Patent Application
20170022974
The present disclosure relates to methods of operating a first wind turbine and a second wind turbine in a situation wherein presence of the first wind turbine affects the wind so that a wake is generated that affects the second wind turbine. The present disclosure further relates to control systems for operating a plurality of wind turbines, and to wind farms including any of such control systems.
Modern wind turbines are commonly used to supply electricity into the electrical grid. Wind turbines generally include a rotor with a rotor hub and a plurality of blades. The rotor is set into rotation under the influence of the wind on the blades. The rotation of a rotor shaft drives a generator rotor either directly (“directly driven”) or through the use of a gearbox. The gearbox (if present), the generator and other systems are usually mounted in a nacelle on top of a wind turbine tower.
Wind turbines are often grouped together in so-called wind farms. In the present disclosure, a wind farm is to be regarded as a cluster of two or more wind turbines. In a wind farm there may be a relatively short distance between wind turbines. Thus, action of the wind on one wind turbine may produce a wake which may be received by another wind turbine. A wake received by a wind turbine may cause loads (particularly vibrations) and/or a reduction of electrical power production in this wind turbine. These loads may damage components of the wind turbine, and this damage may reduce the life and/or the performance of the wind turbine. Therefore, in a wind farm, monitoring is carried out in order to determine possible wake situations and the wind turbines are operated in order to minimize negative effects caused by the wakes.
Currently, it is known that some wake management strategies are defined by simulating operational conditions (e.g. loads, generated power, etc.) for a theoretical layout and parameters of the wind. For every wind turbine in the layout, a set of adjustments in the operation of the wind turbine (e.g. stops, curtailments) is defined for predetermined wind directions in order to perform an operation of the wind turbine as optimal as possible. Optimal operation may be in terms of e.g. producing maximum power, minimizing loads, etc. depending on the main goal pursued at each moment. These adjustments are entered in a control system (e.g. a SCADA control system) which applies them in the wind farm by e.g. sending suitable control signals to e.g. the corresponding pitch systems, brakes, etc.
An optimizing algorithm may be used to improve the performance of at least part of the wind farm based on the interaction between wind turbines. Wake effects may be detected and evaluated by using one or more theoretical wake models including empirical factors (parameters, constants, etc.) which are predefined based on previous experience at existing wind farms (e.g. from operational data collected from before installation of the wind farm). This way of computing wakes may produce results that greatly diverge from what is actually happening between wind turbines due to variations of characteristics of the environment. For instance, the air density may vary, the geography may change due to e.g. trees growing and/or constructions erected in the vicinity of the wind turbines, etc. Furthermore, the models themselves, which are based on existing wind farms, may not be appropriate for the wind farm under consideration as particular conditions of the existing wind farms may not be applicable. This divergence may cause deficient operation of the wind turbines and the wind farm as a whole.
It is an object of the present disclosure to provide methods and systems for operating wind turbines that at least partially reduce at least one of the aforementioned drawbacks, leading to improved performance of the wind turbines as a whole.
In a first aspect, a method is provided of operating a first wind turbine and a second wind turbine in a situation wherein presence of the first wind turbine affects the wind so that a wake is generated that affects the second wind turbine. The method includes determining one or more parameters of the wind at the first wind turbine, and determining one or more parameters of the wind at the second wind turbine.
The method further includes determining a value of a parameter of a theoretical wake model to determine a current wake model. This value is determined based on the one or more parameters of the wind at the first wind turbine, and on the one or more parameters of the wind at the second wind turbine. The method still further includes optimizing the operation of the first and second wind turbines based on the current wake model.
In some examples, the method may further include determining a divergence between a previously determined wake model and the current wake model, and verifying whether the divergence exceeds a predefined threshold. In the case of a negative result of the verification, the previously determined wake model may be taken as the current wake model.
The previously determined wake model may be either a wake model determined in a previous execution (or iteration) of the method (i.e. based on real parameters of the wind at the first and second wind turbines) or a baseline wake model. For example, the previously determined wake model may be the wake model that has been determined most recently in a preceding execution (or iteration) of the method.
