DK201570195A1 - Control system for wind turbine having multiple rotors - Google Patents

Abstract

A method of controlling a wind turbine system, the wind turbine system comprising a plurality of wind turbines mounted to a common support structure, wherein each of the plurality of wind turbines includes a rotor and a power generation system driven by the rotor, the method comprising: obtaining signals indicative of wind speed in at least two spaced locations, each spaced location being on or near to one of the plurality of wind turbines; processing the signals indicative of wind speed to calibrate a wind field model, thereby to produce a simulated spatial wind field around the wind turbine system; and controlling the wind turbine system in accordance with the simulated spatial wind field.

Description

CONTROL SYSTEM FOR WIND TURBINE HAVING MULTIPLE ROTORS

Technical field

The invention relates to a control system of a wind turbine system having multiple rotors and more particularly, but not exclusively, to a control system for estimating a spatial wind field around such a wind turbine system.

Background to the invention A common type of wind turbine is the three-bladed upwind horizontal-axis wind turbine (HAWT), in which the turbine rotor is at the front of the nacelle and facing the wind upstream of its supporting turbine tower.

The spatial wind field can vary considerably over the area covered by the rotor, the spatial wind field representing varying characteristics of the wind, including wind speed, wind veer and wind shear, across a defined area. The rotor and its associated nacelle is preferably operated according to the wind conditions that it is subjected to, to ensure that the wind turbine operates as close to its rated output as possible, and to minimise stress and vibration in the blades of the rotor.

Before wind turbines are installed at a proposed site for a new wind power plant, or ‘wind farm’, measurements are taken to characterise the typical wind conditions at that site. The measurements often include: wind speed; wind direction or inflow angle; turbulence; air density; and wind shear. These measurements are taken at various locations around the site at different times. The measurements are then used to generate a predicted wind field that the wind turbines will be subjected to once installed so that the turbines can be specified appropriately. The predicted wind field may be adjusted to account for factors that influence local wind conditions such as contours around the site, and the roughness of the terrain.

The wind measurements are typically collected using one or more meteorological towers, or ‘met masts’, comprising a range of sensors that gather wind data. The met mast is installed at the site a year or more in advance of installation of the wind turbine so as to characterise wind conditions at the site to generate an accurate predicted wind field. The met masts often remain on the site after the wind turbine has been installed so as to monitor wind conditions during operation of the wind turbines.

Alternative methods for characterising wind conditions at the site to generate a predicted wind field include using algorithms to process historical weather data for the site to predict future conditions, and more recently the use of lidar sensors.

The wind turbines are specified and power-rated according to the predicted wind field. Once commissioned, the operating strategy for the wind turbines is generally not altered, although if wind conditions prove to be more benign than expected the rated power can be increased and/or the predicted lifetime of the turbines can be extended.

Following installation, the wind turbines are controlled dynamically according to instantaneous wind conditions. In conventional wind turbine arrangements, typically wind speed is measured on or near to the nacelle. The measurement is typically low-pass filtered and corrected for wind distortions caused by the nacelle and the rotor. Various operating parameters of the wind turbine such as blade pitch angle and generator power and torque are then optimised according to the measured wind speed, which is assumed to be uniform across the rotor, or to vary in accordance with standardised turbulence models (e.g. Veers or Mann).

It is noted that the standardised models are based on a small selection of individual site measurements and so are to some extent specific to the characteristics of those sites. The models also do not account for varying atmospheric conditions. So, the estimated wind field may be inaccurate. For example, the actual wind field may have a lower coherence than the standardised model, meaning that there is greater variation in wind speed across the area covered by the rotor than this approach predicts.

Wind conditions at the site are continuously monitored in this manner, and in the long term, for example over a period of one year, average conditions around the site can be determined. As noted above, if these conditions are more benign than expected, the wind turbines can be up-rated or have their operating lifetimes extended, in either case providing a gain in the total energy generated by the turbine over its operating lifetime. Conversely, in the unlikely event that the wind conditions are more severe than predicted by the initial site analysis, which tends to be a conservative estimate, a defensive operating strategy can be adopted in which less power is generated in order to protect the turbines from high loadings and the associated stress fatigue that would reduce its operating lifetime.

It is known to support an array of HAWT units from a common support structure, as described, for example, in EP1483501B1. Such a configuration achieves economies of scale that can be obtained with a very large single rotor turbine, but avoids the associated drawbacks such as high blade mass, scaled up power electronic components and so on. However, although such a co-planar multi-rotor wind turbine has its advantages, there are challenges involved in implementing the concept in practice, particularly in controlling the multiple rotors to achieve optimum power production. EP1483501B1 approaches the control strategy by treating each wind turbine of the system as a separate item that is controlled individually.

