AU2014200885B2

AU2014200885B2 - Method for operating a wind turbine

Info

Publication number: AU2014200885B2
Application number: AU2014200885A
Authority: AU
Inventors: Alfred Beekmann; Wolfgang De Boer
Original assignee: Individual
Current assignee: Individual
Priority date: 2009-03-23
Filing date: 2014-02-20
Publication date: 2016-05-26
Anticipated expiration: 2030-03-23
Also published as: AU2014200885A1

Abstract

Abstract The invention concerns a method of operating a wind power installation, wherein the wind power installation is connected to an electrical supply network and in operation, that is to say when wind prevails which is above the initial speed and below the switch-off speed, feeds electrical energy into the supply network and more specifically with the frequency and voltage desired or required by the supply network, wherein in operation of the wind power installation above the rated value or below the switch-off speed the rotor of the wind power installation that has at least one rotor blade rotates and connected to the rotor of the wind power installation is a generator driven by the rotor thereby to generate electrical energy, wherein the wind power installation has means for measuring the frequency prevailing in the electrical supply network and said means for measuring the network frequency are connected to a control device for controlling operation of the wind power installation. According to the invention it is proposed that the power delivered by the generator of the wind power installation to the network is increased quickly and for a short period above the currently prevailing power of the wind power installation if the network frequency of the electrical network is below the desired target frequency of the network by a predetermined frequency value. (Figure 3) Coo 3: ( o? --- ---- 0 co [ZR] Aouanbai 3 q

