NZ617674B2 - Method for operating a wind energy plant in icing conditions - Google Patents

Abstract

The present invention relates to a method for operating a wind energy plant (1), comprising a nacelle (2) having an electric generator for generating electric current, and an aerodynamic rotor (3) having one or more rotor blades (4), the rotor being coupled to the generator. The method comprises the steps of operating the wind energy plant (1) if ice build-up on the rotor blades (4) can be safely ruled out, and of stopping the wind energy plant (1), if ice build-up on the rotor blades (4) is detected, and time-delayed stopping and/or preventing a restart of the wind energy plant (1), if an ice build-up was not detected but is to be expected, and/or time-delayed restart of the wind energy plant (1) if a stop condition, which led to a shut-down of the wind energy plant (1), no longer applies, and an ice build-up was not detected, and an ice build-up or the formation of an ice build-up is not to be expected. steps of operating the wind energy plant (1) if ice build-up on the rotor blades (4) can be safely ruled out, and of stopping the wind energy plant (1), if ice build-up on the rotor blades (4) is detected, and time-delayed stopping and/or preventing a restart of the wind energy plant (1), if an ice build-up was not detected but is to be expected, and/or time-delayed restart of the wind energy plant (1) if a stop condition, which led to a shut-down of the wind energy plant (1), no longer applies, and an ice build-up was not detected, and an ice build-up or the formation of an ice build-up is not to be expected.

Description

Method for ing a Wind Energy Plant in Icing Conditions A problem with wind power installations is in particular the danger due to ice falling or being thrown off. In operation of a wind power installation with iced rotor blades the fact of pieces of ice being flung off can result in danger to the close proximity. When a wind power installation is stopped the danger due to snow and pieces of ice being detached from the wind power installation does not differ substantially from the danger involved with other high structures.

A method of operating a wind power installation having regard to the possibility of icing is described in German laid—open specification DE 103 23 785 A1. Therein ing parameters such as for example the power of the wind power installation in dependence on a boundary condition such as wind speed are lly compared to reference values which occur at the respective wind speed. From deviations between the detected operating parameter and the reference operating parameter it is le to infer ice accretion and suitable protective measure can be initiated, in particular that including ge of the wind power lation.

That procedure is based on the realisation that ice accretion at the rotor blades influences the namics of the blades and thus the rotor, thereby giving deviations in the performance of the installation. They are recognised and evaluated by the described comparison of the ing parameters.

A problem in that respect is that this kind of detection presupposes a mode of operation of the wind power lation, which is as steady as possible, stable and as uniform as possible.

However such ideal conditions frequently do not prevail in the case of light wind which is assumed approximately at wind speeds below 3 or 4 , m/s. In the case of strong wind which is mostly assumed at wind speeds from 20 m/s or 25 m/s the sensitivity of such known s is often inadequate. Accordingly, any ice accretion evaluations are not very reliable or are even impossible.

A similar problem arises if the wind power installation has come to a stop e, in the stopped condition, no operating parameters can be meaningfully compared to predetermined operating parameters. In that case the wind power installation can have come to'a stop for entirely different reasons. They include a, stoppage because of excessively light wind, stoppage because of excessively strong wind, stoppage for nance es and also stoppage due to network failure of the connected electric power supply network into which the wind power installation feeds and from which the wind power installation draws energy for maintaining its operational management. Moreover, stoppage of the wind power installation because of a detected ice accretion is also considered.

Therefore the object of the present invention is to address at least one of the above-mentioned problems. In particular the invention seeks to improve ice accretion detection or icing detection so that ice accretion detection can be carried out even outside a previously secure recognition range of a wind power installation. The invention at least seeks to provide an alternative uration.

According to the invention therefore there is proposed a method of operating a wind power installation according to claim 1.

Accordingly the method is based on a wind power installation comprising a pod having an electric generator for generating electric current and an aerodynamic rotor coupled to a generator and having one or more rotor blades.

Such a wind power installation is operated in a basically known fashion, wherein more ically the rotor rotates if ice accretion on the rotor blades can be reliably excluded. That is the case in ular when the ambient temperature is high and in particular is markedly above +2°C. r, ice accretion is excluded even at low temperatures around ng point if the operating‘parameters in operation of the wind power installation are of the respective value to be ed. That means in particular that, in the part—load range, when therefore there is not sufficient 3O wind for operating the wind power installation at its l power, the power ed by the wind power installation corresponds to the power expected at the prevailing wind speed. In the oad range, when therefore the wind power installation can be operated at nominal power with the prevailing wind, this means that, in the case of a pitch-regulated wind power installation, the set rotor blade angle corresponds to the rotor blade angle to be expected at the prevailing wind speed.

