (UL12, US12) is compared regarding amplitude, frequency and phase position. UD12 show the instant differences; the amplitude difference delta U, the frequency difference delta f and the angle difference between these two voltages DELTA_GAMMAGS. At the beginning of the synchronisation process the stator voltage usually deviates regarding amplitude and frequency from the mains voltage. Point (1) in Fig. 8 depicts how the amplitude equality is obtained by an amplitude control. The frequency of the stator voltage US12 is unknown, since it consists of the unknown mechanical rotational frequency m and the given electrical slip frequency wR. As soon as the stator voltage reaches sufficient amplitude, the frequency ws of the stator voltage can be determined by zero crossovers and the difference to the frequency wG on the controllable value for the slip frequency can be added (2). Thereby the frequency difference becomes 0. In reality the zero crossover recognition reflects also the harmonics in the generator voltage, so it becomes a measuring error regarding the basic frequency of the stator. The switching of the frequency differences causes however at least that the angles difference slowly changes. Thus the angle difference in a further step can be diminished (3). Even if it could be diminished within one scanning step, the reduction is slowly carried out, in order to avoid voltage steps in the rotor. These would release natural oscillations in the rotor and would delay reaching the steady bias point before connecting to the grid (4).
Figure 8: Function of grid synchronisation
Due to the harmonics contents of the amplitude-, frequency-phase controllers during the entire synchronisation process, the process becomes more difficult when operating without rotor position sensor. A reason for that is mutual influencing in the controllers. By optimisation of the parameters in the controllers, a synchronisation time between 0.2 and 0.3 seconds could be achieved despite these circumstances.
Reduction of the current-harmonics caused by the generator
With an extension in the grid side inverter control the 5th and 7th grid harmonics together with the slip harmonics are significantly reduced until the THD value of the line currents reaches a value less than 1%.
Measurements at the 2 MW prototype
Feeding performance independently of fast wind modifications
A variation of the supplied wind energy from the factor 8 (proportionally to the 3rd power of the wind velocity) remains the feeding performance of the generator into the grid constantly on the desired value. These substantial energy fluctuations are converted into rotation energy, whereby the adjustment of the angle the rotor blades (pitch angles) ensures that the generator rate cannot run out of the bias point (1680 rpm).
Dynamics and stiffness of the power adjustment at 1 MW operating point
The very good performance of the wind turbine is reflected also in a stable and fast power adjustment under extreme conditions. Fig. 10a depicts that the power adjustment has a response time of approx. 40 ms equal to approx. two grid periods. Thus a stable operation is ensured also at weak grids. Fig. 6a depicts that the decoupling of active and reactive power is very effective. Fig. 6b depicts that even with the change from under- to over-synchronous operation the power output can be kept constant and the reactive power with 0 kVA is kept constant, which corresponds to a power factor of 1. This fast power adjustment enables a fast response to wind turbulences or changes in the grid load, so that a stable operation is obtained, even under extreme wind or grid conditions, with the grid reactions remaining insignificant.
Grid Synchronisation and Connection
Fig. 11 depicts the sequence where the generator is connected to the grid. The line current IL1 is the total of stator current and grid side static frequency converter current. It is seen that no significant current spikes or inrush current is present during grid connection, when K500 is switched from 0 to 1.
Effect of the harmonics reduction
The two plots demonstrate the effectiveness of the compensation up to the 10th order of the generator and grid harmonics.
As it can be seen at Fig. 13 power fluctrations are controlled effectively so that external disturbances ex. the blade/towerssage is not present in the power signal. Due to this effective powercontrol flicker can be kept at a low level. Alltogether the OptiSpeedTM concept results in high power quality.
Operation, safety and maintenance concept
The operation strategy is as follows. At very low wind speeds the blades of the turbine will be placed in 45 and the rotor will start rotating. At increased wind speed the rotor will accelerate and grid connection of the generator in Star connection will happen very smoothly. The OptiTip control will now bring the angle of the blades to an optimum angle, and speed of the rotor and the power to the grid will follow a power versus speed reference curve. If the wind speed increases further then the speed of the rotor will increase and hence the power led to the grid will increase also. Operating in Star connection of the stator winding means reduced losses in generator, but Star connection is not possible at nominal power, so at a certain power the generator must switch from Star to Delta connection. This is due to the fast and effective control of the rotor currents possible to do within just a few seconds. At nominal wind and at nominal power the power is no longer increased when the wind speed increases. Here the pitch control combined with the buffering effect of the inertia in the rotor enables the power to be controlled constant.
The safety strategy for the OptiSpeed TM system is the same as known from the V29, V29, V39, V42, V44, V47, V63 and V66 turbines. For turbines with individually controlled and actuated pitch system for each blade this concept includes three completely separated control systems. Just one of these is able to bring the turbine out of operation in a controlled manner. The mechanical brake is not a safety component and its function is to help the pitch system to stop rotation completely. For smaller turbines with one common pitch system for the three blades in the rotor, here the brake is a safety component. When such a turbine has to be stopped both the pitch system and the brake is utilised.
The speed of both the rotor and the generator is monitored by the turbine controller. Further a separate and external over speed guard is present. Both of these independent systems constantly monitor the wind turbine. Every six months the turbine needs a scheduled service check. Additionally the turbine is capable to contact Vestas Service department if anything extraordinary happens. This is done through the remote control system connected to almost any Vestas wind turbine. The service department is then capable of checking all sensors in the turbine. This always enables Vestas to bring the wind turbine back in operation within very short while.
Summary and overview
The OptiSpeedTM system is an optimal and powerful system which makes a Vestas wind turbine even more efficient, more flexible with regards to acoustic noise, and the power quality is excellent. The OptiSpeed TM system is offered in the turbines: V52-850kW, V66-1.75/2.0MW, V80-2.0MW. A V80-3.0MW turbine OptiSpeedTM turbine is under development. The above mentioned turbines will not be offered in the USA and in Canada. Here the OptiSlip system is offered. In the coming years Vestas Wind Systems expects to erect turbines all over the world. In the countryside, in mountains, offshore any place where utilisation of the wind energy is economically feasible. The physical dimensions of the turbines seems to be without any limits. Therefore the turbines will grow in size and in power. Technically the OptiSpeedTM is the most optimal and powerful technology ever launched by Vestas Wind Systems.
The contents of the above application description was written, togeher with Vestas Wind Systems, in the following publication
Ehrenberg, J.; Andresen, B.; Rebsdorf, A.:
Windkraftanlagen für den Megawattbereich, Digitale Steuerung eines doppelt gespeisten Asynchrongenerators ohne Lagegeber, Teil 1
Magazine Elektronik 2001, Issue 18, p 60 ... 67
Ehrenberg, J.; Andresen, B.; Rebsdorf, A.:
Windkraftanlagen für den Megawattbereich, Digitale Steuerung eines doppelt gespeisten Asynchrongenerators ohne Lagegeber, Teil 2
Magazine Elektronik 2001, Issue 19, p 78 ... 87