Description

Aloys Wobben Argestrasse 19, 26607 Aurich Method for operating a wind turbine 5 The invention concerns a method of operating a wind power installation and a wind power installation for carrying out the method. As state of the art attention is directed in particular to 'Grid Integration of Wind Energy Conversion Systems', Siegfried Heier, 1998, 10 therein in particular pages 263 ff as well as US No 7 345 373 and WO 01/86143. The most relevant state of the art is document WO 01/86143. That document discloses the teaching of reducing the power of a wind power installation when the network frequency, that is to say the frequency 15 of the network into which the wind power installation feeds its electrical power, exceeds a given value above the target frequency. In the case of Central European networks the target frequency is usually at 50 Hz whereas in the case of US networks it is at 60 Hz. At the same time however there are also always slight network 20 frequency fluctuations which are dependent on how greatly the ratio of the power produced by the energy producers connected to the electrical network is balanced out in relation to the power taken by the consumers, that is to say those who are connected to the electrical network and take electrical energy in order therewith to operate any electrical equipment. 25 If for example the power supply from the generators is above that which the consumers connected to the network are taking in terms of power the network frequency rises and conversely the frequency can also fall below the target frequency, for example 50 Hz, if the power supply offered is less than that which is being taken by the electrical consumers connected to the 30 network. Network management, that is to say the management of producers and also large consumers, is usually implemented by the network operators. In that case network management can provide quite different regulating 2 mechanisms, for example for automatically switching on certain generators (for example gas fired power stations), automatic switch-off of given large consumers or also the use of pumped storage plants and the like. In normal operation even the network management of large supply 5 networks constantly succeeds in keeping the network frequency in the region of the target frequency, in which respect minor deviations are certainly allowed. Those minor deviations however should generally not exceed the region of + 1%o. It will be appreciated that the network management can also involve switching on further networks which are connected to the 10 network in order thereby to feed additional power into the network or to take it from the network and feed it into other networks. For wind power installations, document WO 01/086143 - as already stated above - already teaches reducing the power below the currently available power if a given network frequency value is exceeded, for example 15 a value which is 3%o above the target value of the network frequency (for example over 50 Hz). The document further teaches that, if the frequency continues to rise, the power is linearly further reduced, in dependence on the further rise in the network frequency. 20 Now the object of the present invention is to improve the operation of a wind power installation in comparison with the state of the art and overall to improve the network support of the wind power installation with respect to the network. According to the invention that object is attained by a method having 25 the features of claim 1. Advantageous developments are described by the appendant claims. According to the invention the wind power installation is now no longer switched off on the fall of a given frequency value below the target value of the network frequency, but the wind power installation continues to 30 be operated, more specifically with an increased power, that is to say quickly and for a short period the power is higher than the power which was previously fed into the network. For that purpose for example the rotational energy stored in the moment of inertia of the rotar/generator system is 3 used, that is to say more power is taken briefly from the entire rotor/generator system so that an increased level of power is quickly available immediately upon the network frequency falling below the predetermined target value. That also occurs when the wind power 5 installation had previously fed in at rated power, that is to say it had delivered its normal maximum amount, namely the power that it can take from the wind. The amount by which the power is quickly increased is in a range of up to 10 to 30% of the rated power, preferably about 20% of the rated 10 power. The predetermined frequency value can be established for example by presetting a headband frequency. As soon as the network frequency is below that headband frequency the currently prevailing power of the wind power installation is raised and the power delivered by the wind power installation, 15 more specifically fed into the network, is also raised. In that case the headband frequency is below the desired target frequency of the network by the predetermined frequency value. The predetermined frequency value is preferably greater than 1%o, 2%o or 3%o of the network target frequency. In the case of a 50 Hz network 20 target frequency therefore the system detects when the value falls below the frequency of 49.95, 49.90 and 49.85 respectively. Alternatively or in addition a relative frequency change can also be considered, that is to say a relative frequency drop also denoted by df/dt or a frequency gradient. If the magnitude of such a network frequency change 25 in relation to time is excessively great and therefore the frequency falls excessively quickly the power which is currently to be fed into the network can be briefly increased to support the network. Detecting such a frequency change in relation to time, that is to say df/dt, may make it possible to more rapidly detect a network frequency drop and thus possibly permits faster 30 recognition of the need for network support. Detection of an absolute frequency value, that is to say when the value falls below an absolute predetermined frequency value, and also the change in relation to time, can also be combined. Thus for example it is possible for a fast network 4 frequency drop to be assessed as less critical if the absolute value of the network frequency is above the rated frequency. If in addition or also as an alternative a frequency gradient is detected, it has proven to be desirable to provide for a power increase as 5 from a gradient of 0.1 Hz/s. An amount of change, that is to say a gradient of 0.2 - 7 Hz/s, in particular 0.5 - 2 Hz/s, has proven to be an advantageous range for initiating a power increase. Thus for example 0.2 Hz/s, 0.5 Hz/s, 1 Hz/s, 2 Hz/s and 5 Hz/s are advantageous values. It is to be noted that the detection of a frequency gradient of for example 1 Hz/s usually does not 10 presuppose any measurement over the period of an entire second. Rather, measurement times of 20 ms and less, in particular 10 ms, are suitable measurement times. Shorter measurement times of for example 5 ms or even shorter are also preferred values. In addition both the measurement time and also the underlying amount of change or the underlying frequency 15 gradient can depend on the network target frequency. The above-mentioned values for the frequency gradient and the measurement times provided for same relate to a 50 Hz target frequency. In the case of a 60 Hz target frequency a somewhat greater gradient and/or a somewhat shorter measurement time can possibly be provided. 20 It is also to be mentioned that the short-term power increase can also be used to stabilise or smooth the network frequency or to damp frequency fluctuations. In particular damping of frequency fluctuations can advantageously take account of the frequency gradient. Preferably the short-term power increase is effected, utilising the 25 rotational energy stored in the moment of inertia of the rotor/generator system. That therefore concerns taking kinetic energy which is stored both in the rotating rotor which has one or more rotor blades, and also in the rotating rotor member of the generator. Taking a higher amount of power can be implemented in particular by increasing the exciter current and thus 30 by increasing the generator counter-moment of the generator rotor member. In particular gearless generators with rotor members of large diameter and thus large masses and correspondingly large moments of inertia can store a considerable amount of kinetic energy.