If in contrast ice ac‘cretion on the rotor blades is detected the wind power installation is d. Ice ion is detected for example in that, at an ambient temperature which made ice accretion possible, more specifically in particular at an ambient temperature below +2°C, there are deviations between the actual and the expected power or deviations between the actual and the expected rotor blade angle, which indicate ice accretion. In the part—load range that is usually the case when the actual power is markedly below the expected power, because it is to be assumed that the ice accretion is ng the efficiency of the wind power installation. That is to say the ratio of the electric power generated by the wind power installation to the power afforded in the prevailing wind.

Another detection method for ice accretion is for e monitoring the' natural frequency of the rotor blades in operation. Even that is based on preconditions which do not always adequately occur. Still further methods ‘ are known which can also come up t their limiting s. They include, to give a further example, an optical method which is poor for use in fog or at night.

If now ice accretion has not been detected but is to be expected and is therefore not to be excluded, it is proposed that the wind power installation be stopped in time—delayed relationship. When ice accretion is detected was described above. It is to be expected in particular when the ambient temperature is below a limit temperature, in ular below +2°C. Admittedly still no ice may form at +2°C, but in order to exclude risks due to not detecting or not taking account of a possible ice accretion, it is proposed that that relatively high value of +2°C be taken as a basis.

That also takes account of the fact that measurement uncertainties can 3O occur, that the ature measurement is not effected directly at the potential location of icing and also that the temperature is nced by flow conditions. Alternatively it is also possible to use another value, in particular a limit value of +1°C or +3°C or +4°C.

According to the invention it was recognised thatrstopping the wind power installation in the case of an ice accretion which is not detected but which is to be expected or not to be excluded increases the safety factor for people and s in the region of the wind power lation, in which respect the ing losses of yield turn out to be comparatively slight, measured against the total annual power of the wind power installation.

That is due in particular to the fact that such a stoppage occurs at very high wind speeds which however occur rarely, or at very low wind speeds at which in any caselittle yield can be afforded.

According to the invention it was also recognised that individual pieces of ice are flung off only at greater thicknesses of ice and therefore stoppage of the wind power installation or preventing it from starting up again does not have to be effected immediately but rather can be effected in time-delayed relationship. That reduces any losses of yield, and sometimes even significantly reduces them.

In that case stopping the wind power installation and preventing it from starting up again can be effected under the same preconditions. elayed prevention of restarting of the wind power installation can mean in that respect that a stopped wind power installation is lly not prevented from starting up again, because of the time delay. It therefore starts up again and then ~ within the time delay — possibly comes to a working point at which it is possible for ice accretion to be ed reliably, in particular more reliably than ' in the stopped condition. If in that case the method is successful in reliably detecting that there is no ice accretion, that time delay, as a consequence, has had the result that‘the wind power installation has started again and is operating normally and is producing corresponding output. Without a time delay there would have been the danger that the installation does not start up, freedom from ice would not be detected, and the installation would thus initially remain ently in a stopped condition.

In the present application the expression stoppage of a wind power lation — unless it is clearly specified differently — is used to mean that the installation stops the rotor, or if need be allows it to run in a coasting mode. In that case however the operational management system s in ion, unless further disturbances occur such as for example a network failure, which ts the’operational management system from being maintained. In the event of a network failure state data are stored until the network is restored.

The time delay can start for example from the moment in time or can take account of that moment in time, at which ice accretion was to be expected or could not be excluded. In particular the time delay can begin at the moment when the ambient ature falls below a limit ature.

Additionally or alternatively it is proposed that a stopped wind power installation starts up again in time-delayed relationship when a ge condition — for example because a shadow is thrown, because of oscillation monitoring or also manually such as for example for maintenance - which lead to stoppage of the wind power installation has ceased again and ice accretion was neither detected nor is it to be expected or not excluded.

The time delay begins in particular at the time or takes account of such a time at which the condition occurs, that ice accretion was not detected and is not to be expected. That can mean that, before that moment in time, ice accretion was to be expected or even occurred. It can however also mean that, prior to that moment in time, it was not clear what kind of situation is prevailing. The time delay is ore ed to take account of the fact that, although ice accretion was not detected and is not be expected, there could still be residual ice. Sometimes the observed conditions only indicate that the formation of ice accretion is not to be expected, but information about the presence of an ice accretion can be afforded only with difficulty or not at all. In particular such a condition occurs when ambient temperatures are above and in particular are slightly above a limit temperature such as for example 2°C. At a higher temperature, in 3O particular above 2°C, ice formation is not to be reckoned with. If however ice accretion occurred until a short time ago, that can possibly still be present at least in part. Particularly in that case the result of the time delay is that any ice ion residues can thaw.