Accordingly, to implement dynamic control of each turbine to account for wind speed as described above, a respective sensor attached to each rotor gathers local wind speed measurements, and those measurements are used to optimise control of the rotor to which the sensor is attached. A drawback to this approach is that if one of the sensors fails, the rotor to which it is attached cannot be operated in an optimum manner due to the lack of data, and may even need to be shut down.

It is against this background that the invention has been devised.

Summary of the invention

According to an aspect of the invention, there is provided a method of controlling a wind turbine system. The wind turbine system comprises a plurality of wind turbines mounted to a common support structure, each of the plurality of wind turbines including a rotor and a power generation system driven by the rotor. The method comprises: obtaining signals indicative of wind speed in at least two spaced locations, each spaced location being on or near to one of the plurality of wind turbines; processing the signals indicative of wind speed to calibrate a wind field model, thereby to produce a simulated spatial wind field around the wind turbine system; and controlling the wind turbine system in accordance with the simulated spatial wind field.

This method therefore takes advantage of the increased number of wind speed measurements potentially available in a multi-turbine system by processing all measurements centrally to generate a simulation of the overall wind field around the system. Using multiple measurements to calibrate a wind field model provides a result that more closely reflects actual wind conditions than using a single calibration point. The wind field model may be, for example, a turbulence model such as one of the ‘Mann’ or ‘Veers’ models. Simulating the spatial wind field accurately enables optimised operation of the system for maximised power generation.

The method may comprise continuously updating the simulated spatial wind field on obtaining further signals indicative of wind speed in the at least two spaced locations. In such embodiments, the method may further comprise monitoring the simulated spatial wind field to derive time-averaged estimated wind conditions around the wind turbine system, in which case the method may also comprise estimating from the time-averaged wind conditions peak expected loadings on the rotors of the wind turbine system. Taking averaged measurements in the long term in this manner enables long term management of the system to be optimised, and highlights the accuracy of initial predictions of wind conditions at the site on which the wind turbine system is installed.

The method may comprise estimating loadings on each of the rotors of the system from the simulated wind field. In this case, the method may also comprise estimating fatigue in the rotors based on the estimated loadings. Beneficially, this enables refinement of the predicted lifetime of the rotors before failure. This information can be used to alter operation of the wind turbine system as appropriate, either to preserve the rotors or to extract more energy from the wind.

In some embodiments, at least one of the spaced locations is on a nacelle of one of the wind turbines. This is a convenient arrangement that reflects conventional configurations.

The method may comprise obtaining signals indicative of wind speed at locations on or near to each wind turbine of the wind turbine system. This provides a high number of measurement points with which to calibrate the wind field model for enhanced accuracy.

Controlling the wind turbine system optionally comprises adjusting one or more of: the rated power of each wind turbine in accordance with the simulated spatial wind field; the predicted lifetime of the wind turbine system in accordance with the simulated wind field; and the specification of each wind turbine in accordance with the simulated spatial wind field. Adjusting the specification of the wind turbines may include, for example, upgrading the physical hardware as appropriate. Controlling the wind turbine system may also entail adjusting long term operating strategies such as a blade tip angle strategy. In this way, the method of the invention enables optimised utilisation of the wind turbine system in the long term for maximised total power generation over the lifetime of the system.

The method may comprise analysing transient variations in the simulated spatial wind field over time, for example to improve predictions of peak loadings.

The method may comprise deriving from the simulated wind field a respective local spatial wind field simulation for each wind turbine of the system, each local spatial wind field simulation being weighted for its respective wind turbine. This enhances the accuracy of estimated wind conditions around each individual rotor, and therefore the estimated loadings exerted on them.

In another aspect, there is provided simulation system for simulating a spatial wind field around a wind turbine system. The wind turbine system comprises a plurality of wind turbines mounted to a common support structure, each of the plurality of wind turbines including a rotor and a power generation system driven by the rotor. The simulation system comprises: sensing means arranged to obtain signals indicative of wind speed in at least two spaced locations, each spaced location being on or near to one of the plurality of wind turbines; and processing means arranged to process the signals indicative of wind speed to calibrate a wind field model, thereby to simulate a spatial wind field around the wind turbine system.

The processing means may be implemented on a computing device that is remote from the wind turbine system. This is a convenient arrangement that ensures that the processing power of the processing means is not restricted by practical considerations that might apply if the processing means were integrated with the wind turbine system. Hosting the processing means remotely also allows for control of more than one wind turbine system using a common processing means.