5 Preferably the frequency of the power fed into the network always corresponds to the currently prevailing network frequency. If therefore the network frequency drops, a power increase can be effected, in which case however the frequency of the feed into the network is reduced, adapted to 5 the currently prevailing frequency. Preferably there is proposed a method characterised in that the increase in the power is effected above the currently prevailing power, that is to say also above the rated power, when previously the feed into the network was with rated power. Therefore, even when the wind power 10 installation is operated in the rated mode, a power increase is effected upon a critical drop in frequency. In that respect, it was realised that a rated power which can usually also represent a maximum power at any event for ongoing operation can be exceeded for short-term network support without damage to the wind power installation. 15 In an embodiment it is proposed that the method is characterised in that within a period of 10 to 1000 ms, in particular 20 to 500 ms, preferably 50 to 100 ms, after the frequency value falls below the predetermined frequency value of the network frequency or after the predetermined amount of change is exceeded the power increase is effected and/or the feed is 20 effected with an increased power, that is to say a power which is above the previously set power, for a further period of at least 0.5 sec, preferably at least 1 sec to a maximum of 30 sec, preferably at a maximum about 3 to 6, 8, 10, 12 or 15 sec. In principle a reaction time which is as short as possible, of for example 10 ms, is to be viewed as an idealised value for implementing 25 an increase in power. In particular the time of 10 ms corresponds to a half wave at a network frequency of 50 Hz. A longer response time of up to 1 sec is desirable to prevent an over-reacting or indeed unstable system. In particular values of 50 to 100 ms have proven to be an advantageous compromise. 30 The power increase is required in principle for a short period of time. That period of time usually lasts for at least 0.5 sec but preferably 1 sec and goes up to 3, 6, 8, 10, 12, 15 and a maximum of 30 sec. If an increased power feed of more or markedly more than 10 sec is required, that is no 6 longer generally to be viewed as an instantaneous support measure, but rather an increased power requirement. An effective range for the power increase has proven to be at 3 to 6, 8, 10, 12 or 15 sec. Preferably there is provided a method of controlling a wind park in 5 which each wind power installation is controlled in accordance with a method according to the invention. In particular each wind power installation is adapted to deliver an increased level of power to the network in the case of a frequency dip. In that respect a wind park includes at least two wind power installations but usually markedly more installations like 10 wind power 10 installations. 50 wind power installations or even more. Among all wind power installations in the wind park however only those which are also involved in the described method are to be considered. Preferably in this case too the increase in the power delivered to the network from all wind power installations is effected in unitary and/or central 15 relationship. On the one hand that prevents different installations of a wind park responding at different times and possibly impeding each other. In addition wind parks can be subject to certain conditions such as limit values for coupling to the network if the wind park feeds the power of all wind power installations into the network at a network connection location. Thus 20 for example upper limits for the power fed into the network on the part of the connection line can however possibly also be preset when using a central transformer for same. A central control can take account of such boundary conditions. Sometimes a unitary control of the wind power installations can be helpful, if that is possible with different wind power installations in a wind 25 park. Thus it is possible to implement at least partially unitary control for example in regard to the response times and/or periods of the power increase. If for example in a situation where all or most wind power installations of a wind park are in the rated mode of operation a power increase of all wind power installations should be limited because of a power 30 feed upper limit for the wind park, the control can be effected in such a way that firstly a part of the wind power installations contribute to a power increase and thereafter another part of the wind power installations do so. In addition the level of control and regulating complication and expenditure can 7 be reduced by a central control unit which only delivers the corresponding power target values to each wind power installation in the wind park for example for a power increase. In addition there is proposed a wind power installation adapted to use 5 a method according to the invention. Furthermore there is proposed a wind park which includes a plurality of wind power installations according to the invention and preferably uses a central control method and/or in which the increase in the power of the wind power installations, that is to be delivered to the network, is at least partially unitarily controlled. Central control for the 10 increase in the power to be delivered to the network from all wind power installations can be implemented by a separate central control unit and/or a wind power installation can function as a master so that the other wind power installations depend on that installation. In principle it is also possible for a wind park to be divided into a plurality of park portions, in regard to 15 control procedures, in order for example to bundle together installations of the same or a similar type in each case, in terms of control procedures. To increase the power delivery, not only is utilisation of the rotational energy of the moment of inertia considered, but also as a support or possibly exclusively, it is possible to effect a change in the setting angle of the rotor 20 blades - a change in the pitch angle, referred to as pitching - to increase the wind yield. That is effected in particular when the wind power installation is running under rated load, that is to say is already delivering rated power, and in particular the rotor blades have already been partially pitched to regulate the rated rotary speed. 25 After a power increase the speed of rotation of the rotor can have reduced because kinetic energy has been taken. Particularly in the event of a power increase in the case of a rated load mode of operation, such a reduction can however be less or may not occur at all. A reduction in rotary speed is to be expected in particular in the part-load range and then 30 depends on the level and duration of the power increase, that is to say the power which is additionally delivered.