The baseline wake model may refer to a wake model that has been predetermined based on data other than real parameters of the wind (at the first and second wind turbines) because e.g. no execution of the method has still been performed. The baseline wake model may have been determined based on e.g. data from other wind farms. Once a first execution (or iteration) of the method has been performed, the baseline wake model may not be used anymore. In this case, a wake model determined in a previous iteration or execution of the method may be used instead of the baseline wake model.
Determining parameters of the wind at first and second wind turbines may include determining wind speed and/or wind turbulence and/or wind direction at first and second wind turbines. Other parameters are also possible.
Obtained wind speeds may be used to calculate a value of wind speed deficit, which may be compared to a wind speed deficit obtained from a wake model such as e.g. Jensen model. Similarly, obtained wind turbulences may be used to calculate a value of added wind turbulence, which may be compared to an added turbulence obtained from a wake model such as e.g. Frandsen model.
The current wake model may be determined from a theoretical wake model based on real operational data (parameters of the wind) determined at first and second wind turbines participating in the wake. The real operational data may include e.g. wind speed at first and second wind turbines, from which a real speed deficit may be calculated. A theoretical wake model may include an analytical function expressing the speed deficit depending on a factor (or constant). Determining the current wake model may include calculating a value for the factor/constant that makes the analytical function (of the theoretical wake model) to produce a speed deficit substantially equal to the “measured” real speed deficit. Hence, the current wake model may be seen as a particular version of the theoretical wake model including the recalculated factor/constant. Which parameter or property of the wind is to be taken into account depends on the theoretical model used for describing wake behaviour.
For example, Jensen model includes an analytical function mathematically expressing the speed deficit depending on variables, such as e.g. rotor diameter and thrust coefficient, and on a predefined decay constant. A new value for the decay constant of the Jensen model may be calculated that produces the real speed deficit calculated from the real wind speeds measured at first and second wind turbines. Other theoretical wake models may include other analytical functions including other empirical factors or constants that may also be recalculated based on real parameters of the wind. For instance, Frandsen turbulence model includes an analytical function mathematically expressing added wind turbulence. Other wake models that can be used in the context of the suggested method are e.g. Larsen model and Ainslie model.
These theoretical wake models (Jensen, Frandsen, Larsen, Ainslie, etc.) are well known in the technical field of the present disclosure, so no further details will be provided herein about them.
The proposed method of operating first and second wind turbines is based on determining a current wake model depending on real parameters of the wind measured at first and second wind turbines. Then, the current wake model may be inputted in an optimization process to optimize the operation of the wind turbines. An aspect of the method may thus be that wind turbines are operated more optimally because wake is modeled depending on real operational data (parameters of the wind) rather than on predefined data. In some examples, determining a current wake model describing the behaviour of a wake in a certain wind farm may be done in real-time.
At least some of the parameters of the wind at any of the first and second wind turbines may be determined depending on one or more measurements from a LIDAR associated with the wind turbine. The LIDAR may be arranged in the vicinity of the wind turbine in e.g. a frontal position, such that parameters of the wind received by the wind turbine may be reliably measured.
At least some of the parameters of the wind at any of the first and second wind turbines may also be determined depending on one or more operational characteristics of the wind turbine. These operational characteristics may include at least one of: pitch angle, yaw angle, rotor speed, rotor torque and generated power. Wind turbines may include sensors configured to obtain measures that permit determining such operational characteristics when required.
At least some of the parameters of the wind at any of the first and second wind turbines may also be determined depending on one or more loads measured at the wind turbine. Wind turbines may include load sensors through which the load measurements are obtained.
A particular wind parameter may be determined through any one of the previously described manners or through a combination thereof. In this latter case, the different values obtained for the wind parameter may be averaged such that a more reliable value of the parameter is obtained. The different values of the parameter may be suitably weighted in the averaging depending on an estimated reliability of the algorithm used to determine every value of the wind parameter.
According to examples, the first and second wind turbines may be operated by controlling one or more operational parameters of the wind turbine. Optimum values of the operational parameters (to be controlled) may be obtained either from one or more matrices (or lookup tables), or from one or more functions, or from a combination of both. Optimum values of the operational parameters may be those that maximize parameters of an optimization objective depending on parameters of the current wake model. For example, optimum values of the operational parameters may be those that maximize e.g. the generation of power or the reduction of loads depending on the current wake model.