The processing means optionally comprises a memory that is arranged to store at least one wind field model to be calibrated, in which case the processing means may be further arranged to store spatial wind field simulation data in the memory. In such embodiments, the processing means may be arranged to analyse the stored simulation data to determine average wind field data.

In another aspect, the invention also extends to a computer program product arranged to implement the above method of the invention.

It will be appreciated that preferred and/or optional features of any of the above aspects of the invention may be incorporated alone or in appropriate combination in the other aspects of the invention also.

Brief description of the drawings

So that it may be more fully understood, the invention will now be described, by way of example only, with reference to the following drawings, in which:

Figure 1 is a front view of a multi-rotor wind turbine system;

Figure 2 is a top view of the multi-rotor wind turbine system in Figure 1;

Figure 3 is a schematic view of an embodiment of a control system for the multi-rotor wind turbine system of Figures 1 and 2;

Figure 4 is a schematic perspective view of the wind turbine system of Figures 1 and 2 in combination with a representative wind field; and

Figure 5 is a flow diagram showing a process for optimising control of the wind turbine system of Figures 1 and 2.

Description of embodiments of the invention

Embodiments of the invention provide apparatus and methods for enabling an offline simulation of a spatial wind field around an entire wind turbine system in which a standardised model is calibrated using multiple wind speed measurements.

It should be appreciated that ‘simulated’ in this context may refer to a high resolution animated computer model representing air flow throughout a defined region, for example produced using computational fluid dynamics analysis techniques. Alternatively, the term may refer to a simpler estimation, for example a spreadsheet containing instantaneous values for a range of variables. It is noted that this process is typically implemented offline due to the processing power required.

The standardised models provide additional wind data on top of the raw measurements that enable wind conditions around wind turbines on a wind farm to be estimated more fully. This data includes variables such as length scales, wind coherence and shear distortion. These variables are difficult to measure, but nonetheless impact loadings on the wind turbines, and so to have estimations of the values of such variables is a valuable enhancement to the analysis.

In general, this is achieved by gathering wind speed data at various spaced measurement locations around the system, and then processing the data centrally offline, using the measurements in combination to estimate the spatial wind field across the entire system.

Basic calculations can be performed, such as interpolating between a pair of measurement points to provide an estimation of the wind speed at locations between the measurement points. Also, wind shear and wind veer can be calculated from two or more spaced measurements, thereby dispensing with the need to provide meteorological masts to provide such measurements as in conventional arrangements.

Furthermore, the measurement points can be used as multiple calibration points to calibrate a standardised turbulence model to provide a simulation of the wind field across the entire system. The same turbulence models can be used as in the conventional approach, for example the Mann model or the Veers model. In this way, the offline processing provides information relating to a range of variables, including coherence, length scales and shear distortion, which as already noted are otherwise very difficult to measure.

Processing all of the gathered data centrally to provide multiple calibration points refines the calibration, and so the simulated wind field is more accurate than those provided by conventional approaches in which a single calibration point is used. Moreover, this approach provides a single estimated wind field covering the entire system, in contrast with the conventional approach in which individual local fields are estimated around each measurement point. The wind field estimation is therefore a better representation of the actual wind conditions around the system than the estimations of prior art systems. Each additional measurement point further improves the accuracy of the calibrated turbulence model, and in turn the wind field estimation.

By estimating the wind field around the entire system, local wind fields around each individual wind turbine can be deduced, enabling individual loadings on each wind turbine to be determined. This information can be used in the long term to adjust the operating strategy for each wind turbine as required. The more measurement points that are used, the more accurately the loadings on each wind turbine due to the wind conditions to which the turbine is subjected can be estimated. This enables the power curve for each wind turbine to be improved when the turbine is operating below rated power, while minimising loads when operating above rated power. Also, the likelihood of gusting can be evaluated, either with reference to the turbulence models or using separate analysis.

The area over which the wind field is simulated is theoretically unlimited, as the standardised model can be extrapolated beyond the locations of the measurement points. However, the accuracy of the simulated wind field diminishes with increasing distance from the measurement locations.

Weighted averaging can be applied so that the wind field estimation is adjusted for each individual wind turbine to prioritise the measurements taken closest to that turbine when determining the loadings on that turbine. Therefore, for a system having, for example, four individual wind turbines, four different wind field estimations may be produced, each being weighted for a respective one of the wind turbines. The weight factors to be applied can be taken from the standardised turbulence models mentioned above. Alternatively, once a sufficient amount of data has been collected the coherence of the simulated wind field can be determined, and the weight factors can be derived accordingly. A clear benefit to this approach of using multiple measurement points in a common wind field analysis for the entire system, compared with the earlier described conventional approaches based on operating each turbine wholly independently, is that the wind field locally around each turbine can be estimated. This also applies over the system as a whole. Noting that the wind field can vary considerably over the large area swept by a typical rotor, this approach enables improved power regulation in each wind turbine through more accurate power/torque control and pitch control, to help avoid exceeding the rated power of the turbine. Also, the estimated wind field indicates the prevailing wind direction around the system, and so the yaw angle can be adjusted to account for this.