8 The invention is described in greater detail hereinafter by means of embodiments by way of example with reference to the accompanying Figures. Figure 1 diagrammatically shows a partly opened pod of a wind power 5 installation with a diagrammatic view of the hub and parts of the generator, Figure 2 diagrammatically shows an overview with a gearless rotor/generator coupling arrangement with frequency measurement, Figure 3 shows an embodiment by way of example of a power/frequency characteristic of a wind power installation, 10 Figure 4 shows an alternative embodiment to Figure 3, Figure 5 shows an example illustrating power configurations for a constant power increase, Figure 6 shows an example illustrating power configurations in the event of a power increase which is effected in dependence on the rotary 15 speed of the rotor, Figure 7 shows an example illustrating measurement of a power in the case of a power increase dependent on the rotary speed of the rotor, Figure 8 shows measurement of a power with a constant power increase, and 20 Figure 9 shows possible variations in the power values by which a power increase is to be implemented in dependence on the frequency and for different adjustable maximum values in respect of the power increase. Hereinafter identical references can denote identical components but also components which are similar and not identical. Hereinafter, for the 25 sake of completeness, a wind power installation having a synchronous generator and a gearless design with a full-wave converter is described. Figure 1 diagrammatically shows a pod 1 of a gearless wind power installation. The hub 2 can be seen by virtue of the housing (spinner) being shown partly open. Three rotor blades 4 are fixed to the hub, the rotor 30 blades 4 being shown only in their region near the hub. The hub 2 with the rotor blades 4 forms an aerodynamic rotor 7. The hub 2 is mechanically fixedly connected to the generator rotor 6, which can also be referred to as 9 the rotor member 6 and is referred to hereinafter as the rotor member 6. The rotor member 6 is mounted rotatably with respect to the stator 8. During its rotation relative to the stator 8 the rotor member 6 is supplied with current, usually a direct current, in order thereby to generate a 5 magnetic field and build up a generator moment or generator counter moment which can also be suitably set by that exciter current and can be altered. When the rotor member 6 is thus electrically excited its rotation relative to the stator 8 produces an electrical field in the stator 8 and thus an electrical alternating current. 10 The alternating current produced in the generator 10 which is substantially made up of the rotor member 6 and the stator 8 is rectified by way of a rectifier 12, in accordance with the structure shown in Figure 2. The rectified current or rectified voltage is then converted by means of an inverter 14 into a 3-phase system at the desired frequency. The three-phase 15 current-voltage system produced in that way is stepped up in particular in voltage by means of a transformer 16 to be fed into a connected power network 18. Theoretically it would be possible to omit the transformer or to replace it by a choke. Usually however the voltage requirements in the network 18 are such that it is necessary to step it up by means of a 20 transformer. Control is effected by using a main control 20 which can also be referred to as the main control unit and which forms the highest-level regulating and control unit of the wind power installation. The main control 20 acquires its information inter alia about the network frequency from the 25 subordinated network measuring unit 22. The main control 20 controls the inverter 14 and the rectifier 12. It will be appreciated that in principle it would also be possible to use an uncontrolled rectifier. In addition the main control 20 controls a direct current setting member 24 for feeding the exciter current into the rotor member 6 which is part of the generator 10. The main 30 control 20 modifies inter alia the feed into the network or the working point of the generator when the network frequency falls below a predetermined network frequency limit value. As the generator is operated in rotary speed- 10 variable fashion the feed into the network is effected as described with a full-wave converter formed substantially by the rectifier 12 and the inverter 14. In operation the network voltage and the network frequency is 5 permanently subjected to three-phase measurement by the network measuring unit 22. A new value for one of the three phase voltages is afforded from the measurement operation every 3.3 ms - at any event in the case of a network frequency of 50 Hz. The network frequency is thus detected each voltage half-wave, filtered and compared to the preset limit 10 values. For a 60 Hz system, a value for one of the three phase voltages would be available approximately for every 2.7 ms, more specifically approximately at each zero crossing. Figure 3 shows a diagrammatic example of a frequency configuration and frequency ranges in relation to time, also showing an associated power 15 configuration. It will be seen from Figure 3 that the main control distinguishes in respect of frequency between three operating ranges, mainly the headband range 30, the control band range 32 and the underfrequency range 34. The headband range is the frequency range between the rated frequency fnorn or 20 fN and the headband frequency fdeadband therebeneath. The rated frequency is usually fixedly predetermined such as for example 50 Hz for the European network system or 60 Hz in the US area. The headband frequency fdeadband can be adjusted whereby the headband range can at any event be adjusted in relation to that lower limit. No power increase is provided in the headband 25 range. The control band range 32 extends between the headband frequency fdeadband and the subjacent control band frequency fcontrolband. The control band range can be suitably set by predetermining both the headband frequency fdeadband and also the control band frequency fcontrolband. In the control band 30 range, that is to say when the actual frequency assumes values in the control band range, an increase in the effective power can be effected in dependence on the frequency deviation, more specifically in particular the actual frequency from the headband frequency, by a power increase Pincrease.