In some examples, methods of operating a plurality of wind turbines may be provided. These methods may include detecting one or more pairs of the wind turbines having a first wind turbine and a second wind turbine in a situation wherein presence of the first wind turbine affects the wind so that a wake is generated that affects the second wind turbine. These methods may further include operating, for at least some of the detected pairs of wind turbines, the first and second wind turbines of the pair of wind turbines by performing any of the methods of operating first and second wind turbines described before. At least one of the pairs of wind turbines may be detected depending on a previously determined wake model for the first and second wind turbines of the pair of wind turbines, so that the detection may be based on more real data and therefore may be more reliable. The previously determined wake model may be the wake model determined in the most recent execution (or iteration) of the method.
In a second aspect, control systems are provided for operating a plurality of wind turbines which e.g. may be comprised in a wind farm. These control systems include a processor and a memory. The memory stores computer executable instructions that, when executed, cause the processor to perform any of the previous methods of operating a plurality of wind turbines.
In a third aspect, wind farms are provided including a plurality of wind turbines and any one of the previously described control systems.
Non-limiting examples of the present disclosure will be described in the following, with reference to the appended drawings, in which:
The control system 10 may be connected 12 with the wind turbines T1-T8, such that the control system 10 may receive measurements (e.g. load measurements, wind measurements, yaw measurements, etc.) from sensors associated with some or all of the wind turbines T1-T8. The control system 10 may also send, through the connection 12, proper signals (e.g. set points) to the wind turbines T1-T8 for adjusting their operation as a whole. The control system 10 may include a memory and a processor. The memory may store computer program instructions executable by the processor. The instructions may include functionality to execute one or more examples of a method of operating a plurality of wind turbines, which are described in other parts of the description. The control system 10 may further include a repository 11 for obtaining and storing data related to wind turbines T1-T8, such as e.g. the layout of the wind farm, distances between wind turbines, dimensions of the wind turbines, theoretical wake models, control strategies, etc.
Each or some of the wind turbines T1-T8 may have an individual controller configured to operate the wind turbine depending on individual parameters and control signals (set points) received from the control system 10. These individual controllers may control operational parameters (pitch angle, rotor speed, rotor torque, etc.) of the wind turbine in such a way that set points from the control system 10 and individual requirements are satisfied.
The wind with substantially unchanged speed 203 is shown passing surrounding the rotor of wind turbine T7. The wind with reduced speed 204 is shown influencing the rotor of the wind turbine T7, such that e.g. less power is generated by the wind turbine T7. An excessively reduced wind speed 204 may cause the stop of the wind turbine T7. An objective of an example of a method of operating first and second wind turbines T7, T8 may be maximizing power generation by the wind turbines T7, T8 as a whole, and this may be achieved by e.g. avoiding the stop of the wind turbines T7, T8. Taking this aim into account, the method may suitably vary the operation of the wind turbine T8 in such a way that the wind speed 204 is less reduced so that the stop of the wind turbine T7 is avoided. Turbines T7, T8 may thus be operated by the proposed method in such a way that their performance is improved as a whole.
Jensen model is an example of theoretical wake model which describes the wind speed deficit schematically illustrated by
The wind without added turbulence 209 is shown passing surrounding the rotor of the wind turbine T7. The wind with added turbulence 210 is shown influencing the rotor of the wind turbine T7, such that e.g. higher loads are suffered by the wind turbine T7. Excessively increased wind turbulence 210 may cause too high loads on the wind turbine T7 and its individual operation may be varied accordingly with the aim of reducing such loads. This variation of the individual operation may cause a reduction of power generation or even the stop of the wind turbine T7. An objective of an example of a method of operating first and second wind turbines T7, T8 may be maximizing power generation by the wind turbines T7, T8 as a whole. This objective may be achieved by avoiding the reduction of power generation by wind turbine T7 and/or the stop of the wind turbine T7. Taking this goal into account, the method may cause operation of the wind turbine T8 in such a way that e.g. the added turbulence 208 and its affectation on the wind turbine T7 are attenuated. Turbines T7, T8 may thus be operated by the proposed method in such a way that their performance is improved as a whole. Alternatively, an objective of an example of a method of operating first and second wind turbines T7, T8 may be limiting loads suffered by T7.
