OPTIMIZATION OF A WIND FARM

Abstract

Provided is a method for optimizing an operation of a wind farm. The farm includes wind turbines and each can be adjusted via operating settings, and a farm model depicting the wind farm or part thereof is used. The method comprises an optimization sequence using the farm model, with the steps: specifying an optimization wind direction in the farm model for optimizing the operation of the farm for this wind direction; varying operating settings of at least a first leading turbine of the farm model; determining effects of varying the operating settings of the first leading turbine on at least one downstream turbine of the farm model, which is aerodynamically influenced by the first leading turbine, by means of a wake model; determining a total farm result of the farm model; wherein the operating settings are varied so as to optimize the total farm result.

Claims

1. A method for optimizing an operation of a wind farm, comprising: performing an optimization sequence using a farm model depicting the wind farm or a part of the wind farm, wherein the wind farm includes a plurality of wind turbines and each wind turbine of the plurality of wind turbines has respective operating settings that are adjustable, performing the optimization sequence including: specifying, in the farm model, an optimization wind direction for which the operation of the wind farm is to be improved; varying operating settings of at least a first leading wind turbine of the farm model to improve a total farm result of the farm model; determining, using a wake model, effects of the varying of the operating settings of the first leading wind turbine on at least one downstream wind turbine of the farm model, the at least one downstream wind turbine being aerodynamically influenced by the first leading wind turbine; and determining the total farm result of the farm model; and storing operating settings of a respective wind turbine of the plurality of wind turbines that improve the total farm result.

2. The method as claimed in claim 1, wherein the total farm result of the farm model is a total farm power of the wind farm, and wherein improving the total farm result of the farm model includes maximizing the total farm power.

3. The method as claimed in claim 1, comprising: repeating the optimization sequence including: retaining the optimization wind direction; varying operating settings of at least one further leading wind turbine of the farm model; determining, using a further wake model, effects of varying the operating settings of the at least one further leading wind turbine on at least one further downstream wind turbine, which is aerodynamically influenced by the further leading wind turbine; and retaining the operating settings of the first leading wind turbine unchanged.

4. The method as claimed in claim 3, wherein the at least one further leading wind turbine and the first leading wind turbine are not aerodynamically influenced by one of the at least one further downstream wind turbine and the at least one downstream wind turbine.

5. The method as claimed in claim 1, wherein for the optimization sequence, an optimization wind speed is predefined, and the operating settings that improve the total farm result are stored together with the optimization wind direction and the optimization wind speed.

6. The method as claimed in claim 1, wherein varying the operating settings includes varying one or more of: a rotation speed of an aerodynamic rotor of at least the first leading wind turbine; a blade angle of at least one rotor blade of at least the first leading wind turbine; and an azimuth orientation of a nacelle of at least the first leading wind turbine.

7. The method as claimed in claim 1, wherein operating settings of multiple leading wind turbines are varied at the same time, wherein for each of the multiple leading wind turbines, effects of varying the operating settings of the multiple leading wind turbines on a respective at least one downstream wind turbine, aerodynamically influenced by at least one of the multiple leading wind turbines, are determined by means of the wake model.

8. The method as claimed in claim 1, wherein the wake model is used to determine an induced turbulence, the wake model is used to determine a speed deficit, and/or a total farm power, taking into account a maximum mechanical load of the plurality of wind turbines, is used as the total farm result.

9. The method as claimed in claim 1, wherein for a respective optimization wind direction: the plurality of wind turbines of the wind farm are sorted into a processing order; the optimization sequence is run repeatedly such that for each run, at least one of: the varying of the operating settings, determining the effects of the varying of the operating settings and determining the total farm result is performed; and in the varying of the operating settings of each run, the operating settings of a respective wind turbine according to the processing order are varied such that a first wind turbine of the plurality of wind turbines having the operating settings varied in the first run is first in the processing order, and further wind turbines of the plurality of wind turbines have whose operating settings varied in subsequent runs are subsequent wind turbines of the processing order.

10. The method as claimed in claim 9, wherein in particular the processing order depends on site coordinates of the plurality wind turbines in the wind farm and on the optimization wind direction.