In an embodiment it is proposed that stopping or ting restarting of the wind power installation and also or alternatively causing the wind power installation to start again is effected in dependence on a ice predictive indicator. The ice tive indicator which can also simply be referred to as the tor forms a measurement in respect of the probability of ice accretion and is riately determined or altered. In that respect the ice predictive indicator is so ined or altered that it can give an indication to the probability or is so used without having to represent a probability value in the mathematical sense. The ice predictive indicator is in particular described hereinafter in such a way that a high value points to a high probability of ice accretion and a low value points to a low ility of ice accretion. On the basis of the teaching according to the invention the man skilled in the art can equally. well provide that and implement it in reverse.

The ice predictive indicator is preferably determined in dependence on operating parameters and/or ambient conditions and can also be altered in dependence thereon. ably time is taken into consideration. It is thus desirable if the ice predictive indicator is so altered that it depends on previous values and how far back in time they ed and/0r how long they have already lasted.

In an ment the ice predictive indicator is in the form of a counter. That includes in particular a configuration in which the ice predictive indicator is in the form of les which are implemented in a process computer and which basically can increase and reduce their value in any way within predetermined limits.

Accordingly in an embodiment it is provided that the ice predictive indicator alters its value in a first direction and in particular increases it when ambient conditions and/or operating conditions of the wind power installation favour ice accretion and/or point to ice accretion, in particular if 3O the ambient temperature is below a limit value. That tion takes place in particular in time-dependent relationship so that the value alters successively or continuously with increasing time. If therefore in particular the ambient temperature is below a limit temperature like +2°C, then that value progressively increases with time until it has reached such a high value which can be stored as a limit value and which can be referred to as a stoppage limit value, at which the wind power installation is stopped and at which the wind power installation is prevented from starting up again. If the wind power installation for example is in a condition in which, on the basis of past , it is to be assumed that there is no ice accretion and if the overall situation changes to conditions at which ice accretion can no longer be excluded, then the counter begins slowly to increase. Until it has d the stoppage limit value ~ if it reaches it at all — the time elapsed also depends on the rate of increase of that counter. onally or alternatively it is proposed that the counter changes its values into a second direction, in particular reduces them, when ambient conditions or operating conditions of the wind power installation favour or indicate that ice accretion is not t or is reducing, in particular if the ambient temperature is above the limit temperature.

If therefore for example the situation is one in which an ice accretion is to be assumed to be t or that ice accretion has been detected, or if the situation is unclear and if the situation changes to one in which ice accretion or at least ice accretionoccurrence can be excluded, then the value of the ice predictive indicator, that is to say the counter value, gradually reduces with time. It is reduced until a lower counter limit value is reached, in particular a restart limit value.

The above-described processes in increasing or reducing the ice predictive counter can last for some hours to up to 10 hours or even longer. In that time a situation which points to the possibility of icing and which results in an se in the counter can change to a situation in which it is possible to assume that there is a reduction in ice ion, in particular thawing, or in which there are reliable values which e ice ion. The value of the ice predictive indicator or the counter is in that 3O case reduced again. Equally a reverse situation can occur, in which the counter counts up again. Due to the situation ed therefore the direction in which the value of the ice predictive indicator changes alters.

That takes account of the respective past situation. Therefore preferably one and the same counter is used for the increase and the reduction.

In a further ment it‘ is proposed that an tion in the value, that is to say the value of the ice predictive indicator in the form of a counter, is effected at a speed which is dependent on the ambient conditions and/or the wind power installation ing conditions.

Accordingly therefore the value does not always increase or reduce in the same way with time, but also takes account of a differentiated way of ering the ling conditions.

Preferably in that respect, with a prevailing light wind, that is to say in particular a wind involving wind speeds below 4 m/s, the value is increased more slowly than when a strong wind prevails, more specifically in particular at wind speeds of over 20 m/s if the installation is being operated. That is based on the realisation that, at very high afflux speeds at the rotor blade, due to operation of the installation at high wind speeds, ice accretion can form more rapidly and thus the time until the wind power installation is stopped should be shorter. That can be taken into account by more rapidly increasing the value of the ice predictive indicator which thus rapidly reaches a stoppage limit value. However there is also the possibility of implementing a greater time delay with a prevailing light wind in some other fashion than by the ice predictive indicator in the form of the counter, like for example by means of a reference table or look-up table.