Frandsen model is an example of theoretical wake model which models the effect of added turbulence schematically illustrated by
At block 301, parameters representative of weather conditions at the wind farm site may be obtained from a reference mast or similar arranged at wind farm level. Parameters of the wind (speed and direction) may be systematically measured and obtained, whereas other parameters (wind turbulence, temperature or density) might optionally also be acquired.
These data may be stored in a Database (DB) 315 for later use, such as e.g. for empirical analysis of wind farm data. If just a short time has elapsed since the last execution of block 301, recent data on parameters of the wind and possibly other aspects may be obtained from DB 315 instead of from the reference mast.
At block 302, wake situations in the wind farm may be identified from the ambient data obtained at block 301 and one or more surrogate models associated with the wind farm from DB 315. Situations in that wind conditions (i.e. wind speed and wind direction) are such that a first wind turbine causes a wake affecting a second wind turbine may be detected by processing data from block 301 and data from DB 315. Surrogate model may include geometrical data, i.e. layout of the wind farm including distribution of wind turbines, distances between wind turbines, dimensions of wind turbines, etc. Surrogate model may further include wake model(s) that can be used to model and estimate presence and relevance of wake situations substantially in real time.
As commented in other parts of the description, a theoretical wake model may include parameters characterizing e.g. wind turbines and their distribution, and empirical factors theoretically conditioning the occurrence and magnitude of wakes. The empirical factors may be predefined based on previous experience (empirical data) from known wind farms, thus resulting in a baseline wake model. Block 302 may produce a set of pairs of wind turbines with a wake between them which may distort the operation and power generation of the wind farm as a whole.
Alternatively to using the baseline wake model, a wake model determined in a previous execution of the method may be retrieved from DB 315 and used instead of the baseline wake model for identifying wake situations. A previous execution (or iteration) of the proposed method may have produced a wake model from current real wind data determined at first and second wind turbines and may have stored the previously determined wake model in DB 315 for later use. The wake model retrieved from DB 315 may be the one generated most recently for the first and second wind turbines that are being processed. In that sense, note that different areas of a wind farm may have varying characteristics, so different wake models may be applicable to each of those differentiated areas.
At block 303, one of the pairs of wind turbines detected at previous block 302 may be selected to be processed in subsequent blocks.
At block 304, one or more wind parameters may be determined at first wind turbine which is creating the wake. At block 305, one or more wind parameters may be determined at second wind turbine which is receiving the wake. In both blocks 304, 305, the one or more wind parameters may include one of the wind speed, wind turbulence and wind direction.
Wind speed and/or wind turbulence at a wind turbine may be determined depending on measurements from a LIDAR associated with the wind turbine. Individual controller of a wind turbine may operate the wind turbine by controlling operational parameters such as e.g. pitch angle, yaw angle, rotor speed, torque and generated power based on measurements obtained at the wind turbine. Wind speed and/or wind turbulence may be indirectly determined at a wind turbine from at least some of the operational parameters controlled by the individual controller of the wind turbine. A wind turbine may also include load sensors for determining loads on the wind turbine. Wind speed and/or wind turbulence may be indirectly determined at a wind turbine from load measurements provided by the load sensors.
At block 306, a previously determined wake model having a parameter with a predetermined value for the first and second wind turbines may be retrieved from DB 315. The previously determined wake model may be either a baseline wake model or a wake model determined in a previous execution or iteration of the method.
At block 307, a current wake model may be determined by calculating a value of the parameter of the theoretical wake model based on the obtained wind parameter(s) at first and second wind turbines.
Based on the real wind parameter(s) determined at first and second wind turbines, a real (experimental) magnitude may be obtained. In most cases, this real magnitude may refer to wind speed deficit on the second wind turbine. As commented with respect to
A value of a constant or factor of the theoretical wake model may be calculated. The calculation of the constant or factor may be performed based on the real (experimental) magnitude obtained from the real wind parameter(s) determined at first and second wind turbines. In the case of using the Jensen model, the constant to be calculated may be the “decay constant” which depends on predefined data (roughness of the ground surface and height of the wind turbine tower) in the analytical function of the model. The Jensen model may thus be used to calculate the decay constant in such a way that the analytical function produces a wind speed deficit substantially equal to the obtained real magnitude of the wind speed deficit (depending on real wind speed at first and second wind turbines). This calculated decay constant may be used in subsequent calculations of the proposed method, such as e.g. those aimed at optimizing the operation of first and second wind turbines.