11. The method as claimed in claim 1, wherein in the farm model, for a respective leading wind turbine: an azimuth angle, a blade angle and/or a rotor rotation speed are adjustment values of the respective leading wind turbine; a wind speed and a turbulence are defined or determined as flow conditions; a coefficient of thrust acting on the respective leading wind turbine is determined, using a blade element method, depending on the adjustment values and the flow conditions; an installation power of the respective leading wind turbine is determined, using a blade element method, depending on the adjustment values and the flow conditions; and an induced turbulence and a speed deficit of a downstream wind turbine are determined depending on the coefficient of thrust and the turbulence of the respective leading wind turbine.

12. The method as claimed in claim 11, wherein an azimuth angle, a blade angle and/or a rotor rotation speed of the downstream wind turbine are set as adjustment values, depending on: the adjustment values of the downstream wind turbine, the induced turbulence and the speed deficit.

13. The method as claimed in claim 9, wherein in a total run-through, the optimization sequence is repeated for the plurality of wind turbines until respective operating settings for all leading wind turbines of the wind farm are stored as optimized operating settings, and the total run-through is repeated one or more times and the respective operating settings stored of a preceding total run are used as starting values in a subsequent run.

14. A wind farm, comprising: a plurality of wind turbines, each wind turbine of the plurality of wind turbines having adjustable operating settings, and respective operating settings of the plurality of wind turbines are determined using a method that includes: performing an optimization sequence using a farm model depicting the wind farm or a part of the wind farm, performing the optimization sequence including: specifying, in the farm model, an optimization wind direction for which operation of the wind farm is to be improved; varying operating settings of at least a first leading wind turbine of the farm model to improve a total farm result of the farm model; determining, using a wake model, effects of the varying of the operating settings of the first leading wind turbine on at least one downstream wind turbine of the farm model, the at least one downstream wind turbine being aerodynamically influenced by the first leading wind turbine; and determining the total farm result of the farm model; and storing operating settings of a respective wind turbine of the plurality of wind turbines that improve the total farm result.

Description

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0077] The invention is now described below as an example with respect to exemplary embodiments with reference to the accompanying figures;

[0078] FIG. 1 shows a wind turbine in a perspective depiction;

[0079] FIG. 2 shows a wind farm in a diagrammatic depiction;

[0080] FIG. 3 shows a flow diagram for calculating a wind farm;

[0081] FIG. 4 shows a flow diagram for optimizing a wind farm; and

[0082] FIG. 5 shows a further diagrammatic depiction of a wind farm as explanation of a farm optimization.

DETAILED DESCRIPTION

[0083] FIG. 1 shows a wind turbine 100 with a tower 102 and a nacelle 104. A rotor 106 with three rotor blades 108 and a spinner 110 is arranged on the nacelle 104. The rotor 106 is set in a rotary motion by the wind during operation, and thereby drives a generator in the nacelle 104.

[0084] FIG. 2 shows a wind farm 112 with as an example three wind turbines 100, which may be the same or different. The three wind turbines 100 thus represent in principle an arbitrary number of wind turbines of a wind farm 112. The wind turbines 100 provide their power, in particular the current generated, via an electric farm network 114. The currents or power generated by the individual wind turbines 100 are added together, and usually a transformer 116 is provided which steps up the voltage in the farm in order then to feed this into the supply network 120 at the infeed point 118, generally also known as the PCC. FIG. 2 shows only a simplified depiction of a wind farm 112, which for example shows no control system although naturally a control system is present. Also for example, the farm network for 114 may be constructed differently, in that for example a transformer is also present at the output from each wind turbine 100, to name just another exemplary embodiment.

[0085] The diagram in FIG. 3 is generally based on the assumption that 1 to n wind turbines are present in a wind farm and should be calculated. A wind turbine is abbreviated here as WT and the index extends from 1 to n. Index j indicates the current wind turbine to be calculated or the associated parameters. Index j−1 therefore indicates the previously calculated wind turbine, and accordingly index j+1 indicates the next wind turbine to be calculated. The wind turbines are for this listed symbolically in the farm block 310 and the current wind turbine is marked WT j. A wind turbine may synonymously be described as a windmill or wind generator.