In turn, the actual loadings experienced by the rotors can be determined more accurately for improved long term management. For example, fatigue analysis can be enhanced to provide a more accurate predicted lifetime for each rotor under current conditions. This enables the rated power to be adjusted whilst maintaining a required service life, the predicted lifetime to be extended, or a mixture of the two.

Additionally, the loss of one of the sensors does not entail a total loss of wind data for the associated wind turbine of the system. Instead, the loss of a sensor merely reduces the accuracy of the estimation; the local wind characteristics around each rotor can still be estimated in order to enable continued optimisation of operation.

Furthermore, calibrating a standardised model provides data relating to wind characteristics such as veer and shear to provide an indication of the level of turbulence around the rotors. These factors influence the load exerted on the rotors and so guide the way in which each rotor and nacelle should be operated. Embodiments of the invention provide a more accurate and complete picture of wind characteristics around the system, and therefore the expected loading on each rotor, enabling enhanced optimisation of turbine operation.

For example, a blade pitch control strategy can be adjusted in response to a long term averaged simulated wind field around each rotor, so that the pitch of each blade of the rotor is controlled individually within a specific range that ensures that loadings on the rotor remain within acceptable limits. Alternatively, the pitch of each blade may be controlled so as to maximise the load applied to it at all times to extract a maximum level of energy from the wind.

Alongside this, each turbine comprises a generator, and an operating strategy for controlling the power and torque of the generator can also be adjusted to account for average wind conditions. For example, if the wind speed is generally higher across a rotor than was predicted during commissioning, the generator torque can be raised such that the generator effectively applies braking to the rotor to ensure that the rated output of the wind turbine is not exceeded. Correspondingly, if the wind conditions are more benign than predicted, the generator torque is lowered to maximise energy extraction from the wind.

Blade pitch control and power/torque control are used in parallel so as to maintain the output of the wind turbine as close as possible to its rated power at all times.

To provide context for the above described invention, an illustrative multi-rotor system that is suitable for use with embodiments of the invention is now described with reference to Figures 1 to 3. It should be appreciated that the system of Figures 1 to 3 is referred to here by way of example only, and it would be possible to implement embodiments of the invention into many different types of wind turbine systems.

Referring firstly to Figures 1 and 2, a wind turbine system 2 includes a support structure 4 on which is mounted a plurality of wind turbines 6. In this embodiment, the support structure 4 is a slender tower that is mounted on a foundation embedded in the ground. It should be appreciated that other support structures are possible, for example frame-like structures. Note that the term ‘wind turbine’ is used here in the industry-accepted sense to refer mainly to the generating components of the wind turbine system and as being separate to the support structure 4.

In this embodiment, there are four wind turbines 6, and these are mounted to the support structure 4 in two pairs, each pair including two wind turbines 6 that are mounted to the support structure 4 by a support arm arrangement 10.

The support arm arrangement 10 comprises a mount portion 12 and first and second arms 13 that extend from the mount portion and carry a respective wind turbine 6. As such, each of the support arms 13 includes an inner end 16 connected to the mount portion 12 and an outer end 18 that is connected to a wind turbine 6.

The support arm arrangement 10 is mounted to the support structure 4 at the mount portion 12 so that the support arm arrangement 10 is able to yaw about the vertical axis of the support structure 4. Suitable yaw gearing (not shown) is provided for this purpose. This movement provides a first degree of freedom for the wind turbine 6 with respect to the support structure, as shown on Figure 2 as ‘FT. In embodiments, a rotatable joint is used at the mount portion 12 which does not use a yaw mechanism. In such embodiments, the wind turbines 6 may move around the support structure 4 by use of the thrust force imparted by the wind on the turbines 6.

Each wind turbine 6 includes a rotor 22 that is rotatably mounted to a nacelle 23 in the usual way. The rotor 22 has a set of three blades 24 in this embodiment. Three-bladed rotors are a common rotor configuration, but different numbers of blades are also known; two-bladed configurations are also quite common, for example. Thus, the wind turbines 6 are able to generate power from the flow of wind that passes through the swept area or ‘rotor disc’ 26 associated with the rotation of the blades.