Preferably an increase in the value of the ice tive indicator is also effected more slowly than in ion of the installation with a prevailing strong wind, if the wind power lation was stopped because of an automatic installation stop as in the event of shutdown due to shadow casting or because of a lack of wind or in the event of a manual installation stop such as for example for maintenance, independently of the prevailing wind speed. 3O A further ment proposes, additionally or atively, reducing the value of the ice predictive indicator in the form of a counter more slowly, the lower the ambient temperature is, and in particular . reducing the value proportionally to an integral formed from time over a difference in the ambient ature relative to the limit ature.

That gives a time delay which is correspondingly shorter, the higher the ambient temperature is. In other words, the wind power installation can start up again correspondingly , the warmer that it is. Causing the wind power installation to start up again with a correspondingly r time delay, the higher the t ature is, can also be implemented otherwise than by using the ice predictive indicator as the counter. For example a table, a so—called look-up table, can be provided, which specifies the time delay values for given ambient temperatures.

A further embodiment is characterised in that the wind power installation is coupled to an electric network and is stopped in the event of a network failure while, upon network restoration, that is to say when the network failure is cured, the wind power installation is started up again in dependence on a measurement temperature which depends on the ambient temperature at the network failure and the ambient temperature at the network restoration. That is based on the notion that, for the duration of the network failure, more ically from the beginning thereof to network restoration, information ng to operating parameters and t conditions and in particular ambient temperature are not available or are only limitedly available. In order to be able to better estimate the possibility of ice accretion after the end of the network failure, this procedure adopts a temperature value for the ambient temperature which depends on the temperature at network restoration, that is to say the tly prevailing temperature, and the last detected ambient temperature prior to or at the beginning of the network failure.

Preferably the ement temperature is calculated as a mean value from the ambient temperature at the beginning of the network failure and the ambient temperature upon network restoration, if the network 3O failure is not more than a first failure time, in particular not more than two hours. Here the underlying ation is that the ambient temperature does not change too rapidly and at short failure times consideration of the ambient temperature prior to and after the network failure can already supply meaningful information about the ility of ice accretion. If for example the ambient ature at netWork restoration is 2°C, ice accretion is probable if the ambient temperature at the ing of the network failure was markedly below that, whereas ice accretion is improbable if the ambient temperature at the beginning of the k failure was markedly higher.

It is desirable, in the event of a longer network failure, to provide a ature safety value for determining the ement temperature which can also be referred to as the calculated temperature. Thus it is ed that the measurement temperature be reduced by 2 K in the case of a longer network e, in particular a network failure of over two hours.

In an embodiment the wind power installation is arranged in a wind park and is stopped if at least one further wind power installation of the wind park is stopped because of ice accretion or predictive suspicion thereof. That is based on the realisation that wind power installations behave rather similarly at any event in respect of ice accretion, in the same wind park, because in particular the ambient parameters such as ambient temperature, air humidity and wind speed are similar. However that is also based on the realisation that ice accretion of a wind power installation in the park admittedly does not have to necessarily signify that all other wind power installations in the park also involve ice accretion, but that the probability of ice accretion on the other wind power installations of the same park is high. ect prognoses should therefore occur only rarely and should thus scarcely influence the overall yield of the wind power installation over the course of the year, while the safety aspect, mainly preventing ice from being flung off, can sometimes be icantly increased.

Preferably a wind power installation which was stopped because of a 3O detected ice accretion or suspicion thereof will orient its pod in such a way that a spacing which is as great as possible in relation to regions ered by falling ice, in particular traffic routes and objects, is maintained. In that way not only is the risk due to ice being flung off reduced, but also the danger due to ice purely falling, as can basically also occur in relation to other high building structures.

Preferably a wind power installation is used, which has a heatable wind sensor for measuring the wind speed and heats it at least in the event of a ion of ice accretion. For example it is possible to use a so—called ultrasonic anemometer. Thus, even for the situation involving ice formation which can occur not just on the rotor blades but for example also on the anemometer, that is also intended to provide for wind speed measurement which is still le. Accordingly ice accretion detection means which require a reliable wind speed can then still be used.