At block 308, the previously determined wake model (obtained from DB 315) and the current wake model (determined from current real wind data) may be compared.
At block 309, a verification of whether the comparison performed at block 308 produces a divergence that exceeds a predefined threshold may be performed. In case of positive result of the verification, the method may continue to block 311 at which the current wake model (determined from current real wind data) is provided to block 312 of optimizing the operation of first and second wind turbines. In case of negative result of the verification, the method may continue to block 310 at which the previously determined wake model is taken as the current wake model and therefore provided to block 312 of optimizing the operation of first and second wind turbines.
At block 311, the current wake model (determined from current real wind data) may be also provided to DB 315 for its storage, so that later executions or iterations of the method may re-use the wake model at block 306. An aspect of this re-utilization of wake models determined in previous executions or iterations of the method is that wakes may be estimated based on more real conditions.
In the case that the previously determined wake model and the current wake model (obtained from current real wind data) slightly differ from each other, the previously determined wake model may be taken as the current wake model and therefore used in subsequent calculations. Since the previously determined wake model has been already used (in previous executions of the method) its selection for subsequent calculations may require relatively less computational effort.
At block 312, operation of first and second wind turbines may be optimized taking into account the current wake model, which may have been obtained from current real wind data (at first and second wind turbines) or may be the previously determined wake model, depending on the result of the verification performed at block 309. Optimum operating parameters, such as e.g. pitch, torque, rpms, yaw, etc. may be generated for both first and second wind turbines depending on the current wake model. The optimization may be performed depending on an objective, such as e.g. maximizing power generation, maximizing loads reduction, etc. This optimization may be performed by using matrices or look-up tables having input and output parameters. Input parameters may include those parameters characterizing the current operational state of wind turbines, the operational objective to be achieved, etc. Output parameters may include those operating parameters (pitch, torque, rpms, etc.) to be controlled for operating the wind turbine.
In the case of using the Jensen wake model, the decay constant of the current wake model may be one of the input parameters such that output parameters may be determined depending on the decay constant. Any known optimizing algorithm may be used to select the most optimum output parameters according to the objective to be achieved which is represented in the matrices (or look up tables) in the form of input parameters.
Alternatively to using matrices with input and output parameters, suitable analytical function(s) may be used for optimizing the operation of first and second wind turbines. In this sense, output parameters to be controlled (for operating the wind turbine) may be expressed as a function of input parameters such as e.g. factor or constant of the corresponding wake model. For example, in the case of using Jensen model, an output parameter (pitch, rpms, torque . . . ) may be expressed as a function of the decay constant and other parameters such as e.g. variables characterizing the objective to be achieved (maximum power, maximum loads reduction, . . . ). Any known optimizing algorithm may be used to optimize such analytical functions.
Once optimum operating parameters have been obtained, corresponding control signals (or set points) may be sent to wind turbine actuators to operate the first and second wind turbines according to the obtained optimum operating parameters.
At block 313, a verification of whether all the pairs of wind turbines detected at block 302 have been processed (at blocks 303-312) may be performed. In case of negative result of the verification, the method may loop back to block 303 in order to select a next pair of wind turbines to be processed. In case of positive result of the verification, the method may continue to final block 314.
At block 314, the method may end its execution. From this point, the initial block 300 may be triggered again in order to perform a new iteration or execution of the method, such that the method may continuously iterate under a given frequency, for example.
Although only a number of examples have been disclosed herein, other alternatives, modifications, uses and/or equivalents thereof are possible. Furthermore, all possible combinations of the described examples are also covered. Thus, the scope of the present disclosure should not be limited by particular examples, but should be determined only by a fair reading of the claims that follow.
| Number | Date | Country | Kind |
|---|---|---|---|
| 15382372.9 | Jul 2015 | EP | regional |
