GB2623596A - Wind farm control - Google Patents

Abstract

A method of adaptive induction control for a wind farm is disclosed. The wind farm comprises at least one waked turbine pair comprising a downwind turbine 18 within a wake region of an upstream turbine 16. The method comprises, evaluating a stability-induction relationship to determine a control parameter for the upwind turbine 16 and varying the axial induction of the upstream turbine in response. The stability-induction relation is evaluated based on one or more monitored or predicted atmospheric parameters relating to surface layer stability, and is defined so that a variation in the one or more atmospheric parameters in the absence of variations in both a freestream wind velocity and a wind direction causes a variation in the induction offset. The method allows the power extracted by the pair to be optimised by minimising the wake effect of the upstream turbine 16 on the leeward turbine 18. A method of operating a wind farm, a simulation method, and a wind farm are also disclosed.

Description

WIND FARM CONTROL TECHNICAL FIELD

The disclosure relates to a wind farm control method, an associated controller and wind farm, and simulations to derive parameters for wind farm control.

BACKGROUND

It is known that wind turbines within a wind farm each generate a downstream wake in which the flow conditions, and in particular the flow velocity, differ from the conditions upstream of the respective turbine. As turbines tend to be arranged within a farm so that for many or all wind directions there is at least one turbine disposed within a wake of another turbine, the impact of wake conditions on wind turbine performance has been extensively studied for the purpose of predicting power generation yields. Generally, it is accepted that the wind velocity within the wake is less than the upstream wind velocity.

While a standalone wind turbine may be configured and controlled to extract a maximum amount of power from wind, alternative methods of control have been proposed in the context of maximizing the total power output of a wind farm comprising multiple turbines.

In particular, it has been proposed that the total power output of a wind farm may be improved by controlling upwind turbines to extract less power than a respective maximum, to increase the power that can be extracted by one or more waked turbines (i.e. downwind turbines disposed in a respective wake of an upwind turbine).

One such method of wind farm control is known in the art as "induction control". The expression "induction" (generally interchangeable with "axial induction") refers to the fractional reduction in wind speed (conventionally at the turbine location) relative to the upstream wind speed for a respective wind turbine.

WO 2004/111446 sets out how, according to theory, turbines extract the maximum amount of energy from a fluid if the fluid is decelerated to approximately 2/3 of the upstream wind speed at the location of the turbine, and to approximately 1/3 at a location 1 rotor diameter behind the turbine. The proportion by which the upstream wind speed is reduced to the wind speed in the rotor plane is described by the axial induction factor "a" (e.g. 0.33). WO 2004/111446 refers to conventional practice of designing wind turbines for an axial induction factor of 0.28. WO 2004/111446 discloses a method of wind farm control by which one or more upwind turbines are controlled to have a lower induction factor than waked turbines, in order to realise a net increase in power output from the farm.

SUMMARY

According to a first aspect there is provided a method of adaptive induction control for a wind farm having, for a wind direction, at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate at a respective induction factor; wherein for each turbine, an associated controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method comprising, for one or more waked turbine pairs: determining (e.g. evaluating) a stability-induction relation to determine the control parameter for the upwind turbine in the induction control mode and thereby varying the induction offset; wherein the stability-induction relation is determined (e.g. evaluated) based on one or more monitored or predicted atmospheric parameters relating to surface layer stability, and is defined so that a variation in the one or more atmospheric parameters in the absence of variations in both a freestream wind velocity and a wind direction causes a variation in the induction offset.

The expression "adaptive" as used herein in relation to induction control relates to the variability of the induction offset based on the one or more atmospheric parameters. A method of adaptive induction control may be contrasted with a non-adaptive method which applies a predetermined amount of induction control for a wind farm which is applied irrespective of any local variable conditions relating to surface layer stability (e.g. for a given wind direction).

The method may further comprise transmitting the control parameter to a controller of the upwind turbine, or controlling the upwind turbine according to the determined control parameter.

It may be that the one or more atmospheric parameters is selected from the group consisting of: a value of a wake decay coefficient k for use in a N.O. Jensen wake model, for wind received at the wind farm; a turbulence intensity of wind received at the wind farm; a parameter relating to a temperature gradient in wind received at the wind farm; two or more temperature parameters corresponding to air temperatures at respective heights local to the wind farm; two or more temperature parameters, at least one corresponding to an air temperature at a height local to the wind farm, and at least one corresponding to a local ground or water temperature; an Obukhov length associated with wind received at the wind farm; a gradient Richardson Number, or bulk Richardson Number, for wind received at the wind farm..

The temperature gradient in a layer of wind received at the wind farm may be a temperature gradient in a surface layer, or within the planetary boundary layer.

The method may comprise monitoring two or more temperature parameters from respective temperature sensors at different heights local to the wind farm to determine the one or more atmospheric parameter.

The two or more temperature parameters may be air temperatures as monitored using respective temperature sensors at different heights local to the wind farm. For example, the temperature sensors may be installed on one or more turbines or masts of the wind farm.

The two or more temperature parameters may comprise at least one air temperature as monitored using a respective temperature sensor at a respective height (and optionally at different heights when there are two or more air temperatures monitored), and at least one ground or water temperature as monitored using a respective ground or water temperature sensor.

It may be that the one or more atmospheric parameter is determined based on weather data.

The weather data may be received from a weather data service, such as a weather forecasting or weather monitoring service.

It may be that the control parameter is selected from the group consisting of: a performance target for the respective turbine, for example: a target induction factor; a target thrust coefficient or thrust; a target rotational velocity; a target operating power; or a target operating thrust; a blade angle setting corresponding to a pitch angle of blades of the respective turbine; a generator setting which determines an induced electromotive force of the generator per revolution of the turbine.

It may be that the control parameter is determined by selecting a control law based on the one or more monitored or predicted atmospheric parameter, wherein the control law is for determining the control parameter based on at least one auxiliary variable such as wind velocity or wind direction, wherein the control law is selected from a plurality of control laws which are configured to provide differing profiles of the control parameter for a common operating range of the at least one auxiliary variable.

It may be that the control parameter is determined by reference to a predetermined relation which is a function of the one or more atmospheric parameter, or by reference to a database of predetermined control parameters correlated to the one or more atmospheric parameter.

It may be that the control parameter is determined remotely from the upwind turbine, optionally remotely from the wind farm.

The method may comprise periodically: updating, or determining whether to update, the control parameter based on the one or more monitored or predicted atmospheric parameter; wherein the periodic update or determination of whether to update is performed: at predetermined intervals, for example at intervals of no more than six hours; in response to a threshold change in a monitored parameter relating to an atmospheric condition; or in response to an operator demand.

The monitored parameter may the one or more atmospheric parameters.

According to a second aspect there is provided a method of operating a wind farm comprising, for a wind direction, at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate at a respective induction factor; wherein for each turbine, an associated controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method comprising, for one or more waked turbine pairs: determining the control parameter for the upwind turbine by a method in accordance with the first aspect; and controlling the upwind turbine to operate in the induction control mode based on the control parameter, whereby the induction offset is variable independently of variance of a wind velocity and variance of the wind direction.