Preferably it is proposed that an ice sensor be used, which directly measures ice accretion on the wind power installation, in particular on one or more of the rotor blades. Such measurement can supplement the above-described ice detection modes. It is to be noted that the use of an ice sensor firstly involves ponding capital investment costs. They can possibly be quickly amortised if clear information from such an ice sensor that there is no ice accretion means that the wind power lation can be operated without it having to be otherwise stopped, as a precautionary measure.

Preferably it can be provided in a wind park that only some or one of the wind power installations is fitted with such an ice sensor and information obtained therefrom about ice accretion is itted to other wind power installations in the park, that do not have an ice sensor. In that way the costs of an ice sensor can be distributed among a plurality of installations. Preferably information ed by an ice sensor about an ice accretion is evaluated and in particular stored together with the respectively ling ambient and/or operating conditions of the respective wind power installation in order to improve tion of ice accretion, in particular to individualise same for the tive wind power 3O installation. Ice accretion detection can thus be respectively adapted to the type of lation and the place of erection thereof, in particular by a suitable learning programme.

The invention is described in greater detail hereinafter by way of e by means of embodiments with reference to the accompanying Figures.

Figure 1 shows a perspective view of a wind power, Figure 2 shows the pattern of an ice predictive indicator for different wind speeds, Figure 3 shows the pattern of an ice predictive indicator for two different ambient temperatures, and Figure 4 diagrammatically illustrates the n of an ice predictive indicator of an embodiment in dependence on a temperature pattern by way of example.

Figures 1 shows a wind power installation 1 comprising a pod 2, an aerodynamic rotor 3, rotor blades 4, a spinner 5 and a pylon 6.

Figures 2 plots the pattern of the ice predictive indicator, namely its value, in relation to time for two examples. In accordance therewith a distinction is drawn between prevailing wind speeds which can be identified as strong wind on the one hand and those which can be identified as a light wind on the other hand. In both examples Figure 2 concerns the ion where the wind power installation is in ion and the rotor of the installation is rotating and therefore the installation is not stopped. At the time t1 = 0 an event occurs, which rs upward counting of the ice predictive indicator. That can be for example that the ambient temperature falls below a limit temperature. It is however for example also considered that the temperature is already below the limit temperature and the prevailing wind speed drops to a value such that it must be assumed that there is a light wind situation, or the prevailing wind speed rises to a value such that it must be assumed that there is a strong wind situation.

The value of the ice tive indicator prior to the time t1 is not important. That can be for example of the value 0 or a value is first 3O attributed at all to the ice tive tor at the time t1.

The initial value of the ice predictive indicator can also be viewed as a value which leads to the wind power installation starting up again in another case. That is not involved in the case shown in Figure 2 however so that the value "start" is only shown in brackets.

At any event at the time t1 'there is a condition, by virtue of which the ice predictive indicator is continuously increased with time. The increase is effected for the prevailing strong wind more rapidly than for a prevailing light wind. Thus, with a prevailing strong wind, the ice predictive tor already reaches a value at which the wind power installation is stopped, at the time t2. That value is identified in by the 2 , Figure horizontal broken line noted as Stop. In the example, with a ling strong wind, the ice predictive indicator reaches the criterion for stoppage of the wind power installation after 2 hours. In the case of a light wind the criterion for stoppage of the wind power installation is only reached at the time t3 which in the example is 10 hours.

Figure 2 is a simplified view which is essentially based on the fact that the prevailing ry conditions are substantially steady—state.

Figure 3 also shows two patterns by way of example of the ice predictive indicator, but for the situation where the wind power installation is stopped. At the time t1 which for simplification is specified as 0, there are criteria which have the result that the ice predictive indicator is d. Here too its initial value is firstly not important and it can correspond to the situation in which the installation was stopped, from which reason the "Stop" at the ordinate is shown in brackets. The reduction which can also be referred to an downward ng of the ice predictive indicator is dependent on a differential temperature, namely the current ambient temperature in on to a limit temperature, n the ambient temperature must be greater than the limit temperature. That differential temperature is specified as AT in Figure 3. The illustration is based on the tion that steady—state conditions prevail, namely that the differential temperature AT in the one case is constant at 3 K while in 3D the other illustrated case it is nt at 1 K.

The value of the ice predictive indicator as shown in Figure 3 is reduced in accordance with the integral of the differential temperature over time. In the illustrated examples in Figure 3 the differential temperature is avfl. thus a constant temperature value, namely 3 K in one case or 1 K in the other case, which is integrated over time. In the case of the greater temperature difference of 3 K therefore the ice predictive indicator already reaches the value at which the installation is started again at the time t2, as is shown by the word "Start". In the illustrated example the lation is thus started again after 2 hours.