11 In that case there is an effective power increase which is dependent in particular proportionally on the frequency deviation. Thus the effective power increase Pincrease is also a variable parameter of the control band range. There can thus be an increase in the effective power in dependence on the 5 frequency deviation by an additional power Pincrease of 0% to a preset value Pincreaseset. The maximum increase in the effective power can be preset by means of Pincrease set, wherein Pincrease set can be increased from 0% to Pincrease max in 1 /o steps. The underfrequency range 34 extends downwardly from the control 10 band frequency fcontrolband. When the actual frequency is below the control band frequency fcontrolband then the maximum preset power increase is implemented in the underfrequency range. The power increase Pincrease thus assumes the maximum value which can be for example 10% of the rated power. 15 Figure 3 shows in bold the configuration by way of example of the actual frequency. The frequency initially has the value of the rated frequency form until it drops away at the time to. The configuration of the actual frequency is identified by reference 36. A configuration by way of example of a power to be set is also identified by reference 38. It is to be noted that the 20 power must be at least 4% of the rated power for the control described herein by way of example in order to be able to carry out the desired or at least the desired maximum power increase. The actual frequency 36 drops away at the time to, but is firstly in the headband range 30 so that no power increase takes place. The actual power 25 which is at least 4% of the rated power therefore initially remains constant. At the time ti the actual frequency 36 reaches the headband frequency fdeadband and falls below same. In the illustrated example the power 38 rises linearly with the further drop in the frequency 36. That is to say the power increase Pincrease, namely the respective increase with respect to the initial 30 value PA, is here proportional to the difference between the actual frequency 36 and the headband frequency fdeadband. The proportionality factor is so set here that the power increase Pincrease reaches its maximum value of 10% of the rated power Pn when the frequency reaches the control band frequency 12 fcontrolband. That is the case at the time t 2 . The power increase Pincrease can thus be specified in principle for the control band range with: Pincrease = Pincrease set X PN X (fdeadband - f) / (fdeadband - fcontrolband), insofar as further boundary conditions like also maximum times are observed for a power 5 increase. If the frequency 36 falls further below the control band frequency fcontrolband the power 38 cannot be further increased and thus from the time t 2 firstly remains at a maximum value, namely the initial value plus the maximum value of Pincrease, namely + 10% of the rated power. If the 10 frequency now rises again and at time t 3 exceeds the value of the control band frequency fcontrolband, the power increase is thus also reduced again until the frequency 36 rises at the time t 4 above the headband frequency fdeadband. At that time t 4 the power has then reached the initial value PA again and does not fall any further. 15 It is to be emphasised that Figure 3 shows an idealised configuration and any regulating dynamics are initially disregarded. In addition, in the stated example - contrary to the diagrammatic view - the maximum time for which the power is increased should not exceed 8 sec. It is precisely in the case of smaller power increases however that a prolongation of that time 20 can possibly be considered. It is to be observed that the linear frequency drop and linear frequency rise were selected to illustrate the control diagram and do not necessarily coincide with a frequency characteristic which is usually to be expected in a power supply network. Figure 3 shows a diagram illustrating the configuration of the network 25 frequency and as the reaction thereto the variation in the power feed from a wind power installation. It is moreover to be seen that, at a given time ti, the network frequency dips and more specifically below a given frequency value below the target frequency of about 50 Hz. If the frequency falls below a frequency 30 value of for example 1%o below the target value (and falls still further) the power of the wind power installation is increased above its currently prevailing value, for example by 20% of the currently prevailing power or by up to 30% above the rated power, practically instantaneously, that is to say 13 in an extremely short time and for a short period, that is to say within a few ms, for example 50 to 100 ms or also 500 to 1000 ms, to name a further example. The example in Figure 3 is based on an increase by 10% in the rated power. In the extreme case, when the power is just 4% of the 5 rated power and is increased by 10% of the rated power, at least theoretically it would be possible to implement a power increase by 2.5 times the current power. That can be justified inter alia on the basis that, even with a low power delivery, a comparatively high rotary speed and thus a correspondingly large amount of rotational energy is already stored. Thus 10 for example at 4% rated power it is already possible to reach a rotary speed of about 50% of the rated speed. If many wind power installations behave as described hereinbefore then a large amount of additional power is very quickly made available, with the consequence that the producer/consumer inbalance is very quickly 15 compensated, with the further consequence that the network frequency rises further very quickly and even quickly exceeds its target value. In the illustrated embodiment the increased power feed into the network is effected only for about 2 to 10 sec, preferably only about up to 3 sec, depending on how the frequency behaves. 20 If for example the frequency rises very quickly again then the increased power feed can also be rather reduced again and concluded while in contrast the increased power feed is effected for longer if the underfrequency power feed remains for a longer period of time. Figure 4 also shows the increased power feed for the situation in 25 which the power fluctuates, for example because the wind overall is fluctuating. In addition Figure 4 also otherwise concerns a configuration based on a behaviour which is really to be expected. The frequency 36 is firstly at the rated frequency, namely 50 Hz. At a time to* the frequency 36 then falls off very quickly and also quite soon falls 30 below the headband frequency fdeadband. Figure 4 admittedly also involves the situation where the frequency falls below the headband frequency, but that is detected only after the frequency has fallen below the selected headband frequency after a detection time Atdetect, wherein that detection time is at a 14 maximum 20 Ms. The underfrequency is thus detected at time t 1 * as shown in Figure 4 and the power 38 is thereupon increased. An increase time Atincrease of < 800 ms elapses up to the maximum power increase of Pincrease of 10% rated power above the power still prevailing at the 5 time ti*. When the frequency falls below the selected headband frequency fdeadband the main control, by virtue of internal control functions, provides for a power increase Pincrease of a maximum of 10% of the rated power of the wind power installation from the generator for a preset time trax. The recognition time for the underfrequency is less than 20 ms. The level of the 10 additional power Pincrease is proportionally dependent on the set maximum permissible power increase and the frequency deviation. The power is increased with a fixed gradient of about 250 kW/s - insofar it is considered at any event on the basis of the frequency deviation. In that way, in the case illustrated here, a power increase to the maximum value of a maximum of 15 1 0 % of the rated power of the wind power installation is achieved in < 800 ms. The power increase Pincrease is available over a time of a maximum of 8 sec. After at the latest 8 sec, the effective power of the wind power installation in the illustrated example is restored at about 250 kW/s to the normal, in particular previous working point. 20 Therefore, viewed from the time t1*, the maximum power increase is thus achieved after about 800 ms at the time t 2 *. The maximum increased power which is now set is held until the time t 3 * in order then to gradually fall again until the time t 4 * to approximately the initial value or, in dependence on wind, to a new value. The time from t1* to t 4 *, which can 25 also be referred to as tmax-increase, is a maximum of 8 sec for the illustrated example. It is to be noted that Figure 4 is also a diagrammatic view and precise values including precise time values cannot be exactly read off therein. It is to be observed that the frequency 36 rises again during the 30 power increase, in particular after the time t 2 *, and this can also be attributed to the power increase, that is to say to the power which is additionally fed into the network. Nonetheless that crucially depends on the respective network and the respective wind power installation, and in 15 particular on whether still further wind power installations implement such a power feed into the network. Incidentally in the illustrated example however the frequency does not rise to the rated frequency again within the power increase range. Nonetheless, by virtue of the maximum time 5 achieved, the power increase is reduced and concluded. For the increased power feed into the network, the wind power installation according to the invention uses the rotational energy stored in the rotating system comprising the rotor/generator, by virtue of the moment of inertia. In other words, due to the additional amounts of power taken off, 10 above what is actually predetermined by the power characteristic of the wind power installation, the overall rotor/generator system admittedly continues to rotate, but it loses rotational energy and thus, after the increased power feed into the network, rotates more slowly than previously because more power was taken from the overall system than was delivered by the wind. 15 The behaviour according to the invention of the wind power installation however in particular has the result that the critical underfrequency situations are successfully managed or successfully bridged over by existing wind power installations so that further network management interventions can be initiated within the critical period of time 20 of for example 1 to 8 sec, in particular 1 to 3 sec, after the occurrence of the underfrequency situation, and such network management interventions, after the wind power installation or installations (or entire wind park) has or have fed its or their additional power into the network, intervene in the action thereof and successfully support the network. 25 The technical availability of the power increase Pincrease in the case of a network underfrequency is fundamentally given as from an instantaneous power Pactual of 4% of the rated power. A power increase Pincrease by 10% with respect to the rated power is then possible. A power increase of 200 kW for a wind power installation by way of example is illustrated in principle 30 hereinafter in Figures 5 to 8. In this case 200 kW constitute 10% of the rated power. In principle, it is possible to select between two options for the behaviour in respect of the power increase during frequency support, namely between a frequency-dependent power increase as shown in Figure 5 and a 16 frequency-dependent and rotary speed-dependent power increase as shown in Figure 6. An embodiment which can also be described by reference to Figure 4 and the values of which are specified in Figure 4 can be described as follows. 5 In the case of frequency changes to below the headband the required power increase occurs with a fixed gradient of about 250 kW/s. A power increase Pincrease of up to 10% of the rated power of the wind power installation (WPI) is achieved after about 80 ms. In the case of small frequency changes within the control band and in the power range below 10 500 kW the power gradient is slightly reduced by the generator-induced behaviour upon power changes. The power increase Pincrease is available over a time of a maximum of 6.4 s. After at the latest 7 s the effective power of the WPI is set to the normal working point again at 250 kW/s. The control stabilisation time is dependent on the wind conditions and the installation 15 rotary speed which is set during the implementation time. The transition to the feed of power into the network in normal operation is concluded in about 1 S. Figure 5 shows a target power Porder in relation to time for the situation where no power increase would be implemented. That curve is also included 20 for the purposes of comparison. An underfrequency is detected at the time t 8 in Figure 5 and a power increase of 200 kW is predetermined. That power curve which is basically represented in an angular configuration is identified by Pincrease. That power Pincrease rises at the time tB to that value increased by 200 kW and keeps that value constant until the end time tE and then falls to 25 the value of the normal power target curve Porder. The normal power curve Porder has in the meantime fallen away without that having an influence on the curve Pincrease. The time between the initial time tB and the end time tE is about 8 sec. In addition, a power curve Pactuai is also shown, corresponding to the actually achieved value of the power fed into the network. As shown in 30 Figure 5 therefore the power increase Pincrease over the preset implementation time trax is proportional to the network frequency. That corresponds to a power delivery independently of the rotary speed of the rotor of the wind power installation, that occurs.