The method may comprise: operating the wind farm during an operating period in which surface layer stability conditions change such that there is a change in one or more of: a temperature gradient within a surface layer of wind received at the wind farm; a temperature difference between wind received at the wind farm and a ground or water temperature local to the wind farm; a value of a wake decay coefficient k for use in a N.O. Jensen wake model of the surface layer of wind at the wind farm changes, as calculated based on monitored atmospheric conditions or as determined based on predicted atmospheric conditions; a turbulence intensity of wind received at the wind farm; an Obukhov length, gradient Richardson Number of bulk Richardson Number associated with wind received at the wind farm; wherein the change of surface layer stability conditions causes a corresponding change in the monitored or predicted atmospheric parameter relating to surface layer stability, such that the control parameter for the upwind turbine of the or each waked turbine pair changes during the operating period.

The method may be conducted so that: a change in surface layer stability conditions favours wake recovery; and the corresponding change in the control parameter is to reduce the induction offset; and/or wherein: a change in surface layer stability conditions inhibits wake recovery; and the corresponding change in the control parameter is to increase the induction offset.

According to a third aspect there is provided a method comprising: providing a wind farm model defining a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is modelled to exert a thrust on wind within the model to operate at a respective induction factor; wherein each turbine is modelled based on a variable control parameter relating to an induction factor of the respective turbine, whereby each turbine is operable in an induction control mode in which the control parameter differs relative to a baseline mode to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method further comprising: defining a plurality of simulation cases for the wind farm model by varying each of a set of independent variables, including: freestream wind velocity; wind direction; and an atmospheric parameter relating to surface layer stability; for each simulation case, determining the control parameter for operation of the upwind turbine of at least one waked turbine pair using an optimization procedure having an objective function relating to a total power output of the plurality of turbines; wherein for at least some of the simulation cases, the determined control parameter corresponds to operation of the upwind turbine in the induction control mode to reduce the induction factor of the upwind turbine relative to the baseline mode by a variable induction offset; storing or updating a reference model or database to correlate the determined control parameters for the simulation cases to the atmospheric parameter (and optionally the other independent variables); whereby the reference model or database permits the control parameter for the or each corresponding operational upwind turbine of a wind farm to be determined based on a monitored or predicted value of the atmospheric parameter, so that a variation in the atmospheric parameter in the absence of variations in both a fresstream wind velocity and wind direction causes a variation in the induction offset.

It may be that the atmospheric parameter relating to surface layer stability corresponds to the monitored or predicted atmospheric parameter as defined above with reference to the first aspect. Where the atmospheric parameter relates to a temperature gradient or one or more temperature parameters corresponding to temperatures at respective heights, the wind farm model may be defined to determine the temperature gradient and/or the one or more temperature parameters at locations corresponding to installation locations of corresponding sensors in the associated wind farm (i.e. in the real wind farm that is being modelled, whether this exists or is to be constructed according to the model).

According to a fourth aspect there is provided a non-transitory machine-readable medium storing machine-readable instructions which, when executed by a processor, are configured to cause performance of a method in accordance with any one of the first, second or third aspects.

According to a fifth aspect there is provided a wind farm controller associated with a wind farm comprising a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate according to a respective induction factor; wherein for each turbine, the wind farm controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; wherein the wind farm controller is configured to cause performance of a method in accordance with the first or second aspect; or comprises a non-transitory machine-readable medium storing machine-readable instructions which, when executed by the controller, are configured to cause performance of a method in accordance with the first or second aspect.

According to a sixth aspect there is provided a wind farm controller for a wind farm comprising a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine, each turbine configured to exert a thrust on wind to operate according to a respective induction factor; wherein the wind farm controller is configured to operate each turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the respective turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; wherein for each of one or more waked turbine pairs, the wind farm controller is configured to: determine a stability-induction relation to determine the control parameter for the upwind turbine in the induction control mode and thereby vary the induction offset; determine one or more monitored or predicted atmospheric parameters relating to surface layer stability; wherein the stability-induction relation is defined to be determined based on the one or more atmospheric parameters relating to surface layer stability, and is defined so that a variation in the one or more atmospheric parameters in the absence of variations in both a freestream wind velocity and a wind direction causes a variation in the induction offset.

According to a seventh aspect there is provided a wind farm comprising: a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate according to a respective induction factor; and wherein the wind farm further comprises a wind farm controller in accordance with the fifth aspect or the sixth aspect.

It may be that the wind farm controller is in communication with a remote data source to obtain the one or more atmospheric parameter, or data from which the one or more atmospheric parameter is derivable.

It may be that the wind farm controller is in communication with sensing apparatus installed within the wind farm or local to the wind farm, to monitor the one or more atmospheric parameter or data from which the atmospheric parameter is derivable It may be that the sensing apparatus comprises: one or more temperature sensors; one or more wind velocity sensors; and/or one or more lidar sensors or sonic anemometers, for monitoring an atmospheric parameter relating to atmospheric mixing, turbulence intensity and/or turbulent flux.

The skilled person will appreciate that except where mutually exclusive, a feature described in relation to any one of the above aspects may be applied mutatis mutandis to any other aspect. Furthermore except where mutually exclusive any feature described herein may be applied to any aspect and/or combined with any other feature described herein

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments will now be described by way of example only, with reference to the accompanying drawings, in which: Figures 1 a, 1 b, 2a, 2b schematically show example wind farms layouts and waked turbine pairs; Figure 3 schematically shows a variable wake recovery in a waked turbine pair; Figure 4 schematically shows a velocity profile of wind received at an example wind farm; Figures 5 and 6 are plots of power and thrust curves for various control settings of turbines in an example waked turbine pair; Figures 7 and 8 are plots of optimization results for selecting a value of a control parameter for example waked turbine pairs; Figure 9 is a flow chart of a method of determining a control parameter for variable induction control; Figure 10 is a flow chart of a simulation method for determining control parameters for variable induction control; and Figure 11 schematically shows a non-transitory computer-readable medium and processor.

DETAILED DESCRIPTION

Figures 1a and lb schematically show two respective examples of wind farm layouts 10, 20. In Figure la, the wind farm layout 10 comprises a regular matrix of wind turbines 12 oriented in rows and columns, wherein a spacing between turbines is equal within in each row and within each column. In Figure lb, the wind farm layout 20 comprises a staggered arrangement of wind turbines 22 comprising a plurality of rows. The wind turbines within each row are equally spaced apart, and each row is spaced apart from the adjacent row by the same spacing. However, alternating rows are laterally staggered so that each turbine in a row is laterally offset within its row between the nearest wind turbines 22 of the adjacent row. These wind farm layouts are examples only, and other wind turbine layouts are possible and aspects of the present disclosure apply equally to such layouts.

Each of Figures la and lb show a longitudinal wind direction 2, and a wake path 14, 24 is shown extending from an example upwind turbine 16, 26 of each wind farm layout. As indicated in Figures la and lb, for each wind farm layout there is at least one downwind turbine 18, 28 which falls within a wake region 14 of the respective upwind turbine 16. In the context of the present disclosure, an upwind turbine and a downwind turbine which is within a wake region of the upwind turbine for a given wind direction is referred to as a waked turbine pair.

A wind farm layout may be such that for a plurality of wind directions, or for substantially all wind directions, there is at least one waked turbine pair.

Figures 2a and 2b show the same respective wind farm layouts 10, 20 as shown in Figures la and lb, but with a different wind direction 2'. Figures 1 and 2 together illustrate that the or each downwind turbine 18, 28' in a waked turbine pair may vary depending on the wind direction 2, 2'.