In the case of the smaller ature difference of only 1 K the ice predictive indicator only reaches the value at which the lation can be started again at t3. As the temperature difference here is only one third of the first example, t3 is reached after 6 hours.

With the illustrated patterns an integration time nt is used, which s on the prevailing wind situation. That integration time constant is greater in the case of a prevailing strong wind, more specifically in the illustrated example by the factor of 3, than with a prevailing light wind. Correspondingly, the ice predictive indicator reaches the value at which the installation starts up again, three times as quickly in the case of a strong wind. Those values are specified as t2‘ 40 minutes for a temperature difference AT = 3 K and t3' = 2 hours for a temperature difference of AT = 1 K in Figure 3.

Referring to Figure 4, this shows in an example how the ambient temperature influences the pattern of the ice predictive indicator of an embodiment. For that purpose the upper view in Figure 4 shows the pattern of the ice tive indicator, initially based on an installation operating in an uncertain ice detection range. The illustration correspondingly also applies if the lation is stopped. The lower view shows a notional pattern of the ambient temperature. The illustrated n of ambient temperature was selected for the purposes of clear illustration and does claim to be able to correspond to a possible real ature pattern of an ambient temperature. 3O The present example takes a limit temperature of TG - 2°C as its basis. The actual temperature is firstly about 4°C and is thus above the limit temperature. As the ice predictive indicator is firstly not yet set or involves a start value and the wind power installation is in operation and its rotor is rotating the temperature Initially has no effect on the illustrated pattern of the ice predictive indicator.

At time t1 the temperature reaches the value of the limit temperature and falls further. There is thus basically the risk of ice accretion and the ice predictive indicator thus begins to rise from the time At the time t2 the temperature is below the limit temperature and now rises again. That initially however has no influence on the pattern of the ice predictive indicator and it rises further.

At the time t3 the temperature exceeds the limit ature and rises further continuously. The ice predictive indicator thus does not rise any further from the time t3 e ice accretion or the ence of’ice accretion is no longer to be assumed. Rather, the ice predictive indicator is now d again. e the temperature and therewith also the differential temperature rise that gives an integral thereover, basically a second-order configuration.

At the time t4 the temperature is at a value markedly above the limit ature and initially retains that value. Accordingly there is a reduction in the ice predictive indicator in the form of a linear portion.

At the time t5 the temperature continuously falls and the ice predictive indicator is correspondingly only reduced still more and more slowly.

At the time t6 the temperature reaches the limit temperature again and falls r. Thus the ice predictive indicator is further increased as from the time t6.

At the time t7 the ature rises again but remains below the limit temperature. The ice predictive indicator thus further increases unchanged.

At the time t8 the temperature is still below the limit temperature. 3O Here however the ice predictive indicator has reached the value which leads to stoppage of the wind power lation. That is identified on the ordinate by the word "Stop".

From the time t8 the temperature admittedly r rises but it initially remains below the limit ature. As the installation is already stopped the ice predictive indicator’is not r altered, which is indicated in Figure 4 by a constant value in the upper part.

At the time t9 the temperature has reached the temperature limit value and rises further. The ice predictive indicator is now reduced again but the installation remains stopped. If the ice predictive indicator now reduces further until it reaches the value start which is shown just above the abscissa it can start again, which however is no longer further shown Figure 4.

Expressed in simplified terms the modes of operation for increasing the ice predictive indicator as shown in Figure 2 and for reducing the ice predictive indicator as shown in Figure 3 are combined in Figure 4. Thus those relationships are combined in Figure 4, which corresponds to one embodiment. In principle however the relationships or modes of operation in Figure 2 on the one hand and Figure 3 on the other hand can also be used separately from each other.

In an embodiment it is thus possible to expand ice detection or ice accretion detection by an operating status which can be referred to as ice suspicion or ice prediction. By way f the operating ions in which icing that possibly occurs would not be reliably ised are to be detected. In principle ice accretion detection is effected by monitoring the operating characteristics of the wind power installation and it can thus be limited to the operating range of the wind power installation with power generation. If the wind power installation is not generating any power no compensation with the operating teristics or the management map is also le. Ice accretion detection can thus only limitedly function, under certain conditions.

Those d conditions are now also taken into eration.

They include: - Light wind: here ice detection in the case of operation during very low wind speeds, in particular below about 3 to 4 m/s, by monitoring the operating characteristics is not reliably possible.