In this embodiment, a respective wind sensor 27 is mounted to each nacelle 23, the wind sensors 27 being arranged to provide measurements of wind speed and direction at each sensor location, to enable an estimation of the spatial wind field around the entire system 2 in the manner described above. From this, local wind fields around each of the rotors 22 of the system 2 can be derived and monitored in the long term to determine average conditions and expected fluctuations. Placing the sensors 27 on the nacelles 23 is convenient as it mirrors a conventional sensor configuration. As noted above, the sensors 27 could alternatively be distributed around the system in many other ways to achieve the same effect, namely to provide the spaced measurements that are used to estimate the wind field.

The wind sensor 27 may be a conventional anemometer such as would typically be used in known multi-rotor systems. Alternatively, any other sensing means that is capable of providing an indication of wind speed may be used. For example, a lidar sensor could be used. Such sensors direct a laser beam into a target area and then analyse a returning beam that has been reflected by aerosol particles suspended in the air in order to derive wind speed and direction. The analysis may include, for example, interferometry techniques. Conveniently, a single lidar sensor can scan a region of interest to collect measurements from multiple locations in order to estimate the wind field. Such a scanning lidar sensor could be mounted at any convenient location on the system, for example on one of the nacelles 23 or on the support structure 4. Multiple lidar sensors could be used to provide enhanced resolution and accuracy. A further option is to use load sensors on the rotors 22 to provide an indirect indication of the wind speed, in that the load that the wind applies to each rotor 22 is proportional to the local wind speed when factored for blade pitch angle and wind direction. For example, a strain gauge coupled to a rotor 22 provides an indication of the strain induced in the rotor as a result of the load exerted on the rotor by the wind. Using the known mechanical properties of the rotor 22, such as the geometry and material strength, this indicated strain can be used to determine the wind load, and in turn the wind speed around the rotor at that location. While this is an indirect measurement and so inherently of lower accuracy than direct measurement approaches, load sensors are generally reliable, inexpensive and readily available, and so this approach may be attractive; even if only alongside other measurement methods, for example for error checking or sensor validation. A yet further option is to use the rotor itself as a wind speed estimator, in that the torque applied to the rotor shaft is directly proportional to the average speed of the wind acting over the swept area of the rotor. Therefore, the rotational speed of the rotor is indicative of the average wind speed once system parameters including the inertia of the rotor, the blade pitch angle, the tip speed ratio, and forces opposing rotation such as friction and the generator torque are accounted for. Therefore, the average wind speed can be estimated from the rotational speed of the rotor. As wind acting at the tip of the rotor has more influence on torque than wind acting near the centre of the rotor, there may be a margin of error in the estimation. Therefore, the estimation can be refined by predicting a new rotor speed on altering another system parameter such as the generator torque. Any deviation between the prediction and the actual new rotor speed can be accounted for to improve the estimation. A combination of different types of sensors could also be employed in order to enhance the wind field estimation.

Figures 1 and 2 show the main structural components of the wind turbine system 2, although the skilled person would understand that the illustrated embodiment has been simplified to avoid obscuring the invention with unnecessary detail. Further explanation will now be provided on the system component of the wind turbine system 2 with reference also to Figure 3.

In an embodiment, on a system level, each wind turbine 6 includes a gearbox 30 (hereunder including belt drives or other gearing arrangements) that is driven by the rotor 22, and a power generation system including a generator 32 connected to the gearbox 30 and which feeds generated power to a converter system 34 which converts the power into a suitable frequency and voltage for onward transmission. A pitch control system (not shown) is also provided to control the angle of attack of the blades relative to the wind. The precise configuration of the generator 32 and converter system 34 are not central to the invention and will not be described in detail. However, for present purposes they can be considered to be conventional and, in one embodiment, may be based on a full scale converter (FSC) architecture or a doubly fed induction generator (DFIG) architecture. Furthermore, each of the wind turbines can be considered to be substantially identical, so only one has been labelled fully in Figure 3 for clarity.

In the illustrated embodiment, the power output of the converter 34 of each wind turbine 6 is fed to a distribution unit 40 which has a function to receive power inputs 42 from the wind turbines 6 over suitable cabling 44 for onward transmission to a load 46, which is shown here as the electrical grid.

The skilled person will appreciate that there are various alternative power transmission systems that could be implemented to take power from several generators to a single grid, and so this embodiment is described for illustrative purposes only.

It should be noted at this point that only a single wind turbine system 2 is described here, but that several such systems may be grouped together to form a wind power plant, also referred to as a wind farm or ‘park’. In this case, a power plant control and distribution facility (not shown) would be provided to coordinate and distribute the power outputs from the individual wind turbine systems to the wider grid.