In some wind farm layouts, it may be that a downwind turbine in a first order waked turbine pair also forms an upwind turbine in a second order waked turbine pair. In such circumstances, there may be a compound induction effect on the flow in the two waked turbine pairs, and for this reason the expression "first order" and "second order' etc. is used herein to indicate how many times a particular flow path has passed through a turbine and been subject to induction (i.e. axial speed reduction). The discussion and analysis presented herein is generally in the context of a single waked turbine pair (e.g. a first order pair). However, it will be appreciated that any analysis of induction as disclosed herein can be readily extended by a person skilled in the art to consider compound induction effects (e.g. second and higher order waked turbine pairs).

Figure 3 schematically shows an example wake region 14 (comprising turbulent eddies 15) extending downstream from an upwind turbine 16 towards a downwind turbine 18, with a wind direction 2. The downwind extent 19, 19' of the wake region (e.g. the region in which the wind speed is significantly less than the upstream wind speed) is considered to be variable (as illustrated), and moreover the degree to which wind speed in the wake is less than the upstream wind speed and the rate at which it recovers to the upstream wind speed are also variable; and this phenomenon is referred to in the art as wake recovery. As will be discussed in further detail herein, factors influencing wake recovery include the turbulence intensity of the wind. Wake recovery is discussed in the present disclosure by reference to the downwind extent of the wake over which the wind speed in the wake recovers in velocity. Wake recovery is discussed herein by reference to factors which influence the downwind extent over which the wind speed recovers, which may be assessed or evaluated by reference to any suitable threshold expressed as a function of the upstream wind velocity (e.g. 80%, or even 100%). Accordingly, factors which promote wake recovery tend to reduce the downwind extent over which wake recovery occurs (e.g. to any suitable threshold expressed as a function of the upstream wind speed), whereas factors which suppress wake recovery tend to increase the downwind extent over which wake recovery occurs (e.g. to any suitable threshold expressed as a function of the upstream wind speed). The expression "wake recovery" as used herein does not correspond to a particular mathematical definition for evaluating that wake recovery has occured, and is used to conceptually explain how various factors that impact wake recovery affect the wind speed experienced at a downwind turbine of a waked turbine pair.

Figure 4 schematically shows a plurality of wind turbines 12 of a wind farm 10 disposed above a surface, in this example an ocean surface between atmospheric air and sea water 44. For simplicity, only a first one of the wind turbines 12 is shown with a supporting tower (although it is to be appreciated that each turbine 12 has such a tower).

Figure 4 schematically shows an example velocity distribution 46 within a planetary boundary layer of wind received at the wind farm 10. The planetary boundary layer has a variable height h. In this example, the farm 10 further comprises a monitoring mast 42. In this example, sensing apparatus 48 is provided on the mast 42. As will be described in further detail below, sensing apparatus 48 may comprise sensors monitoring one or more atmospheric parameters, including and not limited to: temperature (including temperature difference); wind velocity, turbulence intensity. A sensing apparatus may include a plurality of sensors at different heights, for example a plurality of temperature sensors for monitoring air temperatures at different heights and/or one or more temperature sensors for monitoring a ground or water temperature. Such a sensing apparatus may additionally or alternatively include sensors for monitoring velocity at different heights (e.g. approximately the same heights as a corresponding plurality of temperature sensors). A sensing apparatus 48 may comprise a sensor for monitoring an atmospheric parameter relating to atmospheric mixing, turbulence intensity, or turbulent flux, such as a lidar sensor or sonic anemometer. For example, it is known to use such sensors (e.g. a sonic anemometer) to determine an Obukhov length using eddy-correlation techniques (as discussed in "the 2015 Sanz Rodrigo paper" referenced elsewhere herein.

Additionally or alternatively to providing sensing apparatus 48 on a mast, such sensing apparatus may be provided on one or more towers of a wind turbine 12, as illustrated in Figure 4.

The expression surface layer as used herein relates to a sub portion of the planetary boundary layer (also known as the atmospheric boundary layer). In the technical field, the planetary boundary layer is understood to be the region in which friction effects associated with atmospheric interaction with a surface influence flow. The surface layer is a relatively thinner layer of fluid closer to the respective surface (e.g. an ocean or land surface), and is the layer in which friction effects are generally constant throughout (rather than decreasing with height). Temperature inversion relates to the planetary boundary layer being covered by a layer of warmer air.

As discussed elsewhere herein, stability in the surface layer (e.g. as represented by the wake decay coefficient k) may be a function of thermal gradients, turbulence and other parameters affecting air received at a wind farm.

As will be discussed in further detail below, the inventors have determined that the induction of one or more upwind turbines (of respective waked turbine pairs) can be controlled to achieve net increases in generated power from the wind farm as a whole, by adjusting the balance of induction and therefore power generation between upwind and downwind turbines, taking into account atmospheric parameters relating to wake recovery.

While it is known to perform induction control (i.e. controlling the induction of an upwind turbine) in order to achieve net gains in power as generated by the wind farm as a whole, existing proposals consider the influence of an upwind turbine on a downwind turbine to only be a function of the respective induction factor and the separation between the rotors The inventors have determined that the optimum amount of induction control to apply varies in dependence on the wake recovery. When wake recovery is strong (e.g. occurs over a relatively short distance), a downwind turbine is less adversely affected by the presence of an upwind turbine, and so the advantages of decreasing the induction factor a at the upwind turbine (and consequently reducing its power output) are commensurately less. In contrast, when wake recovery is relatively poor (e.g. occurs over a relatively long distance), a downwind turbine is relatively more adversely affected by the presence of the upwind turbine, and so it is generally advantageous to have a relatively low induction factor a (and therefore lower power output) at the upwind turbine.

In the present disclosure, an amount of induction control is discussed by reference to a baseline mode of operation and an induction control mode of operation. As noted above, there is a general consensus that in suitable wind conditions, a wind turbine extracts a maximum amount of power when operated to achieve a baseline induction factor, for example 0.33 (noting that this may vary between wind farm locations and wind turbine designs, for example the baseline induction factor may be in the range 0.25-0.4 for any particular wind turbine). A controller for a wind turbine according to the present disclosure is configured to operate a wind turbine in the baseline mode and the induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset. The control parameter is determined (e.g. evaluated) based on a stability-induction relation as is described elsewhere herein.

When an upwind turbine in a waked turbine pair is operated in the induction control mode it extracts less power than it would do in the same conditions in the baseline mode, but as referenced above the waked turbine pair may nevertheless generate more power between them. This is because the velocity of wind provided to the downwind turbine increases as the induction factor of the upwind turbine reduces. This corresponds to the upwind turbine having a purposefully lower thrust coefficient, and hence lower power output, than in a typical condition where it is not waking a downwind turbine, such that the waked pair in combination have a greater power output.

When an upwind turbine of a waked turbine pair is in the induction control mode, the waked turbine pair may have a low induction condition whereby the induction factor of the upwind turbine is relatively lower than the induction factor of the downwind turbine.

The thrust coefficient CT at which a turbine operates is equal to the thrust force exerted on the wind in the streamwise direction of the wind, divided by the product of a rotor swept area A and the dynamic pressure of the upstream wind, %pU2. In equation form, the thrust coefficient CT is: CT -7pU where T is the thrust force, p is the air density, and U is the axial wind velocity.

It is accepted that the thrust coefficient CT is related to the axial induction factor a. As noted in the background, the axial induction factor a is the fractional decrease in wind velocity between the upstream wind and the turbine rotor (i.e. assessed at the turbine rotor). An accepted relationship which holds for the present disclosure is that: CT = 4a(1 -a) Such a definition may be adopted as a definition of the axial induction factor as used herein. Other relationships (e.g. empirical relationships) between the thrust coefficient CT and the induction factor may equally be applied to the present disclosure.