[0086] The method processes the wind turbines from the first wind turbine WT 1 to the last wind turbine WT n. The result amongst others is a speed deficit VD and an induced turbulence TI which act from the front wind turbines WT 1 to WT j−1 on the current wind turbine WT j. For the current wind turbine WT j, therefore the speed deficits VD.sub.1 . . . j−1 and the induced turbulences TI+.sub.1 . . . j−1 are taken into account and a superposition carried out in the superposition block 312. From an environmental condition block 314, the superposition block 312 receives corresponding environmental conditions. The wind speed may also be regarded as an environmental condition, wherein further environmental conditions include shear, i.e., a change in wind speed with location, in particular altitude, and veer, i.e., a change in wind direction with location, in particular altitude. The air density may be another environmental condition.

[0087] All this is taken into account or superposed in the superposition block 312.

[0088] The induced turbulence TI may also be designated the induced turbulence TI+. By superposition of all this information, in the superposition block 312, the total wind speed V acting on the current wind turbine WT j can then be determined, and also the total induced turbulence acting on the current wind turbine WT j. The two values form an input variable for a turbine calculation block 316. In the turbine calculation block 316, a blade element method is carried out to calculate the wind turbine, namely in particular to calculate the power P.sub.el which can be generated. For this, the blade element method, i.e., the turbine calculation block 316, also obtains the environmental conditions from the environmental condition block 314. Also, properties of the wind turbine WT j itself exert an influence, namely data such as properties of the wind turbine as such, i.e., in particular the rotor blades used, and the current operating settings of the wind turbine concerned.

[0089] On this basis then, using the blade element method, the turbine calculation block 316 calculates the electrical power P.sub.el of the current wind turbine. In this way, the current power P.sub.el—which is also known or can be regarded as the output power—is calculated from the aerodynamic variables taking into account the wind farm, namely the wind turbines WT 1 to WT j−1 which potentially stand in front of the current wind turbine WT j. The power P.sub.el thus calculated may be fed back to the current wind turbine WT j as information. The wind turbine WT j concerned may store this data and make it retrievable if required.

[0090] Also, the effects of this current wind turbine WT j on the downstream wind turbines WT j+1 to WT n are calculated. For this, a wake model 318 is used. For this, the wake model 318 obtains the induced turbulence TI acting on the current wind turbine WT j, as calculated by the superposition block 312. The turbine calculation block 316, using the blade element method, also calculates a thrust coefficient C.sub.t which forms a further input variable for the wake model 318. The azimuth deflection or alignment of the wind turbine is also important for the wake model. This azimuth orientation can also be designated a yaw angle. This azimuth angle thus also forms an input variable for the wake model 318. Finally, the position and distance of the respective downstream wind turbines WT j+1 to WT n with respect to the current wind turbine WT j are important, and serve as an input variable for the wake model.

[0091] On the basis of these data, speed deficit VD.sub.j and the induced turbulence TI.sub.j on the downstream wind turbines WT j+1 to WT n are calculated. In this way, from earlier calculations, the current wind turbine WT j has also obtained the speed deficits VD.sub.1 . . . j−1 and induced turbulences TI+.sub.1 . . . j−1 which are relevant for it, namely for the wind turbine WT 1 to WT j−1. So in this way, all wind turbines of the wind farm can be calculated.

[0092] FIG. 4 shows a flow diagram 408 which illustrates a farm optimization. For the farm optimization, a calculation must be repeatedly performed as explained in connection with FIG. 3. For this, the calculation block 410 is shown which in principle contains a sequence explained in FIG. 3. Accordingly, the flow diagram from FIG. 3 is contained as a symbolic miniature in the calculation block 410.

[0093] The flow diagram 408 illustrates in particular an inner loop 412 and an outer loop 414. In the inner loop 412, an optimization sequence is performed for several wind turbines, one after the other. When all wind turbines have been optimized in the repeated optimization sequence, a total run-through has been completed. With the outer loop 414, such a total run-through is repeated in order to improve the result already obtained in a first total run-through, or to establish that it can no longer be improved or no longer significantly further improved.