Since the wind turbine system 2 includes a plurality of wind turbines 6, each of which is operable to generate electrical power as the rotor 22 is driven by the wind, the system includes localised control means that is operable to monitor the operation of respective ones of the plurality of wind turbines 6 and to issue commands thereto.

Each control means is also arranged to process signals received from its respective wind sensor 27, to format raw signals received from the sensor 27 appropriately for central processing.

In this embodiment, the localised control means is provided in the form of a plurality of local control modules 50 that are embodied as respective computing devices each of which is dedicated to an associated wind turbine 6.

The responsibility of the local control modules 50 is to monitor the operation of a specific wind turbine 6 and control the operation of its various components to achieve local control objectives. For example, with reference to a single wind turbine 6 for clarity, the local control module may: monitor rotor speed and control the pitch control system in line with a local pitch control strategy as derived from a local power-speed curve that is specific for that particular wind turbine 6 in order to ensure that maximum power is extracted from the wind during below-rated power operating conditions; control the generator 32 in line with a local torque control strategy in order to limit power production in above-rated power operating conditions, as also derived from said local power-speed curve; and monitor wind speed measurements from a respective wind sensor 27 and transmit them to a centralised control means.

In summary, as a group the local control modules 50 are responsible for controlling the functionality of each wind turbine 6 individually in a way that ignores the interaction between the wind turbine 6 and the rest of the multi-rotor wind turbine system 2. So, the localised control modules 50 are specifically directed to optimising the performance of a respective wind turbine 6 in line with a dynamically estimated wind field together with an associated set of local control objectives and do not take into account how the operation of the other wind turbines 6 or the support structure 2 may influence how the individual wind turbines 6 should be operated as a wider group.

In order to provide a coordinated control strategy, the wind turbine system 2 also includes a centralised control means which is configured to monitor the operation of the wind power system, that is to say the wind turbines 6 and the support structure 4, and to provide centralised control commands to the plurality of wind turbines 6 in order to achieve a set of supervisory control objectives to the wind turbines as a group.

The centralised control means also receives individual wind speed measurements derived from each of the wind sensors 27, and uses those measurements to estimate the spatial wind field around the turbines 6, as described above. This wind field estimation is then used to derive the expected speed and direction of wind striking each blade of each of the rotors 22. This information is used to determine the appropriate blade pitch angle for each blade, as well as the power and torque for the generator 32, to maintain each rotor 22 at a desired point on the local power speed curve. This action can then be implemented in each of the turbines 6 using control commands as described below.

In this embodiment, the centralised control means is provided by a central control module 52 being a computing device incorporated in the central distribution unit 40, although it is noted that in other embodiments the central control module 52 may be separate from the distribution unit 40. Here, the central control module 52 is located on the support structure 4, for example inside the tower or in a housing placed next to the turbine, and includes an integrated wind field processing module 53 that is configured to output an estimated wind field from wind speed measurement inputs. The estimated wind field is output to the central control module, which uses the estimation to produce control commands that will optimise operation of each wind turbine 6 according to the estimated wind field. This may include adjusting the torque of the generator 32, and directing the pitch control system to set the blades to an appropriate angle of attack relative to the wind, or to maintain the required angle of attack according to the local wind direction as derived from the estimated wind field.

The central control module 52 achieves control over each of the wind turbines 6 by providing control commands thereto. As shown in Figure 3, the central control module 52 outputs control commands 54 which are received by each one of the wind turbines 6 and, more particularly, are received by the local control modules 50. The control commands 54 may be of the ‘broadcast’ type of command in which the same command is sent out to each wind turbine 6, or the commands may be of the ‘directed’ type of command in which a specific control command is set to a selected one or more, but not all, of the wind turbines 6.

It will be noted that Figure 3 is a schematic view, so the way in which the control commands 54 and wind sensor 27 readings are transferred to and from the wind turbines 6 is not depicted explicitly. However, it will be appreciated that suitable cabling may exist in the wind turbine system that interconnects the central control unit 52 to the wind turbines 6, and more specifically to the local control modules 50. The interconnections may be direct or ‘point to point’ connections, or may be part of a localised area network (LAN) operated under a suitable protocol (CAN-bus or Ethernet for example). Also, it should be appreciated that rather than using cabling, the control commands 54 may be transmitted wirelessly over a suitable wireless network, for example operating under WiFi™ or ZigBee™ standards (IEEE802.11 and 802.15.4 respectively).