The present disclosure refers to wind turbines as having different operating modes, and the selection of a mode for an upwind turbine to control induction. A mode may be defined by one or more variable control parameters for the turbine, or by one or more control laws for respective control parameters.

Examples of such variable control parameters include a blade pitch setting (corresponding to the pitch angle of the blade relative to the upstream or freestream wind direction), and a generator selling which determines an induced electromotive force of the generator per revolution of the turbine (and thereby effectively corresponds to the resistance or work done by wind on the turbine to rotate the generator). Both the pitch setting and the generator selling are examples of direct control parameters, respectively directly influencing (i) how the blades convert kinetic energy of the wind into a turning force for rotation of the turbine, and (ii) how the generator converts the kinetic energy of the turning turbine into electrical power. Varying the pitch setting varies both the turning force (effectively the "lift" generated by each blade), and thereby causes variation of the rotational speed of the turbine (e.g. when the resisting torque does not change) or the torque transferred to the powertrain (e.g. when the rotational speed does not change), or a combination of the two. Varying the induced electromotive force of the generator per revolution of the turbine effectively varies the power generation per revolution, and thereby the torque required to turn the turbine. Such control thereby permits the rotational speed of the turbine, and thereby the induction factor and thrust coefficient, to be adjusted independently of the blade pitch setting.

By way of example, a generator setting which can be varied to adjust the induced electromotive force of the generator per revolution of the turbine is a gear ratio between the rotating blades and the generator, for example a gear ratio of a gear arrangement (e.g. a gearbox) therebetween. The generator setting may be a generator speed setting, varied using a gear arrangement between the rotating blades and the generator.

The present disclosure also envisages the use of indirect control parameters that may be specified as a target parameter for operation of the respective turbine, and which the respective turbine may achieve by independent variation of one or more settings (e.g. a blade pitch setting and/or a generator settings, as discussed above). Operation of the respective turbine may be adjusted to achieve the respective target parameter based on one or more monitored or predicted parameters associated with the wind turbine and/or wind farm. Indirect control parameters may be any of: (I) a target induction factor for the turbine (which may be evaluated by reference to monitored and/or predicted wind velocities); (ii) a target thrust coefficient or thrust (iii) a target rotational velocity (which may be evaluated by reference to an output of a rotational sensor, or derived from other sensors such as torque and power sensors); (iv) a target operating power (which may be evaluated by reference to a power monitor associated with the turbine); and (v) a target operating thrust (which may be evaluated by reference to a force monitor associated with the turbine, or derived from a relationship with other performance parameters and control settings, such as torque, blade pitch setting and power).

A mode (e.g. a fixed mode) for operating a turbine may define one or more values for control parameters as a constant (or constants) to be maintained through an operating range of the turbine. For example, an example mode for operating a turbine may be defined based on a fixed blade pitch setting (e.g. 10° from a neutral or feathered angle at which the blade does not generate a turning force).

A mode (e.g. a control law mode) for operating a turbine may define a control law for one or more control parameters as a function of a monitored variable. For example, the control law may define a value for blade pitch setting as a variable parameter which is a function of a monitored wind velocity.

There may be a baseline or default mode, which may be defined as either a fixed mode or control law mode as indicated above. Additionally, there may be one or a series of induction control modes which are defined to vary the operation of the wind turbine relative to the baseline mode to reduce an induction factor for the respective turbine. For example, each induction control mode may be defined to superpose a value on a corresponding baseline value according to the baseline mode, for example by adding a constant. For example, there may be 10 induction control modes that each define a control law for the blade pitch setting as a fixed value ranging from 1 to 10 respectively, in addition to a variable value determined based on the baseline mode. By way of example, the baseline mode may vary the blade pitch setting value as a function of (for example) a monitored wind velocity.

Figure 5 is a plot of thrust coefficient vs wind speed for modes 1 and 10 of an example range of respective modes, which may correspond to a baseline mode (mode 1) and a series of 9 induction control modes (2-10). In this example, each induction control mode implements an additional 0.5° constant, such that the constants vary from 0° to 4.5° between the modes. Figure 5 shows the variable profile of thrust coefficient over a range of windspeeds for each of modes 1 and 10 (indicated as "controller 1" and "controller 10" respectively). Figure 5 also shows point values for the respective modes at a common wind speed of 8m/s. In the technical field, the profile of thrust coefficient is known as a thrust curve for the respective controller/control law.

Figure 6 shows a similar plot of power against wind speed for modes 1 and 10. (a "power curve". In the technical field, the profile of power is known as the power curve for the respective controller/control law. Point values are shown for the two modes at a common wind speed of 8m/s.

In this example, it is assumed that the downwind turbine always operates in mode 1 (the baseline mode), which may correspond to maximum extraction of power from the wind. However, the mode for the upwind turbine may vary. Figure 6 shows, using square point markers (overlapping), various power values for the downwind turbine, corresponding to operation of the upwind turbine in each of the respective operating modes 1-10. There is a trade-off in that, a decrease in the axial induction factor of the upwind turbine corresponds to an increase in the thrust and therefore power generation at the downwind turbine, but generally by unequal amounts. Accordingly, there is an optimum amount of induction control to be performed at the upwind turbine. Induction control is performed by selecting the mode for the upwind turbine which corresponds to this optimum.

As noted above, the inventors have determined that conventional induction control can be improved by adaptively applying the induction control based on prevailing conditions relating to atmospheric stability.

In particular, the following worked example demonstrates how the inventors have determined the influence of atmospheric stability on the dynamic control of induction to optimize total power generation. In this simplified example, two turbines Ti (upwind) and T2 (downwind) are considered. The downwind turbine T2 is considered to be fully waked, and separated from upwind turbine Ti by a distance x. The turbines have rotor diameter D. The upwind turbine Ti experiences free stream wind speed U. It is assumed that a N.O. Jensen wake model applies. Each turbine operates according to a selected mode, with two different modes indicated in the equations below using A and B respectively.

The N.O. Jensen wake model is well known in the art. It uses a wake decay coefficient k, which determines how quickly the wind field downwind of a turbine recovers to the upstream conditions (e.g. freestream conditions for a single turbine). The N.O. Jensen wake model is described extensively in the literature, including at section 2.1 of the paper "On the application of the Jensen wake model using a turbulence-dependent wake decay coefficient: the Sexberum case", Alredo Pena et al, Wind Energy 2016; 19:763-779, available online at the following locations (referenced herein as "the 2015 Wind Energy paper": httris:lloni inel i bra rv.wi ley c rnidoi/ful1110.1002iwe.1863; https://doi.oral10.1002Ave.1863 The total power generation, using the same operating modes for both turbines, is therefore: PtAoiActt = PP1 ± PP 2 = PA(U)+P(u) where u2 is the local wind speed at turbine T2, as affected by mode A at turbine Ti. When adjusted so that upwind turbine Ti uses controller B to impart induction control, the total power generation becomes: IL= = 1131 P1712 = 41(11) P142 (U173'2) (1) Induction control is only useful if P total 1> ,A,A o a total Re-arrangement of this condition provides: cy20.42) - (42) > (U) -P1(U) (2) The right-hand side of the above inequality can be readily evaluated by reference to power curves for modes A and B of the turbines. The left hand side may be linearised about the point un = 42 to permit further analysis, resulting in the following inequality: (42 -742) (-dP)A (3) du _ A > P1(U) P1(U) u-uT2 This inequality can be further expanded to express the local wind speeds at turbine T2 as a function of the thrust coefficient at turbine Ti. In particular, by applying the Chain Rule Ida" du = dCt (--act aa it follows that:

A

[(Ct(U))B" (CtUn)Ani)A dun > PA(U) -P(U) (5) t ti=u dam_ '.dtz1 A U=UT2 The expression in square brackets can be readily evaluated based on the thrust curves for the turbines operating the respective modes. The local power curve gradient A (-al))can similarly be readily derived from the power curve for the respective turbine. u d " A The derivative of the induction factor can be determined from re-arrangement and derivation of the equation for the induction factor given above, namely: a = 1/ 2(1 - -and therefore: (let da -= 4 -8a -1 ace aCr 4-8a Finally, from the N.O. Jensen wake model, an expression utilising the wake decay coefficient can be substituted for the derivative of velocity with induction factor, as follows (derivable from equation 1 in section 2.1 of the 2015 Wind Energy paper): 1 u2 2a1 2kx)2 and therefore: du, 2L1 da, (1±2kry

D

The above inequality at (5) can therefore be re-arranged as follows: 1 -2(1 n)A > 13+1,(U) -PPi(U) (6) -(Ct 4-8a / ) 2kx)2 Vau u=iLAI D Tz (4) The left hand side is the product of four components.

Component 1 is known from the thrust curves for the turbines. If induction control is being performed by selection of mode B, then the coefficient of thrust will always be less for mode B, and therefore component 1 will be of negative sign.

Component 2 is always positive because a is always less than 0.5.

Component 3 is always negative, because each variable is always positive (including the wake decay coefficient k), and the only source of a negative sign is the "-2" in the numerator.

Component 4 is always positive, as power increases with velocity for all wind speeds below rated power.

Accordingly, the left hand side is positive when induction control is performed at the upwind turbine Ti. The right hand side will be positive when induction control is performed at the upwind turbine Ti, since the power generated should be less when using mode B to implement the induction control.

By analysing the factors which influence the total power output in this way, the inventors have determined that while many of the parameters determining the net power benefit of induction control are based on the thrust and power curves for the turbines and other fixed quantities such as the rotor separation x and diameter D, the wake decay coefficient k is a variable which influences the above equation. In particular, it can be seen that the left hand side of inequality (5) above becomes greater as the wake decay coefficient k becomes lower.

The inventors have determined that this indicates that the opportunity for advantageous induction control is greater in the case of relative low wake spreading (i.e. for low k).

The wake decay coefficient k is generally related to the stability of the surface layer, which is influenced by each of the atmospheric parameters relating to surface layer stability as discussed herein. For example, the "2015 Wind Energy" paper referenced above discusses a relationship between the wake decay coefficient k and the surface layer stability, by reference to the local atmospheric stability correction 1pm (L) in equation 4 in section 2.1.2, which is expressed for a height h and specific stability condition (measured by the Obukhov length L), as described in the 2015 Wind Energy paper and its references, including reference 15: Obukhov AM. Turbulence in an atmosphere with a non-uniform temperature. Boundary-Layer Meteorology 1971; 2:7-29.

The Obukhov parameter (or Obukhov length) is considered to be an indication of surface layer stability, as is the local atmospheric stability correction referenced above. These are example atmospheric parameters from a wider category of atmospheric parameters which, in the technical field, are considered to relate to (e.g. be a function of or indicate) surface layer stability. The 2015 Wind Energy paper demonstrates the link or relationship between such atmospheric parameters and the wake decay coefficient k (and the turbulence intensity TO. It is understood that other atmospheric parameters have corresponding relationships with surface layer stability.

Other example atmospheric parameters relating to surface layer stability (e.g. being a function of or an indicator of surface layer stability) include: a value of a wake decay coefficient k for use in a N.O. Jensen wake model; a turbulence intensity of wind (the standard deviation of fluctuating wind velocity to mean wind speed) a parameter relating to a temperature gradient in wind. For example, the parameter may be a temperature gradient or temperature difference between two heights, for example as discussed in the "2007 Jensen Paper" referenced below. Such a temperature gradient or temperature difference may correspond to temperatures at two heights within a layer of wind, or may correspond to temperature at a height in the layer of wind and a ground or water temperature; the Obukhov length; the gradient Richardson Number; the bulk Richardson number associated with wind, for example as discussed in "the 2015 Sanz Rodrigo Paper" referenced below; a buoyancy frequency parameter, also known as the Brunt-Vaisala frequency, for example as discussed in "the 1988 Kitaigorodskii Paper" referenced below.

The reference for "the 2007 Jensen Paper" is Jensen, L.E. Analysis of Array Efficiency 25 at Horns Rev and the Effect of Atmospheric Stability In Proceedings of the 2007 EWEC Conference, Milan, Italy, 7-10 May 2007 The reference for "the 2015 Sanz Rodrigo Paper" is J. Sanz Rodrigo. Atmospheric stability assessment for the characterization of offshore wind conditions. Wake Conference 2015. Journal of Physics: Conference Series 625 (2015) 012044.

D01:10.1066/1742-6596/625/1/012044. The 2015 Sanz Rodrigo paper refers to the Obhukov length being determined, for example, by use of sonic anemometer measurements. The 2015 Sanz Rodrigo paper refers to the gradient Richardson Number and bulk Richardson number being determined by reference to temperature and/or velocity measurements, in particular at different heights and/or at a surface (e.g ground or water). The 2015 Sanz Rodrigo Paper refers to the equivalence between various methods for evaluating stability.

The reference for "the 1988 Kitaigorodskii Paper" is S. A. Kitaigorodskii & Sylvain M. Joffre (1988). In search of a simple scaling for the height of the stratified atmospheric boundary layer, Tellus A: Dynamic Meteorology and Oceanographic, 40:5, 419-433, DOI: 10.3402/tellusa.v40i5.11812. The Brunt-Vaisala frequency is given as N where gfl is a buoyancy parameter equivalent to gIT (g is the gravitational acceleration constant, T is temperature), and y is the background stratification dO/dz corresponding to the temperature gradient above the planetary boundary layer (atmospheric boundary layer, ABL). It is considered that y effectively corresponds to stability, and there the Brunt-Vaisala frequency is also related to stability.

Each of the atmospheric parameters above are considered to be indicative of surface layer stability, and thereby wake recovery and/or decay. The discussion above shows how adaptive induction control as a function of wake recovery (and/or turbulence intensity) can be implemented to maximise net power output of a waked turbine pair and/or wind farm in general. Similarly, as the atmospheric parameters discussed herein are indicative of surface layer stability (and thereby wake recovery), suitable stability-induction criterions to achieve such variable induction control can be defined based on any of the atmospheric parameters discussed herein to achieve the corresponding advantages. Any such stability-induction criterion may be a function of one, or more than one atmospheric parameter. Any such stability-induction criterion may be defined to determine a control parameter for a wind turbine as a direct function of the atmospheric parameter, or may be evaluated by first determining an intermediate function (for example a wake decay coefficient), and then determining a control parameter for a wind turbine as a function of the wake decay coefficient.

For example, a monitored temperature gradient in wind approaching a wind farm may be used to derive a predicted wake decay coefficient k, by virtue of theoretical and/or empirical relationships between the temperature gradient and the wake decay coefficient. The wake decay coefficient may then be used to determine a control parameter for a wind turbine corresponding to optimum induction control as discussed elsewhere herein.