17 The following should also be additionally explained in relation to Figure 5: the power of the wind power installation only depends on the network frequency during frequency support. In addition, the power increase Pincrease which is required proportionally to the frequency deviation occurs 5 over the preset implementation time tmax. The total effective power delivery Pactuai is thus the total of the power in accordance with the rotary speed power characteristic at the moment in time of activation of inertia emulation and required power increase Pincrease. The overall effective power delivery is delimited by the maximum apparent power of the wind power installation. 10 Those limits of the wind power installation configurations are shown in the power diagrams in illustrations 7 to 10. In regard to the frequency-dependent and rotary speed-dependent power increase as shown in Figure 6 the power increase achieved, in relation to the preset implementation time, is proportional to the network frequency 15 and also varies in dependence on the rotary speed that is set in respect of the rotor. In dependence on the wind speed and the rotor rotary speed, the power increase is provided adapted to the rotary speed. The nomenclature used in Figure 6 corresponds to that in Figure 5 and an underfrequency is detected at the time tB and the power increase by about 200 kW is effected. 20 In the further variation up to the end time tE the rotary speed decreases and therewith also the target power, without having regard to a power increase, namely Porder. The power increase Pincrease maintains approximately a value of 200 kW above the respective current target power Porder. At the time tE the power increase is then ended and the power Pactuai falls to the value of the 25 target power Porder. In addition as an explanation relating to Figure 6 attention is directed to the following: the power of the wind power installation remains controlled during frequency support with the predetermined rotary speed-power characteristic in dependence on the wind speed. The overall effective power 30 delivery Pactuai over the preset implementation time tmax is thus the sum of the currently prevailing rotary speed-dependent power P and the power increase Pincrease required proportionally to the frequency deviation.