The objective of the central control module 52 is to implement a harmonious control strategy for the group of wind turbines 6 so that their interactions between each other, and the interactions between the wind turbines 6 and the support structure 4 are managed in the most effective way, whilst accounting for the estimated wind field around the system 2. Expressed another way, the central control module 52 applies a higher level control strategy to the operation of the wind turbine system 2, whereas the local control modules 50 apply a lower level control strategy to each respective wind turbine 6 individually. However, both ‘levels’ of the control strategy operate together harmoniously in order to optimise the performance of the wind power system 2, both in terms of absolute power production, production efficiency, and fatigue optimisation.

While the central control module 52 is able to perform low-level calculations so as to implement dynamic control, it may not have sufficient processing power to use the wind speed measurements to calibrate standard turbulence models in the manner described above to provide a high resolution simulation of the wind field around the system 2. For this reason, wind speed data collected by the sensors 27 attached to each of the wind turbines 6 may be forwarded from the central control module 52 to a remote control means 56 that is configured to perform the task of calibrating a standardised turbulence model to derive more accurate estimations of the loading on each rotor 22 than can be estimated in real-time by the central control module 52.

The remote control means 56 is in this embodiment provided as a processor in the form of a remote control module 58 residing on a computer system that is not physically attached to the wind turbine system 2. The remote control module 58 comprises a memory 60 and a calibration module 62. The calibration module 62 is arranged to calibrate standard turbulence models held in the memory 60 to produce a simulated wind field that is representative of wind conditions around the system 2. The simulation is also stored in the memory 60 to allow historical data to be analysed. It is noted that the simulated wind field produced by the remote control module 58 is far more refined than that produced by the central control module 52, as it includes data obtained from the standard turbulence model relating to variables such as coherence, turbulence and shear distortion, for example.

If multiple turbine systems 2 are present on a single site, the remote control module 58 may be common to all of the turbine systems 2 and used to analyse data from each of them to build an estimate of wind conditions across the entire site. It will be appreciated that processing data for the entire site centrally in this way provides enhanced accuracy compared with monitoring each system 2 individually, for the reasons described above.

In other embodiments, the remote control means 56 may be implemented differently. For example, the remote control means 56 may be placed in parallel with the central control module 52, so that the remote control means 56 takes wind speed measurements directly from the sensors 27, rather than through the central control module 52. Alternatively, the remote control means 56 may be integrated with the central control module 52 in a common module that may reside either on a remote computer system, or on the wind turbine system 2.

The remote control module 58 continuously updates the calibrated turbulence models to account for changes in the real-time wind speed data supplied from the central control module 52. The calibrated turbulence models are monitored over time to determine average conditions, and also to refine expecting peak loadings. This analysis can be used to manage the system more effectively in the long term as described above, for example to increase the predicted lifetime of the system 2 if the wind conditions are more benign than expected.

Moving on to Figure 4, a schematic representation of a wind field within an area of interest 62 upstream of the wind turbine system 2 is illustrated. For clarity, the wind turbine system 2 is shown in simplified form, with only the swept area 26 of each rotor 22 and the support structure 4 shown. The varying wind speed within the area of interest 62 is represented using arrows of varying length. The wind field shown illustrates typical variation of wind speed within such an area, although for simplicity the wind direction is shown as uniform within the area of interest 62. Therefore, the wind speed varies significantly throughout the area of interest 62, although an uppermost horizontal portion 64 of the area of interest 62 has generally higher wind speed than a lowermost horizontal portion 66. This corresponds to what would typically be expected in reality, in that wind speed is typically reduced close to ground level due to friction at the surface.

Figure 5 shows in schematic form a control flow 70 for using wind speed measurements to calibrate a standard turbulence model to simulate the wind field around the wind turbine system 2. First, wind sensor measurements are gathered at step 72, for example by the local control modules 50 of the above described wind turbine system 2. The measurements are passed at step 74 to the calibration module 62 of the remote control module 58 in order to process the measurements to calibrate a standard turbulence model, for example a Veers model stored in the memory 60, to derive a simulation of the wind field. Once the wind field has been simulated, local wind field simulations are derived in order to determine the wind speed and direction incident on each individual blade of the rotors 22 and in turn determine at step 76 the loadings on the rotors 22. The rotor loadings are monitored at step 78 over an extended period, and average wind conditions are determined at step 80. The system 2 is then managed at step 82 in response to the average conditions determined by the remote control module 58.

The skilled person will appreciated that modifications may be made to the specific embodiments described above without departing from the inventive concept as defined by the claims.

For example, although in the embodiment of Figure 3 the local control modules 50 are shown as being located within the nacelles 23 of the wind turbines 6, this need not be the case, and embodiments are envisaged in which the local control modules 50 are mounted in different locations, for example on the support arms 13 close to the support structure 4. This may provide the local control modules 50 in a more convenient position for maintenance access.