Figures 7 and 8 show plots of combined power from the upwind and downwind turbines of a waked turbine pair as determined by simulation, with the X-axis indicating the mode number of the upwind turbine (the downwind turbine operating in model, as discussed above). As indicated in the header of each plot, the variables associated with the respective simulation including the turbulence intensity TI (%), wake decay constant k On this example calculated as the turbulence intensity multiplied by 0.004), the velocity upstream of the upwind turbine (U), and the separation between turbines (in multiples of rotor diameter). The evaluation of power and thrust performance is based on a N.O. Jensen wake loss model.

Figure 7 includes a progression of three plots representing identical conditions for velocity and rotor separation (8m/s and 5 rotor diameters (5D) respectively), but with turbulence intensities varying from 6% to 10%. As shown, the peak combined power corresponds to a different mode for the upwind turbine in each case, varying from mode 8 (3.5°) to mode 6° as turbulence intensity varies from 6% to 10%.

Figure 8 shows a similar trend for the same set of conditions, amended only in that the turbine separation is 10 rotor diameters (10D). In this example, the optimum mode number for the upwind turbine varies from 6 (2.5°) to 4 (1.5°).

Figure 9 is a flow chart of a method of adaptive induction control taking into account atmospheric conditions (apart from wind speed and wind direction). The method may be performed by a controller (e.g. a microprocessor or computer executing control instructions), which may be implemented local to or remote from a wind farm. For example, the controller may be located in a hub local to the wind farm and communicatively connected to the respectively turbines. The hub may be located in a tower of a wind turbine, for example. Alternatively, the controller may be in a location remote from the wind farm (for example, at a remote data processing site controlled by the wind farm operator).

Blocks 902, 904, 906 indicate possible sources of an atmospheric parameter relating to surface layer stability, or data from which an atmospheric parameter may be derived. It will be appreciated that a method according to the disclosure may be implemented using only one, or one or more of these sources.

Block 902 provides weather data as may be received by the controller from a forecasting service (for example over the internet). Weather data may include parameters relating to wind speed, wind direction, temperature and the like. Weather data may also include parameters relating to surface layer stability, for example one or more of a wake recovery factor k, a turbulence intensity, a predicted thermal gradient, a predicted air temperature, a predicted ocean surface temperature, an Obukhov length (or parameter), gradient Richardson Number, bulk Richardson Number. Weather data may include data relating to real-time (e.g. prevailing) conditions, for example as compiled based on a meteorological model in conjunction with real-time monitoring data. Alternatively or additionally, weather data may include forecast data, for example including predicted weather data for one or more times, as compiled at an earlier time.

The atmospheric parameter relating to surface layer stability may be determined directly from the weather data (i.e. by virtue of being included in the weather data), or may be indirectly derived. For example, the atmospheric parameter may be a predicted thermal gradient of the planetary boundary layer, and for an ocean wind farm this may be derived based on a theoretical or empirical relationship utilizing one or more parameters of the weather data, including at least the air temperature and ocean surface temperature.

The temperature gradient in the wind received at a wind farm is indicative of the surface layer stability, and therefore the turbulence intensity and wake recovery.

Blocks 904 and 906 relate to local and on-farm monitoring respectively. On-farm monitoring relates to the use of sensors in the immediate vicinity of the farm, for example sensors installed on the wind turbines or masts provided for the wind farm. Local monitoring may additionally include local monitoring from data sources that are separate from the farm but which the same conditions. For example, for an ocean wind farm, a local monitoring source may be a weather station on a nearby offshore platform or nearby land-based station. Local monitoring may include sensors within a 100km radius, or a 50km radius.

Data from each of remote, local and on-farm monitoring may be transmitted to the controller by any suitable means, for example over the internet. On-farm monitoring may be transmitted using a local wired or wireless network.

Block 910 denotes actions performed by the controller as described above. In block 912, the controller determines an atmospheric parameter relating to surface layer stability based on one or more of the weather data 902, local monitoring data 904, and on-farm monitoring data 906.

The atmospheric parameter may be any one or more of (i) a value of a wake decay coefficient k for use in a N.O. Jensen wake model 00 a turbulence intensity of wind received at the wind farm; (iii) a parameter relating to a temperature gradient in a layer of wind received at the wind farm; (iv) two or more temperature parameters corresponding to air temperatures at respective heights local to the wind farm; (v) two or more temperature parameters, at least one corresponding to an air temperature at a height local to the wind farm, and at least one corresponding to a local ground or water temperature; and (vi) an Obukhov length, gradient Richardson Number or bulk Richardson number associated with wind received at the wind farm. Any of these atmospheric parameters may be determined directly from monitored sensor data (e.g. on-farm or local monitoring), or predicted (e.g. derived) from a model which uses such monitored sensor data and/or remotely sourced data (e.g. weather data).

At block 914, the controller evaluates a stability-induction relation for one or more waked turbine pairs of the wind farm, optionally each waked turbine pair. The evaluation of the stability-induction relation may be conducted to determine an optimum control parameter for the upwind turbine in one or more waked turbine pairs, as described above with reference to Figures 7 and 8. As described above with reference to Figures 7 and 8, the stability-induction relation is defined so that a variation in the atmospheric parameter can cause a variation in the optimum mode for the upwind turbine, even in the absence of any variation in both a freestream wind velocity and wind direction. In the examples of Figure 7 and Figure 8, a variation in the turbulence intensity corresponds to a variation in the blade pitch setting over a range of 1° for the upwind turbine as compared with the downwind turbine. This corresponds to the upwind turbine operating in the induction control mode, whereby the induction factor of the upwind turbine is reduced relative to the induction factor in the baseline condition by an induction offset. The induction offset is caused by variation of the respective control parameter. The control parameter in the examples of Figures 7 and 8 is the blade pitch setting, but may equally be any other control parameter as described herein. Operation with the upwind turbine in the induction control mode may correspond to the waked turbine pair being in a low induction condition as discussed elsewhere herein, with the induction factor for the upwind turbine being lower than induction factor for the downwind turbine.

The evaluation of the stability-induction relation is conducted based on the atmospheric parameter. The evaluation may be conducted in a number of ways.

For example, the evaluation may be conducted by the controller executing an optimization in which a model relating to turbine performance is evaluated based on the atmospheric parameter for a plurality of control cases to determine one or more performance parameters associated with each turbine such as induction factor, thrust and/or thrust coefficient, and generated power. The control cases for each waked turbine pair may define a range of different control parameters for the upwind turbine, and the optimization may determine a selected control case based on a determined net power generation of the waked turbine pair and/or the wind farm (e.g. typically selecting the control case that results in the peak/optimum net power generation). The control parameter(s) associated with the control case may then be selected for operation of the or each upwind turbine (block 916).

Alternatively, the evaluation may be conducted by the controller referencing a database or model which correlates the one or more control parameters to one or more independent variables associated with the wind farm, including at least the atmospheric parameter. Additional independent variables may include the wind speed and wind direction, and the identity of the respective turbines of the waked turbine pair. For example, a database may be stored local to or remote from the controller, and the controller may query the database based on independent variables such as (i) the atmospheric parameter, (ii) a monitored or predicted wind speed; (iii) a monitored or predicted wind direction; (iv) an identifying parameter for the respective upwind turbine and/or the associated downwind turbine. The controller then receives the respective control parameter for the upwind turbine from the database (block 916).