[0094] In the inner loop 412, optimization begins with the first wind turbine through to the last wind turbine n. The wind turbines WT 1 to WT n to be calculated are sorted into a processing order which depends on the wind direction. The processing block 416 illustrates this. In principle, the method begins with the first wind turbine, wherein for example this could also be the last, and in general the wind turbine WT i is then the current wind turbine being optimized. For this, an optimization algorithm illustrated by the optimization block 418 changes the current wind turbine WT i to be optimized. For this, the parameters of rotor rotation speed, blade angle and/or azimuth orientation are set or varied as operating settings, namely for the current wind turbine WT i to be optimized. With these adjusted parameters, then the calculation of the wind farm is carried out in the calculation block 410, i.e., carried out as described in connection with FIG. 3. The result is a total farm result, here output as the farm power P.sub.Farm. Depending on this total farm result, the operating parameters can be changed further in the optimization block 418 until an optimum setting has been found for the parameters for this wind turbine, and hence optimum operating settings for the current wind turbine WT i have been found.

[0095] These may then be stored, as illustrated in the total evaluation block 420.

[0096] On the repeat of the inner loop 412, in the calculation in the calculation block 410, new calculation is required however only from the wind turbine i. All wind turbines upstream remain unchanged. This is indicated by the wind turbine WT h which can be shifted correspondingly to the total evaluation block 420, in which the wind turbines already fully optimized are stored.

[0097] When the operating settings of a wind turbine have been optimized, optimization is then performed for the next wind turbine, namely in particular for the wind turbine considered the next in the processing block 416.

[0098] Since in each case only the downstream wind turbines need be calculated, the list of wind turbines to be calculated in the calculation block 410 changes. To this extent, the number of turbines on a list forming the basis for this decreases on each iteration step of the inner loop 412. This is illustrated in FIG. 4 in that an old processing list 422 is changed to a new processing list 424. Accordingly, the new processing list 424 is shorter by at least one wind turbine than the old processing list 422. It should however also be considered that—as FIG. 4 does not show—several wind turbines may be optimized at the same time. In this case, it is also considered that the new processing list 424 may be reduced by more than one wind turbine in comparison with the old processing list 422.

[0099] When all wind turbines of the farm have been optimized, which is the case in particular when the inner loop 412 has been run n times, optimization in the first approximation is completed.

[0100] In order to check whether an adequate optimum has actually been found, the optimization according to the inner loop may be repeated. This is indicated by the outer loop 414. If the multiple run-throughs of the inner loop 412 are repeated, then on the first run of the inner loop this begins again with the first wind turbine, and all wind turbines are then recalculated as described above.

[0101] The result is again a farm power as a total farm result. This farm power of the second run of the outer loop may be compared with the farm power as the result of the first run of the outer loop. In this way, the outer loop 414 may be run several times, and the farm power recorded each time. Then from the recorded farm power levels for each run of the outer loop 414, it can be assessed whether the optimization was sufficient or should be repeated. When there is strong convergence of the farm power from one run of the outer loop 414 to the next, the calculation may be concluded.

[0102] The result is then an optimal setting of the wind farm for at least one wind direction and one wind speed.

[0103] The wind turbine h is the turbine whose settings have just been successfully optimized. It is therefore set aside, i.e., no longer changed, since its effects for this wind direction are now globally constant, i.e., assumed to be unchanged for all further calculations of the wind farm with the same optimization wind direction.

[0104] FIG. 5 illustrates a wind farm 500 with nine wind turbines WT 1 to WT 9. The wind farm 500 may in principle correspond to the wind farm 112 from FIG. 2, which for reasons of clarity shows merely three wind turbines. The nine wind turbines WT 1 to WT 9 of FIG. 5 are shown illustratively. According to FIG. 5, a situation is depicted in which the wind 510 is in principle blowing from the left, as indicated by a large arrow. Accordingly, the azimuth settings of all wind turbines WT 1 to WT 9 are oriented to this wind. Each of the wind turbines WT 1 to WT 9 is depicted symbolically in a top view, with a nacelle and a rotor with two rotor blades standing horizontally in the illustration. In fact, the rotor turns naturally with these rotor blades, and also modern wind turbines today have three rotor blades instead of two.