The responsibilities of processing wind speed measurements, deriving predicted rotor loadings and determining the appropriate action to take may be allocated differently to the manner described above, for example with the local control modules performing a larger proportion of this process. It will be appreciated, though, that the wind field estimation can only be performed by a control module having access to all of the wind speed measurements, although this could be implemented by way of a master local control module that receives measurements from the remaining local control modules and then returns to those remaining modules wind field estimation data.

Furthermore, while the above described system has a single control system, i.e. a single local control module for each turbine connected to a single common centralised control module, the wind field estimation could alternatively be implemented using separate, dedicated apparatus. For example, the output signals from the wind sensors 27 could be transmitted directly to a dedicated central processing unit that is arranged to use the measurements to calibrate a standard turbulence model to derive the wind field estimation, without the need for intermediate local control modules or the involvement of a central control module or remote control module in estimating the wind field. The central processing unit could then determine the appropriate instantaneous configuration for each rotor of the system in relation to the derived wind characteristics, and deliver this information to a separate control module that issues commands to implement the required action, as well as determine long term trends in wind conditions to aid management of the system.

Also, it should be appreciated that although the illustrated embodiment includes four wind turbines mounted to the support structure, this is to illustrate the principle of the proposed spatial wind field estimation apparatus which may be applied to wind turbine systems with more than four wind turbines. Moreover, embodiments are envisaged in which the wind turbines are not paired in groups of two, as in the illustrated embodiment, but are arranged differently and not necessarily having a co-planar relationship.

Claims (18)

1. A method of controlling a wind turbine system, the wind turbine system comprising a plurality of wind turbines mounted to a common support structure, wherein each of the plurality of wind turbines includes a rotor and a power generation system driven by the rotor, the method comprising: obtaining signals indicative of wind speed in at least two spaced locations, each spaced location being on or near to one of the plurality of wind turbines; processing the signals indicative of wind speed to calibrate a wind field model, thereby to produce a simulated spatial wind field around the wind turbine system; and controlling the wind turbine system in accordance with the simulated spatial wind field.

2. The method of claim 1, comprising continuously updating the simulated spatial wind field on obtaining further signals indicative of wind speed in the at least two spaced locations.

3. The method of claim 2, comprising monitoring the simulated spatial wind field to derive time-averaged estimated wind conditions around the wind turbine system.

4. The method of claim 3, comprising estimating from the time-averaged wind conditions peak expected loadings on the rotors of the wind turbine system.

5. The method of any preceding claim, comprising estimating loadings on each of the rotors of the system from the simulated wind field.

6. The method of claim 5, comprising estimating fatigue in the rotors based on the estimated loadings.

7. The method of any preceding claim, wherein at least one of the spaced locations is on a nacelle of one of the wind turbines.

8. The method of any preceding claim, comprising obtaining signals indicative of wind speed at locations on or near to each wind turbine of the wind turbine system.

9. The method of any preceding claim, wherein controlling the wind turbine system comprises adjusting one or more of: the rated power of each wind turbine in accordance with the simulated spatial wind field; the predicted lifetime of the wind turbine system in accordance with the simulated wind field; and the specification of each wind turbine in accordance with the simulated spatial wind field.

10. The method of any preceding claim, comprising analysing transient variations in the simulated spatial wind field over time.

11. The method of any preceding claim, comprising deriving from the simulated wind field a respective local spatial wind field simulation for each wind turbine of the system, each local spatial wind field simulation being weighted for its respective wind turbine.

12. A simulation system for simulating a spatial wind field around a wind turbine system, the wind turbine system comprising a plurality of wind turbines mounted to a common support structure, wherein each of the plurality of wind turbines includes a rotor and a power generation system driven by the rotor, the simulation system comprising: sensing means arranged to obtain signals indicative of wind speed in at least two spaced locations, each spaced location being on or near to one of the plurality of wind turbines; and processing means arranged to process the signals indicative of wind speed to calibrate a wind field model, thereby to simulate a spatial wind field around the wind turbine system.

13. The simulation system of claim 12, wherein the processing means is implemented on a computing device that is remote from the wind turbine system.

14. The simulation system of claim 12 or claim 13, wherein the processing means comprises a memory that is arranged to store at least one wind field model to be calibrated.

15. The simulation system of claim 14, wherein the processing means is further arranged to store spatial wind field simulation data in the memory.

16. The simulation system of claim 15, wherein the processing means is arranged to analyse the stored simulation data to determine average wind field data.

17. A computer program product downloadable from a communication network and/or stored on a machine readable medium, comprising program code instructions that implement a method in accordance with any of claims 1 to 11.

18. A machine readable medium having stored thereon a computer program product in accordance with claim 17.