Although the above examples refer to the selection of a control parameter for the upwind turbine (e.g. a value of a blade pitch setting), the disclosure also envisages the selection of a control law at block 916 in order to determine the control parameter, wherein the control law defines one or more control parameters as a variable quantity dependent on one or more independent variables. For example, as discussed above with reference to Figures 5 and 6, a plurality of control laws may each define a common variable relationship between a variable component of a blade pitch setting and an independent variable (e.g. wind velocity), but each differ by virtue of a constant component to be added to the variable value (e.g. between 0° and 4.5° as discussed above), throughout the range of the independent variable.

As discussed elsewhere herein, any evaluation of the optimum induction control as envisaged herein (whether by direct optimization or by reference to a database) may be applied for a cascade of waked turbine pairs, for example including second and higher order waked turbine pairs.

Figure 9 schematically indicates a communicative link between the determined control parameter 916 and one or more upwind turbines 16 in a wind farm, with Figure 9 schematically showing two waked turbine pairs each comprising an upwind turbine 16 and a downwind turbine 18.

Figure 9 also schematically indicates a communicative link between a turbine 16 and the on-farm monitoring data 906. As described above, sensors may be provided on a turbine, for example on a turbine tower, and sensor data relayed to a controller.

Figure 10 is a flow chart of a simulation method for defining a reference model or database.

Block 1010 denotes independent variables for the simulation method, including at least wind velocity 1012, wind direction 1014, and an atmospheric parameter relating to surface layer stability as discussed above (1016).

Block 1020 denotes steps of the simulation method. The method may be conducted by a controller (e.g. a computer) associated with wind farm control (e.g. used for wind farm control), or a separate controller or computer that is not associated with wind farm control, such as a computer used to generate data that is subsequently provided to a wind farm controller.

In block 1022, a plurality of simulation cases are defined. The simulation cases are defined to produce different permutations of values for the independent variables, including at least wind velocity, wind direction, and the atmospheric parameter. Additional independent variables may be additionally defined, including parameters relating to the spatial separation between the turbines of a waked turbine pair, a position of the downwind turbine in the wake region of the upwind turbine (e.g. centrally disposed or laterally offset relative to the wind direction by a variable amount), and parameters relating to the order of induction effect for the waked turbine pair (i.e. whether the waked turbine pair is a first order, second order, or any higher order waked turbine pair). Such additional independent variables may be specified directly or by reference to other variables. For example, it will be appreciated that such variables may alternatively be defined by specifying the identities of the upwind and downwind turbines in a known layout of wind farm, in conjunction with the wind direction.

The simulation cases may be defined for a particular waked turbine pair, or for a plurality of waked turbine pairs.

The flow chart denotes the defined simulation cases at block 1024.

In block 1026, a control parameter is determined for an upwind turbine of the or each waked turbine pair, by conducting an optimization procedure as described above, having an objective function relating to the total power output of the turbines (i.e. of the turbines of the pair, or for the wind farm as a whole).

As described above, the control parameter may be defined as a direct control parameter such as a blade pitch setting or generator setting, or as an indirect control parameter, such as an induction factor or thrust coefficient. For example, by determining an induction factor or thrust coefficient which corresponds to optimum performance, the simulation model may determine a functional control parameter that can be effective for controlling the selected wind turbines, but is independent of the particular configuration of the respective turbines. For example, the simulation model may define a target thrust coefficient for each upwind turbine in a waked turbine pair. The target thrust coefficient may be applicable to a range of different wind turbines having different relationships between blade pitch settings or generator settings and thrust coefficient. Accordingly, in operation of a wind farm, a separate controller may independently control operation of the respective turbine to vary a direct control parameter (e.g. a blade pitch setting) to achieve the target thrust coefficient. As above, the optimum control parameter may be determined by virtue of selecting an optimum control law which defines the control parameter for the respective conditions (i.e. for the respective independent variables).

The determined control parameters are stored in a reference model or database (block 1030). For example, the reference model or database may be stored in a location where it is remotely accessible (e.g. over the internet). Instances of the reference model or database may be transmitted or copied to controllers for one or more wind farms, which may be local or remote from a wind farm as discussed elsewhere herein.

Block 1040 denotes steps performed by a wind farm controller. In block 1042, the controller monitors or predicts atmospheric conditions to determine independent variables for operation of the wind farm, including one or more atmospheric parameters relating to surface layer stability, and optionally wind velocity and wind direction. The controller may monitor or predict the atmospheric conditions by virtue of receiving data relating to the independent variables, as discussed elsewhere herein, or by monitoring or deriving such parameters from sensor data.

In block 1044, the controller determines a control parameter for each upwind turbine of one or more waked turbine pairs. The control parameter is determined by querying the reference model or database 1030 based on the determined independent variables.

The control parameter may be determined periodically or in response to a stimulus. For example, it may be determined to update, or determine whether to update, the control parameter (i) at predetermined intervals, for example of no more than six hours; and/or (ii) in response to a threshold change in a monitored parameter relating to an atmospheric condition (e.g. a monitored atmospheric parameter relating to surface layer stability, or any other relevant atmospheric condition such as wind velocity and wind direction). It may be determined to update the control parameter in response to an operator demand.

In block 1046, the controller controls the or each upwind turbine of the one or more waked turbine pairs using the determined control parameter.

The controller(s) described herein may comprise a processor. The controller and/or the processor may comprise any suitable circuity to cause performance of the methods described herein and as illustrated in the drawings. The controller or processor may comprise: at least one application specific integrated circuit (ASIC); and/or at least one field programmable gate array (FPGA); and/or single or multi-processor architectures; and/or sequential (Von Neumann)/parallel architectures; and/or at least one programmable logic controllers (PLCs); and/or at least one microprocessor; and/or at least one microcontroller; and/or a central processing unit (CPU), to perform the methods and or stated functions for which the controller or processor is configured.

The controller may comprise or the processor may comprise or be in communication with one or more memories that store that data described herein, and/or that store machine readable instructions (e.g. software) for performing the processes and functions described herein (e.g. determinations of parameters and execution of control routines).

The memory may be any suitable non-transitory computer readable storage medium, data storage device or devices, and may comprise a hard disk and/or solid state memory (such as flash memory). In some examples, the computer readable instructions may be transferred to the memory via a wireless signal or via a wired signal. The memory may be permanent non-removable memory, or may be removable memory (such as a universal serial bus (USB) flash drive). The memory may store a computer program comprising computer readable instructions that, when read by a processor or controller, causes performance of the methods described herein, and/or as illustrated in the Figures. The computer program may be software or firmware, or be a combination of software and firmware.

By way of example, Figure 11 schematically shows a non-transitory computer-readable medium 1102 (e.g. a memory) comprising instructions 1104 which, when executed by a processor 1106, cause performance of a method. The instructions may be to cause performance of any method as described herein. A controller as described herein may comprise a non-transitory computer-readable medium 1102 having such instructions, optionally together with a processor 1106.