[0105] Also, behind each wind turbine WT 1 to WT 9, a wind drag S is depicted symbolically in dotted lines for illustration. Such a wind drag has an approximately helical form. In fact, in the downwind region of each wind turbine, there is not only a clearly helical wind drag but generally turbulence occurs there, namely induced turbulence caused by the respective wind turbine. Also, there is a speed deficit, which in principle indicates that the wind speed is reduced by the wind turbine in front.

[0106] In the exemplary configuration of FIG. 5 therefore, the wind turbines WT 1 to WT 9 are optimized in this sequential order. In the first run of an optimization sequence, wind turbine WT 1 is regarded as the leading wind turbine. All following wind turbines WT 2 to WT 9 may be regarded as downstream wind turbines for this first optimization sequence. The wake of the first wind turbine WT 1 is however evidently of particular relevance only for the fourth wind turbine WT 4. Some effect would also occur for the seventh wind turbine WT 7.

[0107] In the case illustrated in FIG. 5, in which actually three wind turbines stand one behind the other in a row, it is also possible to optimize the first three wind turbines WT 1, WT 2 and WT 3 substantially simultaneously and consider the next three wind turbines WT 4, WT 5 and WT 6 as the downstream wind turbines. In particular wind turbines WT 1 and WT 4 may be considered as a wind turbine pair. The same applies to wind turbines WT 2 and WT 5, and also to wind turbines WT 3 and WT 6.

[0108] Thus in a first optimization sequence, the operating settings of wind turbines WT 1, WT 2 and WT 3 would be adjusted and their optimum found. In a second optimization sequence, the optimization settings of wind turbines WT 1, WT 2 and WT 3 would remain unchanged, namely at the values which resulted from the first optimization sequence.

[0109] In this second run-through of the optimization sequence, the wind turbines WT 4, WT 5 and WT 6 may each be regarded as leading wind turbines, and wind turbines WT 7, WT 8 and WT 9 as downstream wind turbines. In this second run-through of the optimization sequence, the operating settings of these three leading wind turbines WT 4, WT 5 and WT 6 may then be varied in order to find an optimum.

[0110] Finally, in a third optimization sequence or third run-through of the optimization sequence, the operating settings of wind turbines WT 7, WT 8 and WT 9 may be adjusted or set while the operating settings of the other wind turbines WT 1 to WT 6 remain unchanged. In this final run-through of the optimization sequence, there is no need to consider downstream wind turbines.

[0111] It is however also possible that the wind 510 flows onto the wind farm 500 turned through 30°. This is indicated as the second wind 510′ with a dotted arrow. In this case, a joint assessment of the wind turbines WT 1, WT 2 and WT 3 as first leading wind turbines would no longer be possible, or only poorly possible. It may then be provided to optimize one wind turbine after the other. In this case too, wind turbine WT 1 may be considered the leading wind turbine and its operating settings optimized. The other wind turbines may be regarded as downstream wind turbines. The greatest effects may arise, with the wind direction of the second wind 510′, from wind turbine WT 1 onto wind turbine WT 8. However, the first wind turbine WT 1 may be regarded as the first leading wind turbine and the other wind turbines as downstream wind turbines, wherein the calculation will show that perhaps only wind turbine WT 8 is significantly influenced.

[0112] In a second run-through, the operating settings of wind turbine WT 1 may remain unchanged because its operating settings have already been optimized, and wind turbine WT 2 may be considered as the next leading wind turbine.

[0113] In this way, all wind turbines WT 1 to WT 9 may be optimized, in this case in nine runs of an optimization sequence.

[0114] It is then still possible to repeat these nine runs of the optimization sequence, wherein the starting settings of all operating settings of wind turbines WT 1 to WT 9 are the settings found from the first nine runs of the optimization sequence.

[0115] As described herein configuring the operating method of individual wind turbines is no longer as per the prior art, wherein the individual turbine supplies the maximum possible yield, but such that the entire wind farm supplies an increased yield.

[0116] For this, a method is proposed for optimizing the operating settings such as pitch, rotation speed and yaw angle of wind turbines in wind farms as part of a cooperative strategy. Former known turbine control systems use a competitive strategy, in which each wind turbine uses settings which give the maximum power for the individual wind turbine instead of for the wind farm.