— Strong wind: in the case of operation during high wind speeds over about 20 to 25 m/s the sensitivity of the previous detection method falls and/or often cannot, be verified by existing operational experiences.

- Installation ge with the installation ready for operation.

- Network failure.

Thus the previous detection range in respect of ice accretion detection in operation, which can also be referred to as the certain detection range, is limited to wind speeds of between about 4 m/s and 20 m/s, having regard to a safety margin.

Residence durations at low temperatures, mainly ambient temperature below +2°C, increase the icing prediction or ion. At temperatures above +2°C in contrast the suspicion or prediction of icing is reduced again. Likewise in operation of the wind power installation in the certain ice accretion detection range the prediction or suspicion of icing can be d.

The proposed methods e in particular a method which less proposes certain ice ion, but which rather takes account of the possibility of ice occurrence.

Preferably, for operation at wind speeds below 4 m/s, it is assumed that a critical thickness of ice could have formed only within 10 hours.

Accordingly that is taken into account in Table 1 hereinafter under the heading Mode 1.

For operation with a strong wind, because of the higher afflux speeds at the rotor blade, it is assumed that a critical layer of ice could already have formed within 2 hours. Those onships are correspondingly taken into eration in the Table hereinafter as Mode II.

In the case of an automatic installation ge, as occurs for example e of a slight wind or because of shadow—casting shutdown, or in the case of a manual installation stoppage as for example for maintenance purposes, it is assumed that a critical thickness of the layer of ice could have formed within 10 hours. Accordingly that is also taken into consideration in the Table hereinafter under the heading Mode I.

With a network failure, it is often not possible for any wind and temperature data to be recorded by the installation control system.

Howeverfthe last data prior to the network e and the data upon network restoration are available. ng counter states in respect of ice accretionfdetection, in particular the value of the ice prediction indicator, also remain available. The times with network failure are taken into account as follows, in dependence on the duration thereof.

Network failure times of up to 2 hours are taken into account with the mean value from the temperature at the ing of the network failureand the ature upon network restoration in accordance with Mode I as described in the Table below. Therefore with that mean temperatUre value which was also referred to as the determination temperature, an ice predictive indicator is increased or counted upwards if that meah temperature value is below a limit temperature. If it is above that limit temperature the ice predictive indicator is correspondingly reduced. That is correspondingly ed for the on of the network e time, as the underlying time involved.

Network failure times of between 2 and 10 hours are taken into account to cover falls in ature in the meantime, with the mean value from theltemperature at the beginning of the network failure and the temperatLire upon network ation, less 2 K, in accordance with Mode I describedjin the Table hereinafter.

In the case of network failure times for example of over 10 hours, it is assumed that reliable information about the past period of time is not possiblefl: For that purpose, having regard to a safety margin, at all temperatures below +5°C, upon network restoration, it is assumed that there is a? suspicion of icing. The wind power installation therefore initially remains stopped until icing can be excluded.

Implementation for the described Modes I and II is effected by way 3O of a counter which can also be referred to as the ice predictive counter or the predictive counter and which counts upwards upon a prediction or suspicion of icing and counts down again t that suspicion. In that ' 20 case, the times n Mode I and Mode II are different, according to the situation in terms of icing suspicion.

With 30 minutes of operation without ice detection in the certain detection range in respect of ice accretion detection, that is to say upon detection using a power curve method in which the measured power curve is compared to a curve to be expected, the suspicion of icing is d. If therefore certain detection applies, 30 minutes are sufficient, irrespective of the mode used.

In an embodiment, at outside temperature above +2°C, the difference ing +2°C in respect of the currently prevailing outside temperature in relation to time is summed or integrated. Restarting is then effected only after the expiry of a differential temperature-time integral.

Thus for example restarting is effected at 360°Cmin. That can mean for example that restarting is effected after 6 hours at an ambient temperature of +3°C or after 2, hours at an ambient temperature of +5°C. In Mode II restarting is y effected for that case after in.

Mode I/duration Mode II/duration Icing suspicion 600 min 120 min (upward countingL Operation in the certain l detection range (downward counting) 30 min 30 min Differential temperature—time 360°Cmin integral for outside in temperature >2°C (downward counting) i The times specified in the foregoing Table correspond to those for complete increase and reduction respectively in terms of the tive suspicion of icing. Intermediate stages are suitably proportionally evaluated.