18 Figures 7 and 8 show measurements or recordings, corresponding to Figures 6 and 5, of the power target value Pref and the actually set power value Pactual. In that respect the power target value Peef concerns the target power, having regard to the power increase. The power 5 configurations shown in Figure 7 correspond in that respect to a frequency dependent and rotary speed-dependent power increase, similarly to that shown in Figure 6. The power configurations shown in Figure 8 correspond to an only frequency-dependent power increase, similarly to that shown in Figure 5. It is to be observed that Figures 5 to 8 however each represent 10 their own specific configurations. Figure 9 in relation to an embodiment shows various possible frequency-dependent increases of Pincrease in dependence on the selected value of Pincreaseset. The three curves by way of example are identified by Pincrease', Pincrease" and Pincrease". 15 The additional power Pincrease is proportionally dependent on the measured frequency deviation below the headband frequency. The power increase is increased linearly as from the headband frequency fdeadband of 00/a to the preset power increase Pincrease-set upon reaching the control band frequency fdeadband. In addition, when required by the network provider, the 20 preset power increase Pincrease set can be predetermined in 1% steps of the rated power to the maximum permissible power increase Pincrease max, from the rated power. Pincrease set is also not exceeded in the event of a major frequency deviation. Frequency changes occurring during the implementation time cause direct adaptations in respect of the power increase. 25 The ratio Pincrease / Prated in /0 can be illustrated, in dependence on the actual frequency or measured frequency fmeas and in dependence on the value Pincrease set which is specified in %, with the following formula: (Pincrease/Prated) [ 0 /0] - ((fdeadband-fmeas)/(fdead band -fcontroi band) )XPincrease set [ 0 /0] 30 Table 1 specifies characteristic values or setting ranges for an installation by way of example. In principle the headband frequency can be identified as fdeadband and the control band frequency as fcontrolband. The power 19 increase can be identified as Pincrease or Pextra and the rated power as PN or Prated. In the line 'maximum power increase' it is possible to select between the use Pextra = constant or Pextra = variable, in dependence on whether a frequency-dependent power increase or a frequency-dependent and rotary 5 speed-dependent power increase is to be used. Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated 10 integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps. The reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that the prior art 15 forms part of the common general knowledge in Australia. Further, the reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that such art would be understood, ascertained or regarded as relevant by the skilled person in Australia. 20 20 Table 1 Frequency measurement Frequency resolution 0.01 Hz Frequency accuracy 0.004 Hz Frequency recognition t = 0.02 s Frequency measurement 40 - 70 Hz range 50 Hz network 60 Hz network Maximum frequency fmax = 57 Hz fmax = 67 Hz Frequency range Rated frequency frated = 50 Hz frated = 60 Hz Minimum frequency fmin = 43 Hz fmin = 53 Hz Inertia emulation at underfrequency Maximum implementation 8 s time of the increase Detection time 0.02 s 50 Hz network 60 Hz network Deadband frequency 49 - 50 Hz 59 - 60 Hz fdeadband Control band frequency 47 - 50 Hz 57 - 60 Hz control band Power increase Power increase Pincrease set 0 - 10% with respect to Prated Max. power increase Pincrease 10% with respect to Prated max Setting option In steps of 1% with respect to Prated Normal power Additional power from 0% to 4% Prated Pextra = 0 from 4% to 100% Prated Pextra = const Pextra 10% Max. power increase Prated from 4% to 100% Prated Pextra - Pextra 10% variable Prated Gradient of the power =250 kW/s change dP/dt Recognition time 0.02 s Rise time [for 200 kW] 0.8 s Fall time or control 1.0 stabilisation time s. above - Waiting time to next > 2 x maximum duration of the increase increase 5