Claims (19)

1. CLAIMS1. A method of adaptive induction control for a wind farm having, for a wind direction, at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate at a respective induction factor; wherein for each turbine, an associated controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method comprising, for one or more waked turbine pairs: determining a stability-induction relation to determine the control parameter for the upwind turbine in the induction control mode and thereby varying the induction offset; wherein the stability-induction relation is determined based on one or more monitored or predicted atmospheric parameters relating to surface layer stability, and is defined so that a variation in the one or more atmospheric parameters in the absence of variations in both a freestream wind velocity and a wind direction causes a variation in the induction offset.
2. 2. A method according to claim 1, wherein the one or more atmospheric parameters is selected from the group consisting of: a value of a wake decay coefficient k for use in a N.O. Jensen wake model, for wind received at the wind farm; a turbulence intensity of wind received at the wind farm; a parameter relating to a temperature gradient in wind received at the wind farm; two or more temperature parameters corresponding to air temperatures at respective heights local to the wind farm; two or more temperature parameters, at least one corresponding to an air temperature at a height local to the wind farm, and at least one corresponding to a local ground or water temperature; an Obukhov length associated with wind received at the wind farm; a gradient Richardson Number, or bulk Richardson Number, for wind received at the wind farm.
3. 3. A method according to claim 1 or 2, further comprising monitoring two or more temperature parameters from respective temperature sensors at different heights local to the wind farm to determine the one or more atmospheric parameter.
4. 4. A method according to any preceding claim, wherein the one or more atmospheric parameter is determined based on weather data.
5. 5. A method according to any preceding claim, wherein the control parameter is selected from the group consisting of: a performance target for the respective turbine, for example: a target induction factor; a target thrust coefficient or thrust; a target rotational velocity; a target operating power; or a target operating thrust; a blade angle setting corresponding to a pitch angle of blades of the respective turbine; a generator setting which determines an induced electromotive force of the generator per revolution of the turbine.
6. 6. A method according to any preceding claim, wherein the control parameter is determined by selecting a control law based on the one or more monitored or predicted atmospheric parameter, wherein the control law is for determining the control parameter based on at least one auxiliary variable such as wind velocity or wind direction, wherein the control law is selected from a plurality of control laws which are configured to provide differing profiles of the control parameter for a common operating range of the at least one auxiliary variable.
7. 7. A method according to any preceding claim, wherein the control parameter is determined by reference to a predetermined relation which is a function of the one or more atmospheric parameter, or by reference to a database of predetermined control parameters correlated to the one or more atmospheric parameter.
8. 8. A method according to any preceding claim, wherein the control parameter is determined remotely from the upwind turbine, optionally remotely from the wind farm.
9. 9. A method according to any preceding claim, comprising periodically: updating, or determining whether to update, the control parameter based on the one or more monitored or predicted atmospheric parameter; wherein the periodic update or determination of whether to update is performed: at predetermined intervals, for example at intervals of no more than six hours; in response to a threshold change in a monitored parameter relating to an atmospheric condition; or in response to an operator demand.
10. 10. A method of operating a wind farm comprising, for a wind direction, at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate at a respective induction factor; wherein for each turbine, an associated controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method comprising, for one or more waked turbine pairs: determining the control parameter for the upwind turbine by a method in accordance with any of the preceding claims; and controlling the upwind turbine to operate in the induction control mode based on the control parameter, whereby the induction offset is variable independently of variance of a wind velocity and variance of the wind direction.
11. 11 A method according to claim 10; comprising: operating the wind farm during an operating period in which surface layer stability conditions change such that there is a change in one or more of: a temperature gradient within a surface layer of wind received at the wind farm a temperature difference between wind received at the wind farm and a ground or water temperature local to the wind farm; a value of a wake decay coefficient k for use in a N.O. Jensen wake model of the surface layer of wind at the wind farm changes, as calculated based on monitored atmospheric conditions or as determined based on predicted atmospheric conditions; a turbulence intensity of wind received at the wind farm; an Obukhov length, gradient Richardson Number of bulk Richardson Number associated with wind received at the wind farm; wherein the change of surface layer stability conditions causes a corresponding change in the monitored or predicted atmospheric parameter relating to surface layer stability, such that the control parameter for the upwind turbine of the or each waked turbine pair changes during the operating period.
12. 12. A method according to claim 11, wherein: a change in surface layer stability conditions favours wake recovery; the corresponding change in the control parameter is to reduce the induction offset; and/or wherein: a change in surface layer stability conditions inhibits wake recovery; the corresponding change in the control parameter is to increase the induction offset.
13. 13. A method comprising: providing a wind farm model defining a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is modelled to exert a thrust on wind within the model to operate at a respective induction factor; wherein each turbine is modelled based on a variable control parameter relating to an induction factor of the respective turbine, whereby each turbine is operable in an induction control mode in which the control parameter differs relative to a baseline mode to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; the method further comprising: defining a plurality of simulation cases for the wind farm model by varying each of a set of independent variables, including: freestream wind velocity; wind direction; and an atmospheric parameter relating to surface layer stability; for each simulation case, determining the control parameter for operation of the upwind turbine of at least one waked turbine pair using an optimization procedure having an objective function relating to a total power output of the plurality of turbines; wherein for at least some of the simulation cases, the determined control parameter corresponds to operation of the upwind turbine in the induction control mode to reduce the induction factor of the upwind turbine relative to the baseline mode by a variable induction offset; storing or updating a reference model or database to correlate the determined control parameters for the simulation cases to the atmospheric parameter; whereby the reference model or database permits the control parameter for the or each corresponding operational upwind turbine of a wind farm to be determined based on a monitored or predicted value of the atmospheric parameter, so that a variation in the atmospheric parameter in the absence of variations in both a freestream wind velocity and wind direction causes a variation in the induction offset
14. 14. A non-transitory machine-readable medium storing machine-readable instructions which, when executed by a processor, are configured to cause performance of a method in accordance with any one of the preceding claims.
15. 15. A wind farm controller associated with a wind farm comprising a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate according to a respective induction factor; wherein for each turbine, the wind farm controller is configured to operate the turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; wherein the wind farm controller is configured to cause performance of a method in accordance with any of claims 1-12; or comprises a non-transitory machine-readable medium storing machine-readable instructions which, when executed by the controller, are configured to cause performance of a method in accordance with an of claims 1-12.
16. 16. A wind farm controller for a wind farm comprising a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine, each turbine configured to exert a thrust on wind to operate according to a respective induction factor; wherein the wind farm controller is configured to operate each turbine in a baseline mode and in an induction control mode, wherein a control parameter for operation of the respective turbine differs between the induction control mode relative to the baseline mode, to reduce an induction factor of the turbine in the induction control mode relative to the baseline mode by a variable induction offset; wherein for each of one or more waked turbine pairs, the wind farm controller is configured to: determine a stability-induction relation to determine the control parameter for the upwind turbine in the induction control mode and thereby vary the induction offset; determine one or more monitored or predicted atmospheric parameters relating to surface layer stability; wherein the stability-induction relation is defined to be determined based on the one or more atmospheric parameters relating to surface layer stability, and is defined so that a variation in the one or more atmospheric parameters in the absence of variations in both a freestream wind velocity and a wind direction causes a variation in the induction offset.
17. 17. A wind farm comprising: a plurality of turbines spatially configured so that, for each of a plurality of wind directions there is at least one waked turbine pair comprising an upwind turbine and a downwind turbine within a wake region of the upwind turbine; wherein each turbine is configured to exert a thrust on wind to operate according to a respective induction factor; and wherein the wind farm further comprises a wind farm controller in accordance with claim 15 or claim 16.
18. 18. A wind farm according to claim 17, wherein the wind farm controller is in communication with a remote data source to obtain the one or more atmospheric parameter, or data from which the one or more atmospheric parameter is derivable.
19. 19. A wind farm according to claim 17, wherein the wind farm controller is in communication with sensing apparatus installed within the wind farm or local to the wind farm, to monitor the one or more atmospheric parameter or data from which the atmospheric parameter is derivable, wherein the sensing apparatus comprises: one or more temperature sensors; one or more wind velocity sensors; and/or one or more lidar sensors or sonic anemometers, for monitoring an atmospheric parameter relating to atmospheric mixing, turbulence intensity and/or turbulent flux.