[0117] To determine optimal settings, various analytical wake models are combined with calculation by means of a blade element method, also known as a BEM calculation. In comparable known approaches, instead of the operating settings of the turbine, only auxiliary variables are used, such as the induction factor, which cannot be clearly assigned to a specific operating point. The results achieved there are purely theoretical nature.

[0118] By use of the BEM calculation, the actual operating settings may be used as optimization variables. The result is a significantly more precise prognosis of the achieved power of the individual wind turbine and also of the farm. Also, the effect of increased turbulence intensity on the wind turbines can be tested.

[0119] For the BEM calculation of a wind turbine, firstly the flow conditions for each wind turbine must be determined. The contact flow may be adversely affected by a wake. This concerns wind turbines erected upstream which stand in the wake of another wind turbine. This deterioration, which is perceptible by a reduction in wind speed and an increase in turbulence intensity, can be determined with analytical wake models such as NO Jensen or the Qian model. The preferred method proposed here uses the Qian model which offers the advantage of also taking into account the deflection of the wake by a deliberately incorrect setting of the yaw angle. The other models are also considered. Also, the increased turbulence intensity in the wake can be determined.

[0120] Preferably, it is proposed to take account of a superposition of the wakes of several wind turbines.

[0121] If a wind turbine ‘j’ stands in the wake of several upstream turbines i, the wakes are superposed. The resulting speed deficit is calculated with a superposition formula. Instead of the usual linear or quadratic summing of the speed deficits, here preferably the product formation of the residual speeds is proposed, which has proven to be plausible. For this, the following formula is proposed:

[00001]$U_j=U_{\infty }.Math.\underset{i=1}{\overset{j-1}{.Math.}}.Math.\left(\frac{U_{w,\mathrm{ij}}}{U_i}\right)$

[0122] In the formula, the variables have the following meaning:

[0123] U.sub.∞: undisrupted wind speed

[0124] U.sub.i: disrupted wind speed at wind turbine ‘i’

[0125] U.sub.j: wind speed at wind turbine ‘j’

[0126] U.sub.w,ij: downstream speed caused by wind turbine i at position j, disregarding other wind turbines.

[0127] To calculate the values in a wind farm, a list of wind turbine objects is produced, i.e., a list of the wind turbines in the wind farm. From the coordinates and wind direction, the turbines are sorted by wind flow direction. Now the following procedure can be iterated or repeated for all n wind turbines with j=1 . . . n. This is illustrated in FIG. 3.

[0128] With reference to FIG. 3, the following steps are proposed:

[0129] Step 1: Use of speed deficits and induced turbulences from wakes of wind turbines 1 . . . j−1 on wind turbine j.

[0130] Step 2: From the induced turbulences, speed deficits and environmental conditions, by means of the selected superposition process, determination of the actual flow conditions for the wind turbine concerned, namely wind turbine j.

[0131] Step 3: Performance of a BEM calculation for wind turbine j with the flow conditions and operating settings of wind turbine j in order to determine the coefficient of thrust and the turbine power.

[0132] Step 4: From the coefficient of thrust, induced turbulence and set yaw angle, calculation of the wake effects of turbine j on turbines j+1 . . . n, and storage of the results in the corresponding wind turbines.

[0133] Step 5: Repetition of the sequence for the next turbine and all further turbines. In the first run-through, the influence of the first wind turbine on the second wind turbine and all further downstream wind turbines is determined. In the second run-through, the influence of the first and second wind turbines on the third and all further downstream wind turbines is determined. In the third run-through, the influence of the first three wind turbines on the fourth is determined, and so on up to the last wind turbine.

[0134] It is proposed that this procedure is applied in the flow direction, i.e., the wind direction, and hence the farm is completely calculated after a run-through which includes the repetitions for each wind turbine.

[0135] The sequence is shown in FIG. 3.

[0136] To optimize the operating settings, the following is proposed.

[0137] For each wind turbine, wind speed and wind direction, the rotation speed, blade pitch angle and yaw angle are available as optimization variables. For the entire wind farm, the number of variables is therefore equal to three times the number of wind turbines in the farm. These must also be varied with wind speed and wind direction.