At the transition from operation of the wind power installation with the rotating rotor to the stopped condition and vice-versa the counter states for ice accretion ion and for the predictive ion of icing are suitably transferred or ed. That is intended to ensure that wind power installations are stopped or automatic ting can be prevented even in the case of a prolonged residence time outside a detection range, which is assumed to be certain, in respect of the usual ice accretion detection, because of an uncertain icing condition with a suspicion or prediction of icing. Such a prolonged residence duration es for example one of 10 hour or more under prevailing wind conditions below 4 m/s or a residence duration of 2 hours or more under prevailing wind speeds of over 20 m/s.

Claims (16)

1. A method of operating a wind power installation comprising a pod with an ic generator for ting electric current and an aerodynamic rotor coupled to the generator and having one or more rotor blades, including the steps: - operating the wind power installation when ice accretion on the rotor blades can be certainly excluded, and - stopping the wind power installation when ice accretion on the rotor blades is ed, and - delaying the restarting of the wind power installation when an ice accretion was not detected but is to be expected or cannot be excluded, or - delaying the resumption of operation of the wind power installation until the ice accretion which led to the stoppage of the wind power installation has disappeared and further ice accretion is not detected and ice accretion or the ion of further ice accretion is not to be expected.

2. A method according to claim 1 characterised in that stoppage or prevention of restarting of the wind power installation and/or tion of operation of the wind power installation is effected in dependence on an ice predictive indicator which is determined or altered as a ement in respect of the probability of ice accretion.

3. A method according to any one of the preceding claims characterised in that the or an ice tive indicator is in the form of a counter and - alters its value in a first direction when t ions and/or operating conditions of the wind power installation favour ice ion and/or indicate ice accretion when the ambient temperature is below a limit temperature, and/or - alters its value in a second direction, when ambient conditions and/or operating conditions of the wind power installation promote and/or indicate that ice accretion is not present or is reduced when the ambient temperature is above the limit temperature, and - the limit temperature is just above freezing point and in particular the temperature is in the range of 1 to 4oC, preferably about 2oC.

4. A method according to claim 3 characterised in that a change in the value is effected at a rate dependent on the ambient conditions and/or ing conditions of the wind power installation, and that - the value with a prevailing light wind is increased more slowly than with a prevailing strong wind when the installation is in operation, and/or - the wind power installation with a prevailing light wind is stopped after a greater time delay than with a prevailing strong wind, and/or - the value in each case is reduced more slowly the lower the ambient temperature, wherein the value is proportional to an integral of time over the difference in the t temperature to the limit temperature, and/or - the time delay after which the wind power installation starts up again is shorter, the higher the ambient temperature is.

5. A method according to any one of the preceding claims characterised in that the wind power installation is coupled to an ic k and is stopped in the case of a network failure and upon network restoration restarting of the wind power lation is effected in dependence on a ement temperature dependent on the ambient temperature at the beginning of the network failure and the ambient temperature upon network restoration.

6. A method according to claim 5 terised in that - the measurement temperature is calculated as a mean value from the ambient temperature at the beginning of the network failure and the ambient temperature upon k restoration if the network failure is not more than a first failure time and/or - the measurement ature is calculated as the mean value from the ambient temperature at the beginning of the network failure and the ambient temperature upon network restoration less a temperature safety value, if the network failure was longer than the first failure time.

7. A method according to claim 6 wherein the first failure time is not more than 2 hours.

8. A method according to claim 6 or 7 wherein the temperature safety value is 2 K Kelvin.

9. A method according to any one of the preceding claims characterised in that the wind power installation is arranged in a wind park and is stopped when at least one further wind power installation of said wind park is stopped because of ice accretion or suspicion of ice accretion.

10. A method according to one of the preceding claims characterised in that the wind power installation which is stopped because of detected ice accretion or suspicion of ice accretion s its pod in such a way that a spacing which is as great as possible in relation to s ered by ice fall is maintained.

11. A method according to any one of the preceding claims characterised in that the wind power installation has a heatable wind sensor for measuring the wind speed and the wind sensor is heated when an ice ion was detected and/or when an ice accretion cannot be ed.

12. A method according to claim 11 wherein the heatable wind sensor is an ultrasonic anemometer.

13. A method according to any one of the preceding claims characterised in that the wind power installation has an ice sensor and ice accretion is detected by direct measurement by means of the ice sensor.

14. A wind power installation sing a pod with an electric generator for generating ic current and an aerodynamic rotor coupled to the generator and having one or more rotor blades, wherein the wind power installation is prepared for operation with a method according to any one of the preceding claims.

15. A wind park sing at least one wind power installation according to claim 14.

16. A method of operating a wind power installation substantially as herein before described with reference to accompanying