Claims

1. A method of operating a wind power installation, wherein the wind power installation is connected to an electrical supply network and in operation, that is to say when wind prevails which is above the initial speed and below the switch-off speed, feeds electrical energy into the supply network and more specifically with the frequency and voltage desired or required by the supply network, wherein in opera tion of the wind power installation above the rated value or below the switch-off speed the rotor of the wind power installation that has at least one rotor blade rotates and connected to the rotor of the wind power installation is a generator driven by the rotor thereby to gener ate electrical energy, wherein the wind power installation has means for measuring the frequency prevailing in the electrical supply network and said means for measuring the network frequency are connected to a control device for controlling operation of the wind power installation, characterised in that the power delivered by the generator of the wind power installation to the network is increased quickly and for a short period above the currently prevailing power of the wind power installation if the network frequency of the electrical network is below the desired target frequency of the network by a predetermined fre quency value and/or if the network frequency falls with a frequency gradient, namely with a change in relation to time, the amount of which exceeds a predetermined amount of change, wherein - the quick power increase is effected utilising the rotational energy stored in the moment of inertia of the rotor/generator system, wherein the higher amount of power is implemented by increasing the exciter current and thus by increasing the generator counter moment of the generator rotor member.

2. A method according to claim 1 characterised in that the prede termined frequency value is greater than 1 %. , preferably greater than -22 2 %o , particularly preferably greater than 3 %. of the network target frequency and/or the predetermined amount of change is greater than 0.1 Hz/s, in particular is in the range of 0.2 - 7 Hz/s, and is preferably in the range of 0.5 - 2 Hz/s.

3. A method according to one of the preceding claims character ised in that the frequency of the power fed into the network always corresponds to the currently prevailing network frequency, that is to say the power which is fed in assumes a frequency below the network frequency if the network frequency is less than the target value of the network frequency, for example 50 or 60 Hz.

4. A method according to one of the preceding claims character ised in that the increase in power is effected above the currently pre vailing power, that is to say also above the rated power, if previously the feed was effected with a rated power.

5. A method according to one of the preceding claims character ised in that within a period of 10 to 1000 ms, in particular 20 to 500 ms, preferably 50 to 100 ms, after the frequency value falls below the predetermined frequency value of the network frequency or after the predetermined amount of change is exceeded, the power increase is effected and/or feed is effected with an increased power, that is to say a power which is above the previously set power, for a further period of at least 0.5 sec, preferably at least 1 sec to a maximum of 30 sec, preferably a maximum about 3 to 6, 8, 10, 12 or 15 sec.

6. A method of controlling a wind park comprising at least two wind power installations, in which each wind power installation is con trolled in accordance with a method according to one of the preceding claim s, - 23 characterised in that the increase in the power to be delivered to the network from all wind power installations is unitarily and/or cen trally controlled.

7. A wind power installation adapted to carry out the method according to one of the preceding claims.

8. A wind park wherein the wind park includes a plurality of wind power installations according to claim 7 and/or is adapted to carry out a method according to claim6.