[0138] The optimization is performed successively for wind turbines 1 to n as follows, wherein the current wind turbine is designated wind turbine i. For faster convergence, it is therefore proposed to optimize in an individual step only the settings of an individual wind turbine ‘i’ in a row of wind turbines. The target variable, i.e., the variable to be optimized, is the total farm power. The operating settings of the wind turbines situated downstream of ‘i’ are considered constant. The wind farm must only be recalculated from wind turbine ‘i’, and all upstream wind turbines remain unchanged. The input and output parameters are therefore also constant. The optimization problem can be described by the following equation.

[00002]$\underset{s_i}{\max }.Math.P_{\mathrm{el},\mathrm{Far}.Math.m}=P_{\mathrm{el},i}\left(U_i,{\mathrm{TI}}_i,s_i\right)+\underset{k=i+1}{\overset{n}{.Math.}}.Math.P_{\mathrm{el},j}\left(U_k,{\mathrm{TI}}_k\right)+\underset{h=0}{\overset{i-1}{.Math.}}.Math.P_{\mathrm{el},h}$$s.t..Math.{\mathrm{TI}}_k\le {\mathrm{TI}}_{\max },\forall k\ge i+1,A.Math.o.Math.A^l\left(s_i\right)\le {\mathrm{AoA}}_{\max }^l.Math.\forall l=1,\mathrm{.Math.}.Math..Math.s.$

[0139] In the formula, the variables have the following meaning:

[0140] h: running index over wind turbines upstream of ‘i’

[0141] k: running index over all wind turbines downstream of ‘i’

[0142] s.sub.i: the operating settings of yaw and blade pitch angle and tip speed ratio at wind turbine ‘i’

[0143] n: number of all wind turbines

[0144] TI.sub.k: induced turbulence intensity in wake of wind turbine k

[0145] U.sub.k: disrupted wind speed at wind turbine ‘k’

[0146] P.sub.el,x: electrical power of wind turbine x

[0147] TI.sub.max: maximum turbulence intensity, see IEC standard 61400-1, ed3

[0148] s: number of radius sections

[0149] AoA.sup.l: angle of attack at radius section l

[0150] AoA.sup.l.sub.max: maximum angle of attack at radius section l

[0151] (e.g., stall angle).

[0152] It is also conceivable the other secondary conditions are included in the optimization problem. The observation of load limits may also be directly included, instead of not exceeding a maximum turbulence intensity. In addition, the controller stability may be included.

[0153] Alternatively, it is also possible to optimize the operating parameters of all wind turbines simultaneously. For this, the operating parameters may be changed accordingly so slowly that a reaction of the total farm result can be assigned to the respective change.

[0154] It is however preferred to optimize the operating settings of one wind turbine after another. This may be called a forward method.

[0155] In the forward method, the wind turbine ‘i’ to be optimized is iterated from the first to the last wind turbine: the sequence is therefore run from the first to the last wind turbine, and in each case the currently optimized wind turbine is designated wind turbine ‘i’. When all settings of the first wind turbine have been optimized, it is no longer affected by further changes and the power and wake effect are known. It is therefore removed from the list of wind turbines to be calculated and stored in a second list. The second wind turbine moves up to the first place on the list. The process is repeated until the end of the wind farm is reached. The wind farm is calculated each time only from the wind turbine to be optimized.

[0156] The upstream wind turbines are optimized on the basis of the settings of the downstream turbines which may however change later. Therefore, after optimizing the last wind turbine, the list is refilled and the process begins again with the first. This is repeated until the farm power converges, which usually takes place after a few farm run-throughs.

[0157] The method is illustrated in FIG. 4.

[0158] It has been found that former ideas on sectorial control propose selecting operating modes for an individual wind turbine. When these operating modes are produced, wake effects are not taken into account. In a second step, for each wind turbine and wind direction, an operating mode is selected so that for example the farm power is maximal and the load restrictions are observed. In other words, wake effects are only taken into account as a secondary priority.

[0159] In contrast, in the proposed concept, operating modes are proposed which take account of wake effects in the production process and can thus increase the farm yield.

[0160] The various embodiments described above can be combined to provide further embodiments. These and other changes can be made to the embodiments in light of the above-detailed description. In general, in the following claims, the terms used should not be construed to limit the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the disclosure.